Changhai Ru · Xinyu Liu

Yu Sun Editors

Nanopositioning
Technologies
Fundamentals and Applications
Nanopositioning Technologies
Changhai Ru • Xinyu Liu • Yu Sun
Editors

Nanopositioning
Technologies
Fundamentals and Applications

123
Editors
Changhai Ru Xinyu Liu
Harbin Engineering University McGill University Mechanical Engineering
Harbin, China Montreal, Québec, Canada

Yu Sun
University of Toronto
Department of Electrical and Computing
Toronto, Ontario, Canada

ISBN 978-3-319-23852-4 ISBN 978-3-319-23853-1 (eBook)

DOI 10.1007/978-3-319-23853-1

Library of Congress Control Number: 2015960181

Springer Cham Heidelberg New York Dordrecht London

© Springer International Publishing Switzerland 2016
This work is subject to copyright. All rights are reserved by the Publisher, whether the whole or part of
the material is concerned, specifically the rights of translation, reprinting, reuse of illustrations, recitation,
broadcasting, reproduction on microfilms or in any other physical way, and transmission or information
storage and retrieval, electronic adaptation, computer software, or by similar or dissimilar methodology
now known or hereafter developed.
The use of general descriptive names, registered names, trademarks, service marks, etc. in this publication
does not imply, even in the absence of a specific statement, that such names are exempt from the relevant
protective laws and regulations and therefore free for general use.
The publisher, the authors and the editors are safe to assume that the advice and information in this book
are believed to be true and accurate at the date of publication. Neither the publisher nor the authors or
the editors give a warranty, express or implied, with respect to the material contained herein or for any
errors or omissions that may have been made.

Printed on acid-free paper

Springer International Publishing AG Switzerland is part of Springer Science+Business Media (www.

springer.com)
Contents

1 A Review of Stick–Slip Nanopositioning Actuators. . . . . . . . . . . . . . . . . . . . . 1

Peng Pan, Feiyu Yang, Zhenhua Wang, Bowen Zhong,
Lining Sun, and Changhai Ru
2 Piezoelectric Motor Technology: A Review . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33
Amro Shafik and Ridha Ben Mrad
3 Mechanical Design of High-Speed Nanopositioning Systems . . . . . . . . . . 61
Yuen Kuan Yong and Kam K. Leang
4 Parallel-Kinematic Nanopositioning Stages Based
on Roberts Mechanism. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 123
Sicong Wan and Qingsong Xu
5 Electro-Magnetic Nano-Positioning . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 151
Zhuo Zhang, Yueqing Yu, Xinyu Liu, and Xuping Zhang
6 Modeling of Piezoelectric-Actuated Nanopositioning
Stages Involving with the Hysteresis. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 183
Guo-Ying Gu, Li-Min Zhu, Chun-Yi Su, and Han Ding
7 Tracking Control for Nanopositioning Systems . . . . . . . . . . . . . . . . . . . . . . . . . 213
Kam K. Leang and Andrew J. Fleming
8 Position Sensors for Nanopositioning . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 245
Andrew J. Fleming and Kam K. Leang
9 MEMS Nanopositioners . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 295
Jason J. Gorman
10 Control Issues of MEMS Nanopositioning Devices . . . . . . . . . . . . . . . . . . . . . 325
Y. Zhu, S.O.R. Moheimani, M.R. Yuce, and A. Bazaei

v
vi Contents

11 A Review of Nanomanipulation in Scanning Electron Microscopes . . 347

Devin K. Luu, Chaoyang Shi, and Yu Sun
12 Nanopositioning for Lithography and Data Storage. . . . . . . . . . . . . . . . . . . . 381
Wenbo Zhang, Ming Pang, and Changhai Ru
Chapter 1
A Review of Stick–Slip Nanopositioning
Actuators

Peng Pan, Feiyu Yang, Zhenhua Wang, Bowen Zhong, Lining Sun and
Changhai Ru

Abstract There is a surge in the development of actuators based on the stick–

slip of piezoelectric actuators in the past decade for the advantages of nanoscale
resolution and long travel distances. These actuators, termed PSTA (piezoelectric
stick–slip actuators), are widely used in chip assembly and cell manipulation. This
chapter provides a comprehensive review of PSTA systems. Reported PSTAs are
generalized with two specific principles which can be considered as a framework to
further classify other PSTAs. In addition, the applications of PSTAs are categorized
into two groups according to the number of degrees of freedom (DOF). This chapter
also discusses driving signals to actuate different structures of PSTAs and control
methods to achieve high positioning resolutions.

1.1 Introduction of Stick–Slip Nanopositioning

Nanopositioning actuators have been used in many applications. A number of appli-

cations need both nanoscale resolutions and large travel. For example, in the process
of cell manipulation, it is critical to approach cells and make fine adjustments. This
process requires large travel ranges and high resolution. In addition, actuators with
large travel and high resolution are also effective to improve the efficiency of chip
assembly. Thus, actuators having multi-scale motion ranges and high resolutions
have tremendous potential. A growing demand for high precision and accuracy
has spurred the development of new actuator technologies. Structures of actuators
become simpler and smaller. This improvement enables several actuators to be
combined into one system to achieve new functions.
A commonly observed phenomenon in dynamic systems is that motion is caused
during the transition from static to sliding friction force. This phenomenon termed
stick–slip can be employed to develop actuators. Actuators based on the stick–slip

P. Pan • F. Yang • Z. Wang • B. Zhong () • L. Sun • C. Ru

Research Center of Robotics and Micro System & Collaborative Innovation Center of Suzhou
Nano Science and Technology, Soochow University, Suzhou 215021, China
e-mail: [email protected]; [email protected]; [email protected];
[email protected]; [email protected]; [email protected]

© Springer International Publishing Switzerland 2016 1

C. Ru et al. (eds.), Nanopositioning Technologies,
DOI 10.1007/978-3-319-23853-1_1
2 P. Pan et al.

principle have high accuracy, long travel range, and simpler mechanical structures
[1]. Stick–slip actuators (STAs) often cooperate with piezoelectric actuators (PA).
Piezoelectric actuators are composed of piezoelectric materials such as lead–
PbZirconate Titanate (PZT), lead–PbMagnesium Niobate (PMN), and lead–PbZinc
Niobate (PZN). Because of the inherent properties of piezoelectric materials,
their positioning resolution is high but cannot be infinitely high. In addition,
travel distances of piezoelectric actuators are rather small. Thus, these actuators
typically provide a short travel distance with a high resolution. The combination
of piezoelectric actuators and stick–slip actuators (i.e., piezoelectric stick–slip
actuators) makes it possess a high resolution and a long travel distance [1].
Owning to these advantages, there is a surge in innovations of this type of
actuators. Since the first piezoelectric stick–slip actuator was invented in 1976 [2],
much research was done to advance the development of piezoelectric stick–slip
actuators. Up to today, nearly 60 patents about piezoelectric stick–slip actuators
have been disclosed, and a high number of papers have been published in the
literature [3]. It is worth noting that conventional sawtooth signal is widely applied
to actuate piezoelectric stick–slip actuators. However, this driving signal is not
always effective to actuate piezoelectric stick–slip actuators of different structures.
For example, conventional sawtooth signal is unsuitable for moving perpendicularly.
In order to achieve better performance, many improvements have been made to
construct new driving signals. In the meanwhile, control methods have also been
applied to eliminate intrinsic errors. As progress is made, resolution and travel
distance of piezoelectric stick–slip actuators can reach nanometer resolutions and
centimeter travel distances, and speed can reach mm/s [4].
This review provides a comprehensive discussion of the principle, driving
signals, control methods, and applications of piezoelectric stick–slip actuators. In
Sect. 1.2, the principle of PSTA is discussed. The characteristics of PSTA structures
are discussed in Sect. 1.3. Section 1.4 discusses a series of driving signals. In
Sect. 1.5, control algorithms are compared and the performance of different PSATs
is discussed. Applications based on the stick–slip principle are summarized in
Sect. 1.6. In Sect. 1.7, conclusions and future directions are given.

1.2 PSTA Principle

A generalized view of stick–slip is provided in Fig. 1.1. Body A and Body B are
initially at rest. Fe represents external force applied on Body B. Ff represents friction
force exerted on Body B. When Fe is less than or equal to Ffs (maximum static
friction force), Body A sticks to Body B, and displacement and speed of Body A
are both zero relative to Body B. When Fe is greater than Ffs , Body B slips relative
1 A Review of Stick–Slip Nanopositioning Actuators 3

Fig. 1.1 Principle of stick–slip actuation. (a) Stick–slip phenomenon. (b) Change of external force

Fig. 1.2 Specific principles of PSTA. (a) Principle I. (b) Principle II. (c) Driving signal

to Body A. In dynamic systems, this phenomenon should be handled properly via

controller design [5]. This phenomenon can also be utilized for actuation to achieve
long travel distances and high resolutions. For example, Body A can move step by
step along the right direction, relative to Body B by changing the magnitude and
orientation of external force Fe . In particular, Fe in the right direction less than Ffs
can make Body A sticks to Body B. Fe in the left direction greater than Ffs can make
Body A slip to Body B. Therefore, by alternating the magnitude and orientation of
Fe , Body A can move step by step along the right direction.
Based on this general principle, a number of PSTAs can be originated and can be
classified into two groups according to the criterion of which type of force results in
the actuating movement of the end effector. Figure 1.2 illustrates two specific PSTA
principles that are discussed as follows.
4 P. Pan et al.

1.2.1 Principle I

Figure 1.2a shows the first principle where the actuating movement of end effector
is resulted from friction force in a cycle.
The three objects A, B, C, and D shown from left to right in this figure are
known as ground (supporting frame), piezoelectric actuator (PA), driving object,
and slider, respectively. PA is placed between ground and the driving object, and
slider is regarded as end effector. The motion of PA is under control of applied
voltages shown in Fig. 1.2c. The slider placed on the driving object can slide on
the driving object with the aid of friction force. Actuation details of Principle I are
summarized as follows.
Step 1: PA is in the initial state with no applied voltage.
Step 2: PA extends slowly with a small acceleration under the control of the
first period of applied voltage. Driving object moves rightwards with the same
acceleration. The inertial force applied to the slider depends on the mass of slider
and the acceleration of PA and is less than the maximum static friction force. The
slider sticks to the driving object and moves rightwards which is called “stick”
period.
Step 3: PA retracts quickly with a large acceleration under the control of the
second period of applied voltage. Driving object moves leftwards with the same
acceleration. The inertial force applied to the slider depends on the mass of slider
and the acceleration of PA and is larger than the maximum static friction force.
The slider slips relative to the driving object and is at rest relative to the ground
which is called “slip” period.
Step 4: PA retracts to the initial state when the voltage decreases to zero.
According to the four steps above, the slider achieves a step movement.
Continuously repeating these four steps, the slider can move rightwards. If one
wants to make the slider move leftwards, the shape of voltages should be reversed.
In this principle, the actuating movement of the slider occurs in the “stick”
period. In this period friction force contributes to the actuating movement which
means that friction force is directly related to actuating movement of end effector.
In addition the stick period and slip period both occur between the slider and driving
object. PSTA examples of this principle can be found in [6–8]. PSTA of this specific
principle is further termed piezo friction driving actuator (PFDA).

1.2.2 Principle II

As shown in Fig. 1.2b, in this principle the actuating movement of the end effector
results from inertial forces which depend on the mass of the counter object and the
acceleration of PA in a cycle.
1 A Review of Stick–Slip Nanopositioning Actuators 5

The three objects A, B, C, and D from left to right in this figure are known as
the ground (supporting frame), main object, piezoelectric actuator (PA), and counter
object, respectively. PA is placed between the main object and counter object, and
the main object functions as the end effector. The motion of PA is under control of
applied voltages shown in Fig. 1.2c. The main object placed on the ground can slide
on the ground due to inertial forces. The working load can be applied to the main
object. The counter object suspends in air and is not in contact with the ground. The
actuation details of Principle II are as follows.
Step 1: PA is in the initial state with no applied voltage.
Step 2: PA extends slowly with a small acceleration under the control of the
first period of applied voltage. Counter object moves rightwards with the same
acceleration. The inertial force applied to the main object depends on the mass
of counter object and the acceleration of PA and is less than the maximum static
friction force from the ground. The main object sticks to the ground and is at rest
relative to the ground which is called “stick” period.
Step 3: PA retracts quickly with a large acceleration under the control of the second
period of applied voltage. The inertial force applied to the main object depends
on the mass of counter object and the acceleration of PA and is larger than the
maximum static friction force from the ground. The main object slips relative to
the ground and moves rightwards relative to the ground which is called “slip”
period.
Step 4: PA retracts to the initial state when the voltage decreases to zero.
According to these four steps, the main object achieves a step movement with
respect to the ground. Continuously repeating these four steps, the main object can
move rightwards together with the PA and the counter object. To make the slider
move leftwards, the shape of applied voltages should be changed in a reversed order.
In this principle, the actuating movement of the main object occurs in the
“slip” period. In this period the external force (i.e., inertial force) contributes to
the actuating movement which means that friction force is not directly related to
actuating movement of the end effector. In addition, the stick period and slip period
both happen between the ground and main object. The PSTA examples of this
principle can be found in [9–11]. The PSTA of this specific principle is further
termed piezo inertial driving actuator (PIDA).

1.2.3 Discussion

In Principle I, the actuating movement of end effector is generated by the friction

force. In Principle II, the actuating movement of end effector is generated by the
inertial force. All PSTA can be classified into these two types. Some differences
exist between these two principles. (1) Due to the limitation of maximum static
friction force, the driving force of PSTA in the first principle is small. Differently,
the driving force of PSTA in the second type of principle is large. (2) The resolution
6 P. Pan et al.

of PFDA depends on PA and static friction force. The resolution of PIDA depends
on PA and sliding friction force. However, the sliding friction force is more complex
than static friction force. Without applying control algorithms, the accuracy of
Principle I is higher than that of Principle II. (3) The travel distance of PSTA in the
first principle is limited to the dimension of the driving object, whereas the travel
distance of PSTA in the second principle has no limitation in theory.

1.3 Characteristics of PSTA Structure

Based on the two principles discussed above, many actuators have been developed in
the past decades. These actuators can be grouped together according to the number
of degrees of freedoms (DOF). In particular these actuators can be divided into two
large groups, namely (1) those with one-DOF, and (2) those with multi-DOF.

1.3.1 Flexure Design

Flexure is typically used to transmit displacements without loss, resulting in a high

displacement resolution of the motion. Sometimes the flexure is used to amplify
motion to improve actuation stroke. An adjusting screw is often adopted to make
a small deformation of flexure hinges. This small deformation of flexure hinge
generates elastic potential energy which is released when the PA retracts. Thus, the
released elastic potential energy produces a large inertial force which contributes
to the decrease of backlash. Thus, the flexure is commonly used for transmitting
motions produced by the piezoelectric stack actuator.
The typical flexures of one-DOF are the circular flexure hinge, shown in Fig. 1.3a
and the right angle flexure hinge, shown in Fig. 1.3c. The rotation axis of the
circular hinge is almost invariable, and the motion accuracy of it is high. Elliptical
flexure hinge, derived from the circular hinge, is depicted in Fig. 1.3b. In order to
obtain planar motion, some circular flexure hinges are combined [12]. However,
the travel of planar motion is small which is not suitable for applications requiring
large travels [13]. Right angle flexure hinge is a good alternative to the circular
flexure hinge to achieve large travel. However, its motion accuracy is low for its
deflection of the rotation axis [14]. Figure 1.3d shows a modified flexure hinge
derived from the right angle flexure hinge termed corner-filleted flexure hinge [15].
The flexure hinges mentioned above only have one-DOF motion. To obtain multi-
DOFs motions, spherical flexure hinge, having three axes, was developed and is
shown in Fig. 1.4a. When the diameter of the thinner part of the spherical hinge
increases to a certain degree, it only produces two-DOF motions, shown in Fig. 1.4b.
It is worth noting that the stiffness of the thinner part is rather low, making it easier
to break.
1 A Review of Stick–Slip Nanopositioning Actuators 7

Fig. 1.3 Flexures of one-DOF. (a) Circular. (b) Elliptical. (c) Right angle. (d) Corner-filleted

Fig. 1.4 Flexure of multiple DOFs. (a) Typical three-axis flexure: spherical flexure. (b) Prototype
of a two axis flexure

1.3.2 Actuators with One-DOF

One-DOF actuators can be divided into two groups, one-DOF actuators with
basic structure and one-DOF actuators with auxiliary structure. To improve the
performance of this kind of actuators, auxiliary structure should be adopted.

1.3.2.1 One-DOF Actuators with Basic Structure

Among the actuators performing one-DOF actuation, some actuators only have four
basic objects. The four basic objects referred to in the PFDA are PA, driving object,
supporting frame, and slider (end effector), respectively. The four basic objects
8 P. Pan et al.

Fig. 1.5 Basic structures: (a) PFDA with basic structure [16], (b) PIDA with basic structure where
counter object is suspended [9], (c) PIDA with basic structure where the counter object is supported
[18]

Fig. 1.6 (a) PFDA with auxiliary structures [21], (b) PIDA with auxiliary structures [37]

referred to in the PIDA are PA, main object (end effector), supporting frame, and
counter object, respectively. The cases of PFDA have been referred to in [16, 17],
shown in Fig. 1.5a. The cases of PIDA have been referred to in [9, 18], shown in
Fig. 1.5b, c. The counter object shown in [9] is suspended, whereas the counter
object shown in [18] is supported by the ground with a small friction force which
is known as the supporting frame. Support by the ground with a small friction force
makes PIDA more stable resulting in decrease of the lateral vibration of the counter
object.

1.3.2.2 One-DOF Actuators with Auxiliary Structure

Due to manufacturing imperfections, fitting process, and the nature of the PSTA,
it is difficult to achieve high performance by actuators with basic structures. It is
necessary to adopt auxiliary structures to improve performance. Some auxiliary
structure is applied to amplify the travel distance of PA to obtain large displacements
and acceleration. Examples of amplification in PFDA are discussed in [19–22],
and one of them is presented in Fig. 1.6a. For examples of amplification in PIDA,
one can refer to [23]. Some auxiliary structures are adopted to adjust the friction
1 A Review of Stick–Slip Nanopositioning Actuators 9

force or inertial force to reduce the backlash in the “slip” period and enable the
vertical actuation of PSTA. For example, flexure hinge is applied to increase the
inertial force [24–26]. Originally an adjusting screw is adopted to make a small
deformation of the flexure hinge. This small deformation of flexure hinge generates
elastic potential energy which is released when PA retracts. The released elastic
potential energy produces a large inertial force which contributes to the decrease
of backlash. Examples of adjusting mechanisms in PFDA can be found in [27–35].
Examples of adjusting mechanisms in PIDA can be found in [36–40].

1.3.3 Actuators with Multi-DOF

Multi-DOF actuators are divided into those with parallel mechanisms and those
with serial mechanisms. Parallel mechanism means there is at least an end effector
performing two or more DOFs. Serial mechanism means all the end effectors of
multi-DOF actuators only perform one-DOF.

1.3.3.1 Multi-DOF Actuators with Parallel Mechanism

Among the actuators performing multi-DOF actuation, some end effectors of

actuators can realize multi-DOF movement similar to the parallel robot. These
kinds of multi-DOF actuators are called actuators with parallel mechanism, where
a coupling phenomenon between actuation of some DOFs may exist as shown in
Fig. 1.7. Multi-DOF PSTA with parallel mechanism can be found in [41] where
PSTA can perform three DOF actuation including two linear actuations and a
rotation actuation. Two linear actuations are called x-direction actuation and y-
direction actuation, respectively. There is coupling between x-direction actuation
and y-direction actuation. PFDA with parallel mechanism can be found in [42–46].
There is only one main object in PIDA which is recognized as the end effector.
According to this characteristic, PIDA with multi-DOF are all actuators with parallel
mechanism [9, 47, 48].

1.3.3.2 Multi-DOF Actuators with Serial Mechanism

If each end effector of actuators performs one-DOF actuation, the actuators are
called multi-DOF actuators with serial mechanism which can be seen in Fig. 1.8.
It means that there must be no coupling in this kind of actuators. Multi-DOF PFDA
with serial mechanism was proposed in [49] where PSTA can perform three DOF
actuations including two linear actuations and a rotation actuation. There are three
end effectors, and each of them is actuated by its own set of PAs. Each end effector
conducts one-DOF movement. PFDA with serial mechanism were also reported in
[26, 50–52]. This concept is not suitable for PIDA. Therefore, there is no serial
mechanism in PIDA.
10 P. Pan et al.

a b Slider
Driving
Object
(End effector)

PAs

Piezo Tube Driving

(c) Object
(d)
Driving
Bend PA
Object

Slider
(End effector)
Driving
PAs
Object

Counter Object Main Object

c d
PA

Main Object

PA
Shaft

Base
Counter
Objects Piezoelectric
Bim orphs

Fig. 1.7 Multi-PSTAs with parallel mechanism: (a) 2-DOF PFDAs with one vertical movement
and a rotational movement [46], (b) 3-DOF PFDAs with two horizontal movements and a rotational
movement [41], (c) 3-DOF PIDAs with two horizontal movements and a rotational movement [48],
(d) 2-DOF PIDAs with one horizontal movement and a rotational movement [47]

1.4 Friction Model

The output displacement of PSTA is obtained by the change of friction force

between the driving object and slider. Friction, affected by many factors, such
as the relative sliding speed, acceleration, and displacement, is defined as the
tangential forces between two contact surfaces. To obtain precise displacements,
an appropriate friction model describing the characteristics of friction is critical.
The friction model in the literature can be characterized into two categories: static
friction model and dynamic friction model. The friction described in static friction
model is the function of the relative speed. The friction is the function of the relative
speed and displacement. Many frictional effects, including Column friction, viscous
friction, static force, stribeck friction, pre-sliding displacement, varying static force,
and frictional lag have been demonstrated which contribute to the accurate friction
modeling.
1 A Review of Stick–Slip Nanopositioning Actuators 11

Fig. 1.8 Multi-PFDAs with serial mechanism: (a) 3-DOF PFDAs including 3 linear movements
[26], (b) 3-DOF PFDAs including 2 linear movements and 1 rotational movement [49]

1.4.1 Static Friction Model

In the sixteenth century, Leonardo da Vinci proposed that the value of friction only
depends on the normal load, and the direction of the friction is opposite to that
of the direction of motion. Coulomb then presented the Column model based on
da Vinci’s conclusion. Figure 1.9a shows the characteristics of the Column model
which indicates friction force is not associated with the contact area. In this figure,
fc represents the column friction. This model cannot describe friction when the
relative speed is zero. A new model, called ColumnCviscous model, was developed
via combining the Column model and the viscous model, as shown in Fig. 1.9b.
This model indicates that friction is associated with the value of relative speed.
However, this model also cannot describe the characteristics of friction when the
relative speed is zero. In [53], static friction is considered to be associated with the
external force when the relative speed is zero. It is experimentally demonstrated that
the external force to actuate the object from zero to a speed is larger than zero and
to keep the object sustained at this speed. Thus, the column friction is smaller than
the maximum static friction. The model, representing the combination of Column,
viscous, and static friction, is shown in Fig. 1.9c where fs represents the maximum
friction. Another typical model is demonstrated in Fig. 1.9d. In this model, friction
force decreases with the increase of speed at relatively low speeds [15], which
is defined as the stribeck effect. This model, integrating Column, viscous, static
friction, and stribeck effect, gives an accurate description of the characteristics
of friction at low speed and has been proven experimentally [54]. More detailed
discussion about other friction models can be found in [55]. In the static friction
model, it is widely believed that there is no relative movement between two contact
12 P. Pan et al.

a Friction f b Friction f

+ fc + fc

0 Speed v 0 Speed v

− fc − fc

Column Column+viscous
friction model friction model

c Friction f d Friction f
+ fs + fs
+ fc + fc

0 0

− fc − fc
− fs − fs

Column+viscous+static Column+viscous+static+stribeck
friction model friction model

Fig. 1.9 Four typical static friction models [55]

surfaces when it is at the static friction stage. Although the static friction model
can satisfy the requirements of many applications, it is not suitable for analyzing
the principle of PSTA which requires the analysis of dynamic friction effects such
as pre-sliding displacement, varying static force [56], and frictional lag [57]. Thus,
more complicated dynamic friction models are introduced.

1.4.2 Dynamic Friction Model

From a mechanics standpoint, tangential contact and normal contact are both
flexible by the external force. To accurately characterize the dynamic friction effect,
a number of dynamic friction models have been established.
1 A Review of Stick–Slip Nanopositioning Actuators 13

Fig. 1.10 Dahl friction

model

+ fc

0
Displacement x

− fc

1.4.2.1 Dahl Model

Dahl model is the simplest dynamic friction model which is the basis for other
dynamic friction models. This model was proposed by Dahl [58, 59] after he
conducted frictional experiments. Dahl discovered that contact peak between two
contact surfaces results in a small flexible deformation, similar to that of a mechan-
ical spring. Dahl formulated differential equations to characterize the relationships
between displacement and friction before friction reached up to the maximum
friction, shown in Fig. 1.10. In this model, the tangential compliance concept is
utilized to introduce pre-sliding displacement, avoiding the discontinuity about the
status switching in the static friction models. Although this model can characterize
the pre-sliding displacement and the friction lag, the varying static force and the
stribeck effect cannot be described by this model. Thus, this model is not suitable
for the analysis of PSTA.

1.4.2.2 LuGre Model

In 1995, Canudas de Wit proposed the LuGre model which is based on the Dahl
model [57]. This model is an extension of the Dahl model and is based on the bristle
model [60]. The bristle model has an assumption that the contact surface is made
up of a large amount of bristle which deforms randomly. Different from the bristle
model which describes the random deformation of the bristle, the LuGre model is
based on the average deformation of the bristle, shown in Fig. 1.11. The average
deformation of the bristle is represented by the state variable z. The state-variable
equation is
dz 0
Dv z jvj (1.1)
dt g.v/
14 P. Pan et al.

Fig. 1.11 Bristle model

where g(v) describes the stribeck effect, and g(v) is

fc C .fs  fc / eŒ.jvj=vs / 
ı

g.v/ D (1.2)
0

Friction is generated by the deformation of the bristle, which can be described as

a function of the average deformation of the bristle z and the relative speed v.

dz
f D 0 z C 1 C 2 v (1.3)
dt
where  0 is stiffness of the bristle,  1 is the micro damping coefficient,  2 is
the viscous friction coefficient, fs is the maximum static friction, fc is the column
friction, vs is the stribeck speed, and ı is the empirical constant.
This model gives a comprehensive description of all friction effects including
column friction, viscous friction, static force, stribeck friction, pre-sliding displace-
ment, varying static force, and frictional lag. In addition, this model is continuous
and can describe different friction states smoothly and continuously. There are
six parameters in this model including two dynamic parameters and four static
parameters. Compared with the static parameters, the dynamic parameters are
difficult to be identified. In [61], the dissipation of this model was further analyzed.

1.4.2.3 Leuven Model

In [62], Swevers qualitatively compared simulation results based on the LuGre

model and measured results of the shaft’s pre-sliding motion of KUKA 361 IR.
The results indicate that the LuGre model cannot accurately characterize friction
1 A Review of Stick–Slip Nanopositioning Actuators 15

hysteresis. Thus, a more accurate model, called Leuven model, was developed by
Swevers. This model also adopts the average deformation of the bristle.
  ˇ ˇ 
dz Fd .z/ ˇ Fd .z/ ˇ n
D v 1  sgn ˇ ˇ (1.4)
dt s.v/  Fb ˇ s.v/  Fb ˇ

and friction is given

dz
f D Fh .z/ C 1 C 2 v (1.5)
dt
where n represents the coefficient describing the transition curve of different friction
states, s(v) represents the friction behavior at a constant speed, and s(v) is
 ı

s.v/ D sgn.v/ fc C .fs  fc / e.jvj=vs / (1.6)

Fh (z) is the hysteresis friction force which is a static nonlinear force having non-
local memory [15]. It consists of two parts.

Fh .z/ D Fb C Fd .z/ (1.7)

where Fb represents the original position of the transition curve when the speed
reverses, and Fd (z) represents the transition curve generated in a certain time.
The Leuven model is not only capable of providing a more accurate description
about the pre-sliding motion but also able to characterize the friction behavior
that the LuGre model describes. There are six parameters and three mechanisms
required to construct the Leuven model. In [63], a more detailed analysis and many
modifications of the Leuven model are given.

1.4.3 Summary

The static model is not suitable to analyze PSTA because it cannot describe the
pre-sliding motion, varying static friction, and friction lag. When the PA operates,
generated displacements are small which is typically at micro- and nanoscales. This
small displacement requires the concept of the bristle model for analysis. With the
large variation of the PA’s speed, the change of average deformation of the bristle
starts to have influence on friction. LuGre model and Leuven model are appropriate
for describing the friction behavior of PSTA. Although the Leuven model can give a
more precise description of the friction behavior than the LuGre model, the difficulty
of identifying six parameters and three mechanisms restricts its widespread use. The
LuGre model is widely used for analyzing the friction behavior in the PSTA because
it can provide a precise description of the friction behavior, and it is easier to be
realized [4, 26, 64].
16 P. Pan et al.

1.5 Driving Signals

In this section, different driving signals to actuate PSTA are introduced. As

mentioned above, an inertial force larger than maximum static friction force should
be obtained in order to achieve the “slip” period. To achieve the “stick” period,
the inertial force should be smaller than maximum static friction force. In view of
the change of inertial force during a cycle, the driving signal of PSTA should vary
correspondingly. There are three types of driving signals utilized to actuate PSTA.

1.5.1 Sawtooth Waveform Driving Signal

Sawtooth waveform is widely used in PSTA. Many studies have been conducted
on the sawtooth waveform to improve the performance of PSTA including step
size, speed, positioning accuracy, and repeatability [27, 65–67]. These sawtooth
waveforms can be classified into two types, including conventional sawtooth
waveform (CSW) and modified sawtooth waveform (MSW).

1.5.1.1 Conventional Sawtooth Waveform

CSW has a slow ramp followed by a rapid retraction as shown in Fig. 1.12a.
Anantheshwara et al. [67] studied a series of CSWs ranging from straight to
gradually concave. Few studies have studied convex sawtooth waveforms [66]. The
reason can be that a slower retraction where the turning point is not sharp and may
make the performance of PSTA worse. In [65] different CSWs from concave CSW
to straight CSW and all the way to convex CSW were studied. An unexpected result
was shown that the convex CSW performs better compared to the concave CSW

Fig. 1.12 Sawtooth waveform: (a) Conventional sawtooth waveform. (b) Modified sawtooth
waveform
1 A Review of Stick–Slip Nanopositioning Actuators 17

and straight CSW for the shear piezoceramics which performs as the source of
mechanical energy. According to the simulation done in [65] with respect of the
creeping nature of shear piezoceramics, a steeper retraction is gained in the convex
CSW which can improve the performance of PSTA. Straight CSW and concave
CSW are more suitable than convex CSW for the piezo tube slider because the
creeping nature is comparatively small in the piezo tube slider.

1.5.1.2 Modified Sawtooth Waveform

Modifications have been made to CSW to enhance the performance of PSTA which
leads to the development of MSW as shown in Fig. 1.12b. These modifications are
described in [65] including a flat part of waveform right before retraction and a
certain time delay right after retraction. One of the MSWs is shown in Fig. 1.12b
where the straight ramp is followed by a flat segment of which time length is T1
before retraction and a time delay is T2 followed by the beginning of next period’s
straight ramp. These modifications are effective for shear piezoceramics as well as
for piezo tube sliders. However, not all the values of T1 are effective for enhancing
the performance of PSTA. The performance of PSTA becomes worse when T1 is
too larger. With the optimum value of T1, the forward speed of sliding piece does
not decrease significantly while the sliding piece of PA goes further forward with
a distance d1 . When it retracts, the speed of sliding piece can generate a sufficient
inertial force. Compared with the conventional straight SCW the step size of the
MSW is larger. The forward speed of sliding piece decreases drastically which
cannot generate enough inertial force to obtain the “slip” period when T1 is too
large.

1.5.2 Square Waveform

Square waveform (SW) is a symmetrical driving signal as shown in Fig. 1.13. This
driving signal is usually employed in PSTA of Principle II. There are four steps in a
cycle of the square waveform driving signal.
Step 1: the square waveform signal is in the neutral state which generates no inertial
force.
Step 2: the square waveform signal zooms to a definite value which generates a large
inertial force.
Step 3: the square waveform signal remains unchanged which generates no inertial
force.
Step 4: there is a deep descent to zero in the square waveform signal which generates
a large inertial force in a reverse direction compared to that in step 2.
There are two large inertial forces which are reversed in a cycle of this signal.
However, only one large inertial force is needed in Principle II. This signal is
18 P. Pan et al.

Fig. 1.13 Square waveform:

(a) one cycle, (b) continuous
step

Fig. 1.14 Cycloid waveform:

(a) one cycle, (b) continuous
step

not suitable for PSTA of Principle II which contains one piezoelectric actuator. In
fact, this signal is suitable for PSTA of Principle II including several piezoelectric
actuators where some piezoelectric actuators are employed to adjust the friction
force and some are employed to generate actuating movement [48].
Owning to the symmetry of the square waveform driving signal, the circuit is
easy to design. In addition, this driving signal can result in a large driving force.
However, this signal also can cause problems such as loud noises.

1.5.3 Cycloid Waveform (CW)

Cycloid waveform [68–70] driving signal is also a symmetrical signal as illustrated

in Fig. 1.14. This signal is usually employed to actuate vertical PSTA. This is
because the gravity of the end effector in vertical PSTA makes an impact on the
movement. In particular when the end effector moves upwards, the acceleration of
PA (au < 0) should satisfy us Fn g < au . When the end effector moves downwards,
the acceleration of PA (ad > 0) should satisfy us Fn  g > ad . Thus, different
1 A Review of Stick–Slip Nanopositioning Actuators 19

accelerations are required to obtain the desired movement of vertical PSTA’s end
effector. To obtain a desired step, only one of the inequalities should be satisfied
during each cycle of signal. If both of them are satisfied during one cycle of signal,
an erratic motion would appear. Hence, a sawtooth voltage is likely not suitable
for the actuation of vertical PSTA against gravity because a sawtooth signal can
lead to high accelerations of opposite amplitude at the turning points during each
cycle. To realize the actuation of vertical PSTA, a cycloid signal is designed which
is a signal with a quadratic time dependence. This cycloid signal can generate
enough high acceleration in one direction while the acceleration remains small in
the opposite direction. This performance can successfully actuate vertical PSTA
upwards. To move the vertical PSTA downwards, the polarity of this cycloid signal
can be inverted [31].
The cycloid signal is designed to actuate vertical PSTA which maintains the
simplicity of the mechanical structure of PSTA. The performance of PSTAs gained
by cycloid signal is high that nanoscale resolution can be obtained. However, the
cycloid signal complicates circuit design. High-voltage broadband amplifiers are
required to generate the cycloid signal [32].

1.6 Control Methods of PSTA

There are two modes of motion in PSTA [71] including stepping mode and scanning
mode. The stepping mode means that PSTA moves step by step with a high velocity
and high travel distance when the driving signal is applied to it repeatedly. The
resolution of this mode is restricted to one step. When the difference between the
actual position and target position is less than a step, the PSTA moves slowly until it
reaches the target position. This is known as the scanning mode and the resolution
of this mode is high. Therefore, two modes of control methods were developed
for the PSTA. One kind of control method is implemented in the stepping method
to achieve a high stroke positioning. Another kind is implemented in the scanning
mode to achieve fine positioning [72].

1.6.1 Fine Positioning Control

Control is implemented in the scanning mode to achieve a fine position. These

control methods are standard controllers such as PID, optimal controller, and robust
controllers. These control methods are closed-loop.
20 P. Pan et al.

1.6.2 High Stroke Positioning Control

This control is implemented in the stepping method. The open-loop control which is
based on step counting is not suitable for this control mode. The values of every step
are not identical with respect to the nonlinear phenomenon. In addition the external
disturbances can lead to errors. Thus, closed-loop control methods were developed.

1.6.2.1 Basic Control Method

This control method is illustrated by the basic instruments in programming. When

the resolution of sensor is better than a step, the basic instruments in programming
are as follows:

While jxc  xj  step Do apply a step End While

where xc means the target position, x means the actual position.

When the resolution of the sensor is worse than a step, changes must be made to
the algorithm in order to avoid oscillation. If the resolution is limited to the value of
n  step, it is

While jxc  xj  n  step Do apply n  step End While

This control method is easy to implement which can achieve the goal of reaching
the target position.

1.6.2.2 Dehybridization Control Method (DCM)

This control method, based on the hybrid model of the stick–slip system with
three discrete states and four continuous states, was proposed in [73]. The model
of this control method is described in Fig. 1.15. The idea is described as follows:

Fig. 1.15 Block diagram for the controller design via dehybridization [73]
1 A Review of Stick–Slip Nanopositioning Actuators 21

(1) At first a continuous time model is adopted to approximate this hybrid system
which is called dehybridization. (2) Then a standard continuous control method
is applied to design the controller. A procedure named model reduction procedure
was proposed which is related to the singular perturbation theory [74]. With the
assistance of this procedure all states including discrete states and fast continuous
states are eliminated. Because hybrid nature is removed in this procedure it is termed
dehybridization. There are some prerequisites for this procedure. One of them is that
there must be a repetitive mode in the hybrid system. However, the hybrid model of a
stick–slip system does not possess a repetitive mode. An inner switching controller
is applied to provide a repetitive mode to the hybrid model of stick–slip system.
Then on the basis of feedback linearization a standard continuous control method is
applied.
DCM includes three steps: (1) Providing a repetitive mode to hybrid system;
(2) Dehybridization of the resulting repetitive hybrid system; and (3) A standard
continuous control method is applied. By means of this control method there is
no need to separate two control modes (stepping and scanning) to reach the target
position. However, it is difficult to realize dehybridization. It is necessary to render a
repetitive mode with the hybrid mode of stick–slip system which makes this control
method more complex to implement.

1.6.2.3 Numerical Frequency Proportional Control Method (NFPCM)

The NFPCM was introduced in [75], the principle of which is described in Fig. 1.16.
There are two modes of motion discussed above in PSTA. However, in this control
method two modes of motion, stepping mode and scanning mode, are combined
together in a simple way. The position error is converted into the clock signal of
which the frequency is proportional to the position error. In accordance with the
sign of position error, output signal of the counter increases or decreases. Then this
signal is converted into the driving signal whose frequency is proportional to the
position error and amplitude is constant. When PSTA reaches the target position,
the frequency of driving signal is down to zero.

Fig. 1.16 Principle scheme of numerical frequency proportional control

22 P. Pan et al.

Fig. 1.17 Principle scheme of U/f proportional control

1.6.2.4 Voltage/Frequency Proportional Control Method (V/FPCM)

This control method was proposed in [72] encompassing three existing controllers
which are sign controller, the frequency proportional controller, and the classical
proportional controller, respectively. Figure 1.17 illustrates the principle of this
control method. According to the proportional gains KU , Kf and the absolute value
of error (Xc  X), there are five cases during displacement. When absolute value of
the error (Xc  X) is originally large, the values of voltage U and frequency f are
both set to the state of saturation. In this phase the speed is constant. When the
absolute value of the error (Xc  X) becomes smaller, either the value of voltage U
or the value of frequency f is not set to the state of saturation. The obtained speed
in this phase is not constant which includes four cases.
There is a minimum voltage U0 in this control method. When the amplitude U
of the driving signal is below this minimum voltage U0 , there will be not enough
torque and the “slip” period of PSTA will not be realized. Besides this parameter
there is voltage saturation and frequency saturation in this control method to avoid
overvoltages and restrict the PSTA to operate inside the linear zone of frequency.
Two control modes (stepping and scanning) to reach the target position are not
separated. With the increment of Ku , the static error is down to zero. When Ku
increases, oscillation is generated. It is necessary to utilize a faster controller setup
to make the implementation of a higher Ku possible. In this control method, the static
error can be down to zero and the resolution can be higher than the senor resolution.
The characteristics of different control methods are summarized in Table 1.1.

1.7 Applications in Industry

By virtue of its advantages including high resolution and large travel distance, PSTA
is widely used in industry. This section discusses some of these examples.
1 A Review of Stick–Slip Nanopositioning Actuators 23

Table 1.1 Characteristics of different control methods

Two motion
Control Have been modes are
method realized? separate? Complexity Resolution Static error
BCM Y Y Simple Limited to a /n step 130 nm/10 m
DCM N N Complex Unknown Unknown
NFPCM Y N A little complex Limited to the Zero/50 nm
resolution of sensor
V/FPCM Y N A little complex Better than the Zero/10 m
resolution of sensor

Fig. 1.18 PIShift drive designed by PI and the schematic of the PIShift

1.7.1 PhysikInstrumente (PI)

PI is a company that offers a broad portfolio of precision motion technologies

including PSTA. One of the PSTAs designed by PI is shown in Fig. 1.18. This
product is called PIShift drive which is on the basis of Principle II. Modified
sawtooth driving signal is applied to one PA. The speed can reach more than 5 mm/s
with a driving signal of above 20 kHz. This PSTA works silently at this high
frequency and is self-locking when at rest. It is space-saving with relatively high
holding forces of up to 10 N and has large travel distances.

1.7.2 SmarAct

SmarAct is a company that offers miniaturized positioners. The positioners offered

by SmarAct are all based on the stick–slip principle, almost all of which are based on
24 P. Pan et al.

Fig. 1.19 (a) Linear PSTA designed by SmacAct, (b) a complete system designed by SmacAct

Principle I. The typical product of SmarAct is linear PSTA called SLC-1720 (http://
www.smaract.de/) as shown in Fig. 1.19a. Their 7-DOF manipulator is a complete
system consisting of 6 SLC-1720s and a rotary PSAT as shown in Fig. 1.19b. SLC-
1720 is based on linear crossed-roller slides, and it is known for high rigidity and
straightness. Therefore, it is ideally suitable for micro- or nanopositioning. The
dimensions of SLC-1720 are 22  17  8.5 mm3 and the weight of it is about 13 g.
It has a travel distance of 12 mm with a high resolution as much as 50 nm. The
maximum speed of SLC-1720 can reach 13 mm/s with a high frequency of driving
signal as much as 18.5 kHz. The 7-DOF manipulator is suitable for use inside the
chamber of scanning electron microscope (SEM) and focused ion beam (FIB).

1.7.3 KlockeNanotechnik

KlockeNanotechnik offers a nanomotor (http://www.nanomotor.de/) shown in

Fig. 1.20a to provide a linear motion driven by PA. It consists of a piezo tube,
a cylindrical housing, and a slider in which a free axial hole exits. This free
axial hole can be utilized to transport a tip, an electrode, or a microgripper. This
nanomotor takes advantage of the stick–slip principle, having a positioning stroke up
to centimeters with a maximum speed of up to 2 mm/s and a resolution down to the
atomic scale. The driving force against gravity can reach more than 25 g. The length
of it is half the size of a match stick shown in Fig. 1.20a. This nanomotor is suitable
for ultra-vacuum, low temperature, and even underwater applications. The 3-DOF
manipulator, shown in Fig. 1.20b, is one of the applications of the nanomotor. The
manipulator comprises a titling table and several nanomotors, including a central
nanomotor for z movement. A further advantage of this manipulator is that there is
thermal compensation for the central nanomotor when it points downwards. The tip
placed inside the titling table can be positioned in a range of 5  5  19 mm3 .
1 A Review of Stick–Slip Nanopositioning Actuators 25

Fig. 1.20 (a) Nanomotor designed by KlockeNanotechnik, (b) a 3 DOF manipulator

Fig. 1.21 (a) MM3A-EM designed by KleindiekNanotechnik, (b) a system consisted of several
MM3A-Ems

1.7.4 KleindiekNanotechnik

Products designed by KleindiekNanotechnik (http://www.nanotechnik.com/mm3a-

em.html) are based on the stick–slip principle. One of their manipulators is MM3A-
EM. The dimensions of this manipulator are 60  22  25 mm3 which are easy to be
integrated into electron microscopes. The manipulator consists of 3 PAs combined
by serial mechanism. The first 2-dimensional motions are rotary motion with a speed
of up to 10 mm/s and the resolutions of them are 5 nm and 3.5 nm, respectively. The
last dimensional motion is linear motion with a speed of up to 2 mm/s, of which
the resolution is 0.25 nm. The holding force of it can reach up to 1 N. The system
shown in Fig. 1.21b consists of multiple of these manipulators.
26 P. Pan et al.

Fig. 1.22 (a) ANPX 101 designed by Attocube System. (b) AttoMICROSCOPY designed by
Attocube System

1.7.5 Attocube System

The motion of PSTAs designed by Attocube System is reliable under environments

such as high magnetic fields, ultra-high vacuum, and cryogenic temperatures. One of
their typical PSTAs is ANPX 101 (http://www.attocube.com/) as shown in Fig. 1.22a.
The dimensions of this PSTA are 24  24  11 mm3 with a weight of 20 g. When
operating in the scanning mode, the scan range is 5 m and the resolution can reach
up to sub-nanometer. When operating in the stepping mode, the travel distance can
be up to 5 mm with a minimum step size of 50 nm and a maximum speed of up
to 3 mm/s. Maximum working load of this PSTA can be 1 N. AttoMICROSCOPY
shown in Fig. 1.22b is a system that comprises multi-dimensional manipulators in
the chamber of a microscope to perform physical manipulation tasks.

1.8 Conclusion

This chapter provided an overview of actuators based on stick–slip principle and

piezoelectric actuators. The characteristics of different actuators based on stick-
slip principle are summarized in Table 1.2. This kind of actuators is known for
advantages of long travel distance and high resolution. Piezoelectric effect is the
source of mechanical force, and the combination of inertial force and friction force
is taken as the actuation force in PSTA. Existing PSTAs were categorized into
two groups with two specific principles, namely PFDA and PIDA. The difference
between the two groups is which type of force leads to the actual movement of the
end effector. The applications of PSTAs are discussed on the basis of the number
of DOFs, namely one-DOF actuators and multi-DOF actuators. Since conventional
Table 1.2 Characteristics of different PSTAs
Author PR DS CM RS Size (mm) WD (kg) AM P/S DOF
Wen [48] PIDA SW N/A 40 46  20  14 >1 Y N/A 1-L(H)
Wen [48] PIDA SW N/A 7.5 rad N/A N/A N N/A 1-R
Wen [48] PIDA SW BCM 30, 2.5 urad N/A >0.8 Y Parallel 3-L(H), R
Pohl [27] PFDA CSW N/A 40 30  30  20 1 N N/A 1-L(H)
Niedermann et al. [76] PFDA CSW N/A 10 N/A 0.12 N N/A 1-L(H)
Renner et al. [31] PFDA CW N/A 3 N/A N/A N N/A 1-L(V)
Morita et al. [77] PFDA CSW N/A N/A 15(OD)  10(ID)  11 N/A Y N/A 1-R
Zhang et al. [78] PFDA CSW N/A 26, 0.33 mrad N/A N/A Y Serial 2-L(H), R
Morita et al. [79] PIDA MSW BCM N/A N/A N/A N Parallel 3-L(H), R
Stieg et al. [66] PFDA CW N/A 10 N/A 0.025 Y N/A 1-L(V)
Agrait [29] PFDA CSW N/A 0.5"4# N/A N/A Y N/A 1-L(V)
Breguet and Clavel [75] PFDA CSW NFPCM <5 N/A N/A N N/A 1-L(H)
1 A Review of Stick–Slip Nanopositioning Actuators

Rakotondrabe et al. [72] PFDA CSW V/FPCM <1.2 N/A N/A N Parallel 2-L(H), R
Blackford et al. [46] PFDA CW N/A 10 10  13  38 N/A Y Parallel 2-L(V), R
Howald et al. [45] PFDA CSW N/A 4.8 urad N/A N/A Y Parallel 3-L(V), R
Libioulle et al. [42] PFDA CW N/A 20,10"30# 30  30  25 N/A Y Parallel 2-L(H,V)
Tapson and Greene [51] PFDA CSW N/A 20 35  35  35 0.05 N Serial 2-L(H)
Göken [49] PFDA MSW N/A 20 52  25  20 N/A N Serial 3-L(H), R
Yamagata and Higuchi [39] PIDA N/A N/A 100 N/A >1.3 N N/A 1-L(H)
Darby and Pellegrino [37] PIDA MSW N/A <500 N/A 1 Y N/A 1-L(H)
Yoshida et al. [18] PFDA MSW N/A N/A 150  35  15 0.15 Y N/A 1-L(H)
Claeyssen et al. [23] PIDA N/A N/A <8 40  12  27 1 Y N/A 1-L(H)
(continued)
27
 28

Table 1.2 (continued)

Author PR DS CM RS Size (mm) WD (kg) AM P/S DOF
PIShift drive (http://www. PIDA MSW N/A 300 N/A 1 N N/A 1-L(H)
physikinstrumente.com/)
SLC-1720 (http://www. PFDA CSW N/A 50 22  17  8.5 4/0.1 Y N/A 1-L(H/V)
smaract.de/)
Nanomotor (http://www. PFDA N/A N/A 0.1 18  ˚33 0.025 N N/A 1-L(V)
nanomotor.de/)
MM3A-EM (http://www. PFDA N/A N/A 5,3.5,0.25 60  22  25 0.1 N Serial 3-L, R
nanotechnik.com/mm3a-em.
html)
ANPX 101 (http://www. PFDA N/A N/A 50 24  24  11 0.1 N N/A 1-L(H)
attocube.com/)
PR principle, DS driving signal, CM control method, RS resolution of each actuator, WD working load, AM auxiliary mechanism, P/S parallel
mechanism or serial mechanism
P. Pan et al.
1 A Review of Stick–Slip Nanopositioning Actuators 29

driving signals cannot meet the demand of actuation of different structures of

PSTAs and achieving high performance, different driving signals were presented.
The hysteresis of piezoelectric actuators and back-slash problem in PSTA affect
their performance. Therefore, control methods were discussed. In order to achieve
satisfactory performance, a number of issues must be considered For instance, the
influence that driving signal poses on the performance of PSTA should be studied.
Closed-loop control methods which are easy to implement should be explored to
eliminate the intrinsic errors effectively. The mechanical structure of PSTA should
be optimized including the improvement of auxiliary mechanisms to improve the
dynamic oscillation of PSTA. Finally the thermal effect of PAs should be considered
to reduce the adverse effect for the performance of PSTA.

References

1. P. Ouyang, R. Tjiptoprodjo, W. Zhang, G. Yang, Micro-motion devices technology: the state

of arts review. Int. J. Adv. Manuf. Technol. 38(5–6), 463–478 (2008)
2. A.L. Soderqvist, U.S. Patent 3,957,162, 1976
3. Z. Zhang, Q. An, J. Li, W. Zhang, Piezoelectric friction–inertia actuator—a critical review and
future perspective. Int. J. Adv. Manuf. Technol. 62(5–8), 669–685 (2012)
4. B.W. Zhong, Research on key technologies of cross-scale precision platform based on stick–
slip driving. Dissertation, Harbin Institute of Technology, 2012
5. W.S. Owen, E.A. Croft, The reduction of stick–slip friction in hydraulic actuators. IEEE/ASME
Trans. Mechatron. 8(3), 362–371 (2003)
6. J. Peng, X. Chen, Modeling of piezoelectric-driven stick–slip actuators. IEEE/ASME Trans.
Mechatron. 16(2), 394–399 (2011)
7. J. Li, G. Yang, W. Zhang, S. Tu, X. Chen, Thermal effect on piezoelectric stick–slip actuator
systems. Rev. Sci. Instrum. 79(4), 046108 (2008)
8. Q. Zhang, X. Chen, Q. Yang, W. Zhang, Development and characterization of a novel
piezoelectric-driven stick–slip actuator with anisotropic-friction surfaces. Int. J. Adv. Manuf.
Technol. 61(9–12), 1029–1034 (2012)
9. T. Higuchi, M. Watanabe, U.S. Patent 4,894,579, 1990
10. J.L. Ha, R.F. Fung, C.S. Yang, Hysteresis identification and dynamic responses of the impact
drive mechanism. J. Sound Vib. 283(3), 943–956 (2005)
11. Y.-T. Liu, T. Higuchi, R.-F. Fung, A novel precision positioning table utilizing impact force of
spring-mounted piezoelectric actuator—part II: theoretical analysis. Precis. Eng. 27(1), 22–31
(2003)
12. Z.J. Du, W. Dong, L.N. Sun, The flexure hinge and its application in the precision parallel
robot. J. Harbin Inst. Technol. 38(9), 1469–1473 (2006)
13. J.M. Paros, How to design flexure hinges. Mach. Des. 37, 151–156 (1965)
14. F.S. Eastman, The design of flexure pivots. J. Aeronaut. Sci. (Inst. Aeronaut. Sci.) 4(1) (2012)
15. B. Armstrong-Helouvry, Control of Machines with Friction, vol 128 (Springer Science &
Business Media, 1991)
16. K.-J. Lim, J.-S. Lee, S.-H. Park, S.-H. Kang, H.-H. Kim, Fabrication and characteristics of
impact type ultrasonic motor. J. Eur. Ceram. Soc. 27(13), 4159–4162 (2007)
17. Y. Okamoto, Y. Tanijiri, U.S. Patent 6,092,431, 2000
18. R. Yoshida, Y. Okamoto, T. Higuchi, A. Hamamatsu, Development of smooth impact drive
machine—proposal of driving mechanism and basic performance (in Japanese). J. Jpn. Soc.
Precis. Eng. 65(1), 111–115 (1999)
30 P. Pan et al.

19. S. Chang, S. Li, A high resolution long travel friction-drive micropositioner with pro-
grammable step size. Rev. Sci. Instrum. 70(6), 2776–2782 (1999)
20. Y. Wang, S. Chang, Design and performance of a piezoelectric actuated precise rotary
positioner. Rev. Sci. Instrum. 77(10), 105101 (2006)
21. C.-L. Chu, S.-H. Fan, A novel long-travel piezoelectric-driven linear nanopositioning stage.
Precis. Eng. 30(1), 85–95 (2006)
22. C. Belly, MRI-compliant piezo micro-actuator (2009), http://www.lirmm.fr/uee09/doc/
Student_s_%20presentations/Christian_Belly.pdf. Accessed 7 Jan 2010
23. F. Claeyssen, A. Ducamp, F. Barillot, R. Le Letty, T. Porchez, O. Sosnicki, C. Belly, Stepping
piezoelectric actuators based on APAs, in Proceedings of the Actuator, 2008
24. B.B. Nie, Research on precision rotary positioner based on stick–slip driving. Dissertation,
Harbin Institute of Technology, 2013
25. Z.S.R.W.T. Guoan, Y.M.S. Lining, Designing and dynamic modeling of 1D nanopositioner
based on stick–slip motion principle. J. Mech. Eng. 19, 004 (2012)
26. W. Rong, S. Zhang, M. Yu, L. Sun, A 3D stick–slip nanopositioner for nanomanipula-
tion, in International Conference on Mechatronics and Automation (ICMA) (IEEE, 2011),
pp. 195–199
27. D.W. Pohl, Dynamic piezoelectric translation devices. Rev. Sci. Instrum. 58(1), 54–57 (1987)
28. Y. Okamoto, R. Yoshida, Development of linear actuators using piezoelectric elements.
Electron. Commun. Jpn. (Part III: Fundamental Electronic Science) 81(11), 11–17 (1998)
29. N. Agrait, Vertical inertial piezoelectric translation device for a scanning tunneling microscope.
Rev. Sci. Instrum. 63(1), 263–264 (1992)
30. I. Altfeder, A. Volodin, Low-temperature scanning tunneling microscope with a reliable
piezoelectrical coarse approach mechanism. Rev. Sci. Instrum. 64(11), 3157–3160 (1993)
31. C. Renner, P. Niedermann, A. Kent, A vertical piezoelectric inertial slider. Rev. Sci. Instrum.
61(3), 965–967 (1990)
32. V. Yakimov, Vertical ramp-actuated inertial micropositioner with a rolling-balls guide. Rev.
Sci. Instrum. 68(1), 136–139 (1997)
33. R. Yoshida, Y. Okamoto, K. Ishibashi, Y. Tanijiri, H. Okada, Movable stage utilizing
electromechanical transducer. Google Patents, 1998
34. Q.-h. Lu, Z.-j. Gao, G.-z. Yan, D.-t. Yan, The study on miniature inertial piezo-actuators.
Piezoelectrics Acoustooptics 2, 009 (2004)
35. Sasaki, U.S. Patent 0,036,533 A1, 2007
36. T. Jiang, T. Ng, K. Lam, Optimization of a piezoelectric ceramic actuator. Sens. Actuators A:
Phys. 84(1), 81–94 (2000)
37. A. Darby, S. Pellegrino, Inertial stick–slip actuator for active control of shape and vibration. J.
Intell. Mater. Syst. Struct. 8(12), 1001–1011 (1997)
38. H.-j. Zhang, F. Huang, Piezo impact drive mechanism for precise approach and manipulation.
J. Zhejiang Univ. Eng. Sci. 34(5), 519–522 (2000)
39. Y. Yamagata, T. Higuchi, A micropositioning device for precision automatic assembly using
impact force of piezoelectric elements, in 1995 IEEE International Conference on Robotics
and Automation (ICRA) (1995) pp. 666–671
40. T. Higuchi, K. Furutani, Y. Yamagata, K.-i. Kudoh, M. Ogawa, Improvement of velocity of
impact drive mechanism by controlling friction. J. Jpn. Soc. Precis. Eng. Seimitsu Kogaku
Kaishi 58(8), 1327–1332 (1992)
41. J.M. Breguet, Actuionneurs “stick and slip” pour micromanipulateurs. Dissertation, École
Polytechnique Fédérale de Lausanne, 1998
42. L. Libioulle, A. Ronda, I. Derycke, J. Vigneron, J. Gilles, Vertical two-dimensional piezo-
electric inertial slider for scanning tunneling microscope. Rev. Sci. Instrum. 64(6), 1489–1494
(1993)
43. A. Bergander, W. Driesen, A. Lal, T. Varidel, M. Meizoso, H. Bleuler, J.-M. Breguet,
Position feedback for microrobots based on scanning probe microscopy, in 2004 IEEE/RSJ
International Conference on Intelligent Robots and Systems (IROS) (IEEE, 2004),
pp. 1734–1739
1 A Review of Stick–Slip Nanopositioning Actuators 31

44. B. Blackford, M. Jericho, Simple two-dimensional piezoelectric micropositioner for a scanning

tunneling microscope. Rev. Sci. Instrum. 61(1), 182–184 (1990)
45. L. Howald, H. Rudin, H.J. Güntherodt, Piezoelectric inertial stepping motor with spherical
rotor. Rev. Sci. Instrum. 63(8), 3909–3912 (1992)
46. B. Blackford, M. Jericho, M. Boudreau, A vertical/horizontal two-dimensional piezoelectric
driven inertial slider micropositioner for cryogenic applications. Rev. Sci. Instrum. 63(4),
2206–2209 (1992)
47. H. Zhang, G. Cheng, H. Zhao, P. Zeng, Z. Yang, Two dimensional precise actuator driven by
piezoelectric bimorph impact. J. Jilin Univ. (Eng. Technol. Ed.) 36(1), 67–71 (2006)
48. J.M. Wen, Study on planar inertia piezoelectric moving mechanism. Dissertation, Jilin
University, 2009
49. M. Göken, Scanning tunneling microscopy in UHV with an X, Y, Z micropositioner. Rev. Sci.
Instrum. 65(7), 2252–2254 (1994)
50. R. Brockenbrough, J. Lyding, Inertial tip translator for a scanning tunneling microscope. Rev.
Sci. Instrum. 64(8), 2225–2228 (1993)
51. J. Tapson, J. Greene, A simple dynamic piezoelectric X-Y translation stage suitable for
scanning probe microscopes. Rev. Sci. Instrum. 64(8), 2387–2388 (1993)
52. C. Meyer, O. Sqalli, H. Lorenz, K. Karrai, Slip–stick step-scanner for scanning probe
microscopy. Rev. Sci. Instrum. 76(6), 063706 (2005)
53. A. Morin, New friction experiments carried out at Metz in 1831–1833. Proc. Fr. R. Acad. Sci.
4(1), 128 (1833)
54. C.C. De Wit, J. Carrillo, A modified EW-RLS algorithm for systems with bounded distur-
bances. Automatica 26(3), 599–606 (1990)
55. L.L. Liu, H.Z. Liu, Z.Y. Wu, Z.M. Wang, The application of the friction model in mechanical
system. Adv. Mech. 38(2), 201–212 (2008)
56. V. Johannes, M. Green, C. Brockley, The role of the rate of application of the tangential force
in determining the static friction coefficient. Wear 24(3), 381–385 (1973)
57. C.C. De Wit, H. Olsson, K.J. Astrom, P. Lischinsky, A new model for control of systems with
friction. IEEE Trans. Autom. Control 40(3), 419–425 (1995)
58. P.R. Dahl, Solid friction damping of spacecraft vibrations, in 1995 IEEE International
Conference on AIAA Guidance and Control Conference (1995), pp. 1075–1104
59. P.R. Dahl, Solid friction damping of mechanical vibrations. AIAA J. 14(12), 1675–1682
(1976)
60. D.A. Haessig, B. Friedland, On the modeling and simulation of friction. J. Dyn. Syst. Meas.
Control 113(3), 354–362 (1991)
61. N. Barahanov, R. Ortega, Necessary and sufficient conditions for passivity of the LuGre friction
model. IEEE Trans. Autom. Control 45(4), 830–832 (2000)
62. J. Swevers, F. Al-Bender, C.G. Ganseman, T. Projogo, An integrated friction model structure
with improved presliding behavior for accurate friction compensation. IEEE Trans. Autom.
Control 45(4), 675–686 (2000)
63. V. Lampaert, J. Swevers, F. Al-Bender, Modification of the Leuven integrated friction model
structure. IEEE Trans. Autom. Control 47(4), 683–687 (2002)
64. K. Johanastrom, C. Canudas-De-Wit, Revisiting the LuGre friction model. IEEE Control Syst.
28(6), 101–114 (2008)
65. J. Wang, Q. Lu, How are the behaviors of piezoelectric inertial sliders interpreted? Rev. Sci.
Instrum. 83(9), 093701 (2012)
66. A. Stieg, P. Wilkinson, J. Gimzewski, Vertical inertial sliding drive for coarse and fine
approaches in scanning probe microscopy. Rev. Sci. Instrum. 78(3), 036110 (2007)
67. K. Anantheshwara, N. Murali, M. Bobji, Effect of friction on the performance of inertial slider.
Sadhana 33(3), 221–226 (2008)
68. W.F. Smith, M. Abraham, J. Sloan, M. Switkes, Simple retrofittable long-range x–y translation
system for scanned probe microscopes. Rev. Sci. Instrum. 67(10), 3599–3604 (1996)
32 P. Pan et al.

69. K. Svensson, Y. Jompol, H. Olin, E. Olsson, Compact design of a transmission electron

microscope-scanning tunneling microscope holder with three-dimensional coarse motion. Rev.
Sci. Instrum. 74(11), 4945–4947 (2003)
70. S. Reymond, Low temperature scanning contact potentiometry. Rev. Sci. Instrum. 75(3),
694–698 (2004)
71. J.-M. Breguet, R. Perez, A. Bergander, C. Schmitt, R. Clavel, H. Bleuler, Piezoactuators for
motion control from centimeter to nanometer, in 2000 IEEE/RSJ International Conference on
Intelligent Robots and Systems (IROS) (2000), pp. 492–497
72. M. Rakotondrabe, Y. Haddab, P. Lutz, Voltage/frequency proportional control of stick–slip
micropositioning systems. IEEE Trans. Control Syst. Technol. 16(6), 1316–1322 (2008)
73. B. Sedghi, Control design of hybrid systems via dehybridization (2003)
74. H.K. Khalil, J. Grizzle, Nonlinear Systems, vol. 3 (Prentice Hall, Upper Saddle River, 2002)
75. J.-M. Breguet, R. Clavel, Stick and slip actuators: design, control, performances and appli-
cations, in 1998 IEEE International Symposium on Micromechatronics and Human Science
(MHS) (IEEE, 1998), pp. 89–95
76. P. Niedermann, R. Emch, P. Descouts, Simple piezoelectric translation device. Rev. Sci.
Instrum. 59(2), 368–369 (1988)
77. T. Morita, R. Yoshida, Y. Okamoto, M.K. Kurosawa, T. Higuchi, A smooth impact rotation
motor using a multi-layered torsional piezoelectric actuator. IEEE Trans. Ultrason. Ferroelectr.
Freq. Control 46(6), 1439–1445 (1999)
78. Y. Zhang, W. Zhang, J. Hesselbach, H. Kerle, Development of a two-degree-of-freedom
piezoelectric rotary-linear actuator with high driving force and unlimited linear movement.
Rev. Sci. Instrum. 77(3), 035112 (2006)
79. T. Morita, R. Yoshida, Y. Okamoto, T. Higuchi, Three DOF parallel link mechanism utilizing
smooth impact drive mechanism. Precis. Eng. 26(3), 289–295 (2002)
Chapter 2
Piezoelectric Motor Technology: A Review

Amro Shafik and Ridha Ben Mrad

Abstract Piezoelectric actuators are increasingly used in various nanopositioning

applications and emerging applications where miniaturization is important. This
is due to their unique characteristics including their very high accuracy and short
response time as compared to electromagnetic based motors and actuators and
higher output force compared to electrostatic actuators. Piezoelectric motors use
actuators that take advantage of the converse piezoelectric effect. In this chapter,
these motors are classified into quasistatic and ultrasonic motors (USMs) based
on their working frequency. Several designs from the literature and commercial
suppliers are reviewed and their characteristics are presented. Two examples of
piezoelectric motors are discussed in detail. These include a piezoworm stage and
a USM with segmented electrodes. Future development of these technologies is
also briefly discussed addressing issues such as increasing the output power, the
efficiency, and further miniaturization of these devices.

2.1 Introduction

There is an ever increasing need for ultra-precision positioning technologies.

These ultra-precision positioning technologies require actuators and motors with
nanometer accuracy and with ranges in the order of tens of millimeters. The need
is increasing in many fields including nano-metrology, biomedical and surgical
devices, NEMS and MEMS, micro and nanolithography and semi-conductor man-
ufacture, and adaptive optics [1–6]. Piezo technology is a class of actuation that is
capable of fulfilling this need.
Depending on the specific application area, motors are needed to have a number
of characteristics including small size, light weight, low noise levels, no electromag-
netic interference [7], high accuracy and precision, short response time, high output
force, simple design, and simple operation [8]. Motors and actuators can generally
be divided, depending on their working principle, into six main technologies:
electromagnetic, electrostatic, thermal, osmotic, electro-conjugate fluid (ECF), and

A. Shafik • R. Ben Mrad ()

Mechanical and Industrial Engineering, University of Toronto, Toronto, ON, Canada
e-mail: [email protected]; [email protected]

© Springer International Publishing Switzerland 2016 33

C. Ru et al. (eds.), Nanopositioning Technologies,
DOI 10.1007/978-3-319-23853-1_2
34 A. Shafik and R. Ben Mrad

piezoelectric actuators [8]. Electromagnetic (EM) motors and actuators produce

forces that typically rely on the interaction between a magnetic field that is induced
by a current in the stator coil and the rotor’s permanent magnets. These motors have
a high energy density, but unfavorable scaling laws hinder them from being minia-
turized, even to the millimeter size. Electrostatic actuators, like electromagnetic
actuators, use non-contact forces, which are produced by the interaction between
materials with different charges. Electrostatic actuators have excellent scalability
which makes them popular and widely used in microelectromechanical systems
(MEMS) [9]. However, there are some disadvantages that limit their use, such
as snapping down at the end of the output range, nonlinear output, sensitivity to
the environment (e.g., humidity and ambient gas content), and the output torque
or force for micro scale devices is typically limited to the pico scale. In thermal
actuators and motors, the driving force relies on the mechanical strain of the
material used. Amongst thermal actuators, shape memory alloys (SMA) have the
highest strain rates. They have a relatively good scalability and high output force but
suffer from relatively low response time and limited lifespan because of the plastic
strain that arises with repeated cycling. Osmotic actuators do not require electrical
input power. The driving force, in this case, is developed based on the increase
of the pressure in a vessel, which in turn leads to an expansion of a diaphragm.
This pressure is achieved by the osmotic effect that leads to a flow of liquid in
one direction across a semi-permeable diaphragm. These actuators are suitable for
microfluidics applications and are known to have a slow response time and are
characterized by complex designs and different problems related to fouling and
solute deposition. ECF motors and actuators produce force that is based on the
jetting of a fluid in an electric field. The jet rotates the motor by using vanes on the
rotor. This type of motors has good scalability and good output force up to sizes
of the order of 1 mm3 . Piezoelectric ultrasonic motors (USMs) are one class of
piezoelectric motors. They produce forces that rely on the converse piezoelectric
effect, which converts the electric input into a mechanical strain. This type can
produce useful forces in small-size actuators designed to operate at the resonant
frequency of the stator. Overall, USMs offer larger torque at smaller scales and
have the best potential for meeting the high performance requirements for small and
micro scale devices. USMs can generate torques two orders of magnitude larger
than other technologies at the micro scale [10]. Piezoelectric USMs as well as
other piezoelectric motor technologies are presented in detail in this chapter. These
motors use piezoelectric actuators that are typically incorporated inside mechanical
structures and are repeatedly excited to generate linear or rotary motion.
Piezoelectric motors rely on integrating piezo actuators that generally offer many
advantages such as large bandwidth, low thermal coefficients, high reliability, large
force, unlimited resolution (sub nanometer range), no magnetic field produced,
fast response (microsecond time constant), operation at cryogenic temperatures,
and they come at moderate prices compared to other actuators. Other charac-
teristics include displacements ranging from 0.001 to 0.002 of the piezo length
(0.1–0.2 %), and a maximum operating temperature near 300 ı C for widely
used ceramics. However, they have some disadvantages including high voltage
2 Piezoelectric Motor Technology: A Review 35

Fig. 2.1 Mechanical characteristics of (a) DC motors, (b) Ultrasonic motors

requirements (typically 100 V) increasing with the thickness of the actuator. Creep
and hysteresis nonlinearities at levels ranging from 15 to 20 % are also present in
these devices [11]. In addition, as all ceramics, piezoceramic actuators are brittle
and require special protection and packaging. Nevertheless, piezomotors compete
with magnetic motors when the requirements include small size, fast response, and
high accuracy.
A comparison between USMs and DC motors characteristics was made in [7].
As shown in Fig. 2.1a, the maximum efficiency of a DC motor occurs when the
output torque is minimum and the motor is operating close to the no-load speed.
On the other hand, as shown in Fig. 2.1b, the maximum efficiency of a USM
occurs at a higher torque and lower speed. Therefore, a USM is more suitable
for operation when the operating conditions call for high torque and low speed.
Another comparison between the two motors focused on their transient response
characteristics is shown in Fig. 2.2a [12]. It is evident that a USM can respond
faster than an electromagnetic motor with better transient characteristics. A third
comparison is made with respect to the miniaturization versus the efficiency of
both types of motors [13]. As shown in Fig. 2.2b, the efficiency of electromagnetic
motors (EM) declines rapidly when the diameter of the motor is below 10 mm.
However, the efficiency of USM changes smoothly. In addition, manufacturing of
electromagnetic motors in smaller sizes is difficult because their rotors need to be
surrounded by coils and their structures are more complex than that of a USM.
Hence, miniaturization of USMs is more achievable than that of electromagnetic
motors.
There are many other advantages of using piezoelectric USMs as compared to
DC motors. The USMs have torque densities three to ten times higher than standard
DC motors [8]. Therefore, smaller USMs can potentially replace DC motors and
still exhibit high torque outputs. Reduction in size and weight, simple structure,
and absence of gearing mechanisms lead to accurate positioning resolution, quick
response, excellent controllability, and hard braking. USMs can produce high
36 A. Shafik and R. Ben Mrad

Fig. 2.2 Comparison between electromagnetic motors and USM (a) Basic form of transient
response characteristics (b) Basic form of miniaturization vs. efficiency curve

holding torque even if no power is applied [14]. In addition, no magnetic or

radioactive fields are generated in USMs, and their effects are negligible on them.
The use of ultrasonic vibration leads to silent operation. The applications for USMs
include autofocus systems in cameras [7], micro-robots [8, 10], and micro-actuators
for catheters or tele-surgery operations [15].
Piezoelectricity was discovered by the Currie brothers in 1880 in materials like
Quartz and Rochelle salt. In the 1940s, the discovery of barium titanate (BaTiO3 )
led to the beginning of an industrial era for piezoelectric actuator technology. Early
piezoelectric actuators and motors were mainly used in the 1960s in the watch
industry. In 1965, Lavrinenko developed the first ultrasonic piezomotor [16]. In
the 1970s, Siemens developed a step piezomotor [7], and in 1973, IBM developed
an ultrasonic piezomotor based on longitudinal vibrations of horns [17]. In 1980,
a Langevin vibrator was first used to excite the longitudinal vibration mode in
USMs [18]. A significant advance in piezomotor technology occurred in 1982
and 1983 when Sashida developed the standing wave and traveling wave USMs,
respectively [19, 20]. In 1985, Kumada developed a USM based on a hybrid type of
longitudinal and torsional vibrations that was driven by a single-phase signal [21]. In
1987, Panasonic developed a ring-type traveling wave ultrasonic piezomotor using
Sashida’s design [22]. In 1987, Shinsei© [7] was the first company to commercialize
USMs.
In this chapter, a detailed review of piezoelectric motor technology is presented.
The chapter is organized as follows. The basic working principles of various
piezomotors are discussed in Sect. 2.2. A comprehensive classification of the
piezomotor family is presented in Sect. 2.3. Several linear and rotary designs of
piezomotors are reviewed for quasistatic and ultrasonic types in Sects. 2.4 and
2.5, respectively. Examples of piezomotors are presented in Sect. 2.6, and future
developments are listed in Sect. 2.7.
2 Piezoelectric Motor Technology: A Review 37

2.2 The Piezoelectric Effects

Piezoelectric materials are known to have two effects. The direct piezoelectric effect
is the ability of the piezoelectric material to generate an electric charge when it is
subject to a mechanical stress. The converse piezoelectric effect is the ability of the
material to generate stress when it is subject to an electric field. The direct effect
is utilized mainly in sensors, while the converse effect is utilized in actuators. The
working principle of piezomotors is based on the converse piezoelectric effect which
is based on having one or more piezoelectric actuators integrated into a mechanical
structure. These piezoelectric actuators are powered leading to generating a strain
in the piezo material.
The output strain in piezoelectric actuators is very limited, about 0.1–0.2 % of the
length of these actuators. Therefore, in various applications and in order to achieve
a useful strain, the initial strain is amplified by using amplifying mechanisms or by
simply repeating the small steps of the piezoelectric actuator at a high frequency
leading to macroscopic motion. This latter approach is typically the case with
USMs where the piezoelectric motors are excited repeatedly at one of their resonant
frequencies. Compliant structures are often used as amplifying mechanisms and
ways to channel the strain to the load especially when the motors are operating
at high frequencies and this leads to high precision due to the absence of backlash
and ease in miniaturization since typically there are no moving parts incorporated
into the structure of the motor except for the rotor or moving load.

2.3 Classification of Piezomotors

Piezomotors are classified in the literature according to a number of criteria

including the wave propagation method, the output motion and its direction, the
geometric shape of the stator, and the vibration modes used and their combinations
[8, 14, 15, 23, 24]. The most common classification is based on the driving wave
generation method. Zhao [7] added additional features such as the contact status
between the stator and rotor, the nature of the excitation of the stator by the
piezoelectric actuators (resonant versus non-resonant), the direction of the operating
mode (in-plane versus out-plane), and the number of degrees of freedom of the rotor.
Piezomotors are classified in this report according to the motion type into linear
and rotary motors, and according to the input frequency into quasistatic, in which
the input frequency is relatively low and does not correspond to any of the resonant
frequencies of the motor, and ultrasonic, in which the frequency is above 20 kHz and
the excitation frequency often corresponds to a resonant frequency of the motor.
Quasistatic motors, in turn, are classified into inertial and stepping types based
on their principle of operation. The USM family is subdivided into a number of
subfamilies. Contact and non-contact types USMs. In the contact type, the stator
and rotor are in contact by means of a friction material that transfers the vibration
38 A. Shafik and R. Ben Mrad

Fig. 2.3 Taxonomy of piezomotors

of the stator to the rotor. The non-contact type uses a fluidic medium instead of
a friction material. USMs can be also categorized into micro motors, in which
the size is few millimeters and smaller, and macro motors, which are larger than
three millimeters. USMs can be labeled based on the stator shape and construction,
direction of rotation, number of driving piezo actuators, type of exciting vibration
signal, type of generated driving wave, form of piezo elements, and piezo material
used. This entire taxonomy is shown in Fig. 2.3.

2.4 Quasistatic Motors

Quasistatic motors are a type of piezomotors that work with an input voltage at
a frequency lower than any of the resonant frequencies of the motor. Quasistatic
motors are classified into stepping and inertial types based on their principle of
operation. Most of the quasistatic motors are of the stepping type. In this section,
both types are discussed for both linear and rotary type configurations.
In the stepping types, the motor consists of several piezo actuators that are
integrated within a mechanical structure in a way to generate motion by a sequence
of expand–contract and/or clamp–unclamp cycles. In the inertial types, the motor
works mainly based on the inertia of a moving mass which is connected to a metal
shaft which in turn is attached to a piezo actuator. Generally, quasistatic motors have
higher resolution and force but lower speed than USMs.
2 Piezoelectric Motor Technology: A Review 39

Fig. 2.4 Inchworm stepping sequence

2.4.1 Linear Quasistatic Motors

Inchworm Motors The most common and simplest type of quasistatic motors is
the inchworm motor [25–27]. These motors consist typically of three piezoelectric
actuators; two are used for clamping and one for extension to generate the desired
motion of the load. The inchworm uses a sequential clamping mechanism to inch to
the required position (see Fig. 2.4). Several designs based on the same principle with
a number of variations have been proposed [28–30]. One of them is the caterpillar
motor [31] in which five piezoelectric actuators are used instead of three to provide
extra clamping and extension in order to have a continuous motion (i.e., smoother
motion). The caterpillar motor is more complex than the three-actuator based
inchworm. Inchworm motors are typically pursued when nanometers accuracy is
required [32].
Walking Motors As the name implies, the walking motor uses two or more sets of
piezoelectric actuators for enabling the motion steps or walks. The arrangement of
actuators makes one actuator or set of actuators used for clamping while the other
®
set is used to move the rail [33]. In the PiezoWalk Drive , the system is composed
of two sets of piezoelectric actuators and two legs for pushing the moving element.
Each set consists of two multilayer piezo stack actuators. One set is responsible
for pushing the legs to enable motion, and the other set is used for engaging and
disengaging the legs. One step of the drive piezos moves the runner up to 15 m.
Subsequently, the other clamp engages with the other leg to enable a continuous
motion [34].
®
The PiezoWalk (produced by Physik Instrumente [35]) and the Piezo LEGS(R)
(produced by Piezomotor AB [36]) work based on the same stepping principle but
®
they use different arrangements. In the PiezoWalk motor [35], the clamping and
feeding actions are achieved by longitudinal and shear actuators contained in four
legs. The legs are working in pairs for the clamping and the feeding motion. In
®
the Piezo LEGS , the motor also has four legs, but each of them is a piezoelectric
bimorph actuator [37]. The principle of operation is as follows: at the beginning of
the cycle, all legs are elongated and thus bend. Then, the first pair of legs moves
40 A. Shafik and R. Ben Mrad

towards the right while maintaining contact with the runner while the second pair
retracts and moves to left. The second pair then extends to contact the runner and
pushes it to the right while the first pair retracts and moves to the left. And the cycle
continues [36].
Inertial Motors Inertial motors are composed of a single piezo actuator, a moving
mass, and a drive shaft. This type is classified as a quasistatic motor [24], even
though its operating frequency can reach up to 20 kHz. The basic idea of the
inertial piezomotors is that when the piezo actuator expands and contracts at a
low acceleration, the moving mass and the shaft move together because of friction,
and when the piezo expands and contracts at a high acceleration, the inertia
of the moving mass makes it slide on the shaft. The low acceleration mode is
achieved by increasing the input voltage slowly while the high acceleration mode
is accomplished by decreasing the voltage rapidly. This can be done if the piezo
actuator is controlled with a modified sawtooth voltage waveform. Reversing the
motion is easily obtained if the waveform is mirrored. An example of an inertial-
type piezomotor is the one used in the anti-shaking mechanism in cameras made
® ®
by Konica Minolta . The PIShift inertia drive [35] is an example of a commercial
inertial piezomotor. The actuator expands slowly taking along the runner, but when
it contracts rapidly, the runner cannot follow due to its inertia, as shown in Fig. 2.5.
The main drawback of this motor is the high wear rate that is caused by the sliding
friction.
Typical quasistatic motors incorporate several piezoelectric actuators except the
inertial piezomotors. Quasistatic motors are better than USMs in terms of control,
because both the amplitude of the driving signal and the frequency can be controlled
to achieve a desired performance. These motors also offer higher resolution and

®
Fig. 2.5 Inertial piezomotors (PIShift operating principle)
2 Piezoelectric Motor Technology: A Review 41

Table 2.1 Commercial linear quasistatic piezomotors and some of their key specifications
Speed Accuracy Range
Product Manufacturer Concept Force (N) (mm/s) (nm) (mm)
®
Inchworm EXFO Burleigh Stepping 10 1:5 4 50
®
PiezoWalk Physik Instrumente Stepping 10 20 20 30
®
Piezo LEGS(r) PiezoMotor AB Stepping 450 0:3 <1 20
®
PIShift Physik Instrumente Inertial 10 10 <1 26

force than USMs. Increasing the number of actuators integrated into the motor can
make the motion smoother but this comes at an increased complexity and cost. A list
of commercial quasistatic motors is shown in Table 2.1.

2.4.2 Rotary Quasistatic Motors

Inchworm Motors The development on this type of motors has not been pursued
as much as that of linear inchworm motors. In the case of rotary motors, most of
the attention has been directed towards rotary USMs. The rotary inchworm motors
are well suited as rotary positioners in applications requiring high accuracy and
resolution.
The operating principle for linear and rotary inchworms is the same. The rotary
type has typically two sets of clamps and a mechanism to rotate the motor by
stepping action. One clamp is fixed to the stator (the stationary part of the motor) and
the other clamp is attached to the rotor (the rotating mechanism). Rotary inchworm
piezomotors can be grouped into three types; disc motors—clamp on face of the disc
[38, 39], drum motors—clamp on inside diameter of drum [40], and shaft motors—
clamp on outside diameter of shaft [41, 42].
Various approaches have been used for the clamping and rotation mechanisms.
The rotation mechanism can be achieved by using a torsional piezo actuator [39]
or by using a linear actuator and a flexure to convert the output linear motion into
rotational motion [38, 40, 41]. The problem with using a torsional actuator is the
likelihood of failure of the piezo material since it is subject to a direct shear load.
The use of flexures protects the piezo and provides a constant preload that keeps it
in compression, but this comes at the expense of an increased size and complexity
of the motor [43–45]. Regarding the clamping mechanisms, the clamps used in the
disc and drum motors are flexures driven by linear actuators [38, 40, 41]. In shaft
motors, the clamps are in the form of tubular piezo elements that directly clamp the
shaft. A drawback of this latter type of clamps is the high wear and the likelihood
of failure due to shear forces [42]. Thus, the use of flexures with linear actuators is
preferred for clamping and rotation mechanisms [25].
The selection of the inchworm type to use depends on the application require-
ments. By comparing the three types of rotary inchworms listed above leads to the
42 A. Shafik and R. Ben Mrad

following: The shaft motors have a better response time than the disc and drum
motors because of the lower inertia and their relatively small size. The shaft motors
though provide a smaller torque than the other types. This is so because the torque is
directly proportional to the clamping radius. The clamping radius being the distance
from the axis of rotation to the clamp point. They also have a large step size since
the step size is inversely proportional to the clamping radius. This implies that shaft
motors can reach higher speeds but at the expense of resolution. Therefore, for
applications requiring high torque and resolution, the disc or drum motors are the
better choices. If speed is the main criterion, then the shaft motor is a better option.
Stepping Motors Rotary stepping motors work based on the same principle as the
linear ones. PiezoMotor AB offers a rotary Piezo LEGS motor. The smallest of these
motors has a 17 mm diameter, a stall torque of 30 mNm, an angular speed of up to
28 rev/min, and an angular resolution of 0.1 rad.

2.5 Ultrasonic Motors

USMs are piezomotors that operate based on an input voltage in the ultrasonic
frequency range (usually above 20 kHz) and with the input signal frequency
corresponding to one or more of the resonant frequencies of the motor. USMs come
in different sizes (i.e., micro versus macro) and can be classified into contact and
non-contact motors. In this section, all types are discussed for both linear and rotary
motion.

2.5.1 Working Principle

The theory of operation of piezoelectric USMs is based on the converse piezoelectric

effect. This concept is applied to convert the input voltage to the piezoelectric
ceramic stator into resonant vibrations at a high frequency in the range of 20 kHz–
10 MHz, leading to a linear or rotary motion transmitted to the rotor often through
friction between the rotor and stator [24]. Thus, the working principle of USMs
consists of a two-step process:
1. Electrical energy is converted into mechanical vibrations in the stator, which
includes the piezoelectric elements, leading to mechanical vibrations at ultra-
sonic frequencies.
2. At the stator–rotor interface, the high frequency vibrations of the stator are
transformed into macroscopic rotor motion (translation or rotation) by means
of frictional forces between stator and rotor.
A functional block diagram for rotary USM is shown in Fig. 2.6.
2 Piezoelectric Motor Technology: A Review 43

Fig. 2.6 Functional block diagram for USMs

A key aspect in producing the motion of the motor from the vibrating stator is
how to generate a stator tip motion and subsequently how to transmit it to the rotor.
The friction coupling that often incorporates an initial preload force between stator
and rotor ensures that the stator and rotor are constantly in contact and that the
output is repeatable every stator cycle. The preload is a normal force applied to the
rotor towards the stator. Because of the simplicity of USMs, a widely used motor
control method based is pulse width modulation (PWM) [15].

2.5.2 Linear Ultrasonic Motors

A number of designs have been presented by various research groups and industry.
The most prominent of which are listed below.
Elliptical Push Motor The elliptical push motor is a single actuator USM that
uses geometric coupling of two eigenmodes; the longitudinal vibration mode and
the bending mode (1L2B), to produce elliptical motion at the drive tip of the
piezoelectric actuator. This arrangement has been used in the Paderborn rowing
motor [46] and commercialized by Nanomotion [47] in their HR series. This motor
consists of a rectangular piezoelectric plate and four electrodes. It works by exciting
the motor at the resonant frequencies of the piezo plate. The advantage of this
design is that it eliminates the need for frequency coupling of the two modes
and it uses only one amplifier. However, its main problem is that it is difficult to
increase the mechanical power developed by the actuator by increasing the size
of the piezoelectric plate because the power is directly proportional to the square
of the frequency, and the resonant frequencies decrease as the size of the piezo
increases [48].
Traveling Wave Motor Traveling wave motors, also known as propagating wave
motors, work by exciting piezoceramics that are attached to the stator to generate
two resonant standing waves with a phase shift of 90ı . This results in an elliptical
motion at the surface of the stator, which causes the rotor to move with a speed
determined by the horizontal component of the speed [49]. Traveling wave motors
are usually configured for rotary motor applications (to be discussed in Sect. 2.5.3)
as a circular path is needed for the wave to travel. Examples for linear traveling wave
motors are also available such as the ring and vibrator motors that were developed
44 A. Shafik and R. Ben Mrad

by Sashida [49] in 1982. The main drawback of this motor is the low efficiency
(theoretically less than 50 % due to the requirement of two drive signals to generate
the traveling wave).
-Shaped Motor Uchino and Ohnishi [33] presented the -shaped motor. The
stator has fork-shaped metallic legs and is driven by either one or two multilayer
piezoelectric actuators. A slight difference in the bending mode resonant frequency
of the two legs creates a phase shift between them and allows the motor to move the
rail with a walking motion like a trotting horse. The motor design has two actuators
at two corners of the -shaped frame and is driven by a propagating wave. This is
achieved by driving the two piezo actuators with a 90ı phase difference.
E31 Linear Motor An elliptical motion at the drive tip is used to drive a load and
can be achieved by different concepts and arrangements. Elliptical push motors
(discussed previously) use the 1L2B vibration mode, which is accomplished by
exciting simultaneously two different eigenmodes (longitudinal and bending) on a
piezoelectric plate. Physik Instrumente utilizes the E(3,1), or simply E31 , vibration
mode to a single actuator to create an elliptical motion at the drive tip. The E31
vibration mode is a two-dimensional standing extensional wave mode in which
the x-component is at the third vibration mode and the y-component is at the
first vibration mode. To achieve this vibration mode, the dimensions of the piezo
plate should be such that the X:Z (length-to-height) ratio is around 2:1. The motor
consists only of a piezo plate with two electrodes at the front side and one common
electrode at the back. The motor works in one direction by using one of the front
electrodes to apply a potential across the piezo and changes its direction by using
only the other electrode to apply the potential. Some of its advantages include
simplicity, the highest efficiency amongst linear piezomotors, and scalability [48].
®
The current specifications of the PILine ultrasonic piezomotors [35] are: velocity
up to 500 mm/s, holding force up to 15 N, and resolution of 2 nm.
Nanomotor The Nanomotor [50] is a concept developed by Klocke Nanotechnik
[51] which uses one piezo tube. For fine positioning, the extension is carefully
controlled by the piezo tube, but for coarse positioning the piezo is excited with
an ultrasonic pulse wave. The main specifications include a speed up to 5 mm/s,
driving force of 0.25 N, 10 mm positioning stroke, 20 nm/V fine positioning stroke,
and atomic resolution.

2.5.3 Rotary Ultrasonic Motors

Rotary USMs can be classified according to the type of the wave that is used to
excite the piezoelectric material into two types, standing wave type [52, 53] and
traveling (or propagating) wave type [54–56]. Traveling wave motors are the most
common type [57]. The traveling waves are generated by combining two standing
2 Piezoelectric Motor Technology: A Review 45

waves at the same frequency and amplitude but with a phase difference of 90ı . This
leads to the formation of an elliptical motion at the interface of the stator and the
rotor. A comparison of the two types of USMs is shown in Table 2.2.
Another classification of USMs can be done according to their assembly method,
or the operating mode of the piezoceramic material, commonly using PZT (lead
zirconate titanate) as in [58]. Motors can be classified into bonded type [59] and
bolt-clamped type [52, 56]. The latter type gives higher efficiency and output power
than the former one. This is due to the adoption of the d33 mode of PZT in bolt-
clamped type motors, which have higher transfer efficiency than the d31 mode
typically used in bonded type motors [56, 60]. In the bonded type, the stator is
constructed by gluing PZT ceramic to elastic metal blocks with conductive adhesive.
This construction leads to low torque and low efficiency [61] due largely to the
lower electromechanical coupling factor of the d31 mode of PZT, the PZT and the
adhesive layer fatigue under the intense vibration, and the difference in the vibration
amplitudes of the axial bending mode in axial and circumferential directions. Bolt-
clamped [52] or sandwich-type transducers [57, 61] are thus preferred if the above
listed issues are of concern. However, bolt-clamped motors have issues related to the
high energy consumption. In a number of designs, Langevin-type transducers [54,
59] have been adopted because of their excellent energy efficiency, stability, large
output torque, high reliability, and good environment adaptability [58]. Langevin
transducers [62, 63] are piezo-based transducers that convert electrical energy to
ultrasonic mechanical vibrations with amplified amplitude. The use of Langevin
transducers in USMs resulted in better characteristics and opened new application
areas including aeronautics and astronautics [59]. A review of different macro-
USMs is shown in Table 2.3.
A number of rotary designs have been proposed. They include the following.
Wedge and Traveling Wave The wedge and the traveling wave types are two of the
most prominent designs of rotary USMs designed by Sashida [4]. In the wedge type
motor, a Langevin vibrator is used to produce an amplified ultrasonic wave, which
is used to create elliptical motion at the tip of one or more wedges. The wedge is
pushed against a disc to convert the tip elliptical motion to a continuous motion.
This motor could achieve speeds up to 3000 rpm but suffers from rapid wear. The
traveling wave motor does not have the same wear issue. The operation is the same
as for the linear traveling wave motor which was described in Sect. 2.5.2. This motor
is commercially being used in the auto focus system of the Canon cameras. It is more
compact than traditional motors, has a fast response time, and silent operation.
Twist Connector The twist connector motor uses two vibration modes, longitu-
dinal and torsional. It is called also the mode conversion type motor [23] because
the stator converts the longitudinal vibration that is produced by the piezoelectric
disks to a torsional vibration by an elastic torsion coupler. While this motor is more
complex, it has the advantage of higher power density.
Elliptical Push This type of motor uses the same principle as the linear elliptical
push motors except that the slider is circular instead of being linear [34]. Anorad
 46

Table 2.2 Comparison between standing and traveling wave piezo ultrasonic motors
Standing wave Traveling wave
Advantages
– An efficiency of up to 98 % theoretically [15] and 87 % achieved – Less abrasion on the contact surfaces because the traveling wave drives
practically [10]. the rotor continuously [7].
– Lower cost because of using one vibration source only. – The high frequency operation allows for a very high accuracy to be
– Easier to support the stator at the nodal points or lines. This causes obtained from the rotor or slider (sub-nanometer positioning accuracy)
minimum effects on the resonance vibration [15]. [23].
– Controllable in both rotational directions [15] because the motor is a
dual-signal drive, that means it consists of two sets of transducers that are
excited independently.
Disadvantages
– The standing wave drives the rotor discontinuously, which causes wear – Used in applications where the motor is activated for short time durations
on the contact surfaces [7]. This may lead to a short lifespan due to the as heat dissipation is challenging [23].
repeated stator–rotor collision [10]. – Miniaturization to sub-millimeter scales is difficult due to the difficulty to
– It provides intermittent rotational torque or thrust. However, because of fabricate the complex geometry of the stator, dual-signal drive, and
the inertia of the rotor, the rotation speed ripple is not large [15]. alternating poling pattern required by the transducer [10].
– Lack of control in both directions in many designs [15, 52]. However, – The amplitude of the traveling wave has a detrimental effect on the
some designs already solved this problem, as in [53]. scalability of the design [8], because the amplitude of the wave needed
reduces linearly with the size of the device [8].
– More expensive because two vibration sources are required to generate
one wave, and this leads to lower efficiency (not more than 50%) [15].
A. Shafik and R. Ben Mrad
Table 2.3 Macro-USMs construction and published characteristics (when information is not available, the box is left blank)
Construction
Stator Vibration Type of Input Frequency Speed
Ref geometry Assembly Transducer mode excitation Material voltage (kHz) (rpm) Torque
[52] Cylindrical Bolt-clamped Sandwich (1) Bending Standing PZT-41 200 Vrms 28.1 165 0.45 Nm
wave
[53] Square-tubular Bolt-clamped PZT square Standing 50 V 46 1000 370 Nm
tube wave
[54] Ring Bolt-clamped Langevin (4) Bending Traveling PZT-8H 250 V0p 39 120 1.6 Nm
wave
[55] Cylindrical Bolt-clamped Sandwich (4) Longitudinal Traveling PZT-41 200 Vrms 35.35 290 1.95 Nm
wave
2 Piezoelectric Motor Technology: A Review

[56] Cylindrical Bolt-clamped Sandwich (1) Longitudinal & Traveling PZT-41 200 Vrms 31.14 281 1.2 Nm
composite Bending wave
[57] Square driving Bolt-clamped Sandwich— Bending Traveling PZT-41 200 Vrms 25.1 298 9.3 Nm
feet Driving wave
feet (4)
[58] Square driving Bolt-clamped Sandwich— Longitudinal PZT-41 200 Vrms 23.1 71 12.3 Nm
feet Driving
feet (4)
[59] Ring Bolt-clamped Langevin (4) Bending—in- PZT-8H 200 V0p 48.6 100 0.3 Nm
plane
[60] Square driving Bolt-clamped Sandwich— Bending PZT-41 200 Vrms 20 58 9.5 Nm
feet Driving
feet (4)
(continued)
47
 48

Table 2.3 (continued)

Construction
Stator Vibration Type of Input Frequency Speed
Ref geometry Assembly Transducer mode excitation Material voltage (kHz) (rpm) Torque
[61] Ring Bolt-clamped Nested Bending Traveling PZT-41 150 V 19.2 126 0.8 Nm
stacks (radial) wave
[74] Tubular Bolt-clamped PZT tube Bending d31 D 15e  12 50 Vpp 27.6 400 300 Nm
C/N
[75] Bars V-Type Longitudinal TOKIN—NEPEC 130 Vrms 43 700 0.75 Ncm
6
[76] Ring Bonded Traveling PZT-Pb(Zr/Ti)O3 100 V0p Needle Bearing 60 4.2  104 N.m
wave (C-213 model)
[77] Cross-shaped Bonded Sandwich (8) Bending 24 V 80.7 90 180  104 N.m
(plates)
[78] Disk PZT on SOI Traveling Thin film PZT 10 V 284 2300
wave Zr/Ti ratio of 52/48
(thickness
1000 nm)
[79] Ring Bolt-clamped Langevin (4) 5th in-plane Traveling PZT-8H 200 V0p 48.65 80 0.35 Nm
flexural wave
[70] Ring Bonded Two-sided vibration Traveling PZT-8 150 Vpp 21.95 76 0.5 Nm
PZT modal (6,0) wave
A. Shafik and R. Ben Mrad
2 Piezoelectric Motor Technology: A Review 49

Table 2.4 Commercial rotary ultrasonic piezomotors and their specifications

Torque Speed
Product Manufacturer Concept (mNm) (rpm) Accuracy Weight (g)
®
PILine Physik Instrumente Standing 50 180 35 rad 300
USR30 Shinsei Motors Traveling 50 300 – 17

Fig. 2.7 Driving mechanism

of non-contact USMs

commercialized a motor based on this concept and claims an accuracy of 3 arcsec,

a significant improvement over conventional technologies.
Specifications of two commercial rotary USM produced by two leading compa-
nies in the field, Physik Instrumente and Shinsei Motors, are shown in Table 2.4.
Miniaturizing USMs to milli/micro scale is an ongoing field of research [8, 10,
64, 65]. There are two development directions in the field of micro USMs: 1. Using
small bulk piezo elements [64], or 2. Using piezoelectric thin film technology [43].
In [64, 65], bulk Lead-Zirconate-Titinate (PZT) [64] was compared to a stator
transducer fabricated using PZT thin film [65]. The results showed that the bulk
PZT transducer gives more than thirty times larger output torque. The piezoelectric
factor e31 was used to compare both designs as they used a tube transducer. The
output torque is directly proportional to the e31 value. The e31 piezoelectric constant
had a higher value in PZT bulk transducers (4.1 C/m2 ) as compared to that of PZT
thin film (0.19 C/m2 ). The bulk PZT based ultrasonic micro motor provides a max
speed of 650 rpm and a maximum torque of 0.22 mNm at a 100 Vp-p input voltage
and a 85 kHz driving frequency. The maximum torque produced by the PZT thin
film based motor [65] is only 7 Nm.
Non-contact USMs form another research area in the USMs field. Contact USMs
use the friction between the stator and rotor to transfer the motion from one to the
other which leads to heat generation, lower efficiencies, and shorter life span. When
high rotational speeds are needed, the use of solid–fluid–solid driving mechanisms
instead of the solid–solid interface used in contact type motors was investigated [7].
When using the solid–fluid–solid mechanism, the torque is transferred to the rotor
through a fluid contained in the gap between the stator and rotor leading to the
rotor rotation in the direction of the traveling wave as shown in Fig. 2.7. The main
features of the non-contact USMs are: there are no friction losses as in contact type
motors, longer life spans, higher velocities can be achieved, and smaller torques are
produced. These features make this type of USMs suitable for applications that need
higher velocity and smaller torque as shown in Fig. 2.8.
50 A. Shafik and R. Ben Mrad

Fig. 2.8 Contact versus non-contact USMs characteristics

Increasing the stator vibration velocity and utilizing the axial resonance of the
gap between the rotor and the stator can lead to an increase in the revolution speed
[66]. Other features can also lead to an enhanced performance such as adding an
acoustic reflector over the rotor can increase the maximum speed 1.4 times because
of the existence of two air gaps, the gap between the stator and rotor and the gap
between the rotor and the reflector. By using a double-disk rotor with a small air gap
(0.8 mm) between the rotors, in addition to a reflector, can increase the speed by a
1.6 factor [67]. By using this latter arrangement, a speed of 2000 rpm was achieved
with a very low driving voltage of 3 V [68]. The non-contact type of USMs have
not been commercialized yet. A summary of the characteristics of a number of non-
contact USMs is shown in Table 2.5.

2.6 Case Studies

In this section, we present two case studies as a part of research conducted at the
Mechatronics and Microsystems Design Laboratory at the University of Toronto.
The first case presents an Inchworm (called “Piezoworm”) stage design [32], as an
example of a quasistatic piezomotor. The second case presents a Linear E31 motor
with segmented electrodes [69], as an example of an ultrasonic piezomotor.

2.6.1 Piezoworm Stage

A compact clamp piezoworm stage is integrated into a two-axis positioning system

to be used for tracking profiles with nanometer accuracy. To avoid the backlash
Table 2.5 Summary of non-contact USMs designs and their characteristics (when information is not available, the box is left blank)
Traveling Driving
Ref Gap (mm) Speed (rpm) Torque (N.m) Medium wave mode Size (mm) frequency (kHz) Input voltage
[7] 3000 20  1.7
[7] 1200 30  0.5
[7] 6031 3.5  105 45  1.5 18.9
[67] 0.25 2800 Air 30 (dia.)  0.4 (th.) 25.9
2 Piezoelectric Motor Technology: A Review

[80] 0.05 3200 Air 2 Langevin

transducers
[81] 0.7 3000 B03 mode
[82] 4000 Two B03 modes 5 (dia.)  0.5 (th.) 20.5 30 Vp–p
[83] 2026 184  106 B03 mode 19.95
[84] 5300 3.5 mg.m
[85] 2100 1.3  105
[86] 3569 B22 mode 45.6 20 V
51
52 A. Shafik and R. Ben Mrad

Fig. 2.9 Piezoworm design (a) Stage prototype along one axis, (b) Piezoworm motor

Fig. 2.10 Clamp configurations (a) NC clamp, (b) NU clamp

problem and for high precision alignment, a direct connection is designed to connect
the stage to a commercial slide. Prototypes of the piezoworm stage and motor are
shown in Fig. 2.9.
The piezoworm stage consists of a piezoworm motor, a crossed-roller slide, and a
10 nm resolution encoder, all mounted on a small square base plate of 120  120 mm
dimensions [32]. The piezoworm actuator dimensions are 27  43  25 mm. It has
two clamps, the normally clamped (NC) clamp, which grips at zero voltage, and the
normally unclamped (NU) clamp, which grips when voltage is applied. The clamps
are connected to a middle section by extension flexures which work as return springs
to preload and protect the piezo stacks. The piezo stacks for both clamps are made
of PZT Navy type II which offers good free expansion and high stiffness. Friction
surfaces are covered by ceramic strips to reduce wear. The clamp configurations are
shown in Fig. 2.10. The flexures are almost the same for both clamps and the only
2 Piezoelectric Motor Technology: A Review 53

difference is the bottom hole, which is a tapped hole for a set screw in the NC clamp,
and a through hole in the NU clamp. This has been made for adjusting the set screw
to preload the piezo stacks. When piezo stacks are energized, they expand and push
the clamping surface towards the slide in the NU clamp, and away from the slide in
the NC clamp.
The piezoworm stage with the closed loop control showed its efficacy with
average positioning accuracy of ˙20 nm and an average error in tracking appli-
cations when connected to a two-axis stage of 8 nm, 1.72 m, and 1.85 m, in the
nanometer, micrometer, and millimeter regimes, respectively [32].

2.6.2 Linear E31 Motor with Segmented Electrodes

This motor was developed in order to solve the dead zone problem of ultrasonic
piezomotors and to design a linear USM to integrate into a meso-Milling machine
tool. The dead zone problem is that the output force and speed of USMs are coupled
and the only control parameter is the amplitude of the input voltage.
The motor has been designed based on the E31 vibration mode. E31 is a planar
standing wave mode and 3 and 1 represent the number of half-wave lengths
along the x-axis and the y-axis, respectively. Figure 2.11 shows that the maximum
displacement occurs at the drive tip location.
Stator dimensions were selected based on the aspect ratio of the E31 mode
requirements to be 60  30  9 mm. The material used is PZT (Navy Type II), and
supports are added as shown in Fig. 2.12 [69].
Electrodes were arranged to apply a new segmented electrode concept [69]. The
back surface of the motor has one common drain electrode. The front surface has
multiple electrodes with different dimensions (as shown in Fig. 2.13). Based on
the fact that when the active electrode length changes, the output speed and force
change accordingly, the motor was tested with different active electrode lengths
to decouple the output force and speed. Figure 2.13 shows simulated performance
curves showing speed versus electrode size and force versus electrode size. The
figure clearly shows a region where low speed and high force are feasible.

Fig. 2.11 Simulation of

vertical displacement of the
E31 vibration mode by
ANSYS FE Software
54 A. Shafik and R. Ben Mrad

Fig. 2.12 Model of the piezo

plate with drive-tip and
simplified support structure
and slider

Fig. 2.13 Simulated USM performance vs. active electrode length

The experimental performance characteristics of the motor with 200 Vp and

at a 70 N preload are a maximum speed of 0.223 m/s and a maximum thrust of
36 N. The smallest achievable step in open-loop was 9 nm with an 18 s impulse.
Moreover, the proposed design allowed the speed to vary while keeping the thrust
force relatively constant.
2 Piezoelectric Motor Technology: A Review 55

2.7 Future Developments

Increasing the output power, the efficiency, and miniaturization of USMs continue
to be the main areas of current research in this field. Different ideas have been
investigated for increasing the output torque. Li et al. [70] improved the torque and
torque density by sandwiching the stator with two piezoceramic layers bonded to
its upper and lower surfaces and by adding two disc rotors. The stator has teeth on
both sides. This design is capable of producing two times the torque generated by
conventional USMs (i.e., with only one piezoceramic) using the same piezoelectric
material and dimensions. Lu et al. [71] proposed using shear-type piezoceramics and
utilizing the d15 piezoelectric mode for vibration excitation of the stator instead of
the d31 and d33 modes because of its higher values (as shown in Table 2.6). However,
the fabrication process of piezoceramics operating in the d15 mode is more complex
because the electrode surfaces that are used for polarization are different from those
used for excitation.
Protor
D
Pstator

To improve the characteristics of the motor and its efficiency, a number of key
issues need to be considered: selection of vibration modes and the consistency of
the modal frequencies being used to generate the motion, the placement of the
piezoelectric ceramic and the supporting plane, setting of the preload, and any
influence from interfering modes [72]. One of the main issues associated with
efficiency degradation is heat generation due to friction between the stator and rotor.
Adding a preload, for instance, increases the efficiency of conversion from the stator
to the rotor but it also increases the heat produced. A solution to this problem
was proposed in [73] by selecting a proper lubricant. The efficiency becomes
then dependent on the static preload and the lubricant type used. The maximum
torque and efficiency show a linear relationship with the static preload [73]. Proper
selection of the friction material can also improve the efficiency and the output
power [15]. For instance, selection of PPS (Ryton) as a friction material can increase
the power output and efficiency of the Shinsei USM motor (USR 30) to 7 W and
23 %, instead of 1 W and less than 5 %, respectively, obtained when PAMXD6
material is used instead. These developments clearly encourage other researchers to
continue to investigate new friction materials for better output characteristics.

Table 2.6 Piezoelectric Categories PZT-4 PZT-5A PZT-5H PZT-8

constant values for different
Name 110 200-HD 610-HD 300-HD
materials [71]
d33 (pC/N) 360 400 750 235
d31 (pC/N) 150 190 380 100
d15 (pC/N) 520 550 920 390
The efficiency of USMs is defined as the ratio of the
power transferred to the rotor to the stator power output
[76]
56 A. Shafik and R. Ben Mrad

Solutions to the temperature rise and heat transfer problems [14] are presented
in [15] and [23]. The stator was mounted on a radiator to solve issues caused by
temperature rise. In addition, decreasing the working temperature could be achieved
by utilization of higher quality friction materials [23]. Another solution suggested by
[15] is to use anti-resonant frequency instead of fundamental resonance frequency,
which gives a higher mechanical quality factor and the same mechanical vibration
level without generating heat. By applying this solution, conventional inexpensive
power supply can be used, because of the need for high voltage and low current
drivers, since the admittance is very low for anti-resonance state.
The future trend is still further miniaturization [23, 24, 71]. The main objective
of many development works is to produce a compact, inexpensive USM, with
higher power values and acceptable efficiency. Another challenge is to address the
stator/motor contact problem and associated wear issues [14] and to enhance the
understanding of the dynamics between the stator and rotor.

References

1. Y. Okazaki, Precision positioning control apparatus and precision positioning control method.
U.S. Patent 5,801,939, 1998
2. W. Yao, M. Tomizuka, Robust controller design for a dual-stage positioning system, in Pro-
ceedings of International Conference on Industrial Electronics, Control, and Instrumentation,
vol. 1, 1993, pp. 62–66
3. K. Tsai, J. Yen, Servo system design of a high-resolution piezo-driven fine stage for step-
and repeat microlithography systems, in Proceedings of Annual Conference of Industrial
Electronics Society, vol. 1, 1999, pp. 11–16
4. S. Kwon, W. Chang, Y. Youm, Robust and time-optimal control strategy for coarse/fine
dual-stage manipulators, in Proceedings of IEEE International Conference on Robotics and
Automation, vol. 4, 2000, pp. 4051–4056
5. S. Kwon, W. Chang, Y. Youm, On the coarse/fine dual-stage manipulators with robust
perturbation compensator, in Proceedings of IEEE International Conference on Robotics and
Automation, vol. 1, 2001, pp. 121–126
6. B. Zhang, Z. Zhu, Developing a linear piezomotor with nanometer resolution and high stiffness.
IEEE/ASME Trans. Mechatron. 2(1), 22–29
7. C. Zhao, Ultrasonic Motors: Technologies and Applications (Springer, Berlin, 2011)
8. B. Watson, J. Friend, L. Yeo, Piezoelectric ultrasonic micro/milli-scale actuators. Sens.
Actuators A Phys. 152(2), 219–233 (2009)
9. F. Ba-Tis, R. Ben-Mrad, A 3-DOF MEMS electrostatic piston-tube actuator. J. Microelec-
tromech. Syst. 24(4), 1173–1184 (2015)
10. D.K.C. Liu, J. Friend, L. Leo, A brief review of actuation at the micro-scale using electrostatics,
electromagnetics and piezoelectric ultrasonics. Acoust. Sci. Technol. 31(2), 115–123 (2010)
11. R. Ben-Mrad, H. Hu, Dynamic modeling of hysteresis in piezoceramics, in Proceedings of
IEEE/ASME International Conference on Advanced Intelligent Mechatronics, vol. 1, 2001,
pp. 510–515
12. A. Henke, M.A. Kümmel, J. Wallaschek, A piezoelectrically driven wire feeding system for
high performance wedge-wedge-bonding machines. Mechatronics 9(7), 757–767 (1999)
13. J. Wallaschek, Ultrasonic motor research in Germany—past, present, future, in Proceedings of
the First International Workshop on Ultrasonic Motors and Actuators (2005)
14. J. Wallaschek, Piezoelectric ultrasonic motors. J. Intell. Mater. Syst. Struct. 6(1), 71–83 (1995)
2 Piezoelectric Motor Technology: A Review 57

15. K. Uchino, Piezoelectric ultrasonic motors: overview. Smart Mater. Struct. 7(3), 273 (1998)
16. V.V. Lavrinenko, M. Nekrasov, Piezoelectric motor. Soviet Patent 217509 (1965)
17. H.V. Barth, Ultrasonic drive motor. IBM Technical Disclosure Bulletin 16(7), 2263 (1973)
18. P.E. Vasiliev, V.S. Dvornin, A.V. Kondratiev, V.F. Kravchenko, U.S. Patent 4,240,141 (U.S.
Patent and Trademark Office, Washington, DC, 1980)
19. T. Sashida, Trial construction and operation of an ultrasonic vibration driven motor. Oyo
Butsiuri 6(5), 713–718 (1982)
20. T. Sashida, Motor device utilizing ultrasonic oscillation. U.S. Patent 4,562,374 (U.S. Patent
and Trademark Office, Washington, DC, 1985)
21. A. Kumada, A piezoelectric ultrasonic motor. Jpn. J. Appl. Phys. 24(S2), 739 (1985)
22. Y. Ise, Ultrasonic motor. J. Acoust. Soc. Jpn. 54, 6 (1987)
23. I. Prisacariu, C. C. Filipiuc, A general view on the classification and operating principle of
piezoelectric ultrasonic motors. (2012 International Conference and Exposition on Electrical
and Power Engineering (EPE 2012), 25–27 October, Iasi, Romania)
24. K. Spanner, Survey of the various operating principles of ultrasonic piezomotors, in Proceed-
ings of the 10th International Conference on New Actuators, June 2006
25. S.P. Salisbury, D.F. Waechter, R. Ben-Mrad, S.E. Prasad, R.G. Blacow, B. Yan, Closed-loop
control of a complementary clamp piezoworm actuator. IEEE/ASME Trans. Mechatron. 12(6),
590–598 (2007)
26. S. Salisbury, D.F. Waechter, R. Ben-Mrad, S.E. Prasad, R.G. Blacow, B. Yan, Complementary
inchworm actuator for high-force, high-precision applications. IEEE/ASME Trans. Mechatron.
11(3), 265–272 (2006)
27. P.E. Tenzer, R. Ben-Mrad, A systematic procedure for the design of piezoelectric inchworm
precision positioners. IEEE/ASME Trans. Mechatron. 9(2), 427–435 (2004)
28. D. Roberts, Development of a linear piezoelectric motor based on the inchworm model, in 1999
Symposium on Smart Structures and Materials (International Society for Optics and Photonics,
1999), pp. 705–716.
29. J. Li, R. Sedaghati, J. Dargahi, D. Waechter, Design and development of a new piezoelectric
linear Inchworm actuator. Mechatronics 15(6), 651–681 (2005)
30. J. E. Frank, G. H. Koopmann, W. Chen, G. A. Lesieutre, Design and performance of a high-
force piezoelectric inchworm motor, in 1999 Symposium on Smart Structures and Materials
(International Society for Optics and Photonics, 1999), pp. 717–723
31. T. Pandell, E. Garcia, Design of a piezoelectric caterpillar motor, in Proceedings of ASME
Aerospace Division, vol. 52, 1996, pp. 627–648
32. S.P. Salisbury, R. Ben-Mrad, D.F. Waechter, S.E. Prasad, Design, modeling, and closed-loop
control of a complementary clamp piezoworm stage. IEEE/ASME Trans. Mechatron. 14(6),
724–732 (2009)
33. K. Uchino, Piezoelectric Actuators and Ultrasonic Motors (Kluwer Academic Publishers,
Boston, 1997)
34. “L-104,” Micro Pulse Systems Inc., [Online document], http://www.micropulsesystems.com,
seen on 27 May 2002
35. http://www.physikinstrumente.com/technology/, seen on 5 March 2015
36. http://piezomotor.ports-it.net/technology/, seen on 5 March 2015
37. http://evolution.skf.com/, “Micromotor packs a punch”, online article, seen on February 2015
38. J. Oliver, R. Neurogaonkar, J. Nelson, C. Bertolini, Rotary piezoelectric motor for vehicle
applications. U.S. Patent 5,780,956 (1998)
39. K. Ohnishi, M. Umeda, M. Kurosawa, S. Ueha, Rotary Inchworm-type piezoelectric actuator.
Electr. Eng. Jpn. 110(3), 107–109 (1990)
40. K. Duong, E. Garcia,. Development of a rotary inchworm piezoelectric motor, in Proceedings
SPIE Smart Structures and Materials, vol. 2445, 1995, pp. 782–788
41. S. Gursan, Development of a continuous-motion piezoelectric rotary actuator for mechatronics
and micropositioning applications. MASc Dissertation, University of Victoria (1996)
42. K. Mori, Piezoelectric rotary actuator. U.S. Patent 4,468,583 (1984)
58 A. Shafik and R. Ben Mrad

43. P.E. Tenzer, R. Ben-Mrad, On amplification in inchworm™ precision positioners. Mechatron-

ics 14(5), 515–531 (2004)
44. P. Tenzer, R. Ben-Mrad, Amplification in Inchworm precision positioners, in 1st joint Canada-
US Workshop on Smart Materials and Structures (St. Hubert, Quebec, Canada, 2001),
pp. 77–84, 17–18 September 2001
45. S.P. Salisbury, R. Ben-Mrad, Analytical stiffness estimation for short flexures. Mechatronics
16(7), 399–403 (2006)
46. T. Hemsel, J. Wallaschek, A piezoelectric linear vibration drive for high driving force. J
Vibroengineering 1, 7–12 (1999)
47. J. Zumeris, Ceramic motor. U.S. Patent 6,064,140 (2000)
48. O. Vyshnevsky, S. Kovalev, W. Wischnewskiy, A novel, single-mode piezoceramic plate
actuator for ultrasonic linear motors. Ultrason. Ferroelectr. Freq. Control, IEEE Trans. 52(11),
2047–2053 (2005)
49. T. Sashida, T. Kenjo, An Introduction to Ultrasonic Motors (Oxford Press, New York, 1993)
50. http://www.nanomotor.de/p_nanomotor.htm, seen on April 1, 2015
51. V. Klocke, Atomic precision and millimeter range, in Feinwerktechnik, Mikrotechnik,
Mikroelektronik, vol. 104 (1996)
52. Y. Liu, W. Chen, J. Liu, S. Shi, A cylindrical standing wave ultrasonic motor using bending
vibration transducer. Ultrasonics 51(5), 527–531 (2011)
53. S. Park, S. He, Standing wave brass-PZT square tubular ultrasonic motor. Ultrasonics 52(7),
880–889 (2012)
54. X. Lu, J. Hu, L. Yang, C. Zhao, A novel dual stator-ring rotary ultrasonic motor. Sens. Actuators
A Phys. 189, 504–511 (2013)
55. Y. Liu, J. Liu, W. Chen, S. Shi, A cylindrical traveling wave ultrasonic motor using longitudinal
vibration transducers. Ferroelectrics 409(1), 117–127 (2010)
56. Y. Liu, W. Chen, J. Liu, S. Shi, A cylindrical traveling wave ultrasonic motor using longitudinal
and bending composite transducer. Sens. Actuators A Phys. 161(1), 158–163 (2010)
57. Y. Liu, W. Chen, P. Feng, J. Liu, A rotary piezoelectric motor using bending vibrators. Sens.
Actuators A Phys. 196, 48–54 (2013)
58. Y. Liu, W. Chen, P. Feng, J. Liu, A square-type rotary ultrasonic motor with four driving feet.
Sens. Actuators A Phys. 180, 113–119 (2012)
59. X. Lu, J. Hu, L. Yang, C. Zhao, A novel in-plane mode rotary ultrasonic motor. Chin. J.
Aeronaut. 27(2), 420–424 (2014)
60. Y. Liu, W. Chen, J. Liu, S. Shi, A rotary ultrasonic motor using bending vibration transducers.
IEEE Trans. Ultrason. Ferroelectr. Freq. Control 57(10), 2360–2364 (2010)
61. Y.X. Liu, J.K. Liu, W.S. Chen, X.H. Yang, A rotary ultrasonic motor using radial bending mode
of ring with nested PZT excitation. J. Zhejiang Univ. Sci. A 13(3), 189–196 (2012)
62. Y. Ting, J.M. Yang, C.C. Li, C.C. Yang, Y.C. Shao, P3P-6 modeling and design of a linear
actuator by Langevin vibrators, in Ultrasonics Symposium, 2006. IEEE (2006), pp. 2337–2340
63. E. Moreno, P. Acevedo, M. Fuentes, M. Sotomayor, M. Borroto, M.E. Villafuerte, L. Leija,
Design and construction of a bolt-clamped Langevin transducer, in International Conference
on Electrical and Electronics Engineering, Proceedings (2005), pp. 393–395
64. T. Morita, M.K. Kurosawa, T. Higuchi, Cylindrical micro ultrasonic motor utilizing bulk lead
zirconate titanate (PZT). Jpn. J. Appl. Phys. 38(5S), 3347 (1999)
65. T. Morita, M. Kuribayashi Kurosawa, T. Higuchi, A cylindrical micro ultrasonic motor using
PZT thin film deposited by single process hydrothermal method (/spl phi/2.4 mm, LD 10 mm
stator transducer). IEEE Trans. Ultrason. Ferroelectr. Freq. Control 45(5), 1178–1187 (1998)
66. J. Hu, K. Nakamura, S. Ueha, An analysis of a noncontact ultrasonic motor with an
ultrasonically levitated rotor. Ultrasonics 35(6), 459–467 (1997)
67. Y. Yamayoshi, S. Hirose, Improvement of Characteristics of Noncontact Ultrasonic Motor
Using Acoustically Coupled Two Air Gaps. Jpn. J. Appl. Phys. 50(7) (2011)
68. Y. Yamayoshi, J. Shiina, H. Tamura, S. Hirose, Noncontact ultrasonic motor with two flexural
standing wave vibration disks. Jpn. J. Appl. Phys. 48(9S1), 09KD10 (2009)
2 Piezoelectric Motor Technology: A Review 59

69. J. Lau, S.I. Gubarenko, R. Ben-Mrad, A novel plate-type linear ultrasonic motor with
segmented electrodes, in Proceedings of the 1st VMPT (Montreal, QC, 2012)
70. X. Li, W.S. Chen, T. Xie, J.K. Liu, Novel high torque bearingless two-sided rotary ultrasonic
motor. J. Zhejiang Univ. Sci. A 8(5), 786–792 (2007)
71. C.Y. Lu, J.L. Li, W.Y. Pi, Ultrasonic motors using shear-type piezoelectric ceramics, in 2010
Symposium on Piezoelectricity, Acoustic Waves and Device Applications (SPAWDA), IEEE
(2010, December), pp. 465–469
72. Z. Li, C. Zhao, W. Huang, Z.L. Li, Several key issues in developing of cylinder type 3-DOF
ultrasonic motor. Sens. Actuators A Phys. 136(2), 704–709 (2007)
73. W. Qiu, Y. Mizuno, D. Koyama, K. Nakamura, Analysis of lubricating effect of hybrid
transducer-type ultrasonic motor, in Proceedings of 32nd Symposium on Ultrasonic Electronics
32(2E4-3) (2011, November), pp. 301–302
74. S. He, P.R. Chiarot, S. Park, A single vibration mode tubular piezoelectric ultrasonic motor.
IEEE Trans. Ultrason. Ferroelectr. Freq. Control 58(5), 1049–1061 (2011)
75. S.S. Jeong, T.G. Park, M.H. Kim, T.K. Song, Characteristics of a V-type ultrasonic rotary
motor. Curr. Appl. Phys. 11(3), S364–S367 (2011)
76. Y. Hojjat, M.R. Karafi, Introduction of roller interface ultrasonic motor (RIUSM). Sens.
Actuators A Phys. 163(1), 304–310 (2010)
77. T. Park, S. Jeong, H. Chong, K. Uchino, Design of thin cross type ultrasonic motor. J.
Electroceramics 24(4), 288–293 (2010)
78. G.L. Smith, R.Q. Rudy, R.G. Polcawich, D.L. DeVoe, Integrated thin-film piezoelectric
traveling wave ultrasonic motors. Sens. Actuators A Phys. 188, 305–311 (2012)
79. X. Lu, J. Hu, L. Yang, C. Zhao, Principle and experimental verification of novel dual driving
face rotary ultrasonic motor. Chin. J. Mech. Eng. 26(5), 1006–1012 (2013)
80. J.H. Hu, K. Nakamura, S. Ueha, Characteristics of a noncontact ultrasonic motor using acoustic
levitation, in Proceedings of Ultrasonics Symposium, 1996. IEEE, vol. 1 (1996, November),
pp. 373–376
81. Y. Yamayoshi, S. Hirose, Ultrasonic motor not using mechanical friction force. Int. J. Appl.
Electromagn. Mater. 3, 179–182 (1992)
82. S. Hirose, Y. Yamayoshi, H. Ono, A small noncontact ultrasonic motor, in Proceedings of
Ultrasonics Symposium, 1993. IEEE (1993), pp. 453–456
83. J. Liu, B. Wu, Z. Yang et al., A new type of circular cylindrical non-contact ultrasonic motor.
Acta Acustica 3(2), 113–116 (2001) (In Chinese)
84. Y. Ji, C. Zhao, A new type non-contact ultrasonic motor with higher revolution speed.
Piezoelectrics & Acoustooptics 28, 527–533 (2006)
85. Y. Ji, C.S. Zhao, Cylinder type non-contact ultrasonic motor. J. Nanjing. Univ. Aeronaut.
Astronaut. 37(6), 690–693 (2005)
86. B. Yang, J. Liu, D. Chen, B. Cai, Theoretical and experimental research on a disk-type non-
contact ultrasonic motor. Ultrasonics 44(3), 238–243 (2006)
Chapter 3
Mechanical Design of High-Speed
Nanopositioning Systems

Yuen Kuan Yong and Kam K. Leang

Abstract The performance of a nanopositioning system is tightly coupled to the

quality of the mechanical design. Good mechanical design will minimize most
position errors and improve overall accuracy and performance. Poor mechanical
design, on the other hand, can lead to more errors than the issues associated
with the electronics, control system, and other components. In this chapter, an
overview of mechanical design is presented, where the emphasis on flexure-guided
nanopositioning stages for high-speed nanopositioning. The discussions will focus
on systems driven by piezoelectric actuators such as plate-stacks, which are readily
available from a number of commercial suppliers. Design examples of parallel- and
serial-kinematic scanners are presented to illustrate the design process.

3.1 Introduction

Nanopositioners are positioning devices that capable of providing motions with

nanometer resolution. Nanopositioning systems have become the key enabling
tools for precision motion applications in the fields of micro and nanotechnology
[20, 104]. For example, the piezoelectric tube scanner (one type of nanopositioner)
has helped to create one of the most versatile tools in nanotechnology—the atomic
force microscope (AFM) [10, 11]. Since its invention, the AFM has brought about
significant advances in many scientific and industrial fields including life sciences
[6, 90], semiconductor manufacturing [91, 92], nano-machining, and nanofabrica-
tion [31, 55, 68, 85]. The AFM has revolutionized the way in which biologists
analyze samples at resolution beyond the reach of optical microscopy [82].
Nanopositioners exist in many forms and a majority of them are driven by
piezoelectric actuators. Examples include the aforementioned piezoelectric tube

Y.K. Yong ()

School of Electrical Engineering and Computer Science, The University of Newcastle,
Callaghan, NSW 2308, Australia
e-mail: [email protected]
K.K. Leang
Department of Mechanical Engineering, The University of Utah, Salt Lake City, UT, USA
e-mail: [email protected]

© Springer International Publishing Switzerland 2016 61

C. Ru et al. (eds.), Nanopositioning Technologies,
DOI 10.1007/978-3-319-23853-1_3
62 Y.K. Yong and K.K. Leang

scanner [57], flexure-based nanopositioners [35, 37, 69, 100, 104], piezoelectric
benders [94], inchworm or stick-slip type nanopositioners [54, 62, 65], beam scanner
[73], and mirror tilting positioner [83].

3.2 Scanning Probe Microscopy

A common application of nanopositioners is in the lateral and vertical positioning

stages of scanning probe microscopes (SPMs) such as the AFM. Unlike a traditional
optical microscope that uses light for imaging, an AFM image is formed by scanning
a microcantilever probe over the surface, as illustrated in Fig. 3.1. The AFM is one
of the most versatile microscopes due to its ability to work with conducting and
nonconducting samples in a vacuum, air, or water [10]. The probe is a microma-
chined cantilever with a sharp tip protruding towards the sample surface. When the
probe is brought into contact with the surface, the tip-to-sample interaction causes
the cantilever to deflect vertically. This deflection is measured and used to construct
an image of the sample. The AFM essentially “feels” the surface with a tiny, finger-
like cantilever. In a vacuum, resolution of an AFM is on the order of 0.01 nm. With
such high resolution, an AFM can generate topographical images of atoms, as well
as to control, manipulate, and alter the properties of matter at the nanoscale.

Fig. 3.1 The operation of an atomic force microscope

3 Mechanical Design of High-Speed Nanopositioning Systems 63

Fig. 3.2 Two positioning schemes for SPMs: (a) scan-by-sample and (b) scan-by-probe

The positioning of the probe tip relative to the sample can be achieved with
two basic configurations: (a) scan-by-sample or (b) scan-by-probe as shown in
Fig. 3.2. In the scan-by-sample configuration, the nanopositioner, flexure-based
design shown equipped with three piezo stacks, moves the sample relative to a
fixed probe. The x and y axes piezos position the sample along the lateral direction
(parallel to the sample surface) and a z axis stack moves the sample vertically. The
deflection of the cantilever is measured optically, by reflecting a laser beam off
the end of the cantilever onto a nearby photodetector. Alternatively, in the scan-by-
probe arrangement shown in Fig. 3.2b, a nanopositioner moves the probe relative
to a fixed sample both laterally and vertically. In scan-by-probe systems, the laser
and photodetector are required to move with the cantilever; however, this can be
avoided by incorporating sensing elements into the cantilever itself, such as using
piezoresistive, piezoelectric, or capacitive elements.
There are three basic operating modes of an AFM: contact, noncontact, and
tapping mode. In contact mode, the probe interacts with the sample at very close
range where the dominant force on the tip is repulsive. In this mode, the deflection
of the AFM cantilever is sensed and a feedback controller is used to maintain a
desired deflection. The spring constant of a contact mode AFM cantilever varies
between 0.001 and 10 N/m.
Soft samples such as living cells have a contact stiffness comparable to, or
less than, the cantilever stiffness, therefore, they may be deformed or damaged
during contact mode operation. Noncontact mode avoids direct sample contact by
exploiting attractive Van der Waals forces. In this mode, the AFM tip is hovered
above the surface (at approximately 50–150 Å). By oscillating the tip, the effective
stiffness of the microcantilever is effected by the force gradient of the attractive
forces. The effective stiffness can be related to the sample topography by measuring
or regulating the amplitude, phase, or resonance frequency of the probe. In general,
noncontact mode AFM provides lower resolution than contact mode but does not
pollute or damage the sample. Noncontact mode can also be used to measure long-
range forces such as magnetic or electric fields in samples such as hard disk media
or charged insulators.
64 Y.K. Yong and K.K. Leang

For high-resolution imaging of soft samples such as living cells, polymers, and
gels, tapping mode AFM is the preferred method. In this mode, the AFM cantilever
is oscillated near its resonance frequency (50 kHz–1 MHz) using a piezoelectric
actuator. As the AFM tip is brought into contact with the surface, the tip lightly
touches or taps the surface. When the cantilever intermittently contacts the surface,
the oscillating behavior is altered by the energy loss during the tip-to-sample
interaction. The change in energy is monitored and used to construct an image of
the surface.
Precision positioning is needed in many AFM applications. In particular, precise
position control in both the lateral and vertical directions is needed to hold the probe
at a desired location or to track a desired motion trajectory. For instance, when the
AFM is used to create quantum dots (2–80 nm in size), accurate position control of
the indenter tip is needed as the probe position directly affects the size, spacing,
and distribution of the nanofeatures. Even 2–4 nm variation in size and spacing
of the nanofeatures can drastically alter their properties [40]. Additionally, high-
speed control of the probe’s movement is needed for high throughput fabrication,
imaging, and metrology. Without accurate motion control along a specific trajectory
at high speed, oscillations can cause the tip to collide with nearby features, leading
to excessive tip-to-sample forces and imaging artifacts. Large forces can damage the
probe tip or soft specimens such as cells. Thus, accurate position control is critical
in an AFM.

3.3 Flexure-Guided Nanopositioner

Flexure-guided nanopositioners have emerged as the design of choice for high-

speed, high-accuracy nanoscale positioning. These nanopositioners have recently
appeared in several commercial AFMs [72–74]. The key advantage of flexure-
guided nanopositioners is their capability of high-speed scanning and low cross-
coupling effects. With an increase in application for high-speed nanopositioning,
such as for monitoring fast pace biological cell events [6] and high-speed nano-
metrology [29], flexure-guided nanopositioners are becoming increasingly popular
and critical, particularly for high-speed SPM systems [6, 37, 69, 104].
Flexure-guided systems exploit the advantages of compliant mechanisms
(flexible mechanisms), where a flexible monolithic joint deforms elastically to
offer accurate and repeatable fine motions. These flexible structures replace the
traditional joints in rigid-link mechanisms, and offer motion guiding and mechanical
amplification as illustrated in Fig. 3.3. Flexure-guided mechanisms are commonly
machined from a single piece of material (monolithic) and no assembly of parts
(i.e., between links and joints) is required. Consequently, the number of parts
of a flexure-guided mechanism is substantially low, resulting in a low weight.
These advantages enable the design of compact, light, and fast nanopositioners.
Furthermore, the absence of moving and sliding joints provide a considerable benefit
where problems such as wear, backlash, friction, and the need for lubrication are
3 Mechanical Design of High-Speed Nanopositioning Systems 65

Fig. 3.3 Example compliant mechanism: (a) schematic of flexure and mechanical amplifier, (b)
example flexure-based long-range serial-kinematic nanopositioner, and (c) example flexure-based
long-range parallel-kinematic nanopositioner

eliminated [28, 44, 96]. Thus, flexure-guided nanopositioners provide repeatable,

reliable, and smooth motions to fulfill the requirement of accurate nanoscale
positioning for many nanoscale applications.
66 Y.K. Yong and K.K. Leang

Flexure-guided mechanisms are categorized into two main configurations: serial

or parallel (see examples in Fig. 3.4). Both serial- [3, 4, 35, 38] and parallel-
kinematic [37, 49, 67, 69, 95, 96, 100] configurations have been used to design
flexure-guided mechanisms. Depending on design requirements, each configuration
has its own advantages and disadvantages. A serial-kinematic system is created (1)
by stacking one piezo-stack actuator in series with another actuator to obtain the cor-
responding displacements [3], or (2) by nesting one flexure-guided nanopositioner
into another [6, 35, 38]. Typically, high mechanical bandwidth can only be achieved
in one motion axis [38]. This is mainly due to the relatively heavy weight that the
base actuator would have to carry. Another disadvantage is the inability to measure
and correct for parasitic motions caused by the cross-coupling effect from other
axes [35]. Nevertheless, the cross-coupling effect can be minimized with carefully
designed flexures and mechanisms [35, 38]. For raster scan purposes, the serial-
kinematic design with one high-bandwidth stage is sufficient. Additionally, the
serial-kinematic configuration is more cost effective as only one high-bandwidth,
high-power piezo-amplifier is required [35].
Parallel-kinematic configurations have been used in commercially available
nanopositioners [72, 74, 83]. Parallel structures offer high motion accuracy and
high mechanical stiffness along the actuation direction, leading to high resonance
frequencies. All actuators can be located relative to ground (base), thus reducing
the inertia of the moving sample platform. Parallel structures generally have
symmetrical configurations and they are less sensitive to temperature variations
which could change the kinematics of the structures due to material expansion or
contraction. All these characteristics of parallel structures make them attractive for
accurate, fast nanopositioning applications. However, the cross-coupling between
the x and y axes of parallel mechanisms is more difficult to deal with compared to
serial mechanisms. Nevertheless, with properly designed flexures, nanopositioners
with cross-coupling as low as 70 to 35 dB can be designed [41, 42, 83, 97, 100].
Although high-speed parallel-kinematic nanopositioners require high-power, high-
bandwidth piezo-amplifiers for both x and y axes, the fast axis of a parallel
nanopositioner can be chosen arbitrarily for rastering in AFMs [37, 69, 97, 100].
The parallel-kinematic configuration can also be used for emerging non-raster
scan methods such as spiral-scan [52], cycloid-scan [102], and Lissajous-scan
patterns [8, 88, 106].

3.4 Design Considerations

High-speed nanopositioners can be employed for video-rate SPM imaging, where

the imaging rate can be as high as 30 frames per second [6, 38]. The frame
rate establishes the required scan rate for the fast-axis motion. For example,
for a 100100 pixels image, the fast-axis scan rate should be at least 3 kHz.
The parasitic motions along the x, y, and z axes should be within 1 % or less
( 40 dB). The resolution of the nanopositioner should be in the sub-nanometer
3 Mechanical Design of High-Speed Nanopositioning Systems 67

Fig. 3.4 Examples of serial- and parallel-kinematic nanopositioners. (a) A serial-kinematic

system which is constructed by nesting one flexure-guided mechanism. The y-axis is the fast axis
in this design. This is a reconstructed figure of Ando et al. [4]. (b) A parallel-kinematic system
where all actuators are relative to ground. The fast axis can be either the x- or y-axis. The original
figure can be found in [100]. (c) Lumped parameter models of both configurations
68 Y.K. Yong and K.K. Leang

range. The scan range should be approximately 10 10 m or more. In order to

achieve the above characteristics and to avoid image artifacts, discussed above,
it is necessary to consider (1) the mechanical stiffness of the nanopositioner and
flexure design, (2) actuator properties and drive electronics, (3) material selection,
and (4) manufacturing techniques.

3.4.1 Flexure Hinges

Flexures are important mechanical elements that commonly used in ultra-precision

positioning devices to guide motions with nanometer accuracy. When forces are
applied, these hinges generate motions by elastically deform their structures. Flex-
ure hinges are increasingly popular for designs requiring monolithic manufacturing,
smooth repeatable motions, low inertial mass, frictionless, and zero backlash.

3.4.1.1 Commonly Used Flexure Hinges in Nanopositioning

There are many types of flexures which provide unique properties to suit different
positioning requirements. Examples of flexures which provide in-plane rotational
degree-of-freedom (DOF) are basic beam flexure [69, 100], circular [44, 99, 104],
corner-filleted [47], double-hinged [35, 89], and conical [47, 48, 80]. The out-of-
plane stiffness of these flexures are high, however, if two degrees-of-freedom is
desired, the depth of these flexures can be reduced to allow out-of-plane rotation
(see Fig. 3.5b). There are also a multiple degrees-of-freedom flexure as illustrated
in Fig. 3.5c where this hinge is free to rotate about all three axes [46].
Among all the aforementioned in-plane flexures, circular flexures provide the
most accurate rotational motions because of the relatively small shift of the center
of deflection. Circular flexures have a small in-plane rotational motion (bending),
therefore, there are suitable for applications which require accurate positioning
over a relatively small range [104]. For high displacement-orientated applications,
corner-filleted flexures are more suitable due to their relatively low bending
stiffness. However, there exhibit significant shift of center of deflection, making
them less attractive for high precision-orientated applications [44, 104]. Flexures

Fig. 3.5 Flexure degrees of freedom: (a) one, (b) two, and (c) multiple degrees-of-freedom
3 Mechanical Design of High-Speed Nanopositioning Systems 69

Fig. 3.6 A long, slender

beam subjected to forces and
moments

with a good combination of a large deflection angle and high rotational precision
are the cross-axis pivot, cartwheel, and butterfly-type flexures [78, 93]. These types
of flexures can be difficult to model due to the complexity of the structures.
Many methods can be used to derive a closed-form solution for the deformation
of a flexure. One of these methods is the Castigliano’s second theorem [44] (also
known as Castigliano’s displacement theorem). This method is a very effective
method that allows calculation of the deformations of elastic bodies under the action
of external and reactions loadings. The theorem is only valid for materials that
deform elastically and obey Hooke’s law for stress and strain.
According to the Castigliano’s theorem, the linear displacement ui and angular
deformations i of an elastic body at a point in a given direction (i D x; y; z) is the
partial derivative of the total strain energy U with respect to the force Fi and Mi
acting at that location, i.e.:

@U
ui D ; (3.1)
@Fi
@U
i D : (3.2)
@Mi

For a long, slender beam in Fig. 3.6 that is subjected to bending, shearing, axial load,
and torsion, the total strain energy is [44]

U D Ubending C Ushearing C Uaxial C Utorsion : (3.3)

and
Z Z
My2 Mz2
Ubending D Ubending;y C Ubending;z D ds C dsI (3.4)
L 2EIy L 2EIz

Z Z
˛Vy2 ˛Vz2
Ushearing D Ushearing;y C Ushearing;z D ds C dsI (3.5)
L 2GA L 2GA
70 Y.K. Yong and K.K. Leang

Z
Nx2
Uaxial D Uaxial;x D dsI (3.6)
L 2EA
Z
Mx2
Utorsion D Utorsion;x D ds; (3.7)
L 2GJ

where Mi is the moment, Vi is the shear force, and Nx is the axial force. G D 2.1C/E

is the shear modulus, Ii is the second moment of inertia, J is the torsional moment
of inertia, ˛ is the shear coefficient, and ˛ D 6=5 for a rectangular cross-section
[16, 60, 107].
For a one-DOF, constant-width flexure hinge for which loadings are acting at
location 1, as illustrated in Fig. 3.7, deformations at location 1 can be expressed into
a matrix form by rearranging Eqs. (3.1)–(3.7) and factoring out all the common load
parameters. This gives

a y b y
x
x
t
x
t t(x) 1 4 2 5 3
t(x)
L x
r L/2 r
L

c y d
y

x x
c
t t
x x
t(x) t(x)
1 3 1 3

2 2

r
L/2
L=2r L

Fig. 3.7 Dimensions of commonly used flexures. (a) Basic beam, (b) corner-filleted, (c) circular,
and (d) ellipse
3 Mechanical Design of High-Speed Nanopositioning Systems 71

8 9 2 38 9
ˆ
ˆ u1x >
> C1;xFx 0 0 0 0 ˆ
ˆ F1x >
>
ˆ
ˆu > > 6 0 7ˆˆF > >
ˆ
< 1y > = 6 C1;yFY C1;yMz 0 0 7ˆ< 1y > =
6 7
1z D 6 0 C1;z Fy C1;z Mz 0 0 7 M1z ; (3.8)
ˆ
ˆ > 6
> 7ˆ >
ˆ
ˆ u1z >> 4 0 0 0 C1;zFz C1;zMy 5 ˆ ˆ
ˆ F1z >
>
>
:̂ >
; :̂ >
;
1y 0 0 0 C1;y Fz C1;y My M1y

where C1;z Fy D C1;yMz and C1;y Fz D C1;zMy . The in-plane compliances are
Z
1 L
1
C1;xFx D dxI (3.9)
Eh 0 t.x/
Z
12 L
x2 ˛E
C1;yFy D dx C C1;xFx I (3.10)
Eh 0 t.x/ G
Z
12 L
x
C1;yMz D dxI (3.11)
Eh 0 t.x/3
Z
12 L
1
C1;z Mz D dx: (3.12)
Eh 0 t.x/3

The out-of-plane compliances are

Z
12 L
x2 ˛E
C1;zFz D dx C C1;xFx I (3.13)
Eh3 0 t.x/ G
Z
12 L
x
C1;zMy D dxI (3.14)
Eh3 0 t.x/
12
C1;y My D C1;xFx : (3.15)
h2

Note that the torsional deformation 1x is usually insignificant for nanoposition-
ers discussed in this chapter. Therefore 1x is neglected in Eq. (3.8). The in-plane
compliance C1;yFy and out-of-plane compliance C1;zFz also include the shearing
term ˛E C
G 1;xFx
. For a relatively long beam compared to its height, the shearing
effect can be ignored according to the Euler–Bernoulli-beam model. This model
assumes the planar cross section remains perpendicular to the neutral axis after the
external bending has been applied. Shearing stresses and deformations are ignored.
However, for a short beam, the shearing effect needs to be taken into consideration.
Deformation at location k can also be found by introducing a fictitious force Fk
and moment Mk such that

@U
uk D ; (3.16)
@Fk
@U
k D : (3.17)
@Mk
72 Y.K. Yong and K.K. Leang

However for the design analysis of most nanopositioners, deformations/stiffness at

location 1 are the main design focus. Readers are referred to [44] for closed-form
equations of deformations/stiffness at other locations of flexures.
Flexures that are commonly used in both serial- and parallel-kinematics nanopo-
sitioners are the planar types, i.e., basic beam, corner-filleted [44, 47], circular
[44, 61, 87, 99], and elliptic [44, 80] as shown in Fig. 3.7. The closed-form
compliance equations for the above flexure types can be found in [44]. These
equations are presented here for convenience.
For a basic beam flexure with constant thickness, that is t.x/ D t for 0  x  L,
as shown in Fig. 3.7a, the in-plane compliance equations are

L
C1;xFx D I
Eht
4L3 ˛L
C1;yFy D C I
Eht3 Ght
6L2
C1;yMz D I
Eht3
12L
C1;z Mz D (3.18)
Eht3 :
The out-of-plane compliance equations are

4L3 ˛L
C1;zFz D 3
C I
Eh t Ght
6L2
C1;zMy D I
Eh3 t
12L3
C1;y My D : (3.19)
Eh3 t
For a corner-filleted flexure as shown in Fig. 3.7b where the thickness is
8 h p i
ˆ
< t C 2 r  x.2r  x/ ; x 2 Œ0; r
t.x/ D t; x 2 Œr; L  r (3.20)
:̂ p
t C 2fr  .L  x/Œ2r  .L  x/g; x 2 ŒL  r; r;

the in-plane compliances are

" r #
1 L  2r 2.2r C t/ 4r
C1;xFx D Cp arctan 1 C  I
Eh t t.4r C t/ t 2

3 4.L  2r/.L2  Lr C r2 /
C1;yFy D
Eh 3t3
3 Mechanical Design of High-Speed Nanopositioning Systems 73

p
t.4r C t/Œ80r4 C 24r3 t C 8.3 C 2 /r2 t2 
C p
4 t5 .4r C t/5
p
t.4r C t/Œ4.1 C 2 /rt3 C t4 
C p
4 t5 .4r C t/5
q
.2r C t/3 .6r2  4rt  t2 /arctan 1 C 4rt
C p
t5 .4r C t/5
40r4 C 8Lr2 .2r  t/ C 12r3 t C 4.3 C 2 /r2 t2
C
2t2 .4r C t/2
t4
2.l C 2 /rt3 C 2 4L2 r.6r2 C 4rt C t2 /
C C
2t2 .4r C t/2 t2 .2r C t/.4r C t/2
.2r C t/Œ24.L  r/2 r2  8r3 t C 14r2 t2 C 8rt3 C t4 
 p
t5 .4r C t/5
r )
4r
 arctan 1 C I
t
(
6L
C1;yMz D .4r C t/ŒL.2rt /.4r C t/2
Eht3 .2r C t/.4r C t/2
p
 4r2 .16r2 C 13rt C 3t2 / C 12r2 .2r C t/2 t.4r C t/
r )
4r
 arctan 1 C I
t
( "
12 2r
C1;z Mz D L  2r C t.4r C t/.6r2 C 4rt C t2 /
Eht3 .2r C t/.4r C t/3
r #)
p 4r
C 6r.2r C t/2 t.4r C t/arctan 1 C : (3.21)
t

The out-of-plane equations are

( "
12 .L  2r/.L2  Lr C r2 / t 2.L  2r/
C1;zFz D C Lr log p
Eh3 3t 2r C t t.4r C t/
r #)
4r
 arctan 1 C I
t
  r
6 t t
C1;zMy D 3
L.L  2r/ C 2r t log  2.L  2r/
Eh t 2r C t 4r C t
74 Y.K. Yong and K.K. Leang

r #)
4r
 arctan 1 C I
t
 r r 
6 t t
C1;y My D 2L  4r D t C 4.2r C t/ arctan : (3.22)
Eh3 t 4r C t 4r C t

p circular flexure hinge as shown in Fig. 3.7c, the thickness t.x/ D t C 2Œr 
For
x.2r  x/. The in-plane compliances are
" r #
1 2.2r C t/ 4r
C1;xFx D p arctan 1 C  I
Eh t.4r C t/ t 2

3 8r3 .44r2 C 28rt C 5t2 /
C1;yFy D 2.2 C /r C t C
4Eh.2r C t/ t2 .4r C t/
p
C .2r C t/ t.4r C t/
80r4 C 24r3 t C 8.3 C 2 /r2 t2 C 4.1 C 2 /rt3 C t4
 p
t5 .4r C t/5
r )
8.2r C t/4 .6r2 C 4rt C t2 / 4r
 p arctan 1 C I
t5 .4r C t/5 t
"
24r2
C1;yMz D t.4r C t/.6r2 C 4rt C t2 /
Eht3 .2r C t/.4r C t/3
r #
p 4r
C 6r.2r C t/2 t.4r C t/arctan 1 C I
t
C1;yMz
C1;z Mz D : (3.23)
r
The out-of-plane compliance equations are

24r2 t C1;zFz
C1;zFz D log I C1;zMy D I
Eh3 2r C t 2r
 r r 
6 t t
C1;y My D 4.2r C t/ arctan  t : (3.24)
Eh3 t 4r C t 4r C t

For an elliptical flexure as shown in Fig. 3.7d, with the thickness

2 s 3
 
2x 2 5
t.x/ D t C 2c 41  1 1 ; (3.25)
c
3 Mechanical Design of High-Speed Nanopositioning Systems 75

where c is a constant, the in-plane compliances are

" r #
1 4.2c C t/ 4c
C1;xFx D p arctan 1 C  I
4Ehc t.4c C t/ t
(
3L3
C1;yFy D tŒ96c5 C 96c4 t C 8.11 C 4 /c3 t2
16Eht3 c3 .2c C t/.4c C t/2
r
2 3 4 5 t
C32.1 C /c t C 2.2 C 5 /ct C t   4 .2c C t/4
4c C t
r )
2 2 4c
.6c C 4ct C t /arctan 1 C I
t

6L2
C1;yMz D
Eht2 .2c C t/.8c2 C t2 /
" r #
2 2 6c.2c C t/2 4c
 6c C 4ct C t C p arctan 1 C : (3.26)
t.4c C t/ t

The out-of-plane compliances are

(
3L3 .2c C t/.4c2 C 4ct C t2 /
C1;zFz D 3 3
2.4  /c2 C 4.1 C /ct C p
16Eh c t.4c C t/
" #)
2c
 2arctan p  I
t.4c C t/
( " # )
3L2 2c C t 2c
C1;zMy D p 2arctan p C  : (3.27)
2Eh3 c t.4c C t/ t.4c C t/

Double-hinged flexure as depicted in Fig. 3.8 has been widely used in nanoposi-
tioning systems to guide the displacement of the sample platform [32]. It is also used
to restrain parasitic (out-of-plane and off-axis) motions so that the displacement
stage only moves along the actuation axis. Recently a detailed design analysis of the
double-hinged flexure for high-speed AFM serial-kinematic nanopositioning stage
is reported in [35]. They pointed out that when a corner-filleted flexure is displaced
in the vertical direction, the majority of the vertical displacement is caused by shear
deformation of the center section of the flexure. Therefore, a simple way to increase
the out-of-plane stiffness and consequently the resonance frequency is to thicken
the center section of the beam flexure. This converts the beam flexure into a double-
hinged flexure with a “rigid” center connecting link as shown in Fig. 3.8.
76 Y.K. Yong and K.K. Leang

Fig. 3.8 Flexure design for increased out-of-plane stiffness. Left: conventional corner-filleted
beam flexure; and right: serial-compliant double-hinged flexure with thickened center section

In this case, the cross-sectional area and second moment of inertia values are
A.x/ D ht.x/ and I.x/ D ht.x/3 =12, respectively. For example, the thickness of the
flexure in Fig. 3.8 is
8 h p i
ˆ
ˆ t C 2 r  x.2r  x/ ; x 2 Œ0; a
ˆ
ˆ
ˆ
ˆ
ˆ
ˆ
ˆ
t h p i x 2 Œa; b
;
ˆ
ˆ
ˆ t C 2 r  .l  x/.2r  l C x/ ; x 2 Œb; c
<
t.x/ D t C 2rh i ; x 2 Œc; d ; (3.28)
ˆ
ˆ p
ˆ
ˆ t C 2 r  .l  g/.2r  l C g/ ; x 2 Œd; e
ˆ
ˆ
ˆ
ˆ
ˆt
ˆ ; x 2 Œe; f 
ˆ h i
:̂ t C 2 r  pg.2r  g/ ; x 2 Œf ; L

where a D r, b D l  r, c D l, d D L  l, e D d C r, f D L  r, g D L  x, t
and l are thickness and length of the thin section of the flexure, r is the fillet radius,
t C 2r D T is the thickness of the thickened section, and L is the length of the entire
flexure. For this case, the compliance is determined by first determining the total
strain energy while using the thickness function t.x/ in the area A.x/ and second
moment of inertia I.x/ expressions. Again, the coordinate system is placed on the
free end for simplification and to allow for direct integration. For instance, the total
strain energy for bending due to a point load is
Z Z
L
M.x/2 L
˛V.x/2
UD 3
dx C dx;
2E ht.x/
0 0 2Ght.x/
12
Z Z
12Fy2 L x2 ˛Fy2 L 1
D dx C dx: (3.29)
2Eh 0 t.x/3 2Gh 0 t.x/
3 Mechanical Design of High-Speed Nanopositioning Systems 77

Taking the partial derivative with respect to the applied force Fy gives the displace-
ment
Z Z
@U 12Fy L
x2 ˛Fy L
1
uy D D 3
dx C dx: (3.30)
@Fy Eh 0 t.x/ Gh 0 t.x/

Then, the in-plane (and out-of-plane) stiffness is then calculated numerically by

taking the ratio of the force to deflection.

3.4.1.2 Boundary Conditions of Flexures

The in- and out-of-plane compliance/stiffness is then calculated by taking the ratio
of deflection to force/moment or vice versa in Eq. (3.8). Displacement and loading
conditions of a flexure are dependent on its support conditions. The commonly used
support conditions for beams are depicted in Fig. 3.9.
For example of a fixed-guided beam flexure (in which the moving end of the
flexure remains parallel to the fixed-end) with concentrated end-load, which is a
common boundary condition in nanopositioning design, the resultant moment is
Fy L=2. By using Eqs. (3.8), (3.18) and (3.19), the flexure displacement uy due to
applied force Fy and moment Mz D Fy L=2 is

uy D C1;yFY Fy C C1;yMz Mz
 3   2
4L ˛L 6L Fy L
D C Fy C
Eht3 Ght Eht3 2
 3 
L ˛L
D C Fy : (3.31)
Eht3 Ght

Therefore, the stiffness ky is

 3 1
Fy L ˛L
ky D D C : (3.32)
uy Eht3 Ght

Similarly the vertical displacement uz is

uz D C1;zFz Fz C C1;zMy My
 3   2
4L ˛L 6L Fz L
D C Fz C
Eh3 t Ght Eh3 t 2
 3 
L ˛L
D C Fz ; (3.33)
Eh3 t Ght
78 Y.K. Yong and K.K. Leang

Mz = 0

Fy L

Cantilever

Mz = - Fy L/2

Fy L

Fixed-guided

Mz = - (Fy /2)(L/2/2)
=- Fy L/8

Fy Fy /2
L L/2
Full model Half model
(fixed-guided equivalent)
Fixed-fixed

Fig. 3.9 Support conditions of a beam with concentrated end-load

and the stiffness kz is

 3 1
Fz L ˛L
kz D D C : (3.34)
uz Eh3 t Ght

The axial displacement ux and stiffness kx are

L
ux D Fx I (3.35)
Eht
Eht
kx D : (3.36)
L
3 Mechanical Design of High-Speed Nanopositioning Systems 79

Table 3.1 Stiffness Cantilever Fixed-guided Fixed-fixed

equations of a basic beam Eht Eht 4Eht
kx
flexure with loading and hL 3 i1 hL 3 i1 hL i1
4L ˛L ˛L L3 ˛L
support conditions as shown ky Eht3
C Ght
L
Eht3
C Ght 16Eht3
C 4Ght
Fig. 3.9 h i1 h i1
4L3 ˛L L3 ˛L 4L ˛L 1
kz Eh3 t
C Ght Eh3 t
C Ght 16Eh3 t
C 4Ght

Fig. 3.10 A common design

of a parallel-kinematic
nanopositioner

Note that the ratio between the x- and y-axes stiffness (neglecting shearing) is
2
kx
ky
D Lt . This ratio is usually large for a long, slender flexure beam, which
provides excellence reduction of orthogonal motion. The ratio between the z-
2
and y-axes stiffness is kkyz D ht . Therefore, large flexure height h is one of
the key factors for increasing the out-of-plane stiffness. However, it comes at an
expense of a higher profile nanopositioner which increases the effective mass, thus
reducing the actuation resonance. This effect may not be attractive for high-speed
nanopositioning systems.
Using the similar method discussed above, stiffnesses of a basic beam for other
support conditions, such as cantilever and fixed-fixed, are derived and exhibited in
Table 3.1. The same procedures can be used to derive stiffnesses of flexures with
other profiles as discussed in Sect. 3.4.1.
A common nanopositioning design constructed from connecting several flexures
in parallel is shown in Fig. 3.10. The effective stiffness is Nki (i D x; y; z), where N
is the number of flexures. Generally speaking, the effective lateral stiffness kx and
ky can be increased by (1) increasing the number of flexures and (2) decreasing the
flexure length. The effective vertical stiffness kz can be increased by (1) and (2), and
also by using double-hinged flexures as previously discussed and shown in Fig. 3.8.

3.4.1.3 Commonly Used Flexures in MEMS

In linear micromachined structures such as gyroscopes [2] and MEMS nanoposi-

tioners [53, 56], the suspension systems are designed to be compliant (flexible) along
the desired motion axes, and stiff along other directions. Most suspension systems
utilize several fixed-guided flexures arranged in parallel as shown in Fig. 3.10 to
80 Y.K. Yong and K.K. Leang

achieve the desired stiffness and resonance frequency. One disadvantage for the
fixed-guided beam flexure is that, when experiencing large lateral deflection, the
axial tension stress in the flexure increases significantly. This axial stress induces a
nonlinear force-deformation relationship in the flexure [2]. To avoid this nonlinear
stiffness behavior, a variation of the fixed-guided flexure, such as the folded (also
known as U-shaped flexure), double-folded and crab-leg flexures are used to reduce
the axial stress components [21, 63].
A folded flexure as shown in Fig. 3.11a is constructed by connecting two fixed-
guided flexures in series. The two constrained points of the flexure are located at
the same side, and the two unconstrained ends are free to expand or contract in all
directions. Therefore, the lateral deflections does not result in axial loading. Since
a folded flexure consists of two fixed-guided flexures in series, the stiffness of the
folded beam is
1 1 1
D C
ky;folded ky ky
2L3
D ;
Eht3
Eht3
ky;folded D ; (3.37)
2L3

where ky D Eht3 =L3 (neglecting shearing) as provided in Table 3.1. Similarly,

Eht
kx;folded D I (3.38)
2L
Eh3 t
kz;folded D : (3.39)
2L3
For example shown in Fig. 3.11a, four folded flexures are arranged in parallel to
create the suspension system. The effective stiffness of the system is 4kfolded , i.e.:

2Eht3
Ky;foldedsys D I (3.40)
L3
2Eht
Kx;foldedsys D I (3.41)
L
2Eh3 t
Kz;foldedsys D : (3.42)
L3
A double-folded flexure consists of two folded flexures connected symmetrically
in parallel is shown in Fig. 3.11b. The stiffness of a double-folded beam is
2kz;folded , i.e.:
3 Mechanical Design of High-Speed Nanopositioning Systems 81

Fig. 3.11 MEMS suspension a

systems utilize (a) folded
flexures, (b) double-folded
flexures, and (c) crab-leg
flexures

Eht3
ky;doublefolded D I (3.43)
L3
Eht
kx;doublefolded D I (3.44)
L
82 Y.K. Yong and K.K. Leang

Eh3 t
kz;doublefolded D : (3.45)
L3
Note that for small deflection, the folded flexure provides half the stiffness of a
fixed-guided flexure, while the double-folded flexure provides the same stiffness as
that of the fixed-guided flexure.
For the suspension system that is constructed from two double-folded flexures as
shown in Fig 3.11b, the effective stiffnesses are

2Eht3
Ky;dfsys D I (3.46)
L3
2Eht
Kx;dfsys D I (3.47)
L
2Eh3 t
Kz;dfsys D : (3.48)
L3
Another commonly used flexure in MEMS applications is a crab-leg flexure. The
schematic of a crab-leg suspension system is shown in Fig. 3.11c. Crab-leg flexures
are used in resonant-structure suspension systems such as MEMS gyroscopes [63].
The crab-leg flexure consists of a shin and a thigh segment. The thigh segment has
length La and thickness ta . The shin segment has length Lb and thickness tb . The
height of the flexure is h. For a suspension system constructed by four fixed-guided
flexures (Fig. 3.10), both the kx and ky are scaled by the same term 4Eht=L, therefore
it is impossible to specify kx and ky independently. The crab-leg suspension system
allows the two lateral stiffnesses kx and ky to take on values independent to each
other [63]. The detailed stiffness derivation of the crab-leg suspension system can
be found in [21]. Their stiffness equations are presented here for convenience.

Ehtb3 .4Lb C La /
Ky;crablegsys D I (3.49)
Lb3 .Lb C La /
Ehta3 .Lb C 4 La /
Kx;crablegsys D I (3.50)
La3 .Lb C La /
 3
where D Ib
Ia
D tb
ta
.

3.4.1.4 Finite-Element-Analysis

Even though analytical expressions of flexures serve as a useful guideline in design,

finite-element-analysis (FEA) is still a popular and powerful numerical tool for
solving complicated engineering problems, including static and modal analyses,
piezoelectricity, thermal analysis, coupled thermal-electric analysis, etc. Frequency
response functions, which are useful for studying cross-coupling and out-of-plane
3 Mechanical Design of High-Speed Nanopositioning Systems 83

motions, can also be estimated using FEA package such as ANSYS. By careful
consideration of mesh quality and convergence of results, accuracy within 15 %
between FEA and experimental results can usually be achieved.

3.4.2 Piezoelectric Stack Actuators

3.4.2.1 Comparison of Piezo-Stack Actuated Nanopositioners

State-of-the-art flexure-guided piezo-driven stages [71], both direct drive serial-

kinematic [5, 35, 38] and parallel-kinematic [70, 104, 105] configurations, have
been developed with mechanical resonances in the tens of kHz range. These designs
employ stiff and compact stack piezoactuators. The actuation resonance frequency
for a stack piezoactuator held fixed on one end can be approximated by Kenton [33]
s
1 3E
fa D Hz; (3.51)
2 Lp piezo

where Lp is the length of the actuator, E is the elastic modulus, and piezo is the
actuator’s density. Substituting in values for E D 33:9 GPa and piezo D 8000 kg/m3
and assuming an achievable stroke (R) of 1 m per mm of piezo length, it is shown
that the maximum scanning frequency is fa D 567;460 R1 Hz m. When mass mend
is added to the free end of the actuator, the actuation resonance frequency reduces to
v
u
1 u
AE
fa D t Lp
Hz; (3.52)
2 1
3
ALp piezo C mend

where A is the cross-sectional area.

For comparison, the relationship between range and resonance frequency for
a variety of commercial and custom nanopositioners is shown in Fig. 3.12 [33].
The range is plotted with respect to the resonance frequency for each stage when
provided. When full details are not available for multi-axis positioners, it is assumed
that the resonance frequency is provided for the stage with the largest displacement,
and therefore; the largest range is plotted with respect to the lowest resonance
frequency. Theoretically, the maximum attainable range R for a given actuation
(longitudinal) resonance is R D 567;460 f 1 (as derived in [33]). This value is for
a fixed-free piezoactuator with a modulus of elasticity of 33.9 GPa and a density of
8000 kg/m3 assuming 1 m of travel per mm of piezo length. This relationship is
shown in Fig. 3.12 as a dashed line. The commercial and custom nanopositioners
in Fig. 3.12 are well below this theoretical limit. A trend line depicted as a solid
line shows that for commercial and custom nanopositioners, the relationship is
approximately R D 30;688 f 0:916 .
84 Y.K. Yong and K.K. Leang

104 Custom
MadCityLabs
Physik Instrumente
Thor Labs
Npoint
103 Schitt er & Stemmer, 2004 [46]
Tube scanner Queensgate Instruments

Yong et al., 2009 [61] R = 30,688f -0.916

Li & Xu, 2010 [73] R² = 0.7647
PKxy - w/ mech. amp.
PKxy - w/ mech. amp. R= 567,460f -1
Kenton & Leang, 2012 [20] et al., 2007 [37]
102 SKy - direct drive PKxy - direct drive
Range (µm)

Kim et al., 2005 [114]

SKx - direct drive Kenton & Leang, 2012 [20]
SKy - direct drive SKx - direct drive
SKz - direct drive et al., 2007 [37]
SKz - direct drive
1
10
Yong et al. [21] Kenton et al., 2011 [28]
PKxy- direct drive Yong & Moheimani, 2010 [38] SAz- direct drive
PKxyz - direct drive

Humphris et al., 2005 [41]

Yong et al. [21]
Tuning fork
100 PKxy- direct drive
Picco et al., 2007 [23]
Picco et al., 2007 [23]
Tuning fork
Tuning fork
Ando et al., 2008 [33] Rost et al., 2005 [57] Kenton & Leang, 2012 [20]
SKxyz - direct drive SKxyz - shear piezo SKz - direct drive
-1
10
101 102 103 104 105 106
Resonant frequency (Hz)

Fig. 3.12 High-performance commercial and custom nanopositioners plotted as range R with
respect to resonance frequency f . The solid line represents a linear least-square-error line fit
to the data points. The dashed line represents the theoretical first mechanical resonance in the
actuation mode for a fixed-free piezoactuator (assuming 1 m of travel per 1 mm length). SK serial-
kinematic, PK parallel-kinematic, SA single axis, x, y, z refers to axis being referenced

Piezo-stack actuators are widely used to drive high bandwidth, flexure-guided

nanopositioners, where repeatable sub-nanometer resolution is a requirement.
Piezoelectric materials generally produce small strains, i.e., approximately 0.1 %
of their original length. However, piezoelectric ceramic actuators can generate
mechanical stresses in the order of tens of megapascals. That is, they can provide
tens of newtons of force over a millimeter squared area. The response time of a
piezoelectric actuator is determined by its molecular structure [39]. Due to the
fact that molecular changes occur over a very small length scale, piezoelectric
materials can respond very quickly to stimuli, e.g., an acceleration rate of 104 g can
be achieved by a typical piezo-stack actuator [67]. Piezoelectric materials are quite
suitable for applications that require fast actuation.
A piezo-stack actuator is constructed by “stacking” and gluing multiple thin
layers of piezoelectric ceramics (as thin as 100 m) together such that the polar-
ization direction is aligned with the direction of stroke and the blocking force [83].
All layers are electrically connected in parallel. This enables the actuator to be
operated at 100 V or less, and to achieve a stroke of approximately 0.1 % of
the stack’s length. The actuator also generates high actuation forces. The force
3 Mechanical Design of High-Speed Nanopositioning Systems 85

generated by a piezo-stack actuator is proportional to its d33 constant, as opposed

to d31 that is smaller and often used in piezoelectric actuators. By combining a
piezo-stack actuator with a flexure-guided amplification mechanism it is possible
to realize nanopositioning platforms that can achieve a relatively large stroke with
sub-nanometer positioning resolution [41, 43, 100].

3.4.2.2 Preloading

Because a piezo-stack actuator is made by gluing together multiple layers, the

actuator is highly sensitive to tensile forces (pulling forces). Tensile forces can
cause cracking of the ceramic material as well as separation of the layers. For a
non-preloaded actuator, manufacturers often suggest that the tensile load should
be within 5–10 % of the compressive load limit in order to avoid damaging the
stack [83]. During high-speed operations, the piezo-stack experiences significant
pushing and pulling forces due to the inertial forces cause by its own mass and
the additional mass that it carries. A preload must be applied to compensate for
the tensile force, thus protecting the piezo-stack actuator. Despite these challenges,
stack actuators are used extensively in high-speed nanopositioning designs because
of their high stiffness, fast response, and high force generation.
For flexure-guided mechanisms, a preload is commonly applied to the piezo-
stack actuator using flexures (see Fig. 3.13). This is an elegant approach since
it eliminates the need for additional components such as mechanical springs.
The preload on the piezo-stack actuator is achieved when it is first installed
into the flexure-guided structure by pressing the piezo-stack against the flexures
using fasteners (see Fig. 3.13a) or metal shims. Other preloading methods include
applying weights to deform the flexure elastically, allowing a piezo-stack actuator to
be placed in its designated position. Weights are then released to restore the original
position of the flexures (see Fig. 3.13b). The flexures will be slightly deformed and
the resultant reaction force will hold the actuator in place.
The required preload can be calculated from the sample mass and its acceleration
(computed from the known scan trajectory) using Newton’s second law. A procedure
for estimation of preload is discussed in [38].

3.4.2.3 Travel Range

The travel range of a piezo-stack actuator reduces when the stiffness of its flexures
increases. One can imagine the flexures as a spring load (see Fig. 3.14) and the total
available displacement of a piezo-stack actuator can be estimated as [83]
 
kT
LD Lo ; (3.53)
kT C ks
86 Y.K. Yong and K.K. Leang

Fig. 3.13 Preloading a piezo-stack actuator. (a) Preloading using flexures. The deformation of the
flexures are exaggerated in the diagram. (b) Preloading using fastener

Fig. 3.14 Displacement of a

piezo-stack decreases as the External spring
spring load increases. compliance
Original diagram can be Δxmax
found in [83]
Stroke

0
Force F block

where L is the displacement with external spring load, Lo is the nominal

displacement without external force, kT is the piezo-stack’s stiffness, and ks is the
external spring load. In most nanopositioner designs, this effect is insignificant
since the stiffness of the flexures is designed to be 1/20th to 1/5th the stiffness of
the piezo-stack actuator [38, 100, 104]. However, for nanopositioner designs where
a large travel range is a priority, the consideration of total displacement in Eq. (3.53)
may become significant.
3 Mechanical Design of High-Speed Nanopositioning Systems 87

a c

Fig. 3.15 Commonly used amplification mechanisms. (a) Schematic of a level-type mech-
anism. The amplification ratio can be estimated as r D .a C b/=a. (b) Schematic
of a bridge-type mechanism. The amplification ratio can be estimated by r D
 p
l2 sin2  C di .2l cos   di /  l sin  =di . (c) FE simulated displacement of the level-type
mechanism. (d) FE simulated displacement of the bridge-type mechanism

Free stroke of most piezo-stack actuators is approximately 0.1 % of their length,

which is insufficient for applications where a large displacement is a necessity.
Multiple piezoactuators per degree-of-freedom have been used to increase travel
range and bandwidth, but at the cost of increased power to drive the piezoactuators
at high frequencies [5, 70]. Designs which involve mechanical amplification have
been implemented to increase the range without having to increase the actua-
tor’s length [25, 26, 32, 36, 49, 71, 95, 100]. A parallelogram flexure guiding
system as shown in Fig. 3.15a, c is a commonly used amplification mechanism.
The amplification ratio of this mechanism can be estimated as r D .a C b/ =a. The
bridge-type amplification mechanism, illustrated in Fig. 3.15b, d, is also commonly
used to amplify displacement of a piezo-stack actuator [36, 45, 50]. For this
bridge-type amplification mechanism, the actuator is extended initially, and then
retracted to provide the input displacement. The amplified output displacement of
this mechanism is perpendicular to the direction of the actuator. The amplification
ratio of this mechanism can be estimated as [45],
q 
rD l2 sin2  C di .2l cos   di /  l sin  =di : (3.54)
88 Y.K. Yong and K.K. Leang

Figure 3.15c, d illustrate the finite-element (ANSYS software) simulated displace-

ments of the two flexure-guided amplification mechanisms, respectively.
Incorporation of a mechanical amplification mechanism into a nanopositioner is
an effective way of increasing its achievable stroke. However, the added mass of
the mechanical amplifier together with the flexible linkages lowers the mechanical
resonance of the structure, which leads to a lower bandwidth. To achieve a high
resonance frequency, the design of the nanopositioner has to be compact and
rigid [35, 37]. Consequently, the length of the flexures and the amplification levers
have to be small. However, short flexures and levers will reduce the overall motion
of the device. This forces the designer to reach a compromise between the travel
range and the bandwidth.

3.4.2.4 Self-Heating

During dynamic operation a piezoelectric actuator experiences self-heating, which

is known to increase with actuation frequency and amplitude. It is also well
known that electromechanical and electrical properties of a piezoelectric actuator
can vary with temperature. Therefore, good thermal management is necessary for
predictable performance, as well as to prevent premature failure and depoling of
the piezoelectric material used. Heat management can be challenging due to the
low thermal conductivity of the piezoelectric material. Creative heat sink designs
can be employed to minimize the heating, but should not hinder the motion of the
actuator. When an piezoelectric actuator is packaged in a metal case, the air gap can
act as an insulator. Several vendors offer specialized actuator configurations, that
are based on methods such as the “ThermoStable” technique, in order to improve
heat management [7].

3.4.2.5 Electrical Considerations

Due to the highly capacitive nature of piezo-stack actuators, amplifiers with large
current and power dissipation capabilities are needed to drive the actuators at high-
speed [22]. Fleming [22] and Leang and Fleming [38] cover some of the electrical
and drive issues related to high-speed operation of piezo-stack actuators. The current
Ip and the associated power dissipation Pd in a linear amplifier can be estimated by
Leang and Fleming [38]

Ip D sCVp (3.55)
Pd D Ip Vs  Vp ; (3.56)

where s is the Laplace variable, C is the piezo-stack capacitance, Vp is the load

voltage and, Vs is the supply voltage. For a nanopositioner that is to be operated at
high speeds, selection of a piezo-stack actuator with a small capacitance is prudent.
This will ensure that the actuator can be driven by an amplifier with sufficient
3 Mechanical Design of High-Speed Nanopositioning Systems 89

current and power dissipation capabilities. Typically, a smaller piezo-stack actuator

possesses a smaller capacitance, but it has a lower stiffness and can provide a
smaller stroke. These issues should be factored into the selection of a piezo-stack
actuator for high-speed nanopositioning. Yong et al. [104] and Kenton [33] reports a
comparison of piezo-stack actuators with various cross-sectional areas versus their
maximum travel ranges and stiffness.

3.4.3 Material Considerations

The dynamic performance of a high-speed flexure-guided nanopositioner is relied

on its material properties. Not only that, the choice of material for its supporting base
also plays a major role in high-speed nanopositioner design. Material properties of
great importance in dictating the dynamic performance of a nanopositioner are the
Young’s modulus of elasticity E, density , the coefficient of thermal expansion
(CTE). To facilitate the fabrication method of choice for ultra-precision machining
in nanopositioning, that is the wire electrical-discharge-machining (WEDM), mate-
rials with sufficient electrical conductivity should also be taken into consideration
during the design process. In summary, materials deemed suitable for realizing a
high speed and thermally stable flexure-guided nanopositioner should have high
E= ratio, low CTE, and high .
Some commonly used materials in nanopositioning design are aluminum alloy
Al 7075 and Al 6061, titanium alloy, invar alloy, and super invar alloy. Their
properties are detailed in Table 3.2. Aluminum alloys such as Al 7075 and Al 6061
are relatively inexpensive and easily machined. They have been used to fabricate
the vast majority of flexure-guided nanopositioners [17, 36, 49, 71, 100]. Aluminum
has a high CTE, as stated in Table 3.2. Thus, it is not suitable for applications
where time and temperature changes are the important factors. However, if the entire
measurement can be completed in a short timeframe (i.e., in seconds or minutes)
with adequate temperature control system, in a laboratory environment, aluminum
alloys are deemed the most practical design material.

Table 3.2 A comparison of different material properties [15]

CTE E=
Material E (GPa) (g/cm3 ) (ppm/K) (GPa.cm3 /g)
Aluminum alloy (6061) 70 2:71 23:6 25:8
Aluminum alloy (7075) 72 2:81 23:6 26
Stainless steel 190 7:92 17:3 24
Titanium alloy (Grade 5) 106–114 4:42 8:8 24–26
Invar alloy (Invar 36) 148 8:1 1:3 18:4
Super Invar 148 8:1 0:3 18:4
90 Y.K. Yong and K.K. Leang

Stainless steel is often used for fabricating the rigid base due to its high
Young’s modulus. Titanium alloys have a similar E= value compared to aluminum
alloys. They have a much lower CTE value. Thus, titanium alloys are suitable for
applications where positioning accuracy needs to be maintained over a long period
of time under varying temperature conditions. However, titanium is not always
preferred over aluminum, because it is a more difficult material to machine [15].
Invar, a nickel–iron alloy, which exhibits even smaller values of CTE com-
pared to titanium, can be used for applications with extreme thermal constraints.
However, Invar alloys are considerably more expensive and somewhat difficult to
machine [76]. These alloys are highly magnetic due to their large iron and nickel
content [13]. Therefore, they are not suitable for applications where magnetic
characteristic is a consideration.

3.4.3.1 Thermal Drift

Thermal expansion of the material from which a mechanism is made of can cause
significant drift in its motion. Material expansion is inversely proportional to the
melting point of a material [14]. The change of length (from l0 to lf ) of a solid
material for a change in temperature (from T0 to Tf ) is given by

lf  l 0
D ˇ.Tf  T0 /; (3.57)
l0

where ˇ is the CTE and has units of (ı C)1 or K1 . For nanoscale motions, thermal
effects cannot be ignored [58]. Careful material selection and good mechanical
design are effective methods of minimizing thermal effects. For example, the CTE
for aluminum is 23  106 =ı C, while for Super Invar alloy is only 0:3  106 =ı C,
over 70-times lower. Effective practices also include carefully matching the stage
material with the materials of surround support structures. Also, materials with high
thermal conductivity quickly reach thermal equilibrium, thus minimizing transient
distortion during thermal expansion.

3.4.4 Manufacturing Techniques

The appropriate manufacturing technique for a given design depends on the scale
of the positioning stage and the material from which the mechanism is to be
machined. The positioning accuracy of a nanopositioner depends on the precision
with which its flexures are machined [66]. In particular, it is crucial to ensure all
cuts of flexures have parallel sides without tapering. For example, the flexures of
the nanopositioner in Fig. 3.16 have to be as identical as possible to avoid parasitic
motions (or out-of-plane motions), caused by uneven stiffness of flexures.
3 Mechanical Design of High-Speed Nanopositioning Systems 91

Fig. 3.16 Single axis

nanopositioner. Flexures
located at both sides of the
actuator have to be as
identical as possible to avoid
parasitic motions due to
asymmetric flexural stiffness

Standard milling, turning, and drilling techniques are best suited for metals, such
as aluminum, titanium, and steel. These techniques are best for feature sizes above
1 mm, and they can achieve tolerances on the order of ˙0.001 in ( ˙25 m).
Some of the high-precision machining methods that are often employed to construct
nanopositioning platforms are discussed here.

3.4.4.1 Wire Electrical-Discharge-Machining

Wire Electrical-Discharge-Machining (EDM) is one of the most frequently used

machining techniques for fabricating monolithic structures such as flexure-guided
nanopositioners, owing to its ability to machine complicated shapes with relatively
fast cutting speed and accuracy [27, 64]. This method of machining was developed
in the 1940s and is based on the erosion of a metallic material in the path of electrical
discharges that form an arc between a continuously moving electrode tool (wire)
and the workpiece. Dielectric fluid (deionized water) is ejected into the sparking
area to flush away the eroded particles [64, 81]. Wire diameters of approximately
100 m are often used. Traditional machining techniques are used to remove the
bulk of the stock before performing EDM. Dimensional accuracy on the order of
˙12 m can be achieved using the EDM process without severely tapering sides
and a good surface finish (1.95 Ra or less) [1]. Note that Ra is the arithmetic mean
of the absolute roughness profile ordinates [79].

3.4.4.2 Diamond Tool Machining

This is an ultra high-precision machining method which provides high geometrical

accuracy, good surface quality, capability to machine complicated profiles and can
be used to cut very hard materials [75]. However, this machining process can be
labor-intensive, leading to high fabrication cost.
92 Y.K. Yong and K.K. Leang

3.4.4.3 Laser Machining

Laser machining operations have large production rates. Therefore, they are used
extensively in mass-production of industrial apparatus. The laser machining tech-
nique operates on the simple principle of heating and melting of workpieces.
However, there are possible defects which could affect the performance of a
nanopositioner. For example, cuts may not have parallel sides, a workpiece may
have recast layers, i.e., materials on both sides of the cut may re-solidify after
cooling, leading to different properties for these layers, and cracks may appear on
surface of a workpiece due to poor surface finishing [19].

3.5 Design Example of High-Speed Lateral Nanopositioners

3.5.1 Parallel-Kinematic Nanopositioner

This section presents a design example of a high-speed parallel-kinematic nanopo-

sitioner as depicted in Fig. 3.17. Due to its mechanical symmetry in xy-plane,
the nanopositioner can be operated in both raster and non-raster scan modes
[51, 102, 106]. A piezoelectric stack actuator is used to drive each motion axis of
the nanopositioner. Left-spring or beam-like flexures are used to guide the sample
stage along the direction of motion of the actuator, while restraining other parasitic
motions.
To preload the x and y piezoelectric stacks, weights are first applied to deform
the flexure elastically, allowing a piezoelectric stack actuator to be placed in its
designated position. Weights are then released to restore the original position of the
flexures. The procedure results in the application of the necessary preload to the
piezoelectric stack.

Fig. 3.17 A
parallel-kinematic
xy-nanopositioning stage
3 Mechanical Design of High-Speed Nanopositioning Systems 93

3.5.1.1 Effective Stiffness and Displacement Estimations

A key design specification for a high-speed AFM nanopositioner is a very high

first mechanical resonance frequency along the actuation direction [35, 100]. This
requirement, however, often leads to a reduction in the travel range of the stage.
This is due to the inevitable increase in the structural stiffness when
p the resonance
frequency is increased. This trade-off is best described by !n D k=m, where k and
m represent the effective stiffness and mass of the stage, respectively, and !n is the
resonance frequency of the mechanical system. The following study of the effect of
flexure dimensions on the nanopositioner’s resonance frequency and displacement
was instrumental in reaching a compromise between the speed and travel range of
the stage.
The stiffness of individual flexure was derived using strain energy method [9] and
Castigliano’s second theorem [44, 47]. These methods have been commonly used
in flexure design literature [35, 46, 95, 96]. The in-plane displacement of a beam
flexure can be defined as,
2 3 2 32 3
ux Cx;Fx 0 0 Fx
4 uy 5 D 4 0 Cy;Fy Cy;Mz 5 4 Fy 5 ; (3.58)
z 0 Cz ;Fy Cz ;Mz Mz

where the displacements and loads are as shown in Fig. 3.6. The compliance values
are derived in [35], as Cx;Fx D L=AE, Cy;Fy D L=3EI C ˛L=GA, Cy;Mz D Cz ;Fy D
L2 =2EI, Cz ;Mz D L=EI. The corresponding notation is described in Table 3.3.
For a fixed-guided beam, the resultant moment Mz at the guided end of the beam
due to the applied Fy at that point is Mz D Fy L=2. Note that the proposed flexures
used in this work are short flexures. Therefore, the shear-induced deformation term
˛L=GA in Cy;Fy should not be ignored. Inclusion of the shear term is known to result
in different deformations in flexures [99]. From Eq. (3.58), the displacement of the
flexure along the y-direction is obtained as:

Table 3.3 Notation

Symbol Definition
t Beam thickness
L Beam length
h Beam height
A Cross-sectional area, A D ht
E Young’s modulus
v Poisson’s ratio
G Shear modulus, G D E= Œ2 .1 C v/
I Second moment of inertia about the z-axis, I D ht3 =12
J Polar moment of inertia, J D ht h2 C t2 =12
˛ Shear correction factor
94 Y.K. Yong and K.K. Leang

uy D Cy;Fy Fy C Cy;Mz Mz
 3   
L ˛L Fy L3
D Fy C  : (3.59)
3EI GA 2 2EI

Substituting I D ht3 =12 and A D ht into Eq. (3.59), and rearranging the equation,
the stiffness of the flexure along the y-direction is determined as:
"  3  #1
Fy 1 L ˛ L
ky D D C : (3.60)
uy Eh t Gh t

Since we know E, G, and h, the parameter that determines the flexure stiffness is L=t.
Flexures of the xy-nanopositioner in Fig. 3.17 are arranged in parallel. Therefore, the
general stiffness equation of the xy-nanopositioner along their direction of actuation
is ks D N ky , where N is the number of flexures. The maximum displacement of
the nanopositioner Ds is dependent on the piezoelectric stack actuator and it is
estimated as:
 
kp
Ds D l ; (3.61)
kp C ks

where l is the maximum stroke of the unloaded piezoelectric stack actuator and kp
is the stiffness of the piezoelectric actuator.

3.5.1.2 Resonance Frequency Estimation

The lateral resonance frequencies along the direction of actuation are estimated
using the algebraic eigenvalue method [30]. Assuming the system is a two-degree-
of-freedom undamped system as shown in Fig. 3.18, we may write

MRx C Kx D 0; (3.62)

Fig. 3.18 Each actuation axis of the xy-nanopositioner is modeled as a two-degree-of-freedom

undamped system. mp and kp are the mass and stiffness of the piezoelectric stack actuator. ms and
ks are the mass and stiffness of the nanopositioning nanopositioner
3 Mechanical Design of High-Speed Nanopositioning Systems 95

   
mp 0 kp C ks ks
where M D , and K D . mp and kp are the mass and
0 ms ks ks
stiffness of the piezoelectric stack actuator. Also, ms and ks are the mass and stiffness
of the nanopositioner. Substituting x D M1=2 q into Eq. (3.62) and multiplying the
resulting equation by M1=2 yields

R C M1=2 KM1=2 q.t/ D 0:

Iq.t/ (3.63)

Let KQ D M1=2 KM1=2 , by substituting q.t/ D vej!t , where v is a constant vector,

into Eq. (3.63) we obtain

Q D ! 2 v:
Kv (3.64)

If we replace ! 2 with  in Eq. (3.64), the solutionspfor  are the eigenvalues

p of
Q The resonance frequencies are f1 D 1 =2 and f2 D 2 =2 .
the matrix K.
Note that the smaller of the two frequencies f1 and f2 will be the first translational
mode of the system and is denoted as fres .

3.5.1.3 Parametric Study

In this section, a parametric study of the xy-nanopositioner design is presented. In

order to have a quantitative analysis of the xy-nanopositioner design, the relationship
among N, L=t, ks , Ds , and fres is plotted in Fig. 3.19. The number of flexures, N is
stepped from 1 to 20 and L=t is varied from 3 to 15.
As the ratio of L=t increases, the stiffness and resonance frequency decrease
while the displacement increases. This is an unavoidable trade-off between stiff-
ness/resonance frequency and displacement. Intuitively, the resonance frequency
increases when the number of flexures N increases. However, the additional mass of
each flexure lowers the resonance frequency. For example, as shown in Fig. 3.19, at
L=t D 6 and N D 10, fres is increased to 12 kHz, which is a threefold increment in
resonance frequency compared to that of a single flexure (N D 1). However, when
N is increased to 20, fres is improved only by a factor of four at the expense of the
displacement range being reduced by 20 %. Therefore, the resonance frequency of
the nanopositioner may not be increased significantly by introducing more flexures
into the design.
For the xy-nanopositioner design, Ks was set to approximately 10 N/m to
ensure that adequate preload was applied to the piezoelectric stack actuator.
To achieve approximately 10 kHz resonance frequency without sacrificing too much
displacement, L=t D 9 and N D 12 were selected in the design.
96 Y.K. Yong and K.K. Leang

Fig. 3.19 Displacement Ds ,

stiffness Ks , and resonance
frequency fres plots of the
xy-nanopositioner in relation
to the ratio L=t and the
number of flexure N

3.5.1.4 Finite-Element-Analysis

The analytical equations derived in Sects. 3.5.1.1 and 3.5.1.2 were based on a single
axis mass-spring model. They serve as a useful tool for initial parametric design
study of the nanopositioner. In this section, a finite-element (FE) model of the
xy-nanopositioner is presented. The FE simulations were used to obtain a better
estimation of the stiffness and resonance frequencies of the nanopositioner.
3 Mechanical Design of High-Speed Nanopositioning Systems 97

Fig. 3.20 FE simulated first mechanical resonance modes along the actuation direction of the
xy-nanopositioner

Meshing was first generated using ANSYS “auto-mesh” tool. Meshing near the
flexure was then refined manually to obtain more precise results. The material of
the nanopositioner is aluminum alloy Al7075 with Young’s modulus of 72 GPa. For
modal analysis, motions of the base, in which the xy-nanopositioner was attached,
were fixed during simulation. The simulated first resonant frequencies of the stage
along the x and y axes both are at 12.7 kHz as illustrated in Fig. 3.20.
For stiffness analysis, a point force was applied to each actuation direction of
the nanopositioner and the corresponding displacement was recorded. The stiffness
is calculated as force over displacement. The simulated stiffness are 13.1 N/m for
both lateral axes.

3.5.1.5 Piezoelectric Stack Actuators

Driving piezoelectric stack actuators at high frequencies amounts to a significant

challenge. It is well known that the larger the capacitance of a piezoelectric stack
actuator, the larger the current required to drive the actuator at high speed [38].
The required current may exceed the current limit of the driving amplifier. Thus,
the piezoelectric stack actuators must be chosen carefully to ensure their suitability
for the specific system. The piezoelectric stack actuators selected to drive the
stage along the x and y directions are Noliac SCMAP07. These actuators have a
capacitance of 380 nF. Table 3.4 contains detailed specifications of these actuators.
Three PiezoDrive PDL200 amplifiers were used to drive the piezos. The maximum
operating frequency of driving a triangular signal can be estimated as f 3 dB D
0:1=190 C [59]. With the setup described above, it is estimated that the fast lateral
axis of the nanopositioner can be driven with a triangular signal up to 1.4 kHz, at
the full voltage range of 200 V, without exceeding the current limit of the amplifier.
98 Y.K. Yong and K.K. Leang

Table 3.4 Specifications of the piezoelectric stack actuators

Dimension Capa. Stroke @ 200 V Stiff. Res. freq.
Piezo (mm) (nF) (m) (N/m) (kHz)
X, Y 5510 380 11.8 283 220

Fig. 3.21 Displacements and cross-coupling motions of the xy-nanopositioner

3.5.1.6 Experimental Outcomes

To evaluate the displacement range and coupling between various axes of

the nanopositioner, a 100-Hz sine-wave of 200 V was applied to each of the
piezoelectric stack actuators. Displacements and cross-coupling motions of the
nanopositioner were measured using the ADE Technologies 8810 capacitive sensors
and results are plotted in Fig. 3.21. The measured full displacement range of the x
and y axes are 6.5 m and 6.6 m, respectively. When the x-axis is driven by the
200-V sine-wave, the induced cross-coupling motions at the y and z axes are 2.5 %
(31.8 dB) and 1.58 % (36 dB) of the x displacement, respectively. When the y
axis is driven, it induces 3.1 % (29.9 dB) and 0.53 % (45.5 dB) in the x and z
displacement, respectively. Although these couplings are quite small, they can be
made even smaller by using a feedback control technique, such as that reported
in [101].
Frequency responses of the x- and y-axes of the nanopositioner were mea-
sured using a dual channel spectrum analyzer (HP 35670A). Figure 3.22 plots
the frequency response functions of the two axes. The first dominant mode for the
two lateral axes are 13.7 kHz and 11.6 kHz, respectively. The measured resonance
frequencies are within the 10 % range of those predicted by finite-element analysis.
This is most likely due to the machining errors when the nanopositioner was
being manufactured, and the variations in stiffness of piezoelectric stack actuators.
Nevertheless, the measured resonances are in reasonable agreement with the FE
simulations.
3 Mechanical Design of High-Speed Nanopositioning Systems 99

Fig. 3.22 Measured frequency responses of the x- and y-axis. The first dominant peak are at
13.7 kHz and 11.6 kHz, respectively

a b

Sample platform
Sample platform
ky x
cx ky
cy
y kx
cy
kx y
cx x

Fig. 3.23 Simplified models for two-axis scanning: (a) serial- and (b) parallel-kinematic config-
uration. The spring and damping constants include the effects of the piezoactuators and added
flexures in each direction

3.5.2 Serial-Kinematic Nanopositioner

The design of a high-speed serial-kinematic nanopositioner is described in this

section as an illustrative example. It is pointed out that for scanning in two
directions, there are two basic configurations: serial- and parallel-kinematics as
illustrated in Fig. 3.23. In a serial-kinematic system, for example, the design used by
several commercial vendors of scanning stages and in [4, 35], there is exactly one
actuator (and sensor) for each degree of freedom (see Fig. 3.23a). One disadvantage
of this design is the inability to measure (and correct for) parasitic motion such
100 Y.K. Yong and K.K. Leang

as runout or guiding error. Although the serial configuration is simple to design,

a penalty is that high resonance frequency can only be achieved in one axis. But,
this also implying that the high-speed axis requires high power and wide bandwidth
performance, reducing overall cost compared to a parallel-kinematic design.
In the simplified model shown in Fig. 3.23, the effective stiffnesses and damping
effect for both the serial- and parallel-kinematic configurations include the flexures
and the actuators along each direction. To achieve high resonance frequencies, the
effective stiffnesses should be as high as possible while achieving the desired range
of motion. The effects of inertial force generated by the sample platform during
scanning must also be taken into account. The flexures must provide enough preload
to avoid exposing the stack actuator to damaging tensional forces.
In the design of a serial-kinematic nanopositioner, while the resonance frequency
of the fast axis is of primary concern, the slow-axis resonance frequency can
essentially be ignored. For example, the scan rate of the slow axis is one-hundredth
the scan rate of the fast axis when acquiring a 100  100 pixel AFM image.
Therefore, the fast-scan axis can be designed independently without any significant
consideration for performance implications on the slow-scan axis. As previously
stated, for scanning-type applications, one lateral axis operates much faster than the
other, and thus the serial-kinematic configuration is well-suited for these types of
applications [6, 38].
For the design presented below, the objective is to create a nanopositioner for
high-bandwidth applications that include video-rate scanning probe microscopy.
The proposed design offers approximately 9  9  1 m range of motion with kHz
bandwidth. Vertically-stiff, double-hinged serial flexures are employed to guide the
motion of the sample platform to minimize parasitic motion (runout) and off-axis
effects (refer to example in Fig. 3.8).

3.5.2.1 Flexure Design for Lateral Positioning

The basic layout for a serial-kinematic design is shown in Fig. 3.24, where the high-
speed (x-axis) stage is nested inside of the low-speed (y-axis) stage. The stage body
is typically manufactured from 7075 aluminum or Invar using the wire EDM process
to create a monolithic platform. The sample platform is located on the x-stage body,
and vertical motion is achieved with a piezo-stack actuator embedded into the x-
stage body [6]. Compliant flexures with improved vertical stiffness to minimize out-
of-plane motion are used to guide the motion of the sample platform. The flexures
are strategically-placed to minimize the sample platform’s tendency to rotate (x , y ,
z ) at high frequencies. Also, the stage is designed to ensure that the first resonance
in all three axes are axial (piston) modes, rather than off-axis modes which can
severely limit performance.
q ui (i D x; y; z), the single DOF mechanical resonance
For translational motion,
is given by fui ;0 D 21 mkii , where mi and ki are the effective translational mass
and stiffness, respectively. Likewise
q for rotational motion, i (i D x; y; z), the
1 ki
first resonance is fi ;0 D 2 Ji
, where Ji and ki are the effective mass moment
3 Mechanical Design of High-Speed Nanopositioning Systems 101

Fig. 3.24 A serial-kinematic

x-axis piezo Sample platform
nanopositioner (top view),
showing the lateral stages and
(high-speed stage) (with z-axis stage)
details of the sample platform
and z-stage

y-axis piezo
(low-speed stage)
x

d kfx

mx ux
b
Top kpx
a Jz
e
y θz
x
kfθz + kfθz
Jy
f
c θy
z, θz
kfθy + kfθy
Side
z g
y, θy x, θx x mz uz
kpz kfz

Fig. 3.25 Generic flexure-guided stage simplified to single degree-of-freedom systems modeling
the dominant modes

of inertia and rotational stiffness, respectively. To insure that actuation modes

occur before the out-of-plane modes, the strategy taken is to optimize the stage
geometry and flexure configuration so that the out-of-plane stiffness-to-mass ratios
(kz =mz , ky =Jy , kz =Jz ) are higher than the actuation stiffness-to-mass ratio kx =mx .
Figure 3.25 shows the simplification of a high-speed x-stage into single DOF
systems to model four of the dominating resonance modes. The top and side views
are broken down to show the effective springs and masses affecting the body for (d)
actuation ux , (e) and (f) rotation z and y , and (g) vertical uz modes. Damping is
omitted for convenience.
102 Y.K. Yong and K.K. Leang

The vertical stiffness of the x- and y-stages is increased by (1) increasing the num-
ber of flexures, (2) utilizing shorter (effective length) flexures, and (3) converting the
flexures from constant rectangular cross-section beam flexures to a serial-compliant
double-hinged flexure with a “rigid” center connecting link (see Fig. 3.8).
The stiffness of a flexure is defined as the ratio of a load F and the resulting
displacement u. The displacements and loads are: translational displacement ui ,
rotational displacement i , translational force Fi acting on a point in the i direction,
and moment Mi (torque T) acting about the i axis (i ), respectively, where i D x; y; z.
Figure 3.6 illustrates the corresponding directions of the displacements and loads
acting on the free end of a fixed/free cantilever beam which models a beam
flexure. The in- and out-of-plane compliances for a fixed/free beam is derived
using Castigliano’s second theorem [16, 44, 84]. The compliance equations are then
used to derive equations for the actuation and vertical stiffness ki of a fixed/roller
guided beam shown in Fig. 3.26(a1)–(a3). It is pointed out that the fillet radius
is considerably smaller compared to the flexure length and therefore has minimal
effect on the flexure stiffness. For this reason, to simplify the flexure stiffness
equations in this initial analysis, the compliance equations are derived for a beam
with a constant cross-sectional thickness.

(a1) (b1) y
y x Rigid link
x

uy
Flexure hinges
Fy

(a2) (b2)
z
x

Fy

uz (b3)
Mz1
θz1 1
Fz
uy1
(a3)
Fy1
Mj

Fi

Fig. 3.26 Corner-filleted and center-thickened flexures showing loads and deformations: (a1) top
and (a3) side views showing displacement caused by force Fi , for i D x; y, in a fixed/guided
end configuration, and (a3) loads acting on the free end of a fixed/free beam for a corner-filleted.
(b1) top view, (b2) top view with applied load, and (b3) expanded view of corner-filleted flexure
hinge
3 Mechanical Design of High-Speed Nanopositioning Systems 103

For a fixed/free beam of rectangular cross section the total strain energy is

U D Uaxial C Utorsion C Ubending C Ushear

Z L 2 
F T2 M2 ˛V 2
D C C C dx; (3.65)
0 2AE 2GJ 2EI 2GA

where L is the beam length, A is the cross-sectional area of the beam, h is the height,
t is the thickness, E is the Young’s modulus, G D 2.1C/ E
is the shear modulus, 
h  4
i
is the Poisson’s ratio, J D ht3 13  0:21 ht 1  12h
t
4 is the approximate torsional
3
moment of inertia [107], I D ht12 is the second moment of inertia about the vertical
z axis, V is the shear force, and ˛ is the shape factor for the cross section used in the
shear equation (for a rectangular cross section ˛ D 6=5) [16, 60, 107].
Applying Castigliano’s second theorem, the displacement of a point in a given
direction ui , i is the partial derivative of the total strain energy with respect to the
applied force, i.e.:

@U @U
ui D I i D : (3.66)
@Fi @Mi

From here the compliance is simply found by dividing the displacement by the
applied load, i.e.:

ui i
Cui ;Fj D I Ci ;Mj D : (3.67)
Fj Mj

For example, the compliance of the rectangular cross section fixed-free beam in
Fig. 3.6 due to a point load in the y direction starts with the total strain energy
Z Z
L
M.x/2 L
˛V.x/2
UD dx C dx; (3.68)
0 2EI.x/ 0 2GA.x/

where A.x/ and I.x/ are constant. The coordinate system is placed on the free end
of the flexure as shown in [44] where the shear is V.x/ D Fy and moment is M.x/ D
Fy x. The total strain energy for the applied load is
Z Z
Fy2 L ˛Fy2 L Fy2 L3 ˛Fy2 L
UD x2 dx C dx D C : (3.69)
2EI 0 2GA 0 6EI 2GA

Therefore, the resultant displacement is

@U Fy L3 ˛LFy
uy D D C ; (3.70)
@Fy 3EI GA
104 Y.K. Yong and K.K. Leang

and the compliance is

uy L3 ˛L
C22 D D C : (3.71)
Fy 3EI GA

The compliances are then used to form the compliance matrix C which is
T
defined as the ratio of the displacement U D x y z z y x for a given load
T
L D Fx Fy Mz Fz My Mx , hence the displacement vector is
8 9 2 38 9
ˆ
ˆ ux >
> C11 0 0 0 0 0 ˆ
ˆ Fx >
>
ˆ
ˆu > > 6 0 7 ˆ
ˆ >
>
ˆ
ˆ >
y> 6 C22 C23 0 0 0 7ˆ ˆ F >
>
ˆ
< > = 6 ˆ
7< >
y
=
z 6 0 C23 C33 0 0 0 7 Mz
D6 7 : (3.72)
ˆ
ˆ u >
z> 6 0 0 0 C44 C45 0 7ˆ ˆ Fz >
>
ˆ
ˆ >
> 6 7ˆ >
ˆ 
ˆ y> >
>
4 0 0 0 C45 C55 0 5ˆ ˆ >
ˆ My >
>
:̂ ; :̂ > ;
x 0 0 0 0 0 C66 Mx

For a constant cross section fixed/free beam the compliances are C11 D AE L
,
3 ˛L 2 4L 3 ˛L 6L 2 12L
C22 D 3EI C GA , C23 D 2EI , C33 D EI , C44 D Eh3 t C GA , C45 D Eh3 t ,C55 D Eh3 t ,
L L L

and C66 D GJ L
. For a long slender beam, shear strain has little effect and therefore
can be ignored in C22 . For a short beam with a significant height-to-length aspect
ratio, such as the vertical displacement of the flexure shown in Fig. 3.26(a2), much
of the deflection is in shear, and therefore it can not be ignored.
The displacement vector equation presented above is used to solve for the
actuation stiffness ky and vertical stiffness kz of a fixed/guided flexure beam, i.e.,
Fi =ui D ki . Torsional stiffness is not investigated because the x rotational mode is
largely dependant upon the vertical flexure stiffness when the flexures are placed
at the corners of the stage body. Figure 3.26(a3) shows the applied load and the
expected deflection curve of the flexure in both the (a1) actuation direction and
(a2) vertical direction. The active load being applied to the flexure is the in-plane
force Fi . The resultant moment Mj D Fi L=2 is caused by the roller-guided end
constraint. Therefore, the flexure displacement in the actuation direction uy due to
the applied force Fy and moment Mz D Fy L=2 is

uy D C22 Fy C C23 Mz D C22 Fy  C23 Fy L=2

 3 
L ˛L L L2
D Fy C  : (3.73)
3EI Ght 2 2EI

Taking the ratio of the applied load to the displacement, the actuation stiffness
(neglecting shear) is
 3 1
Fy L ˛L 12EI
ky D D C Š 3 : (3.74)
y 12EI Ght L
3 Mechanical Design of High-Speed Nanopositioning Systems 105

Using the same method, the displacement of the flexure in the vertical direction uz is

uz D C44 Fz C C45 My D C44 Fz  C45 Fz L=2

 3 
4L ˛L L 6L2
D Fz C  : (3.75)
Eh3 t Ght 2 Eh3 t

Similarly, the vertical stiffness is

 1
L3 ˛L
kz D C : (3.76)
Eh3 t Ght

Because of the high aspect ratio in the vertical direction, shear can not be ignored.
It is pointed out that Eqs. (3.74) and (3.76) are used to design the flexures to meet
certain specifications, such as lateral and vertical stiffness.
Overall, the effective vertical stiffness can be improved to increase the out-of-
plane stiffness by (1) increasing the number of flexures n, (2) decreasing the flexure
length L, and (3) thickening the center section of a beam flexure to create a serial-
compliant double-hinged flexure. Additionally, flexure placement is important to
help increase rotational stiffness. Increasing the length (and width) of a stage and
placing flexures at the corners of the moving platform increase rotational stiffness of
the platform. However, the cost of increasing the size of the platform is increasing
overall mass, thus lowering the mechanical resonance.

3.5.2.2 Results: Comparison Between FEA and Experimental Results

The first three modes for the x and y stages were predicted using the frequency tool in
COSMOSWorks (FEA). It is assumed that the resonances of the y-stage would not
be excited by the dynamic motion of the inner nested x-stage. This allows the design
shown in Fig. 3.27a to be broken down into the low-speed y-stage [Fig. 3.27(b1)–
(b3)] and high-speed x-stage [Fig. 3.27(c1)–(c3)]. The boundary faces of each stage
(shown hatched) have a fixed boundary condition. All contacting components are
bonded together with compatible mesh. The meshing is done at “high quality”
with refined meshing at the flexure fillets and pivot points (0.25 mm minimum
element size on surfaces). The materials used and their corresponding mechanical
properties are as follows: aluminum: E D 72 GPa,  D 0:33, D 2700 kg/m3 ; steel:
E D 200 GPa,  D 0:28, D 7800 kg/m3 ; piezo-stack: E D 33:9 GPa,  D 0:30,
D 8000 kg/m3 ; and alumina: E D 300 GPa,  D 0:21, D 3960 kg/m3 , where
the modulus for the piezo-stack was calculated from the stiffness and blocking
force. The predicted first mechanical resonance for the y- and x-stage are 5.96 kHz
and 25.9 kHz, respectively, all of which are in the corresponding stage actuation
direction as preferred. Simulated FEA frequency response is done using the Linear
Dynamic (Harmonic) tool in COSMOSWorks. A constant amplitude sinusoidal
force is applied in the actuation direction at the corners of the piezoactuator/stage
106 Y.K. Yong and K.K. Leang

Low-speed
a y-stage
(b1)-(b3)

Vertical
High-speed z-stage
x-stage
(c1)-(c3)

b1 b2 b3
y z qz

f1;y 5.96 kHz f2;y 8.3 kHz f3;y 10.6 kHz

c1 c2 c3
x qy z=qy

f1;x 25.9 kHz f2;x 32.5 kHz f3;x 39.1 kHz

Fig. 3.27 Finite-element analysis results showing first four modes: (a) high-speed scanning stage;
(b1)–(b3) low-speed y-stage; (c1)–(c3) high-speed x-stage. Each stage section is designed to have
the first mechanical resonance to occur in the actuation direction

interfaces. The force generated is assumed proportional to the applied voltage.

A global modal damping ratio of 0.025 is applied to simulate the damping of
aluminum alloy and to produce a gain of 20 dB. Figure 3.28(a1), (b1) show the
predicted frequency response plots for the x- and y-axes with the resonant peaks
occurring at 25.9 kHz and 5.96 kHz, respectively. Overall, the FEA results and the
experimental measurements show good agreement.

3.6 Design Example of a Vertical Nanopositioner

There are increasing demands for vertical nanopositioning systems in applications

such as high-speed vertical stage for scanning probe microscopy [23, 34, 98],
objective focusing stages [24, 77], nanofabrication [85, 86], mask, and wafer surface
inspection [12]. Typically, flexures are used to provide preloads to piezoelectric
stack-driven nanopositioners [100, 105], and to minimize off-axis motions from the
actuation direction [34, 98].
3 Mechanical Design of High-Speed Nanopositioning Systems 107

High−speed (x−axis) Low−speed (y−axis)

(a1) FEA Predicted (b1) FEA Predicted
−50
−60
Mag. (dB)

Mag. (dB)
−80
−100
−100
−120
5
−150 5
1,000 10,000 10 1,000 10,000 10
(a2) Inductive sensor on y−stage (b2) Inductive sensor on stage base
0
−20
Mag. (dB)

Mag. (dB)
−40
−50
−60
−80
5
−100 5
1,000 10,000 10 1,000 10,000 10

Fig. 3.28 (a1)–(a2) Comparison of predicted and measured frequency response functions for the
high-speed stage (x-axis), (b1)–(b2) the low-speed stage (y-axis). The vertical dashed line is used
to compare the experimentally measured results to the FEA predicted first resonance peak

Fig. 3.29 Assembled and exploded views of the z-nanopositioner design

A design example of a high-speed vertical nanopositioner [98] is presented here.

The z-stage is mounted on the xy-stage presented in Sect. 3.5.1 (see Fig. 3.29)
to achieve the three translational DOFs (degree-of-freedoms) for high-speed
atomic force microscopy. The vertical nanopositioner utilizes a diaphragm flexure
108 Y.K. Yong and K.K. Leang

a b

c d

Fig. 3.30 Diaphragm flexure of the z-nanopositioner. (a) Diaphragm flexure. (b) Curved-beam
model. (c) Cross-sectional bending and torsional moments of the beam. (d) Beam element

in Fig. 3.30a to guide and preload the z-axis piezoelectric stack actuator. The
diaphragm flexure is constructed from eight curved beams. The curved beams are
fixed at one end and guided at the other. To estimate the effective stiffness of
the diaphragm flexure, a point load Fz is applied to the guided end. The total strain
energy of the curved beam flexure (neglecting shearing) is

ZL ZL
Mr2 M2
UD ds C ds; (3.77)
2EI 2GJ
0 0

where Mr and M as shown in Fig. 3.30c are the cross-sectional bending and
torsional moments, which vary with , E is the Young’s modulus, G D E=2 .1 C /
is the shear modulus,  is the Poisson’s ratio, I D h3 t=12 is the second moment
of inertia, J D ht3 C h3 t =12 is the polar moment of inertia, ds is the length of
a beam element, and L is the length of the beam. According to the Castigliano’s
second theorem, the deflection ız of the curved beam flexure can be calculated by
taking the partial derivative of U with respect to Fz , that is

ZL ZL
@U 1 @Mr 1 @M
ız D D Mr ds C M ds: (3.78)
@Fz EI @Fz GJ @Fz
0 0
3 Mechanical Design of High-Speed Nanopositioning Systems 109

The equilibrium equations for moment along the x and y directions are [18]

Mr cos   M sin  C Fz y D 0; (3.79)

Mr sin  C M cos  C Fz .R  x/ D 0: (3.80)

Solving for Mr and M ,

Mr D Fz y cos   Fz .R  x/ sin ; (3.81)

M D Fz y sin   Fz .R  x/ cos : (3.82)

The following procedure is employed in order to perform the integration of

Eq. (3.78) over the variable x in the Cartesian coordinates [18]. The curve of the
beam is a circular profile, therefore:

x2 C y2 D R2 ; (3.83)
p
y D R2  x2 ; (3.84)

where R is the radius of curvature of the beam. By differentiating Eq. (3.84), we

have
dy x
D p : (3.85)
dx R2  x2

As shown in Fig. 3.30c, the length of a beam element ds is

s
p  2
dy
ds D dx2 C dy2 D dx 1 C : (3.86)
dx

And sin ./ and cos ./ are expressed as

dx 1
sin  D Dr  2 ; (3.87)
ds dy
1 C dx

dy dy=dx
cos  D Dr  2 : (3.88)
ds dy
1 C dx

By substituting Eqs. (3.81), (3.82), (3.86)–(3.88) and @Mr =@Fz , @M =@Fz into
Eq. (3.78), and performing integration from x D R cos . =4/ (where =4 is the
angle of the beam) to x D R, we have
110 Y.K. Yong and K.K. Leang

 
2 T1 T2
ız D Fz R C ; (3.89)
EI GJ

where T1 and T2 are

ZR
1
T1 D  q dx; (3.90)
R2
R
p R2 x2
2

ZR
Rx
T2 D  q dx: (3.91)
2
R
p
.R C x/ R2Rx2
2

From Eq. (3.89), the stiffness of the curved beam is Kz D Fz =ız . The diaphragm
flexure consists of eight curved beams arranged in parallel. Therefore, the total
stiffness of the flexure is
8Fz
Kf D : (3.92)
ız

For the chosen flexure dimensions and material as documented in Table 3.5, Kf is
calculated to be 5.17 N/m.

3.6.1 Resonance Frequency

Resonance frequency of the single-mounted z-nanopositioner can be estimated by

modeling it as a two DOF spring-mass-damper system [69] as shown in Fig. 3.31.
In this model, the upper half of the piezoelectric stack actuator is lumped to the
effective mass of the flexure and sample, the lower half of the actuator is lumped
to the effective mass of the support structure, that is the xy-nanopositioner. Also, c1
and c2 represent the effective damping coefficients, and k1 and k2 are the effective
stiffnesses. We have m1 D mp =2Cms , m2 D mp =2Cmf , c1 D cs , c2 D cp Ccf , k1 D
ks , and k2 D kp C kf . The subscripts p, f , and s stand for piezoelectric stack actuator,
diaphragm flexure, and support structure, respectively. By using the lumped spring-
mass-damper system, two resonances (f1 and f3 ) and one anti-resonance (f2 ) can be
obtained [69], as
s
1 k1
f1  I
2 m1 C m2
3 Mechanical Design of High-Speed Nanopositioning Systems 111

Table 3.5 Dimensions and Diaphragm flexure Values

material properties of flexure
and piezoelectric stack Radius of curvature, R (mm) 2.99
actuator Thickness, t (mm) 0.38
Height, h (mm) 1.1
Effective mass, mf (g) 0.32
Stiffness, kf (N/m) 5.17a
Young’s modulus, E (GPa) 72
Poisson’s ratio,  0.33
Density, (kg/m3 ) 2770
Piezo Values
Dimension (mm) 336
Mass, mp (g) 0.41
Stiffness, kp (N/m) 65
Young’s modulus, Ep (GPa) 45
Poisson’s ratio, p 0.3
Capacitance, Cp (nF) 70
Travel range @ 200 V (m) 5.5
Support structure (xy-stage) Values
Mass, ms (g) 5.72
Vertical stiffness, ks (N/m) 81b
Density, (kg/m3 ) 7500
a
Obtained using Eq. (3.92)
b
Obtained using FE simulations

s
1 k1
f2  I
2 m1
s
1 .m1 C m2 / k2
f3  : (3.93)
2 m1 m2

Here, f1 is the vertical resonant mode of the xy-nanopositioner, and f3 is the resonant
mode of the preloaded piezoelectric stack actuator with flexures. The calculated
resonance frequencies f1 and f3 of the single-mounted z-nanopositioner design are
17.8 kHz and 61 kHz, respectively.

3.6.2 Finite-Element-Analysis

ANSYS was used to perform finite-element (FE) analysis of the z-nanopositioner.

The flexure is deformed using a piezoelectric stack actuator that generates the
requisite vertical displacement. A magnet is glued onto the flexure as illustrated in
Fig. 3.32. The magnet’s function is to hold a metal sample holder. It has a diameter
112 Y.K. Yong and K.K. Leang

Fig. 3.31 A 2-DOF

spring-mass model of the
z-nanopositioner where m, c,
and k are mass, damping
coefficient, and stiffness,
respectively. The subscripts p,
f , and s are referred to the
piezoelectric stack actuator,
diaphragm flexure, and
support structure (the
xy-nanopositioner)
respectively

Fig. 3.32 (Finite-element simulations. (a) First vertical resonant mode of the xy-nanopositioner:
19.7 kHz. (b) First vertical resonant mode of the z-nanopositioner: 65 kHz

of 5 mm and a thickness of 1 mm. The flexure is made of aluminum alloy (Al7075).

The piezoelectric actuator is a Noliac SCMAP06 stack actuator. Dimensions and
material properties of the actuator, which were used to construct the ANSYS model,
can be found in Table 3.5.
ANSYS simulations predicted that the first vertical resonance frequency of the
xy-nanopositioner is at 19.7 kHz. The vertical resonant mode of the z-nanopositioner
appears at 65 kHz (see Fig. 3.32). The simulated stiffness along the z-axis is
5.15 N/m. The analytical resonance frequencies and stiffness are in reasonable
agreement with the FE simulations.

3.6.3 Travel Range

To estimate travel range of the nanopositioner, we may use the formulation

L D Lo  kp = kf C kp [83, 100], where L is the displacement with external
spring load, Lo is the maximum nominal displacement of the piezoelectric
3 Mechanical Design of High-Speed Nanopositioning Systems 113

stack actuator without an external spring load, kf is the flexure stiffness, and kp
is the piezoelectric stack actuator stiffness. The estimated displacement of the
z-nanopositioner is 5.1 m.

3.6.4 Inertial Cancellation

The z-nanopositioner is mounted to a parallel-kinematics xy-nanopositioner pre-

sented in Sect. 3.5.1 and illustrated in Fig. 3.29. When the z-nanopositioner is only
comprised of a single-nanopositioner, as is the case in [97, 103] (and illustrated
in Fig. 3.33a), the fast movements of the piezoelectric stack actuator tend to
induce impulsive forces which propagate through, and excite the resonant modes
of the xy-nanopositioner [5, 104]. This resonant mode, which often lives at a
lower frequency than that of the vertical axis, in turn, causes vibrations in the z-
nanopositioner. The frequency response of the single-nanopositioner configuration
shows that the vertical resonant mode of the xy-nanopositioner is the dominant mode
with a phase drop of more than 45ı . This mode may need to be suppressed by
means of feedback control to avoid vibrations. In this configuration, bandwidth of
the vertical feedback loop is limited by the low resonance frequency of the xy-stage.
Equation (3.93) shows that the vertical resonant mode of the xy-stage, f1 , can be
increased by increasing the vertical effective stiffness ks of the xy-nanopositioner.
This can be achieved by adding more beam flexures to the xy-nanopositioner.
However, these flexures will increase the effective mass of the device which in turn
will reduce the resonance frequency. Furthermore, the additional flexures result in a
reduction in the xy travel range. To avoid this, a dual-mounted z-nanopositioner
configuration, illustrated in Fig. 3.33b, is used to counterbalance the impulsive
forces of the z-actuator [34, 35, 98]. This inertial counterbalance method avoids
the need to change ks through mechanical design. The bottom z-actuator cancels
the impulsive force exerted by the top actuator which, when the mass is balanced
properly, eliminates the first vertical resonant mode. Other inertial counterbalance
techniques can also be found in [5].
The effectiveness of the dual-mounted nanopositioner is sensitive to the weight
difference between the opposing piezoelectric stack actuators. This is illustrated in
Figs. 3.33(b2), (b3). If the sample and the dummy weights are perfectly balanced,
the vertical mode of the xy-nanopositioner will not be triggered. In this case, the
dominant mode is the vertical mode of the z-nanopositioner where, by design, is at
least 4 times higher than that of the xy-nanopositioner. Thus, in this case, bandwidth
of the vertical feedback loop is limited by the resonant mode of the z-nanopositioner.
However, in practice, it is rather difficult to perfectly balance the weights. It is more
likely that the dummy sample is slightly lighter or heavier than that of the sample
sitting on top of the z-nanopositioner (see Fig. 3.33b). Nevertheless, if the weights
are well-balanced, the effect of the xy-stage’s dynamics on the vertical positioning
stage will be negligible, possibly consisting of a very small peak in magnitude and
a phase variation of less than 45ı . In such a case, the dynamic mode of the xy-stage
will not be the dominant mode.
114 Y.K. Yong and K.K. Leang

b1

b2

b3

Fig. 3.33 Frequency responses of different z-nanopositioner configurations. (a) Single-nano-

positioner configuration without inertial cancellation. (b) Dual-mounted z-nanopositioner configu-
ration: (b1) Weights are perfectly balanced. (b2) Weight of the sample is slightly heavier than that
of the dummy. (b3) Weight of the dummy sample is slightly heavier than that of the sample
3 Mechanical Design of High-Speed Nanopositioning Systems 115

Fig. 3.34 Experimental setup of the nanopositioner. The z-nanopositioner is mounted to the xy-
nanopositioner

3.6.5 Experimental Results

The experimental setup is shown in Fig. 3.34. A Polytec PSV-300 laser scanning
vibrometer was used to measure the z-axis frequency response and travel range of
the nanopositioner. A 100-Hz, 200-V peak-to-peak sinusoidal signal, which was
amplified by a PiezoDrive PDL200 amplifier, was applied to drive the piezoelectric
stack actuator, and the measured full-range displacement of the z-nanopositioner is
5 m. The displacement is plotted in Fig. 3.35.
A pseudo-random input signal with a frequency in the range of 1–100 kHz was
applied to the piezoelectric stack actuator. The measured frequency response of
the z-nanopositioner are plotted in Fig. 3.35a. The weight of the dummy sample
is 0.03 g, which is approximately the weight of the sample. The frequency response
of the z-nanopositioner in Fig. 3.35a shows a pole-zero pattern, indicating that the
dummy mass is slightly lighter than the sample. The measured first vertical resonant
mode of the single-mounted z-nanopositioner is 17.1 kHz. Due to the additional
weights of the dual-mounted z-nanopositioner, the vertical resonant mode of the
xy-nanopositioner drops from 17.1 to 15 kHz. However, its resonant peak has a
magnitude of only 1.6 dB and a phase drop of less than 15ı . Due to its negligible
profile, this mode will not be the dominant mode that dictates the vertical feedback
bandwidth. As shown in the measured frequency response, the dominant mode in
the z-axis is the 60-kHz resonant peak. As a result, a controller may be designed to
suppress the 60-kHz resonant peak in order to obtain an even higher bandwidth.
116 Y.K. Yong and K.K. Leang

Fig. 3.35 (a) Measured frequency response. (b) Travel range of the nanopositioner

3.7 Conclusions

This chapter described the design process for high-speed flexure-guided nanoposi-
tioning stages with a focus on atomic force microscopy applications. It discussed
important design considerations which include balancing trade-off between range
and speed (resonance frequency), taking careful consideration of flexure designs,
actuator properties and drive electronics, material selections, manufacturing tech-
niques, and using FEA tools to optimize mechanical resonances. Design examples
of serial- and parallel-kinematic nanopositioning stages were presented. A design
example of a vertical nanopositioning stage with inertial counterbalance technique,
which passively suppresses the low vertical resonant mode of the system was also
presented.

Acknowledgements Y. K. Yong would like to acknowledge Prof. Reza Moheimani for many years
of support, guidance, and valuable feedback, her colleagues at the Laboratory for Dynamics and
Control of Nanosystems for providing meaningful discussions and creating an incredible research
environment, and the University of Newcastle for their commitment to a young researcher. She also
wishes to thank the Australian Research Council for supporting her research work. K.K. Leang
acknowledges A.J. Fleming for many years of collaboration and for providing the PiezoDrive
amplifiers used in the experiments. Additionally, he thanks the National Science Foundation for
supporting his research work.
3 Mechanical Design of High-Speed Nanopositioning Systems 117

References

1. N.M. Abbas, D.G. Solomon, M.F. Bahari, A review on current research trends in electrical
discharge machining (EDM). Int. J. Mach. Tool. Manuf. 47, 1214–1228 (2007)
2. C. Acar, A. Shkel, MEMS Vibratory Gyroscopes: Structural Approaches to Improve Robust-
ness, 1st edn. (Springer, New York, 2009). doi:10.1007/978-0-387-09536-3
3. T. Ando, N. Kodera, E. Takai, D. Maruyama, S. Kiwamu, A. Toda, A high-speed atomic
force microscope for studying biological macromolecules. Proc. Natl. Acad. Sci. U. S. A. 98,
12468–12472 (2001)
4. T. Ando, N. Kodera, T. Uchihashi, A. Miyagi, R. Nakakita, H. Yamashita, K. Matada, High-
speed atomic force microscope for capturing dynamic behavior of protein molecules at work.
e-J. Surf. Sci. Nanotechnol. 3, 384–392 (2005)
5. T. Ando, T. Uchihashi, T. Fukuma, High-speed atomic force microscopy for nano-
visualization of dynamic biomolecular processes. Prog. Surf. Sci. 83(7–9), 337–437 (2008)
6. T. Ando, T. Uchihashi, N. Kodera, D. Yamamoto, A. Miyagi, M. Taniguchi, H. Yamashita,
High-speed AFM and nano-visualization of biomolecular processes. Pflügers Arch. Eur. J.
Physiol. 456(1), 211–225 (2008)
7. APC International Ltd: Piezo-Mechanics: An Introduction. APC International Ltd., Pleasant
Gap (2003)
8. A. Bazaei, Y.K. Yong, S.O.R. Moheimani, High-speed Lissajous-scan atomic force
microscopy: scan pattern planning and control design issues. Rev. Sci. Instrum. 83(6), 063701
(10pp.) (2012)
9. F. Beer, E.R. Johnston, Mechanics of Materials (McGraw-Hill, New York, 1992)
10. G. Binnig, C.F. Quate, Atomic force microscope. Phys. Rev. Lett. 56(9), 930–933 (1986)
11. G. Binnig, H. Rohrer, The scanning tunneling microscope. Sci. Am. 253, 50–56 (1986)
12. G. Borionetti, A. Bazzali, R. Orizio, Atomic force microscopy: a powerful tool for surface
defect and morphology inspection in semiconductor industry. Eur. Phys. J. Appl. Phys. 27,
101–106 (2004). doi:10.1051/epjap:2004129
13. K.H.J. Buschow, F.R. de Boer, Physics of Magnetism and Magnetic Materials, (Kluwer
Academic Publishers, New York, 2004)
14. W.D. Callister, Materials Science and Engineering: An Introduction (Wiley, New York, 1994)
15. F. Cardarelli, Materials Handbook: A Concise Desktop Reference, 2nd edn. (Springer,
London, 2000)
16. R.R. Craig, Mechanics of Materials, 2nd edn. (Wiley, New York, 2000)
17. M.L. Culpepper, G. Anderson, Design of a low-cost nano-manipulator which utilizes a
monolithic, spatial compliant mechanism. Precis. Eng. 28(4), 469–482 (2004)
18. T. Dahlberg, Procedure to calculate deflections of curved beams. Int. J. Eng. Educ. 20(3),
503–513 (2004)
19. N.B. Dahotre, S.P. Harimkar, Laser Fabrication and Machining of Materials (Springer,
New York, 2008)
20. S. Devasia, E. Eleftheriou, S.O.R. Moheimani, A survey of control issues in nanopositioning.
IEEE Trans. Control Syst. Technol. 15, 802–823 (2007)
21. G.K. Fedder, Simulation of microelectromechanical systems. Ph.D. thesis, Electrical
Engineering and Computer Sciences, University of California at Berkeley (1994)
22. A.J. Fleming, A megahertz bandwidth dual-amplifier for driving piezoelectric actuators and
other highly capacitive loads. Rev. Sci. Instrum. 80(1–7), 104701 (2009)
23. A. Fleming, Dual-stage vertical feedback for high-speed scanning probe microscopy. IEEE
Trans. Control Syst. Technol. 19(1), 156–165 (2011)
24. A.J. Fleming, K.K. Leang, Design, Modeling and Control of Nanopositioning Systems
(Springer, London, 2014).
25. P. Gao, S. Swei, Z. Yuan, A new piezodriven precision micropositioning stage utilizing flexure
hinges. Nanotechnology 10, 394–398 (1999)
118 Y.K. Yong and K.K. Leang

26. D.C. Handley, T.F. Lu, Y.K. Yong, Workspace investigation of a 3dof compliant micro-motion
stage, in The Eighth International Conference on Control, Automation, Robotics and Vision,
ICARCV, pp. 1279–1284 (2004)
27. K.H. Ho, S.T. Newman, S. Rahimifard, R.D. Allen, State of the art in wire electrical discharge
machining (WEDM). Int. J. Mach. Tools Manuf. 44(12–13), 1247–1259 (2004)
28. L. Howell, Compliant Mechanisms. (Wiley, New York, 2001)
29. A.D.L. Humphris, B. Zhao, D. Catto, J.P. Howard-Knight, P. Kohli, J.K. Hobbs, High speed
nano-metrology. Rev. Sci. Instrum. 82(4), 043710–5 (2011)
30. D.J. Inman, Engineering Vibration. Prentice Hall Inc., Upper Saddle River (1996)
31. M.S. Johannes, D.G. Cole, R.L. Clark, Atomic force microscope based nanofabrication of
master pattern molds for use in soft lithography. Appl. Phys. Lett. 91(12), 123111 (2007)
32. M. Jouaneh, R. Yang, Modeling of flexure-hinge type lever mechanisms. Precis. Eng. 27,
407–418 (2003)
33. B.J. Kenton, Design, characterization, and control of a high-bandwidth serial-kinematic
nanopositioning stage for scanning probe microscopy applications. Master’s thesis, Univer-
sity of Nevada, Reno (2010)
34. B.J. Kenton, A.J. Fleming, K.K. Leang, Compact ultra-fast vertical nanopositioner for
improving scanning probe microscope scan speed. Rev. Sci. Instrum. 82(12), 123703 (2011)
35. B.J. Kenton, K.K. Leang, Design and control of a three-axis serial-kinematic high-bandwidth
nanopositioner. IEEE/ASME Trans. Mechatron. 17(2), 356–368 (2012)
36. D. Kim, D. Kang, J. Shim, I. Song, D. Gweon, Optimal design of a flexure hinge-based XYZ
atomic force microscopy scanner for minimizing abbe errors. Rev. Sci. Instrum. 76, 073706
(2005)
37. J.H. Kindt, G.E. Fantner, J.A. Cutroni, P.K. Hansma, Rigid design of fast scanning probe
microscopes using finite element analysis. Ultramicroscopy 100, 259–265 (2004)
38. K.K. Leang, A.J. Fleming, High-speed serial-kinematic AFM scanner: design and drive
considerations. Asian J. Control (Special issue on Advanced Control Methods for Scanning
Probe Microscopy Research and Techniques) 11(2), 144–153 (2009)
39. D.J. Leo, Engineering Analysis of Smart Material Systems. (Wiley, Hoboken, 2007)
40. D. Leonard, M. Krishnamurthy, C.M. Reaves, S.P. Denbaars, P.M. Petroff, Direct formation
of quantum-sized dots from uniform coherent islands of InGaAs on GaAs surfaces. Appl.
Phys. Lett. 63(23), 3203–3205 (1993)
41. Y. Li, Q. Xu, Modeling and performance evaluation of a flexure-based xy parallel microma-
nipulator. Mech. Mach. Theory 44(12), 2127–2152 (2009)
42. Y. Li, Q. Xu, Development and assessment of a novel decoupled xy parallel micropositioning
platform. IEEE/ASME Trans. Mechatron. 15(1), 125–135 (2010)
43. H.C. Liaw, B. Shirinzadeh, Neural network motion tracking control of piezo-actuated flexure-
based mechanisms for micro-/nanomanipulation. IEEE/ASME Trans. Mechatron. 14(5),
517–527 (2009). doi:10.1109/TMECH.2009.2005491
44. N. Lobontiu, Compliant Mechanisms: Design of Flexure Hinges (CRC Press, Boca Raton,
2003)
45. N. Lobontiu, E. Garcia, Analytical model of displacement amplification and stiffness opti-
mization for a class of flexure-based compliant mechanisms. Comput. Struct. 81, 2797–2810
(2003)
46. N. Lobontiu, E. Garcia, Two-axis flexure hinges with axially-collocated and symmetric
notches. Comput. Struct. 81, 1329–1341 (2003)
47. N. Lobontiu, J.S.N. Paine, E. Garcia, M. Goldfarb, Corner-filleted flexure hinges. Trans.
ASME J. Mech. Des. 123, 346–352 (2001)
48. N. Lobontiu, J.S.N. Paine, E. O’Malley, M. Samuelson, Parabolic and hyperbolic flexure
hinges: flexibility, motion precision and stress characterization based on compliance closed-
form equations. Precis. Eng. 26, 183–192 (2002)
49. T.F. Lu, D.C. Handley, Y.K. Yong, C. Eales, A three-DOF compliant micromotion stage with
flexure hinges. Ind. Robot 31(4), 355–361 (2004)
3 Mechanical Design of High-Speed Nanopositioning Systems 119

50. H.W. Ma, Y. Shao-Ming, L.Q. Wang, Z. Zhong, Analysis of the displacement amplification
ratio of bridge-type flexure hinge. Sens. Actuators A Phys. 132(2), 730–736 (2006)
51. I.A. Mahmood, S.O.R. Moheimani, Fast spiral-scan atomic force microscopy. Nanotechnol-
ogy 20(36), 365503 (2009)
52. I.A. Mahmood, S.O.R. Moheimani, B. Bhikkaji, A new scanning method for fast atomic force
microscopy. IEEE Trans. Nanotechnol. 10(2), 203–216 (2010)
53. M. Maroufi, A.G. Fowler, A. Bazaei, S.O.R. Moheimani, High-stroke silicon-on-insulator
MEMS nanopositioner: control design for non-raster scan AFM. Rev. Sci. Instrum. 86(2),
023705 (12pp) (2015)
54. R. Merry, N. de Kleijn, M. van de Molengraft, M. Steinbuch, Using a walking piezo actuator
to drive and control a high-precision stage. IEEE/ASME Trans. Mechatron. 14(1), 21–31
(2009)
55. S. Miyake, M. Wang, J. Kim, Silicon nanofabrication by atomic force microscopy-based
mechanical processing. J. Nanotechnol. 2014, 102404 (2014)
56. A. Mohammadi, A.G. Fowler, Y.K. Yong, S.O.R. Moheimani, A feedback controlled mems
nanopositioner for on-chip high-speed afm. IEEE J. Microelectromech. Syst. 23(3), 610–619
(2014)
57. S.O.R. Moheimani, Invited review article: accurate and fast nanopositioning with piezoelec-
tric tube scanners: Emerging trends and future challenges. Rev. Sci. Instrum. 79(7), 071101
(2008)
58. B. Mokaberi, A.A.G. Requicha, Compensation of scanner creep and hysteresis for AFM
nanomanipulation. IEEE Trans. Autom. Sci. Eng. 5(2), 197–206 (2008)
59. Newcastle Innovation Ltd., Industry Development Centre, University Drive, Callaghan NSW
2308, Australia: PiezoDrive PDL200 Manual, www.piezodrive.com
60. S.R. Park, A mathematical approach for analyzing ultra precision positioning system with
compliant mechanism. J. Mater. Res. Process. Technol. 164–165, 1584–1589 (2005)
61. J.M. Paros, L. Weisbord, How to design flexure hinges. Mach. Des. 37, 151–156 (1965)
62. J. Peng, X. Chen, Modeling of piezoelectric-driven stick-slip actuators. IEEE/ASME Trans.
Mechatron. 16(2), 394–399 (2011)
63. A.P. Pisano, Y.H. Cho, Mechanical design issues in laterally-driven microstructures. Sens.
Actuators A Phys. 23(1–3), 1060–1064 (1990). Proceedings of the 5th International
Conference on Solid-State Sensors and Actuators and Eurosensors {III}
64. A.B. Puri, B. Bhattacharyya, An analysis and optimisation of the geometrical inaccuracy due
to wire lag phenomenon. Int. J. Mach. Tools Manuf. 43, 151–159 (2003)
65. M. Rakotondrabe, Y. Haddab, P. Lutz, Development, modeling, and control of a micro-
/nanopositioning 2-dof stick-slip device. IEEE/ASME Trans. Mechatron. 14(6), 733–745
(2009)
66. J.W. Ryu, D.G. Gweon, Error analysis of a flexure hinge mechanism induced by machining
imperfection. Precis. Eng. 21, 83–89 (1997)
67. S. Salapaka, A. Sebastian, J.P. Cleveland, M.V. Salapaka, High bandwidth nano-positioner: A
robust control approach. Rev. Sci. Instrum. 73(9), 3232–3241 (2002)
68. R.B. Salazar, A. Shovsky, H. Schonherr, G.J. Vancso, Dip-pen nanolithography on
(bio)reactive monolayer and block-copolymer platforms: Deposition of lines of single
macromolecules. Small 2(11), 1274–1282 (2006)
69. G. Schitter, K.J. Åstrom, B. DeMartini, P.J. Thurner, K.L. Turner, P.K. Hansma, Design and
modeling of a high-speed AFM-scanner. IEEE Trans. Control Syst. Technol. 15(5), 906–915
(2007)
70. G. Schitter, P.J. Thurner, P.K. Hansma, Design and input-shaping control of a novel scanner
for high-speed atomic force microscopy. Mechatronics 18, 282–288 (2008)
71. F. Scire, E. Teague, Piezodriven 50-m range stage with subnanometer resolution. Rev. Sci.
Instrum. 49(12), 1735–1740 (1978)
72. See Asylum Research: http://asylumresearch.com for information on flexure-based nanopo-
sitioning platforms
120 Y.K. Yong and K.K. Leang

73. See Mad City Labs Inc: http://www.madcitylabs.com/ for information on flexure-based
nanopositioning applications
74. See Park Systems: www.parkafm.com for information on the use of flexure-based nanoposi-
tioning platforms in commercially available AFMs
75. E. Shamoto, T. Moriwaki, Ultraprecision diamond cutting of hardened steel by applying
elliptical vibration cutting. CIRP Ann. Manuf. Technol. 48(1), 441–444 (1999)
76. M. Shiga, Invar alloys. Curr. Opin. Solid State Mater. Sci. 1(3), 340–348 (1996).
doi:10.1016/S1359-0286(96)80023-4
77. V. Shilpiekandula, K. Youcef-Toumi, Dynamic modeling and performance trade-offs in
flexure-based positioning and alignment systems, Chapter 25, in Motion Control, ed. by F.
Casolo (InTech, Croatia, 2010), pp. 481–496
78. S.T. Smith, Flexures: Elements of Elastic Mechanisms. (Gordon and Breach, New York,
2000)
79. G.T. Smith, Cutting Tool Technology: Industrial Handbook. (Springer, London, 2008)
80. S.T. Smith, V.G. Badami, J.S. Dale, Y. Xu, Elliptical flexure hinges. Rev. Sci. Instrum. 68(3),
1474–1483 (1997)
81. T.A. Spedding, Z.Q. Wang, Study on modeling of wire EDM process. J. Mater. Process.
Technol. 69, 18–28 (1997)
82. D.J. Taatjes, A.S. Quinn, J.H. Rand, B.P. Jena, Atomic force microscopy: high resolution
dynamic imaging of cellular and molecular structure in health and disease. J. Cell. Physiol.
228(10), 1949–1955 (2013)
83. The World of Micro- and NanoPositioning (PI catalog). Physik Instrumente, Karlsruhe (2005)
84. S.P. Timoshenko, History of Strength of Materials. (McGraw-Hill Book Company, New York,
1953)
85. A.A. Tseng, Advancements and challenges in development of atomic force microscopy for
nanofabrication. Nano Today 6(5), 493–509 (2011)
86. A.A. Tseng, S. Jou, A. Notargiacomo, T. Chen, Recent developments in tip-based nanofabri-
cation and its roadmap. J. Nanosci. Nanotechnol. 8, 2167–2186 (2008)
87. Y.M. Tseytlin, Notch flexure hinges: an effective theory. Rev. Sci. Instrum. 73(9), 3363–3368
(2002)
88. T. Tuma, J. Lygeros, V. Kartik, A. Sebastian, A. Pantazi, High-speed multiresolution scanning
probe microscopy based on Lissajous scan trajectories. Nanotechnology 23(18), 185501
(2012)
89. S. Wadikhaye, Y.K. Yong, S.O.R. Moheimani, B. Bhikkaji, Control of a piezoelectrically
actuated high-speed serial kinematic AFM nanopositioner. Smart Mater. Struct. 22(2), 025030
(12pp.) (2014)
90. C. Wang, V.K. Yadavalli, Investigating biomolecular recognition at the cell surface using
atomic force microscopy. Micron 60(0), 5–17 (2014)
91. D.L. White, O.R. Wood, Novel alignment system for imprint lithography. Rev. Sci. Instrum.
18(6), 3552–3556 (2000)
92. K. Wiesauer, G. Springholz, Fabrication of semiconductor nanostructures by nanoindentation
of photoresist layers using atomic force microscopy. J. Appl. Phys. 88(12), 7289–7297 (2000)
93. P. Xu, Y. Jingjun, Z. Guanghua, B. Shusheng, Y. Zhiwei, Analysis of rotational precision
for an isosceles-trapezoidal flexural pivot. J. Mech. Des. 130(5), 052302 (9pp.) (2008).
doi:10.1115/1.2885507
94. Y.K. Yong, A. Fleming, Piezoelectric actuators with integrated high-voltage power electron-
ics. IEEE/ASME Trans. Mechatron. 20(2), 611–617 (2015)
95. Y.K. Yong, T.F. Lu, The effect of the accuracies of flexure hinge equations on the output
compliances of planar micro-motion stages. Mech. Mach. Theory 43, 347–363 (2008)
96. Y.K. Yong, T.F. Lu, Kinetostatic modeling of 3-RRR compliant micro-motion stages with
flexure hinges. Mech. Mach. Theory 44(6), 1156–1175 (2009)
97. Y.K. Yong, S.O.R. Moheimani, A compact XYZ scanner for fast atomic force microscopy
in constant force contact mode, in IEEE/ASME International Conference on Advanced
Intelligent Mechatronics, Montreal (2010)
3 Mechanical Design of High-Speed Nanopositioning Systems 121

98. Y.K. Yong, S.O.R. Moheimani, Design of an inertially counterbalanced Z-nanopositioner

for high-speed atomic force microscopy. IEEE/ASME Trans. Nanotechnol. 12(2), 137–145
(2013)
99. Y.K. Yong, T.F. Lu, D.C. Handley, Review of circular flexure hinge design equations and
derivation of empirical formulations. Precis. Eng. 32(2), 63–70 (2008)
100. Y.K. Yong, S. Aphale, S.O.R. Moheimani, Design, identification and control of a flexure-
based XY stage for fast nanoscale positioning. IEEE Trans. Nanotechnol. 8(1), 46–54 (2009)
101. Y.K. Yong, K. Liu, S.O.R. Moheimani, Reducing cross-coupling in a compliant XY nanoposi-
tioning stage for fast and accurate raster scanning. IEEE Trans. Control Syst. Technol. 18(5),
1172–1179 (2010)
102. Y.K. Yong, S.O.R. Moheimani, I.R. Petersen, High-speed cycloid-scan atomic force
microscopy. Nanotechnology 21(36), 365503 (2010)
103. Y.K. Yong, B. Bhikkaji, S.O.R. Moheimani, Analog control of a high-speed atomic force
microscope scanner, in IEEE/ASME International Conference on Advanced Intelligent
Mechatronics, Budapest (2011)
104. Y.K. Yong, S.O.R. Moheimani, B.J. Kenton, K.K. Leang, Invited review article: high-speed
flexure-guided nanopositioning: mechanical design and control issues. Rev. Sci. Instrum.
83(12), 121101 (2012)
105. Y.K. Yong, B. Bhikkaji, S.O.R. Moheimani, Design, modeling and FPAA-based control of a
high-speed atomic force microscope nanopositioner. IEEE/ASME Trans. Mechatron. 18(3),
1060–1071 (2013)
106. Y.K. Yong, A. Bazaei, S.O.R. Moheimani, Video-rate Lissajous-scan atomic force
microscopy. IEEE Trans. Nanotechnol. 13(1), 85–93 (2014)
107. W.C. Young, R.G. Budynas, Roark’s Formula for Stress and Strain, 7th edn. (McGraw-Hill,
New York, 2002)
Chapter 4
Parallel-Kinematic Nanopositioning Stages
Based on Roberts Mechanism

Sicong Wan and Qingsong Xu

Abstract This chapter presents the design of parallel-kinematic nanopositioning

stages with large workspace and low crosstalk. Based on compliant Roberts
mechanism, the design of multi-axis parallel compliant stage is synthesized. A new
XY nanopositioning stage is studied in detail. Pseudo-rigid-body model (PRBM) is
developed to build the quantitative models of the compliant Roberts mechanisms.
In addition, finite-element analysis (FEA) is carried out to validate its performance.
A prototype is fabricated and tested through experimental studies. Results show
that the XY stage delivers a work range larger than 12 mm in each axis. Moreover,
the parasitic motion of the stage in the non-working direction is less than 1.7 % of
the motion stroke. It indicates that the designed stage meets the requirements of a
large stroke and high precision linear guiding mechanism, which demonstrates the
feasibility of the proposed design ideas using the Roberts mechanism. Moreover,
a feedback control using PID control algorithm is implemented to demonstrate the
positioning performance of the developed XY stage. The reported ideas can also be
extended to the design and control of other types of micro-/nanopositioning systems.

4.1 Introduction

With the development of technology, the micropositioning and displacement tech-

nology has played an important role in the defense industry, microelectronics
engineering, aerospace technology and biological engineering, and other fields
[9, 12, 22, 28, 29, 32, 33]. It directly affects the whole machinery industry technology
and technique development, determines the quality of machining accuracy and
performance of products, and also affects the reliability and stability of the device.
Compliant mechanism can transmit motion and force by using the material’s
elastic deformation capability. As compared with rigid-body mechanism, compliant
mechanism eliminates the problems such as clearance, friction, and wears [8].
In the last few years, many scholars have proposed different flexible displacement

S. Wan • Q. Xu ()
Department of Electromechanical Engineering, Faculty of Science and Technology,
University of Macau, Avenida da Universidade, Taipa, Macau, China
e-mail: [email protected]; [email protected]

© Springer International Publishing Switzerland 2016 123

C. Ru et al. (eds.), Nanopositioning Technologies,
DOI 10.1007/978-3-319-23853-1_4
124 S. Wan and Q. Xu

mechanism. Generally, the compliant mechanism whose workspace is larger than

1 mm is known as a large-stroke compliant mechanism, while the compliant
mechanism whose workspace is less than 1 mm is called a small-stroke compliant
mechanism. According to the flexible hinge type, the notch-type flexible hinge
mechanism is difficult to obtain a large-stroke movement because of the limited
flexible deformation. However, according to the flexible structure types, stiffness
value of the flexible mechanism which consists of redundant symmetric structures
is very sensitive to the driving force. Therefore, it is also difficult to achieve a
large-stroke motion. Appropriately increasing the size of the flexible structure can
increase the workspace, but the installation space will limit the size of the stage.
On the other hand, because it is hard to make a precision micropositioning
system with both a large workspace and multi-degree of freedom at the same
time, the applications of compliant mechanism are restricted [13, 21, 27, 30]. For
instance, many applications require a compliant mechanism which has a larger
range of work. To name a few, the micro-coordinate measuring machines, adaptive
adjustment mechanism in astronomical telescope, MEMS processing systems,
binary optical processing equipment, and so on [6, 16]. Hence, the large-stroke
and multi-DOF micropositioning stage has drawn extensive attentions of some
recent research. For example, Tan et al. [18] proposed a multi-leaf configuration
of expanded-motion-range flexure hinges. Zhao et al. [34] designed a new large-
deflection pivot based on curved flexure element. Xu [25] designed a novel compact
long-stroke precision positioning stage based on flexible leaf springs. Trease [20]
proposed a large-displacement compliant joint, and Choi et al. [5] designed a large-
displacement precision XY positioning stage by using the cross strip flexure joints
and over-constrained mechanism.
According to previous work [2, 7, 15, 17, 19, 26], the motion accuracy of such
mechanisms is generally about 1 % of its workspace. In traditional design, there
is no strict compensation for the coupled displacement among axes. So, such
structures exhibit a large amount of coupling displacement. For example, Lin and
Lee [15] put forward a comprehensive flexible structure of a large-stroke compliant
mechanism, which is a flexible mechanism based on pseudo-symmetrical structure.
Simulation results show that its workspace is 2.34 mm. However, the amount of
coupling displacement in non-working direction is almost 390 m. That is, the
cross-axis coupling amounts to 16.6 %. Therefore, to achieve a precise positioning,
it is necessary to reduce the crosstalk value.
What’s more, it is challenging to devise a compliant micropositioning stage with
a stroke longer than 10 mm in each working axis. One reason lies in that the size
of the flexible component is the decisive factor which determines its workspace
directly [24]. It is not easy to devise an XY stage have both large workspace and
compact dimension simultaneously [24]. In recent research, a compact stage is
designed in [11], which has a dimension of 100  100 mm2 . Yet, the stage can
just provide a workspace of 2  2 mm2 . In contrast, the XY stage designed in [14]
produces a workspace of 20  20 mm2 , but it is contributed by a large physical
dimension of 432  432 mm2 .
4 Parallel-Kinematic Nanopositioning Stages Based on Roberts Mechanism 125

To this end, the conceptual design of a new decoupled XY micropositioning

stage is proposed in this research. It can deliver a large stroke and possesses a
relatively compact structure thanks to a two-layer design. Specifically, the Roberts
mechanism is introduced to design a linear guiding mechanism. Considering that a
single Roberts mechanism can only provide the linear motion of a single point, two
Roberts mechanisms working in parallel is employed to construct a flexible linear
guiding mechanism. By cascading two newly proposed linear compliant guiding
mechanisms, a symmetrical structure design yields a novel large-stroke compliant
parallel-kinematic XY micropositioning stage based on Roberts mechanism. To the
knowledge of the authors, no work has been conducted to devise a multi-axis
micropositioning stage using the Roberts mechanism in the literature.
The main contribution of this work is the conceptual design and verification
of a large-stroke parallel-kinematic compliant XY micropositioning stage based
on Roberts mechanism, which has not been attempted in previous literature.
The remaining parts of the paper are organized as follows. Section 4.2 presents
the conceptual design of the new XY stage. Analytical modeling of the XY stage
is conducted in Sect. 4.3. These models are verified by conducting finite-element
analysis (FEA) simulations in Sect. 4.4. Section 4.5 describes a fabricated prototype
and experimental investigations. Concluding remarks are summarized in Sect. 4.6.

4.2 Conceptual Design of Large-Range XY Compliant

Mechanism

In this section, the traditional Roberts mechanism is introduced. As a traditional

rigid-body mechanism, it can convert rotational motion into straight-line motion
approximately. Based on this mechanism, two novel compliant parallel linear
guiding mechanisms are proposed.

4.2.1 Background of Roberts Mechanism

Roberts mechanism was first proposed by Richard Roberts (1789–1864) [1]. Its
schematic is shown in Fig. 4.1a. When the link lengths are designed as R1 D R3
and the central blue triangle is isosceles (i.e., the two non R2 sides are of equal
length), the trajectory of point P will become an approximate straight line, as shown
in Fig. 4.1b.
126 S. Wan and Q. Xu

Fig. 4.1 Schematic diagram of the Roberts mechanism. (a) Traditional Roberts mechanism [23];
(b) structure of the Roberts mechanism

Fig. 4.2 (a) Flexure Roberts mechanism; (b) two new compliant Roberts linear guiding
mechanisms

4.2.2 Proposal of New Compliant Linear Guiding Mechanism

By replacing the conventional kinematic pairs of traditional rigid-body Roberts

mechanism with flexible leaf springs, a new kind of compliant linear guiding
mechanism is obtained, as shown in Fig. 4.2a. However, a single Roberts mechanism
can only provide a single point of linear motion. To overcome this issue, two Roberts
mechanisms connected in parallel can be employed to generate a compliant parallel
linear guiding mechanism. According to the joining ways, two different mechanisms
are obtained as shown in Fig. 4.2b.
4 Parallel-Kinematic Nanopositioning Stages Based on Roberts Mechanism 127

Fig. 4.3 A novel serial XY

flexure mechanism. (a) XY
mechanism; (b) lightweight
optimization process

4.2.3 Design of New Compliant Serial XY Linear Guiding

Mechanism

Based on the two new flexible linear guiding mechanisms, a novel XY mechanism
is designed as shown in Fig. 4.3a. It is observed that the two compliant Roberts
mechanisms are connected in series. In addition, the proposed design is a two-layer
structure, and it can yield decoupling motion in X axis and Y axis. The two-layer
design enables the generation of a relatively compact dimension design.
In order to reduce the weight and to improve the resonant frequency, a
lightweight optimization process is applied to the serial XY guiding mechanism
as illustrated in Fig. 4.3b. The resulted serial XY mechanism is employed in the
following design of an XY stage.

4.2.4 Design of New Compliant Parallel-Kinematic XY

Micropositioning Stage

As a conceptual design, four serial XY linear guiding mechanisms are used to

construct a parallel XY micropositioning stage, which owns a 4-PP (P stands for
prismatic joint) parallel mechanism. The CAD model of the overall structure for an
XY stage is shown in Fig. 4.4. This structure meets the requirement of the integrated
machining and easy assembly. In addition, the lightweight optimization process
reduces the mass of the platform while improves the dynamic performance of the
128 S. Wan and Q. Xu

Fig. 4.4 A new compliant parallel XY large-stroke micropositioning stage. (a) Top view;
(b) isotropic view; (c) side view

system. Moreover, the idea of symmetrical structure is also adopted in the stage
design. Such a symmetrical design can compensate for the coupling displacement
to some extent, as revealed in the later discussion.
The specific dimension design is introduced in the following section by estab-
lishing theoretical models.

4.3 Analytical Modeling of the XY Stage

4.3.1 PRBM of the Compliant Roberts Mechanism

As discussed above, the parallel-kinematic XY compliant stage consists of four

serial XY compliant Roberts mechanisms. For each single compliant Roberts
mechanism, the kinematic principle is illustrated in Fig. 4.5. When an external force
is applied to the coupler point in the extension direction, the two flexure beams are
deformed to generate an approximate straight-line motion at the coupler point.
To calculate the stiffness of a Roberts mechanism, the pseudo-rigid-body model
(PRBM) can be used. The concept of PRBM is to model the force-displacement
relationship of a compliant element using an equivalent rigid-body mechanism with
the compliance modeled as torsional springs around the joints. This allows the vast
knowledge of traditional mechanism kinematics to be applied to the design of a
compliant element.
4 Parallel-Kinematic Nanopositioning Stages Based on Roberts Mechanism 129

Fig. 4.5 Kinematic principle

of a flexible Roberts L3
mechanism

L2 b3
L4

COUPLER POINT d

The PRBM of a single beam is shown in Fig. 4.6a. In addition, the PRBM of
a single flexure Roberts mechanism is shown in Fig. 4.6b, where the two flexure
beams are replaced by four identical torsional springs and two rigid links [8]. The
overall stiffness for this group of flexure joints can be obtained through

F
KD ; (4.1)
x
where x and F are the displacement and applied force for the movable plate,
respectively. These two parameters are obtained by the following equations:
r3
x D r2 .cos 2  cos 20 / C .cos 3  1/ C b3 sin 3 (4.2)
2
4K Œ.2  h32 / 2 C .2h42  h32 / 4  .1 C h42  h32 / 3 
FD ; (4.3)
r2 sin 2 C h32 . r23 sin 3  b3 cos 3 /

where b3 is the perpendicular distance from the center of link 3 to the coupler point
(see Fig. 4.6b). ri , i , and i0 (i D 2; 3; 4) are the pseudo-rigid-body (PRB) length,
final angle, and initial angle of link i, respectively. The values for r2 and r3 are
defined as l, as shown in Fig. 4.6b. In addition, h32 , h42 , and K are the kinematic
coefficients defined by the following equations [3]:

ı3 !3 r2 sin.4  2 /
h32 D D D (4.4)
ı2 !2 r3 sin.3  4 /
ı3 !4 r2 sin.3  2 /
h42 D D D (4.5)
ı2 !2 r4 sin.3  4 /
Ech3
K D K‚ ; (4.6)
6l
130 S. Wan and Q. Xu

Fig. 4.6 The PRBM of compliant Roberts mechanism. (a) The PRBM of a single beam; (b) the
PRBM of a single Roberts mechanism; (c) RBM for the Roberts mechanism
4 Parallel-Kinematic Nanopositioning Stages Based on Roberts Mechanism 131

where is the characteristic radius factor and K‚ is the stiffness coefficient. For
this PRBM approximation, the values of and K‚ are set as 0.852 and 2.65,
respectively [10]. The terms 2 , 3 , and 4 are the differences of link angles
with respect to their initial positions.
In order to determine the required force for a given displacement, the instan-
taneous angles of the three links should be known. Given an input displacement,
all the angles can be found by solving (4.2) and the explicit four-bar equations
simultaneously [8]. Then, these angles can be substituted into (4.3) to find the
required force. The resulting stiffness K of the flexure joint is determined by (4.1).
For an XY stage design as shown in Fig. 4.4a, a stiffness model is derived as
shown in Fig. 4.7. Then, the stiffness of the XY stage seen at the input end can be
computed as

Kin1 D 8K: (4.7)

So, Fmax in one-sided direction can be calculated as follow.

Fmax  8K  Dmax : (4.8)

Fig. 4.7 Stiffness model of

the XY stage
132 S. Wan and Q. Xu

4.3.2 Dynamics Modeling

For a single compliant Roberts mechanism, four torsional springs (Kd1 D Kd2 D
Kd3 D Kd4 ) and two lumped masses of the two flexible links (lm2 D lm4 ) are added
in the PRBM, as shown in Fig. 4.6c [1, 2, 4, 8]. Considering that m2 D m4 , J2 D J4 ,
and r2 D r4 , lm2 can calculate as:

1
Ji D mi li (4.9)
12
m2 J2
lm2 D lm4 D C 2; (4.10)
4 r2

where m2 D m4 D l  c  h  AL  7075 with AL  7075 denoting the density of the

material.
The kinetic energy for the PRBM system can be expressed as follows [4]:

1
TD meB u2
2
   
1 2 1 2 1 2 1 2 1 2 1 2
D lm2 vB C lm4 vC C m3 vm3 C J3 !3 C mb3 vb3 C Jb3 !3 ;
2 2 2 2 2 2
(4.11)

where meB is the equivalent mass of a single compliant Roberts mechanism, mi and
Ji are the mass and moment of inertia of link i (i D 3; b), respectively. lm2 and lm4
are the lumped masses of the two flexible links, which are given in [31]. u is the
velocity of the coupler point, !3 is the angular velocity of link 3, and vB , vC , vm3 ,
and vb3 are the velocities of the corresponding points, respectively.
From the kinematic analysis of the rigid-body motion of a four-bar linkage, it is
known that

VB D VC D f .2 ; uP / (4.12)
Vm3 D g.2 ; uP / (4.13)
Vb3 D h.2 ; uP / (4.14)
!3 D q.2 ; uP /: (4.15)

Therefore, the equivalent mass of a single compliant Roberts mechanism can be

obtained as [1]:

1
meB D ˛.lm2 C lm4 / C ˇm3 C mb3 C .J3 C Jb3 /; (4.16)
2
4 Parallel-Kinematic Nanopositioning Stages Based on Roberts Mechanism 133

where
r3
D tan 2 C b3 (4.17)
2
r32
˛D (4.18)
42 cos2 2
 
b3 2
ˇ D 1 (4.19)

 
b3 2
D 1 : (4.20)
2

In order to generate a high bandwidth of the control system, a high resonant

frequency of the stage is required. Based on Lagrange’s equation, the free motion of
the XY stage can be described by the dynamic equation

.8meB C mmov /  qR .t/ C Kin q.t/ D 0; (4.21)

where mmov is the mass of the movable plate of the XY stage and q D Œx yT is the
displacement of the movable plate in two working directions.
Based on the theory of vibration, the modal equation can be derived as

ŒKin  !i2 .8meB C mmov /ˆi D 0; (4.22)

where the eigenvector ˆi (for i D 1 and 2) represents a mode shape and the
eigenvalue describes the corresponding natural cyclic frequency, which can be
obtained by solving the characteristic equation. Because the mass matrix and
stiffness matrix of the stage are positive definite, the natural frequency of this XY
stage in each working axis is computed as
s
1 Kin
fi D : (4.23)
2 8meB C mmov

4.4 Model Verification with FEA Simulations

In this work, the XY stage is designed to generate a stroke over ˙7 mm in each

working direction as an illustration. For the convenience of calculation, the angle
parameters are selected as 20 D 40 D 75ı , r3 D 30 mm, and 30 D 180ı .
Moreover, the natural frequency of the system is constrained to be greater than
40 Hz. Such constraint is imposed by using the dynamic model (4.16). Based on the
design criteria (4.2), (4.3), (4.8), and (4.16), the model described above can be used
to determine the unknown geometric parameters of the structure. As a case study,
134 S. Wan and Q. Xu

Table 4.1 Geometry Parameter (Symbol) Value

parameters for the mechanism
Length of flexible links 2 and 4 (l2 , l4 ) 36 mm
Width of flexible links 2 and 4 (c2 , c4 / 10 mm
Thickness of flexible links 2 and 4 (h2 , h4 ) 0.5 mm
Length of flexible link 3 (l3 ) 30 mm
Width of flexible link 3 (c3 ) 10 mm
Thickness of flexible link 3 (h3 ) 3 mm
Length of flexible link b3 (lb3 ) 36 mm
Width of flexible link b3 (cb3 ) 10 mm
Thickness of flexible link b3 (hb3 ) 6 mm
Initial angle of links 2 and 4 75ı

the chosen geometry parameters which satisfy the design criteria are summarized
in Table 4.1. The flexure parallel linear guiding stage is designed with the main
parameters: l2 D l4 D 38 mm, b3 D 36 mm, h D 0:5 mm, and c D 10 mm.

4.4.1 Static Performance Analysis with FEA

The quantitative models are developed under the ideal situations. In order to test
the performance of the designed stage and to verify the accuracy of the derived
models, FEA simulation is conducted by using ANSYS software package. The
material of the stage is selected as Al-7075 alloy, which is easy to machine, and
its main characteristic parameters are: Young’s modulus D 71.7 GPa, yield strength
D 503 MPa, Poisson’s ratio D 0.33, and density D 2:81103 kg/m3 .
The quantitative models as developed in previous section predict that, the work
range of the XY stage in each direction is ˙7 mm. The deformation result is shown
in Fig. 4.8b. In addition, Eq. (4.9) evaluates that the maximum force required to
drive the stage is Fmax D 180:5 N. However, the FEA simulation result shows that
the required actuation force is 160 N. Taking the FEA result as the benchmark, it is
seen that the established quantitative models predict the stage performance with a
deviation of 12.8 % with respect to FEA result.
Moreover, when one input displacement arrives at 7 mm, the deformation result
of FEA static simulation is illustrated in Fig. 4.8a. It shows the compliance of the
Y-axis direction when . The yield strength of the Al-7075 is Π D 503 MPa. As
can be seen from Fig. 4.9, the induced maximum stress is 485.25 MPa < Π D
503 MPa, which shows a safety factor of 1.27 of the material, meeting the strength
condition.
In addition, Fig. 4.10 shows the compliance of the Y-axis direction when
ıx D 0. As compared with the FEA result, the result of the developed quantitative
model is 5.6 % lower. The discrepancy mainly comes from the small deformation
assumptions of the PRBM.
4 Parallel-Kinematic Nanopositioning Stages Based on Roberts Mechanism 135

Fig. 4.8 Static FEA results of the XY stage under different conditions. (a) Deformation result
when ıy D 7 mm and ıx D 0 mm; (b) deformation result when ıy D 7 mm and ıx D 7 mm

On the other hand, the FEA results as illustrated in Fig. 4.11 show the parasitic
motion in Y-axis, which is induced by the actuation in X-axis. Figure 4.11a exhibits
the FEA results of the symmetrical and asymmetric structure designs. It is obvious
to observe that as compared with the asymmetric design, the symmetrical structure
design can significantly reduce the parasitic motion in the non-working direction by
over 60 times.
The new Roberts mechanisms can produce a straight-line motion under the
working situation. As shown in Fig. 4.11 and Table 4.2, when the stage delivers the
full work range in the working direction (i.e., 7 mm in X-axis), the central point’s
parasitic motion in the non-working direction of the output stage (i.e., Y-axis) is
0.64 m. It means that the maximum cross-axis displacement caused by the work
direction is only 0.009 % in theory.
136 S. Wan and Q. Xu

Fig. 4.9 FEA results of stress distribution

Fig. 4.10 The motion displacement comparison between FEA and calculated results

4.4.2 Dynamic Performance Analysis with FEA

The FEA results of dynamic performance analysis are shown in Fig. 4.12. It is
observed that the first-three natural frequencies are 32.3 Hz, 32.5 Hz, and 58.7 Hz,
respectively. As described in the previous section, the designed natural frequency
is 40 Hz. This reveals that the developed quantitative dynamic model predicts the
stage’s natural frequency with a deviation of 18.75 % with respect to FEA result.
Besides, the modal shapes of the first two modes are shown in Fig. 4.12a, b,
which are the translations along the two working axes. The third one as shown
in Fig. 4.12c is the rotation of the output platform in the working plane. The first
six resonant frequencies are shown in Table 4.3. The similar values of the first
4 Parallel-Kinematic Nanopositioning Stages Based on Roberts Mechanism 137

Fig. 4.11 Comparison of parasitic motions for the XY stage. (a) FEA simulation results of
parasitic motions of the XY stage with and without symmetrical structure design; (b) experimental
results of parasitic motions of the XY stage with symmetrical structure design

two natural frequencies indicate that the XY stage has almost the same dynamics
performance in the two working directions, which benefits from the parallel-
kinematic design of the stage. In addition, the frequency of the rotational mode
is almost twice higher than the first-two modes, which confirms that the stage has
two desired degrees of freedom.
138 S. Wan and Q. Xu

Table 4.2 FEA simulation results of the parasitic motion with and without symmet-
rical structure design
Working The parasitic motion in Y-axis The parasitic motion in Y-axis
displacement in without symmetrical structure with symmetrical structure
X-axis (mm) design (m) design (m)
0 0 0
1 6.0361 0.0839
2 12.091 0.1551
3 17.656 0.2297
4 22.879 0.3552
5 29.362 0.4517
6 35.529 0.5429
7 41.076 0.6371

4.5 Prototype Fabrication and Experimental Study

4.5.1 Prototype Fabrication

The assembled CAD model of the designed XY stage is shown in Fig. 4.13a.
The XY stage has been fabricated using Al-7075 alloy through the wire electrical
discharge machining (WEDM) process. The prototype is shown in Fig. 4.13b, which
is driven by two voice coil motors (VCMs). The output positions are measured by
two Keyence laser displacement sensors. The controller is constructed by NI CRIO
9022 and 9018 FPGA module. Control algorithm is realized by NI LABVIEW
software package.
Since the FEA result reveals that Fmax D 160 N, two VCM motors (model:
NCC05-18-060-2X, from H2W Techniques, Inc.) are selected to drive the stage.
They can provide a sufficient large output force of 194.6 N with a stroke of 12.7 mm.
The total dimension of the stage is 244  244 mm2 . So, the area ratio of the stage is
0.32 %. It is seen that the proposed stage exhibits a relatively large area ratio, which
means that the space is fully used by the two-layer design.
The two-layer design enables the generation of relatively compact physical size
in comparison with a monolithic design. However, the shortcoming of two-layer
design lies in that assembly is needed to develop the XY stage. Considering that the
fixing holes can be manufactured with a high precision of ˙5 m, a low level of
assembly error is expected. The performance of the developed XY stage is tested
through the subsequent experimental studies.
4 Parallel-Kinematic Nanopositioning Stages Based on Roberts Mechanism 139

Fig. 4.12 FEA simulation results of modal analysis. (a) The first resonant mode; (b) the second
resonant mode; (c) the third resonant mode
140 S. Wan and Q. Xu

Table 4.3 The first six Mode Frequency (Hz)

frequencies of resonant
modes 1 32:33
2 32:53
3 58:76
4 212:74
5 324:33
6 326:42

Fig. 4.13 (a) CAD model

and (b) fabricated prototype
of the XY micropositioning
stage
4 Parallel-Kinematic Nanopositioning Stages Based on Roberts Mechanism 141

4.5.2 Experimental Investigations

In this section, experimental studies are carried out to test the static and dynamic
performance of the developed XY micropositioning stage.
First, the stroke of the XY stage is tested by applying a 0.1-Hz, 7-V sinusoidal
voltage signal to drive the each VCM at a time. For instance, by applying the signal
to the X-axis VCM as shown in Fig. 4.14a, the output displacement in X and Y axes
are shown in Fig. 4.14b, c, respectively.

a 10
Input voltage (V)

−5

−10
0 1 2 3 4 5 6 7 8 9
Time (s)
b x 104
1
Displacement x (μm)

0.5

−0.5

−1
0 1 2 3 4 5 6 7 8 9
Time (s)
c
100
Displacement y (μm)

−100

−200
0 1 2 3 4 5 6 7 8 9
Time (s)

Fig. 4.14 Experimental results of X-axis motion range testing. (a) Voltage applied to VCM;
(b) output displacement in X-axis; (c) output displacement in Y-axis
142 S. Wan and Q. Xu

It is seen from Fig. 4.14 that the motion range in X-axis is 12.4 mm, and
the caused coupling motion amounts to 195:0 m. This indicates a crosstalk of
195 m
12:4 mm
100 % = 1.57 % between the axes. Similar results are produced when the
VCM in Y-axis is actuated. That is, the motion range in Y-axis is 12.2 mm, and the
coupling motion in X-axis is 202.3 m, which exhibits a crosstalk of 202:3 m
12:2 mm
100 %
= 1.66 % between the two working axes.
Due to the manufacturing error and assembly error of the XY stage, the experi-
mental results (1.57 % and 1.66 %) of crosstalk is much bigger than the FEA simu-
lation result (0.09 %). If the machining and assembly precision can be improved, the
discrepancy between the simulation result and experiment result will be reduced. In
practice, to reduce the crosstalk values and to achieve precise positioning, suitable
feedback control algorithm can be implemented to cooperatively control the output
displacements in the two working axes of the XY micropositioning stage.
Next, the dynamic performance of the XY stage is examined by applying a swept-
sine voltage signal to drive each VCM alternately. By driving the X-axis VCM using
a swept-sine voltage with the frequency ranging from 0.1 to 1000 Hz, the produced
X-axis motion response is recorded by the laser displacement sensor. The frequency
response is obtained by FFT algorithm and the result is shown in Fig. 4.15. It is
found that the stage exhibits a natural frequency of 18.6 Hz in the X-axis. Moreover,
the natural frequency in the Y-axis is obtained as 19.5 Hz, which is similar to the
X-axis value.

100
Magnitude (dB)

50

−50
10−1 100 101 102 103

2000
Phase (degrees)

1000

−1000

−2000
10−1 100 101 102 103
Frequency (Hz)

Fig. 4.15 Frequency response of the XY stage in X-axis direction

4 Parallel-Kinematic Nanopositioning Stages Based on Roberts Mechanism 143

4.5.3 Experimental Study with PID Control

To demonstrate the positioning performance of the XY stage, closed-loop control

is implemented using PID control. For the purpose of designing the PID controller
of the stage, only the x-axis motion control is handled in this research, because the
y-axis motion is similar to x-axis performance.
The input/output datasets are used to identify the plant transfer function by
estimating the model from the frequency response data. The frequency responses
obtained by experimental study are shown in Fig. 4.15. The first resonant mode
occurs around 16.6 Hz. According to the Bode plot as shown in Fig. 4.15, a second-
order plant model is identified as shown below.

5s2  1:8  104 s C 2  107

G.s/ D : (4.24)
1:2s2 C 36s C 2  104

The model matches the system dynamics well in the frequencies up to 100 Hz. In
order to capture high frequency dynamics accurately, a much higher order model is
required to be identified.

4.5.3.1 Simulation Study

According to the transfer function (4.24), a PID control is realized in MATLAB and
Simulink, as shown in Fig. 4.16. The PID control parameters are adjusted according
to the PID Tuner in the MATLAB/Simulink. In order to get a combination property
of the PID control, according to automatic adjustment of the control parameters,
the manual tuning is also applied. Then, the PID control parameters Kp , Ki , Kd , and
filter coefficient (N) are 0.006, 37.282, 0.006, and 730337.058, respectively.
The purpose of the control simulation is to test the feasibility of the PID
controller. Figures 4.17 and 4.18 show the system responses to a step input and a
sinusoidal input, respectively, where the dashed lines represent the reference inputs
and the solid lines represent the output responses. It can be seen that the system

num(s)
+– PID(s)
den(s)
PID Controller Transfer Fcn

Sine Wave
Function
Scope

Step

Fig. 4.16 MATLAB simulation model with PID control

144 S. Wan and Q. Xu

Input
a
displacement (mm)
output

0.5

0.5 0.51 0.52 0.53 0.54 0.55 0.56 0.57 0.58 0.59 0.6
time (s)
b
displacement (mm)

0.5

0.5 0.55 0.6 0.65 0.7 0.75

time (s)

Fig. 4.17 Simulation results of step response with PID control. (a) Step input signal response;
(b) step response error

performance is not very good, but the steady state error is small as shown in both
cases. The simulation results are shown in Table 4.4.

4.5.3.2 Experimental Study

After the simulation study with the PID control, experimental study is carried out on
the prototype XY stage. The system is driven by two VCMs. The output positions
are measured by two Keyence laser displacement sensors. Control algorithm is
realized by NI LABVIEW software package. The diagrams of the control algorithm
program are shown in Fig. 4.19.
In order to test the performance of the system, different types of signal are chosen
as the input signal of the PID control system. The sampling frequency of the laser
sensor is set as 1000 Hz. To remove the noises of sensor output signal, a second-
order low-pass filter with a cut-off frequency of 5 Hz is adopted in data acquisition
process.
4 Parallel-Kinematic Nanopositioning Stages Based on Roberts Mechanism 145

Input
output
a
displacement (mm)
1

0.5

−0.5

−1
0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5
time (s)

b
0.02
displacement (mm)

0.01

−0.01

−0.02

−0.03
0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5
time (s)

Fig. 4.18 Simulation results of sinusoidal tracking with PID control. (a) Sinusoidal tracking
result; (b) sinusoidal tracking error

Table 4.4 Simulation results of the PID controller

Step response Sinusoidal response
2 % settling time Steady-state error Steady-state error bound
Maximum overshoot (%) (s) (mm) (mm)
16 0.4 0.002 ˙0.028

The experimental result of step response is shown in Fig. 4.20. In addition,

Fig. 4.21a–c show the system responses to 0.1-Hz, 0.5-Hz, and 1-Hz sinusoidal
inputs, respectively, where the blue dashed lines represent the reference inputs and
the red solid lines represent the output responses. A summary of experiment results
with different inputs are shown in Table 4.5.
In view of Tables 4.4 and 4.5, it is observed that as compared with the
simulation result, the experimental study produce similar trend of results. However,
the experimental results exhibit a time delay of about 50 ms. With sinusoidal
reference input, the delay time is proportional to the frequency of the input signal.
This time delay is mainly induced by the bandwidth of the control system. In
addition, the delay may also arise from the low-pass filter (5-Hz cut-off frequency)
which is used in the data acquisition process.
In order to identify the positioning accuracy, one part of the stable error signal
has been extracted. This part of signal has 1000 sample points. Then, it is used to
146 S. Wan and Q. Xu

Fig. 4.19 NI LABVIEW control algorithm program. (a) Front panel; (b) program diagram

Input
output
1.2
1
displacement (mm)

0.8
0.6
0.4
0.2
0
−0.2
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
time (s)

Fig. 4.20 Experimental results of step response with PID control

4 Parallel-Kinematic Nanopositioning Stages Based on Roberts Mechanism 147

Input
a output
displacement (mm) 1.5
1
0.5
0
−0.5
−1
−1.5
0 1 2 3 4 5 6 7 8 9 10
time (s)
Input
output
b 1.5
1
displacement (mm)

0.5
0
−0.5
−1
−1.5
0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5
time (s)
Input
c 1.5
output

1
displacement (mm)

0.5
0
−0.5
−1
−1.5
0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5
time (s)

Fig. 4.21 Experimental results of sinusoidal tracking with PID control. (a) 0.1-Hz sinusoidal
input; (b) 0.5-Hz sinusoidal input; (c) 1-Hz sinusoidal input

fit a normal distribution curve as shown in Fig. 4.22. According to the ISO standard
as shown in (4.25), the positioning accuracy should be ˙3 . So, the positioning
accuracy of the stage is 23 D 0:96 m  1 m.
sP
n
iD1 .xi xN /2
D where W n D 1000: (4.25)
n1
148 S. Wan and Q. Xu

Table 4.5 Experiment results of the PID control system

Step response
Type of signal Maximum overshoot (%) 2 % settling time (s) Steady-state error (mm)
Step input 2.5 0:46 0:003
Sinusoidal response
Steady-state error bound Mean absolute error Root mean square error
(mm) (mm) (mm)
0.1-Hz sine wave ˙ 0.02 0:012 0:015
0.5-Hz sine wave ˙ 0.1 0:067 0:107
1.0-Hz sine wave ˙ 0.2 0:130 0:156

Fig. 4.22 Testing results of the stage positioning accuracy

4.5.4 Discussion on Stage Performance

The experimental results demonstrate a large stroke of the developed XY micro-

positioning stage. The crosstalk between the two working axes may be induced
by the fabrication errors of the stage parameters, and assembly errors of the stage
and motors. In addition, due to the stroke limit of the adopted VCM motors, the
experimental result of the XY stage stroke is slightly lower than the analytical
result. The desired range of ˙7 mm can be achieved by employing VCMs with
larger stroke.
The experimental testing reveals that the XY stage exhibits the natural frequen-
cies of 18.6 and 19.5 Hz in X- and Y-axes, respectively. Such values are smaller than
the FEA results of 32.3 and 32.5 Hz in the two working axes, respectively. These
discrepancies are mainly caused by the mass of the moving coil of VCM motor,
which is not considered in FEA simulation. In the future, a higher natural frequency
4 Parallel-Kinematic Nanopositioning Stages Based on Roberts Mechanism 149

will be achieved by implementing an optimal design of the XY stage parameters.

In addition, control schemes will be implemented to achieve a precise positioning
control of the XY stage.

4.6 Conclusion

A new type of parallel-kinematic XY micropositioning stage based on flexure

Roberts mechanism has been designed, analyzed, fabricated, and tested in this paper.
FEA simulation results show that this kind of compliant mechanism not only has
good performance in reducing the parasitic motion in non-working direction, but
also can produce a large workspace of 14  14 mm2 . So, it meets the requirements
of a large stroke, high precision linear guiding mechanism. The large stroke and
small crosstalk have been verified by experimental studies. Experimental testing
demonstrates a natural frequency around 20 Hz in both working axes. Moreover, it
owns a relatively compact structure thanks to a two-layer mechanism design. It is
found that it is feasible to devise a large-stroke micropositioning stage based on the
Roberts mechanism. At last, a PID system was designed to control the X–Y motion
of stage, and it get a high manipulate precision (the positioning accuracy is 1 m).
Future work involves architectural optimization and precision control of the XY
stage for pertinent applications.

Acknowledgements The work was supported by the Macao Science and Technology Develop-
ment Fund under Grant Nos.: 070/2012/A3 and 052/2014/A1, and the Research Committee of the
University of Macau under Grant Nos.: MYRG083(L1-Y2)-FST12-XQS and MYRG078(Y1-L2)-
FST13-XQS.

References

1. A.G. Ambekar, Mechanism and Machine Theory (Prentice-Hall of India, New Delhi, 2007)
2. S. Awtar, A.H. Slocum, Constraint-based design of parallel kinematic XY flexure mechanisms.
J. Mech. Des. 129(8), 816–830 (2007)
3. W. Chen, J. Jiang, W. Chen, J. Liu, A novel flexure-based uniaxial force sensor with large range
and high resolution. Sci. China Technol. Sci. 58(8), 1940–1948 (2013)
4. W. Chen, J. Jiang, J. Liu, S. Bai, W. Chen, A passive eddy current damper for vibration
suppression of a force sensor. J. Phys. D Appl. Phys. 46, 075001 (2013)
5. Y.J. Choi, S.V. Sreenivasan, B.J. Choi, Kinematic design of large displacement precision XY
positioning stage by using cross strip flexure joints and over-constrained mechanism. Mech.
Mach. Theory 43(6), 724–737 (2008)
6. K.B. Choi, H.J. Lim, G.H. Kim, J.J. Lee, A flexure-based scanner for a fully bidirectional
operation driven by a differential piezo force. Proc. IME C J. Mech. Eng. Sci. 228(7),
3186–3199 (2014)
7. D. Devaprakasam, S.K. Biswasa, Design of a flexure for surface forces apparatus. Rev. Sci.
Instrum. 74(3), 1228–1235 (2003)
8. L.L. Howell, Compliant Mechanisms (Wiley, New York, 2001)
150 S. Wan and Q. Xu

9. Z.B. Huang, W.J. Ge, L.E. Ma, Research status quo of flexible mechanism and its application
prospect in the field of bionic. J. Mach. Des. Res. 20(21), 279–252 (2004)
10. N.B. Hubbard, J.W. Wittwer, J.A. Kennedy, D.L. Wilcox, L.L. Howell, A novel fully
compliant planar linear-motion mechanism, in Proceedings of ASME 2004 International
Design Engineering Technical Conferences and Computers and Information in Engineering
Conference, pp. 1–5, Salt Lake City (2004)
11. D. Kang, K. Kim, D. Kim, J. Shim, D.G. Gweon, J. Jeong, Optimal design of high
precision XY-scanner with nanometer-level resolution and millimeter-level working range.
Mechatronics 19(4), 562–570 (2009)
12. Y. Lei, B.Y. Chen, Y.Z. Yang, Nanoscale research present situation and development trend of
micro bench. J. Zhejiang Univ. Technol. 23(1), 72–82 (2006)
13. Y.M. Li, Q.S. Xu, Modeling and performance evaluation of a flexure-based XY parallel
micromanipulator. Mech. Mach. Theory 44, 2127–2152 (2009)
14. Z. Li, J. Yu, A novel large-range XY compliant parallel micro-manipulator. J. Changzhou Univ.
25(4), 14–19 (2013)
15. Y.T. Lin, J.J. Lee, Structural synthesis of compliant translational mechanisms, in Proceedings
of 12th IFToMM World Congress, pp. 1–5 (2007)
16. S. Polit, J. Dong, Development of a high-bandwidth XY nanopositioning stage for high-rate
micro-/nanomanufacturing. IEEE/ASME Trans. Mechatron. 16(4), 724–733 (2011)
17. X.D. Shi, J.L. Li, L.W. Wang, Design of large displacement and high precision linear stage
based on laminated flexure hinge and ultrasonic motor. Manuf. Technol. Mach. Tool 2,
104–106 (2007)
18. K. Tan, G.H. Zong, S.S. Bi, Z. Yu, Multi-leaf configuration of expanded-motion-range flexure
hinges. ProcDual Use Technologies & Products 3, 38–39 (2007)
19. X. Tang, I.M. Chen, Q. Li, Design and nonlinear modeling of a large-displacement xyz flexure
parallel mechanism with decoupled kinematic structure. Rev. Sci. Instrum. 77(11), 115101
(2006)
20. B.P. Trease, Y.M. Moon, S. Kota, Design of large displacement compliant joints. J. Mech. Des.
127(4), 788–798 (2004)
21. H. Wang, X.M. Zhang, Input coupling analysis and optimal design of a 3-DOF compliant
micro-positioning stage. Mech. Mach. Theory 43, 400–401 (2008)
22. X.K. Wang, D. Wu, C.Y. Cheng, J. Mech. Eng. 10(5), 570–576 (1999) (Chinese)
23. Wikipedia (2007) http://commons.wikimedia.org/wiki/File:Roberts_linkage.gif
24. Q.S. Xu, New flexure parallel-kinematic micro-positioning system with large workspace. IEEE
Trans. Robot. 28(2), 478–491 (2012)
25. Q.S. Xu, Design, testing and precision control of a novel long-stroke flexure micropositioning
system. Mech. Mach. Theory 70, 209–224 (2013)
26. Q.S. Xu, Design and development of a compact flexure-based XY precision positioning system
with centimeter range. IEEE Trans. Ind. Electron. 61(2), 893–903 (2014)
27. Q.S. Xu, Y.M. Li, Analytical modeling, optimization and testing of a compound bridge-type
compliant displacement amplifier. Mech. Mach. Theory 46, 183–200 (2011)
28. S.L. Xu, D. Chen, Precision ultra-precision positioning technology and its application. J. Mech.
Eng. 8(4), 73–75 (1997) (Chinese)
29. Y.D. Yan, T. Sun, S. Dong, With new drive micro displacement worktable design and
experimental research. J. Mech. Eng. 37(5), 5 (2001)
30. Y.K. Yong, T.F. Lu, Relationship among input-force, payload, stiffness and displacement of a
3-DOF perpendicular parallel micro-manipulator. Mech. Mach. Theory 45, 756–771 (2010)
31. Y.Q. Yu, L.L. Howell, C. Lusk, Y. Yue, M.G. He, Dynamic modeling of compliant mechanisms
based on the pseudo-rigid-body model. J. Mech. Des. 127, 760–765 (2005)
32. Z.J. Yuan, X.K. Wang, Precision and Ultra-Precision Machining Technology, pp. 1–3.
(Machinery Industry Press, Beijing, China, 1999)
33. H. Zhao, Z.Q. Lv, The present situation and development trend of piezoelectric micro-
positioning stage. J. Mod. Mach. 4, 33–35, 52 (2001)
34. S.S. Zhao, S.S. Bi, G.H. Zong, J. Yu, New large-deflection flexure pivot based on curved flexure
element. Chin. J. Mech. Eng. 45(4), 8–12 (2009)
Chapter 5
Electro-Magnetic Nano-Positioning

Zhuo Zhang, Yueqing Yu, Xinyu Liu, and Xuping Zhang

Abstract Electro-magnetic actuation technology has the potential of providing

millimeters of actuation range with nanometers of positioning resolution, and
recently has received increasing attention in the field of nano-positioning. Compared
with other types of nano-positioning systems that are driven by electrostatic,
thermal, and piezoelectric actuators, the electro-magnetic nano-positioning systems
are able to achieve large power density with low input voltages, are capable of
operating in large distances with virtues of linear response, and are featured with
multi-axis actuation capability and with high bandwidth. Therefore, a variety of
electro-magnetic nano-positioning systems have been developed for various nano-
scale applications. Although electro-magnetic nano-position systems are equivalent
in positioning objects in nano-scale, they could be different in size, configuration,
degrees of freedom, fabrication process, and control scheme for being suited
to certain application requirements. However, electro-magnetic nano-positioning
systems generally consist of three essential parts: an electro-magnetic actuator, a
guiding mechanism, and a positioning stage. Instead of covering all the aspects
on either the various types of systems or the applications, this chapter presents
the common and general principles of electro-magnetic nano-positioning systems:
actuation, guiding mechanisms, kinematics, dynamics, control, and fabrication. The
typical systems and applications in the literature are employed as examples to
illustrate the common principles. The aim of the chapter is to provide readers with
insight and guidance on the development of electro-magnetic nano-positioners.

5.1 Overview

Compared with other actuation types available such as electrostatic, thermal,

and piezoelectric, etc., electro-magnetic actuations require low operation voltages,
moderate power density, large operating distances, linear response, multi-axis
capability, and high bandwidth [1]. These advantages make electro-magnetic nano-

Z. Zhang • Y. Yu • X. Liu • X. Zhang ()

Department of Engineering and Aarhus School of Engineering, Aarhus University,
Inge Lehmanns Gade 10, 8000 Aarhus C, Denmark
e-mail: [email protected]

© Springer International Publishing Switzerland 2016 151

C. Ru et al. (eds.), Nanopositioning Technologies,
DOI 10.1007/978-3-319-23853-1_5
152 Z. Zhang et al.

Fig. 5.1 Schematic diagram of the principle of probe-based data storage [6]

positioning systems increasingly attractive over the recent decades. There are
emerging applications and techniques that could benefit from nano-positioning
technology, such as the scanning probe microscopy (SPM) [2], probe-based data
storage (PDS) [3, 4], and nano-scale electro-machining (nano-EM) [5], etc. This
chapter is devoted to an overall introduction of the fundamental principles and
guidance on the analysis, design, control, and fabrication of the electro-magnetic
nano-positioning technology, i.e., nano-positioning apparatus driven by electro-
magnetic actuators.
One application is the PDS that has been proposed in the late 1990s. This
technology has emerged from SPM, where a surface is scanned and manipulated
at the atomic level to form topographic images [6, 7]. This data storage system
is MEMS-based, which is comprised mainly of two separate layers, as shown in
Fig. 5.1 [6]. The upper layer depicted in Fig. 5.1 is the medium layer, on which data
will be recorded or removed, whereas the other layer carries an array of probes (or
heads, usually several thousands) that is used to read/write/erase data via scanning
and contact with the medium. There are two main scanning configurations for PDS,
i.e., (a) probe-scanning, where the probe moving relative to a fixed medium layer,
and (b) medium-scanning, where the medium layer is movable while the probe layer
remains stationary. Usually the tip size of the probes is atomic-scale, resulting in the
data bits are approximately 15 nm in diameter and the tracking pitch of 30 nm. A
pit that was written into the surface and the original surface represent logical “1”
or “0,” respectively. In order to achieve data storage with high performance such as
ultrahigh storage density, high speed, and suited to portable applications, the nano-
positioning system must meet a stringent set of requirements. For example, to ensure
extremely high data throughputs, a sufficient scanning range, high bandwidth, as
well as reliable nano-scale positioning accuracy are needed. In addition, the portable
storage applications also require lower energy consumption and vibration resistant
capabilities.
5 Electro-Magnetic Nano-Positioning 153

Fig. 5.2 Nano-scale

electro-machining process [8]

Another important emerging application benefited from the nano-positioning

technology is probe-based nano-scale electro-machining (nano-EM). This technique
modifies a surface at nano-scale through removing conductive material from the
work surface. The process is illustrated in Fig. 5.2 [8]. The gap between the probe
and the work surface is filled with dielectric. When sufficient high electric potential
is applied, the dielectric will breakdown resulting in removal of local material from
the work surface. In nano-EM, the positioner must control the probe-tip location and
orientation relative to the surface in all six degrees of freedom (DOFs), i.e., three
translation DOFs: x, y, and z and three rotational DOFs:  x ,  y , and  z . In addition to
having the nanometer-level or better resolution, the nano-EM process also requires
good repeatability and short execution time to make the technology practical and
competitive.
Electro-magnetic nano-positioning technology enables us to move large or small
objects in nanometer-level or better precision, however, the actual positioner is
not necessarily be on nano-scale. The dimensions range from macro- [9] to meso-
[8] and even micro-scales [10, 11] based on various application purposes. Macro-
scale nano-positioners can easily operate over large distances (100s microns to
centimeters) and exert large forces (N-kN) with nanometer precisions. One typical
macro device is the magnetic levitation system [12–14], which achieve nanometer-
level precision thanks to its non-contact nature. It eliminates mechanical friction,
stiction, backlash, and hysteresis that are detrimental to precise control. In addition,
the macro-sized system is well capable in integrating sensing and feedback systems
in order to stabilize their motion and ensure nano-scale precision. However, the
sensing and feedback control add complexity to the design and assembling of the
system. The large mass leads to low natural frequencies and bandwidth. This limits
applications where high throughput is needed. On the other hand, macro-scale nano-
positioners often consume more energy, and exhibit more thermal growth that may
cause thermal instability and errors. A macro six-axis nano-positioning device with
magnetic levitation technology is shown in Fig. 5.3 [9].
Although macro-sized devices are capable of positioning objects in nano-scale,
they are not amenable to massively parallel operation resulting in low throughput
154 Z. Zhang et al.

Fig. 5.3 Photograph of the six-axis maglev nano-positioning device [9]

Fig. 5.4 Image of an

x-axis meso-scale
electro-magnetically driven
nano-positioner developed by
MIT. The dime is included for
size reference [8]

rates. The great mass and volume limit the use for portable applications. Recent
years, researchers from Massachusetts Institute of Technology (MIT) developed a
six-axis meso-scale nano-positioner driven by electro-magnetic actuators, as shown
in Fig. 5.4 [8]. This device is only centimeters in size with the central stage driven
by three sets of integrated electro-magnetic actuators. Due to the reduced size, it is
possible to run several nano-positioners in parallel to achieve high throughput rates.
Moreover, the reduction in mass makes it much easier to attain higher mechanical
5 Electro-Magnetic Nano-Positioning 155

Fig. 5.5 Structure of probe-based data storage (PDS) with electro-magnetic micro x–y stage and
its working principle [10]

bandwidth of the system. Also, meso-scale nano-positioners exhibit less thermal

growth, cost less energy, and cause smaller thermally-induced errors accordingly
than macro counterparts. However, several challenges must be overcome as the
size is significantly scaled down. For example, design and integration of meso-scale
multi-axis electro-magnetic actuators are totally different from macro devices. The
assembling of position and orientation sensors is a big challenge.
With the maturation of MEMS (Micro-Electro-Mechanical System) technology,
electro-magnetic nano-positioner can be scaled down to millimeter scale. Figure 5.5
shows a micro-scale nano-positioner for PDS application. The size of the media
platform is 5  5 mm2 . The micro-scale positioner is compatible and portable.
However, due to their small size it is difficult, if not impossible, to integrate sensors,
and it is challenging to design a micro stage with up to six DOFs.
Despite their differences in size, the macro, meso, and micro electro-magnetic
nano-positioner share a common operation principle and structure. The positioning
system mainly consists of a stage, electro-magnetic actuation parts, and a guiding
part that links the actuation parts to the stage. The purpose of this chapter is to
introduce common principles on how electro-magnetic actuations could generate
stage motions with nanometer precisions. Based on the principles and different
applications, we will also point out key considerations on design, control, and
fabrication. This chapter will not be confined to certain types of electro-magnetic
nano-positioning systems, instead, it will provide potential readers with the basic
principles, methodologies, and fundamental considerations on this topic.
156 Z. Zhang et al.

5.2 Principle of Electro-Magnetic Actuation

5.2.1 Actuation Force Generation

In general, the electro-magnetic actuation parts consist mainly of one or several pairs
of permanent magnets and coils. Upon the application of a current in the coil, the
associated magnetic field of the coil is generated. In presence of the magnetic field
generated by permanent magnets, the output actuation forces are generated thanks
to the attraction and/or repel forces between coils and permanent magnets.

5.2.1.1 Solenoid

Solenoid is often used along with a permanent magnet to generate actuation forces
based on magnetic attraction and/or repulsion effect. Solenoid behaves like a long
bar magnet of which the north and south pole of the magnetic field is governed by
right-hand rule, as shown in Fig. 5.6. There are a variety of applications benefited
from solenoid such as electro-magnetic lifting machine, the electric bell, the electro-
magnetic relay, and telephone earpiece, etc.
Based on Ampere’s Circuital Law, the magnetic field of solenoid is expressed as

B D 0 IN=L (5.1)

where 0 is called the permeability of free space, I the input current, N the number
of turns, i.e., number of loops the solenoid has, and L the length of solenoid.
The strength of the magnetic field can be increased by the four approaches: (1)
increasing the current; (2) increasing the number of turns; (3) placing a soft iron
rod/core inside the solenoid; and (4) pushing the turns of the wire closer to make
north and south poles closer. In practice, for positioning applications, the magnetic
strength (actuation force) is governed by the control of input currents. In addition
to the magnitude of the passing current, the direction of the current determines the
direction of north and south poles and the direction of actuation accordingly.
The main shortcoming of the solenoid-based actuation is the difficulty to
fabricate and assemble meso- and even micro-scale solenoids. They are used

Fig. 5.6 Solenoid and

right-hand rule
5 Electro-Magnetic Nano-Positioning 157

alone in macro-scale magnetic levitation systems for nano-positioning. Moreover,

solenoid-based electro-magnetic actuators can only achieve one-axis motion, which
limits their uses when multi-axis actuation is desired. These drawbacks are induced
mainly by the spatial structure of the solenoid. To overcome this, planar-coil electro-
magnetic actuation schemes have been employed in nano-positioning systems. To
help reader better understand the actuation principles, the physical principle of
Lorentz force and Ampere’s force is briefly introduced as below.

5.2.1.2 Lorentz Force and Ampere’s Force

In presence of only a magnetic flux B, and electric charge q moving with a velocity
v experiences a magnetic force Fm . The magnetic force is called Lorentz force and
is expressed as

Fm D qv  B (5.2)

The magnetic force Fm affects only the direction of motion of q, not the kinetic
energy. Because the direction of magnetic force is perpendicular to the velocity of
q and does no work. Electric current is motion of charge. To make Lorentz force
practical, Ampere’s force has been employed which refers to the force exerts on the
current-carrying wire in a magnetic field. The Ampere’s force, total force acting on
the wire, is
Z
FD IdL  B (5.3)
wire

where I represents electric current and dL refers to differential elements of length

of the wire. Figure 5.5 shows a planar-coil-based electro-magnetic micro x–y stage,
in which the Lorentz force/Ampere’s forces used to actuate the stage. Usually,
the magnetic field is created by permanent magnets, and thus the magnitude and
direction of the magnetic force are determined by electric current.

5.2.2 Configurations
5.2.2.1 Magnet-Moving Versus Coil-Moving

In generating actuating force, two configurations of electro-magnetic actuator have

been adopted: magnet-moving and coil-moving. Moving magnet actuations operates
with the coil fixed and the magnets moves for actuation, whereas the moving coil
systems works by fixing the magnets to the base and the coils are moved for
actuation. A typical moving magnet system is shown in Fig. 5.7 [15], where the
coils are fixed and the magnets are mounted on a flexure guiding mechanism. Fixing
158 Z. Zhang et al.

Fig. 5.7 (a) Photograph of a prototype nano-positioner for probe-based data storage and (b) the
thermal position sensors [15]

the coils is ideal for thermal dissipation and therefore higher operating currents
are allowed and larger actuation forces are achieved. However, the magnets must
be bonded onto the flexure mechanisms, which is a challenging work especially
for meso- and micro-scale devices. Another drawback inherent in magnet-moving
design is that the magnets are in general a big mass which results in lower bandwidth
of the actuators.
In contrast, for coil-moving configurations, as shown in Fig. 5.5, it is easier
to achieve high bandwidth since the mass of coils can be decreased by virtue of
micro-fabrications. However, for applications that require sufficient precision, the
heat dissipation from the moving coil through the flexure mechanism will cause
thermal-induced errors. To reduce the thermal-based errors, the operation current
must be kept in lower magnitude for coil-moving configurations. As a result, such
configurations can only provide less actuation force compared with magnet-moving
configurations.

5.2.2.2 Single-Axis and Multi-Axis

Most of applications for nano-positioning require at least two degree-of-freedom,

such as planar scanning. Some applications require up to six degree-of-freedom
(DOF), such as nano-EM. For this purpose, electro-magnetic actuators must have
one-axis or multi-axis capabilities, and need to be appropriately arranged. In
general, for the purpose of achieving the positioning stage with multiple DOFs,
either the choices of more actuators with less axes or less actuators with more
5 Electro-Magnetic Nano-Positioning 159

Fig. 5.8 A six DOFs nano-positioning device driven by six single-axis actuators [9]

axes is feasible. Fewer axes for each actuators will facilitate design of actuators
with a compromise that the design of guiding mechanism and actuator arrangement
become complex. On the contrary, more axes for each actuators may add complexity
of actuator design but make design of guiding mechanism and actuator arrangement
with ease. However, to the best knowledge of the authors, no efforts have been
devoted to the development of electro-magnetic actuators having more than two-
axis.
The six-axis magnetic levitation nano-positioning device shown in Fig. 5.3
has six independent single-axis electro-magnetic actuators, of which the detailed
arrangement is seen in Fig. 5.8 [9]. In order to achieve six-DOF positioning of the
stage, three actuators are combined to generate in-plane motion, and the rest of
the three actuators are responsible in producing out-of-plane motions. Another six-
DOF stage developed by MIT is shown in Fig. 5.4, where only three actuators with
double-axis are incorporated. Each actuator provides two mutually perpendicular
axial motions, i.e., in-plane and out-of-plane, as shown in Fig. 5.9. The three ways
actuations simplified the guiding mechanisms that connect the actuators and the
center stage. As a tradeoff, it adds complexity in design, fabrication, and assembling
of each actuators.
As illustrated in Fig. 5.10 [16], in achieving both in-plane and out-of-plane
motions, two separated coil layers are stacked. The stacked configuration is
benefited from the planar coil-moving scheme. The actuation force is generated
thanks to Lorentz force/Ampere’s force via interaction of the coil current with the
permanent magnet field.
160 Z. Zhang et al.

Fig. 5.9 A six DOFs

nano-positioning device
driven by three double-axis
actuators [8]

Fig. 5.10 Schematic diagram

of a double-axis
electro-magnetic actuator
[16]

5.3 Compliant Guiding Mechanisms

Guiding mechanisms enable motion guidance of individual actuators to form the

stage motion. There are two main categories of the guiding mechanisms: (1)
the rigid mechanisms which use prismatic and rotational kinematic joints for
bearing and guidance, as shown in Fig. 5.8; (2) the compliant mechanisms (i.e.,
flexure mechanisms) which transfer or transform motion, force, or energy via
elastic deformations, as shown in Figs. 5.5 and 5.9. Compared with rigid guiding
mechanisms, the compliant counterparts are more favorable particular in meso- and
micro-scale applications. Since fabrication and assembling of rigid mechanisms
are challenging and rigid mechanisms will incur backlash, inaccuracy in motion,
frictions, and wear because of the rigid prismatic and rotational joints. In contrast,
5 Electro-Magnetic Nano-Positioning 161

the compliant mechanisms have advantages in two categories: cost reduction (part-
count reduction, reduced assembly time, and simplified manufacturing processes)
and increased performance (increased precision, increased reliability, reduced wear,
reduced weight, and reduced maintenance) [17]. These virtues ideally fulfill the
requirements as the size of electro-magnetic nano-positioner decreases. Moreover,
this compliant structure enables stabilization without the feedback aids. There
are also some challenges associated with compliant guiding mechanisms such
as hysteresis, sensitivity to overloading, thermal sensitivity, and low range of
motion, etc. In this section, we will provide readers with key methodologies and
considerations in designing the compliant guiding mechanisms.

5.3.1 Design Considerations

Recent efforts in compliant guiding mechanisms are based on intuitions, for which
the beams with constant sectional area are usually used. Upon operation and
analysis, small and linear deformations are assumed. In designing the compliant
guiding mechanisms for electro-magnetic nano-positioning systems, the finite-
element analysis is used to obtain the force-deformation relationship, transmission
ratio, thermal response, and stress, etc. In what follows, we list some key design
considerations that can be used as a reference in achieving high nano-positioning
performances for a variety of applications.

5.3.1.1 Distributed and Lumped Compliant Mechanisms

Distributed compliant mechanisms are those structures that they are flexible at every
position, whereas the lumped compliant mechanisms are those consists of both
rigid and flexible parts. For distributed compliant mechanisms the final position
is determined by accumulation of deformations at every position. One typical
geometry of distributed compliant mechanisms is the beam with uniform cross-
sectional areas. As shown in Fig. 5.9, three sets of double-bent beams are used to
support the stage and actuators and guide motions. In contrast, the final position
of a lumped compliant mechanism is guided thanks to deformation of flexible
hinges [18]. Different types of hinges provide unique properties to suit different
requirements. Here we present three commonly used flexible hinge forms, i.e., the
circular, the corner-filleted, and the elliptical hinges, as shown in Fig. 5.11 [19].
The circular hinge is superior to the others in achieving accurate rotational
motions with high out-of-plane deflections. The center of deflection can be esti-
mated to be at the center of the hinge. However, the high precision is only
limited to a relatively small range. The corner-filleted hinge has relatively low
bending stiffness. The performances of elliptical hinges are in between the circular
and corner-filleted ones. For the same deflection, elliptical have lower maximum
stress than corner-filleted ones resulting in a longer fatigue life. Besides the three
162 Z. Zhang et al.

Fig. 5.11 Different types of hinges. (a) Circular, (b) corner-filleted, and (c) elliptical [19]

Fig. 5.12 Compliant parallel-guiding mechanisms in positioning application [19]

Table 5.1 A comparison of different flexure hinges [19]

Stiffness
Flexures In-plane Out-of-plane Motion accuracy Max. stress
Circular High High High High
Corner-filleted Low Low Low Low–med
Elliptical Low–med Low–med Med–high Low
Parabolic Med–high Med–high Med–high Med
Hyperbolic High High High Med–high

commonly used hinges, there are other flexure hinges available such as parabolic
and hyperbolic. Details of the performances of those hinges are listed in Table 5.1
[19].
One typical compliant mechanism called compliant parallel-guiding mechanisms
has been used in positioning applications, as shown in Fig. 5.12 [19], in which two
opposing links remain parallel throughout the mechanisms motion. In this regard,
5 Electro-Magnetic Nano-Positioning 163

the parallel-guiding mechanisms produce alone the translation motion without a

rotation motion exerting on the stage. To this end, the opposing links must be
fabricated with the same length and form a parallelogram. This compliant four-bar
mechanism can be either fully- or partially-compliant designed.

5.3.1.2 Stiffness and Resonant Frequency

The stiffness of the compliant guiding mechanisms determines the range of motion
and resonant frequencies of the nano-positioner. For general cases, the compliant
guiding mechanisms are used only confined to the linear elastic assumptions, where
the positioning system can be treated as a linear mass-spring model. With this, the
static displacement and resonant frequency of the system are given by [8]

F D kı (5.4)
r
1 k
f D (5.5)
2 m

The static displacement and resonant frequency can be obtained for a value of
stiffness. The stiffness of simple structures, such as uniform cross-sectional beams,
can be derived using either Euler–Bernoulli theory or Castigliano’s second theorem.
For complex structures, finite-element methods are often used to obtain the stiffness.
As can be seen in Eq. (5.4), for a force-limited actuation, a stiffer flexure leads to
small range of motion. Equation (5.5) shows that for a given moving mass, a stiffer
structure increases resonance frequency. Because the flexure structure deforms and
stores potential energy, the resonance frequency will limit the bandwidth. As a
result, the compliant guiding mechanisms should be designed neither too flexible
nor too stiff. Another important concern is that the stiffness is sensitive to the
fabrication errors, which may cause errors in the motion range and bandwidth.

5.3.1.3 Transmission Ratio and Coupling

The transmission ratio is defined as the ratio of the stage to actuation displacement
in a given direction [8]. This concept is important in nano-positioning systems
since structure of large transmission ratio will amplify the errors of the actuation
causing large positioning errors in stage. The compliant guiding mechanisms may
be designed to enable motion/error de-amplification for nano and even sub-nano-
positioning applications. However, this de-amplification treatment will result in a
limited range of motion.
For most cases, flexure-based systems will give rise to coupling problems, which
is caused by non-zero stiffness of the flexures perpendicular to the direction of
actuation. This will definitely change the transmission ratios and lead to additional
parasitic motions.
164 Z. Zhang et al.

5.3.1.4 Thermal Issues

For nano-motions, the thermal effects cannot be ignored [19]. As mentioned

previously, for coil-moving configurations, the heat generates via Joule heating in
the coils and dissipates through radiation, convection, and conduction. Among these,
heat conduction to compliant guiding mechanisms predominates. The increased
temperature will lead to thermal expansion effects, resulting in significant drift
and positioning error. Careful material selection and mechanical arrangement are
effective ways to minimize thermal expansion effects. Using the materials with
small coefficient of thermal expansion (CTE) ensures less expansion with increased
temperatures. Moreover, materials of high thermal conductivity should be selected
in case that quick thermal equilibrium will be achieved. Proper arrangement of the
compliant guiding mechanisms is essential in eliminating the thermal effects, which
is presented in the following section.

5.3.1.5 Configurations

The configurations of compliant guiding mechanisms are considered at two aspects:

function and performance. First, the configurations must fulfill the requirement that
several actuations produce stage motions with certain number of DOFs. This is
achieved by determining how many actuators with certain number of axes are used
and how they should be allocated. Second, the arrangement of the compliant guiding
mechanisms must ensure the positioning performance.
As mentioned previously, the thermal effects of the coil-moving actuator result
in thermal drift and significant errors. In addition to a proper selection of materials,
a symmetry structure is often employed for passive thermal compensation purpose,
as shown in Fig. 5.13. In this scheme, the thermal drift and positioning errors are
minimized because of the symmetric configuration. As a result, greater current can
be applied on the coils and therefore the actuators can produce larger force without
a loss of positioning error of the stage.

Fig. 5.13 Nano-positioner flexure concepts. (a) Single-bent beam, (b) double-bent beam, and (c)
bow-flexure [8]
5 Electro-Magnetic Nano-Positioning 165

Figure 5.13 also illustrates three possible symmetric configurations, i.e., single-
bent beam, double-bent beam, and bow-flexure [20, 21]. The geometric constraints
limit the total length of the beams of the single-bent beam concept, which results
in higher stiffness, higher resonant frequency, and lower transmission ratio over
the other two concepts. The double-bent beam and bow-flexure share similar
performances.
Compliant guiding mechanisms have the virtue of shock and vibration resistance
because of their compliance. On the other hand, however, the compliant mechanisms
must be designed stiffer to be robust against position errors caused by vibration
and shock. To eliminate the shock and vibration effects on position errors and
allow sufficient compliance for operations simultaneously, a mass balance concept
is preferred. The nano-positioner in Fig. 5.7 is designed based on this concept,
of which the one-dimensional illustration is shown in Fig. 5.14a [22]. In this
configuration, if we apply an actuation force in the positive y-direction, as shown
in Fig. 5.14b the compliant mechanisms will pull the table down in negative y-
direction, making the dynamic effects of the actuator and the table balanced. When
under shock conditions, as shown in Fig. 5.14c, an acceleration will be exerted on
the whole structure. If mactuator  x1 D mtable  x2 , then the pivot, table, and actuator
do not move relative to the frame. With this, the operation precisions of the nano-
positioning system will not be affected by external shock and vibrations.

Fig. 5.14 (a) One-dimensional illustration of a mass balance positioner. (b) Rotation of pivot
and displacement of the actuator and table due to the application of a force to the actuator. (c)
Illustration of the forces and torques acting on the pivot resulting from the acceleration of the
frame of the positioner [22]
166 Z. Zhang et al.

5.3.2 Design Methods

The design of compliant guiding mechanisms is essential in achieving high perfor-

mance for various positioning applications. However, current design of compliant
guiding mechanisms stems from designer’s intuitive and trial and errors. A lack of
reliable and systematic design methods makes it not possible to design the compliant
guiding mechanisms with optimal performance. Moreover, current designs are
based on linear assumption of material properties. This is only adapted to small
deformation applications and results in only a limited range of positioning motions.
Thus in this section, we present two major design methods that is useful in the large
deformation design and optimal design of the compliant guiding mechanisms.

5.3.2.1 Pseudo-Rigid-Body Model

Pseudo-rigid-body model is an efficient tool for analysis and design of compliant

mechanisms with the assumption of large and nonlinear deflections [17]. Although
there are other approaches available for large deflection problems such as elliptic
integral and nonlinear finite-element methods, the formulations are cumbersome
and solutions are time-consuming. The pseudo-rigid-body model, on the other hand,
provides a rather simple method for large and nonlinear deflection modeling in that
the rigid-link mechanism theory can be used to analyze and design the compliant
mechanisms. The essential of the modeling concept is: the deflection of compliant
mechanisms is modeled using rigid-body components, and the lumped springs
are employed to simulate the equivalent force–deflection characteristics. Different
compliant mechanisms require different pseudo-rigid-body models to accurately
predict the deflection path and force–deflection relationships. In achieving this, the
key issues of pseudo-rigid-body modeling are to decide where to place the pin joints
and the value of the spring constants. In what follows, we present two common
flexible segments that may be useful in design the compliant guiding mechanisms
when large and nonlinear deflections are needed.
First we consider the small-length flexural pivots, as shown in Fig. 5.15 [17].
The beam consists of two segments: one is short and flexible, and the other is
long and rigid. This beam can be treated as lumped compliant mechanism if the
small segment is significantly shorter and more flexible than the large segment. In
this case the pseudo-rigid-body model can be defined as shown in Fig. 5.16 [17],
where the two rigid links joined at a pin joint. The pin joint is called characteristic
pivot. The pivot can be put at any point along the flexible segment with no loss of
accuracy because the length of the flexible segment is much small compared to that
of the rigid link. For simplicity, however, the characteristic pivot is usually placed
at the center of the short segment. Apparently, the accuracy of the model will be
improved for the decreased length of flexible segment. The spring constant value can
be found from elementary beam theory. The pseudo-rigid-model is more accurate if
the bending is the dominant loading in the flexural pivot. If the transverse and axial
5 Electro-Magnetic Nano-Positioning 167

Fig. 5.15 Small-length flexural pivot [17]

Fig. 5.16 Simple pseudo-rigid-body model of the small-length flexural pivot [17]
168 Z. Zhang et al.

Fig. 5.17 (a) Fixed-guided flexible segment and (b) its pseudo-rigid-body model [17]

loads become significant, greater errors will be introduced using this simple pseudo-
rigid-body model. Detailed descriptions of this model are presented in reference
[17].
The electro-magnetic nano-positioning stages are often guided by the fixed-
guided a flexible segment, as shown in Fig. 5.17a, of which one end of the beam
is fixed while the other maintains at a constant angle. The associated pseudo-
rigid-model is shown in Fig. 5.17b. Details of the pseudo-rigid-model can be
found in [17]. It should be noted that for analysis and design of certain compliant
mechanisms, the pseudo-rigid-body model is not unique. Actually, all the compliant
mechanisms can be thought of as a sequence of infinite number of rigid bodies
linked by springs. In modeling the compliant mechanisms, one has to make a
compromise between the accuracy of the pseudo-rigid-body model and complexity
of analysis and design. Fewer springs are preferred in pseudo-rigid-body model
if it fulfills the accuracy requirements. As a result, the pseudo-rigid-body model
conducts the analysis and design of the compliant mechanisms based on the
methodologies used for rigid-body mechanisms. The modeling method is not only
5 Electro-Magnetic Nano-Positioning 169

suited to the kinematics but also the dynamics of compliant mechanisms. The main
drawbacks of the pseudo-rigid-body model is that it only applies to compliant
structure with regular geometry such as beams with constant cross-sectional area
so that the elementary beam theory can be used to determine the stiffness constant.

5.3.2.2 Optimal Synthesis with Continuum Models

As mentioned, pseudo-rigid-body model and beam theory are suited only for
structure that has regular geometry. However, for compliant mechanisms design,
the structural geometry is not necessarily to be regular. Analysis, modeling, and
design of compliant mechanisms with irregular geometry can be accomplished by
using continuum model instead [23–25]. Based on continuum solid mechanics,
structural optimization techniques can be used for compliant mechanisms design.
By specifying the desired force–deflection behavior, materials, and design domain
(i.e., the region the mechanism to be designed must fit), the structural optimization
method generates the optimal solution automatically. The optimal design can be
based on many design criteria such as maximum stiffness, minimum weight,
suitable flexibility, and natural frequency, etc. For irregular geometries, the elastic
deflection behavior is governed by partial differential equations, of which closed-
form solutions are not possible. These equations are solved using numerical methods
such as finite-element method, finite difference method, boundary element method,
etc. The detailed introduction of this method can be found in [17].

5.4 Kinematics and Dynamics

Previous sections introduce the design issues of the two main components of the
electro-magnetic nano-positioning systems, i.e., the electro-magnetic actuators and
the guiding mechanisms. In the analysis and design of a whole electro-magnetic
nano-positioning system, however, the primary issue is the kinematic analysis, for
example, deciding how many actuators with single- or multi-axes are required
in achieving stage motions with certain DOFs, and how the actuators should be
arranged. In view of control, it is important to understand that how these actuators
should be controlled in achieving desired motion. In addition to kinematics, some
dynamics issues should also be taken into account.

5.4.1 Degrees of Freedom

The DOFs of the positioning system is essential in the study of kinematics and
dynamics. In general, stage motion with several DOFs requires at least an equal
number of properly arranged independent actuation inputs [26]. Unlike traditional
170 Z. Zhang et al.

Fig. 5.18 Working principle of a six-axial positioning stage [27]

mechanisms that is capable of achieving rotations by directly using the rotation

actuators. The rotation motion of the electro-magnetic nano-positioning stage is
only accomplished based on the couple, because the electro-magnetic input is
always a force rather than a torque. Thus a combination and arrangement of
these actuation forces becomes the primary design consideration. Figure 5.18 [27]
shows a typical combination and arrangement of actuations for six-axial stage
applications. The upper three in-plane actuation forces are combined to generate in-
plane motions, i.e., from left to right, x- and y-axial rectilinear motions, and rotation
with respect to z-axis. The lower three vertical actuation forces are combined to
generate out-of-plane motions, i.e., from left to right, rotations about x and y axes,
and translation along z axis. These actuation forces are combined in symmetric
manner to create independent motions of the stage, i.e., three independent axial
motions and three independent rotations about each axis. Thus the one-to-one
mapping between actuation forces and stage motions can be built. Similar layouts
of the actuation forces are presented with details in [26].

5.4.2 Kinematics

The kinematics describes the relationship between the stage motions and actua-
tion input motions. Different from conventional mechanisms kinematics such as
kinematics of robotics that the kinematics are governed by geometric relationships
caused by the rigid links, the kinematics of electro-magnetic nano-positioning
systems are difficult to define because the motions are mainly governed by the
stiffness of the compliant guiding mechanisms. In other words we need to consider
the relationship between input actuation forces and resultant stage motions rather
5 Electro-Magnetic Nano-Positioning 171

than only the relationship between input and output motions. In current research,
assumptions of linear deformations of compliant guiding mechanisms, small stage
displacements, and constant actuation forces over the range of motion are made.
Euler’s beam theories and Castigliano’s second theorem can be used to derive the
stiffness of compliant guiding mechanisms that has regular geometries. For irregular
geometries, closed-form solutions may be challenging. Numerical solutions such as
finite-element analysis may be used to determine the stiffness. Because of these
assumptions, matrices are employed for building the input–output linear kinematic
mappings.
As in the six-axis positioning stage layout, here, we employ the stiffness matrix
K of the compliant guiding mechanism which relates the actuation force F on the
stage to the stage displacements x, as in (5.6) [8]. The actuation force on the stage
is related to the actuation force acting on the paddle through force transformation
matrix T, and the actuation force on the paddle is related to the input vector i
(voltage, current, etc.) through actuation matrix, Ka . Ka is derived from the electro-
magnetic analysis. With this, the actuation force on the stage is given by (5.7) [8].

F D Kx (5.6)

F D TKa i (5.7)

Equating (5.6) and (5.7), we derive the system input–output relationships, as

in (5.8) [8]. For the objective stage motions, one can calculate the required input
vectors using (5.8) if these matrices are invertible. Because the stage center of mass
is on the plane of the horizontal actuations, the horizontal and vertical motions are
decoupled.

i D .TKa /1 Kx (5.8)

5.4.3 Dynamics

With the assumption of linear system and lumped stage mass and damping, the
dynamics of the system is given by [8]

M xR C BPx C Kx D TKa i (5.9)

where M, B, and K are the mass, damping, and stiffness matrices of the system.
Coupling between the in-plane and out-of-plane motions generates unwanted
erroneous motions and add complexity of control design. It can be decoupled by
placing the stage center of mass on the plane of the horizontal actuations. However,
as will be presented in later section, in certain applications, the coupling effect
cannot be eliminated in this way, and it could only be minimized via control
172 Z. Zhang et al.

methods. For most of the nano-positioning systems that are not lumped mass and
guided by compliant mechanisms, it is difficult to establish closed-form dynamics
equations. However, the dynamics of the system can be identified using simulations
and experiments.

5.5 Control

The control of electro-magnetic nano-positioning systems varies with respect to

different applications. Thus in this section, we present the control schemes on the
typical application of electro-magnetic nano-positioning application, the probe-data
storage device, as in Figs. 5.1 and 5.7. In this system [28], a MEMS-based micro-
scanner with x/y-directional motions is used to position the storage medium relative
to the array of probes so that data is written/read/erased. Each actuator consists of
a pair of magnets and a solenoid in the middle. Upon application of a current, one
magnet is attracted and the other is repelled by the solenoid, forcing the shuttle to
move in the same direction. If the direction of current is altered, the shuttle will
be pushed in opposite direction. Before introducing the control schemes, we first
present sensors that are used to provide the positional information of the micro-
scanner.

5.5.1 Position Sensors

5.5.1.1 Thermal Position Sensors

The working principle of the thermal positional sensors is depicted in Fig. 5.19
[29]. The devices have the form of “U”-shaped cantilever, which consists of two
supporting legs and one resistive heater in-between. The devices are made from
single-crystal silicon using standard bulk micromachining technology. The resistive
heater is made from moderately doped silicon, whereas the legs are made from the
highly doped silicon. When applied a voltage on the legs, current flows through
the heater resulting in temperature increases of the device. To ensure a uniform
temperature increases along the resistive heater, the legs are usually fabricated with
equal length.
Figure 5.19c, d illustrates the working principle of the thermal position sensor.
To use the device as a position sensor, it should be positioned right above the surface
of the object of interest and the surface of the sensor should be parallel to the object
surface. Since the larger the overlapped area between the sensor and the underlying
object leads to more efficiency in cooling, the temperature of the sensor should be
lower than that of the less overlapped case. On the other hand, the resistance of
the sensor increases with an increased temperature and decreases with decreased
temperature. This variation in resistance can be monitored by looking at the current
5 Electro-Magnetic Nano-Positioning 173

Fig. 5.19 (a) Schematic diagram of the thermal position sensor. (b) Application of a voltage on the
legs results in current flow through the resistive heater. (c) The current causes the heat generation
of the resistive heater, which is conducted to the underlying surface through the air layer. (d) As
the underlying surface is moved towards the left, more heat is transferred to the surfaces, which
results in temperature decreases of the resistive heater and its electrical resistance accordingly [29]

flowing through the beams. When the sensor is applied by a constant voltage, the
change in overlapped area can be thus indirectly detected by measuring the current
passing through the sensor. To obtain the position information, the long axis of the
resistive heater should be aligned parallel to the direction of motion. Hence, the
position can be sensed by monitoring the current flowing through the sensor. Usually
for both x and y axial position sensing, two sensors are combined and arranged
in orthogonal directions independently. The thermal position sensor of an electro-
magnetic nano scanner is shown in Fig. 5.7b.
The single thermal sensor is capable of achieving a satisfactory resolution.
However, the sensing results are affected by fluctuation in ambient temperatures
and ageing effects owing to dopant diffusion or oxidation. These effects can be
minimized by employing a pair of same thermal sensor and make a differential
measurement. As shown in Fig. 5.20a, a pair of sensor positioned with an equal
overlapped area will give rise to approximately equal cooling of the resistive
heaters and therefore equal temperatures and resistances of the devices. As the
underlying object displace towards the left, the difference in overlapped areas leads
to difference in temperature and resistance of the resistive heater accordingly. In
this arrangement, the fluctuation in ambient temperature and ageing effects for both
sensors are approximately equal and thus the difference between the resistances will
not be affected.

5.5.1.2 Medium-Derived PES

The thermal position sensors provide universal position information with nanometer
resolutions. The probe-storage device, on the other hand, provides an alternative
174 Z. Zhang et al.

Fig. 5.20 Differential measurement set-up using a pair of matched thermal sensors. (a) The test
object is positioned between the two thermal sensors leading to equal cooling, and results in
approximately equal resistances of the two sensors. (b) Because the object is moved towards the
left, the cooling of the left sensor increases and the right sensor decreases, and this give rise to a
decrease in resistance of the sensor on the left (R1 ) and increase of the sensor on the right (R2 ) [29]

position sensing technique that is convenient in measuring the deviation of the

read/write microcantilever tip from the track center. In addition to data fields, certain
storage fields (called servo field) and the corresponding cantilever tips are reserved
for generation of position signal, as in Fig. 5.21a. The servo field is shown in
Fig. 5.21b, where the circle represents the indentations inscribed prior to the regular
operation of the device. The x-direction distance of indentations in all bursts is equal
to the bit pitch (BP). The cross-track distance between indentations center line of
the same burst is equal to the track pitch (TP), whereas the distance between burst
centers of A and B (or C and D) is TP/2 [30].
These prewritten indentations provide medium-derived position-error signal
(PES), which measures the position deviation in cross-track direction. In regular
operations, the track centerline of the data fields is aligned with that of the burst
C. Bursts A and B are combined to create the in-phase signal (I-signal), and burst
C and D are used to generate quadrature signal (Q-signal). Moreover, the PES is
obtained over discrete time step which is needed by the scanner to move a BP in the
x-direction. Thus the accurately data writing or reading operations along x-direction
is ensured by medium-derived PES. Compared with the thermal position sensors
that are capable of offering absolute position information, the medium-derived PES
provides only the relative position deviations in cross-track directions. In other
words, the thermal position sensors have a larger range of operation, whereas
the medium-derived PES confines to a limited range of sensing. The two sensors
provide similar resolutions.
5 Electro-Magnetic Nano-Positioning 175

Fig. 5.21 Concept of the medium-derived PES [30]

5.5.1.3 Noise Characteristics

In the absence of external disturbances, positioning accuracy is markedly affected

by the noise characteristics of the sensors. For the thermal position sensor, the
predominant noise sources are Johnson noise and 1/f -noise of the silicon resister,
whereas the predominant noise sources of the medium-derived PES are medium
and channel noise [30]. Moreover, the thermal position sensor suffers from drift and
sensitivity variations. The noise spectral characteristics of both the thermal position
sensor and medium-derived PES are shown in Fig. 5.22 [30]. As can be seen from
Fig. 5.22, there is a significant low-frequency component of the thermal position
sensor. This will result in the measurement particularly noisy at low frequencies. In
contrast, the medium-derived PES shows good low-frequency fidelity. It should be
noted that beyond a certain frequency, performance of the thermal position sensor
is better than that of the medium-derived PES.
176 Z. Zhang et al.

Fig. 5.22 The characteristics of the thermal position sensors and the medium-derived position-
error signal (PES) [30]

5.5.2 Control Schemes

The control schemes are based on actual requirements in positioning. As with the
PDS devices, the control system has mainly two functions. First, the scan table is
to be located to a target track from an arbitrary initial position. This procedure is
known as seek-and-settle. Second, from the target position, the medium scanner
should be moved with constant velocity along the length of the track along which
the writing/reading are operated. The probes should be maintained on the center of
the target track while operation. This is achieved by the track-follow procedure.

5.5.2.1 Cross-Coupling

In this planar configuration of the electro-magnetic nano-positioning system, x-

and y-directional motion is achieved independently by their respective actuators.
However, there are cross-coupling components between the axes in the system
owing to the compliant guiding mechanisms. As shown in Fig. 5.23 [31], the
cross-coupling is nonlinear and position-dependent. In design of control system,
the cross-coupling is usually seen as the disturbance that has to be eliminated by the
control.
5 Electro-Magnetic Nano-Positioning 177

Fig. 5.23 Cross-coupling of x- and y-axes [31]

5.5.2.2 Seek-and-Settle

The seek-and-settle procedure is essential in data access time, and thus minimizing
the duration in this procedure becomes an important control objective. Since
medium-derived PES is only capable of providing deviation information in cross-
track direction, in this seek-and-settle control procedure the thermal position sensors
are thus employed to form a close-loop control scheme. Detailed control architecture
can be found in [32].

5.5.2.3 Track-Follow

Different from the seek-and-settle control that the control of x- and y-directional
motions of the scanner is equivalent, the track-follow control exhibits different
control requirements in the two axes. As mentioned earlier, the x-direction scan
should be kept at constant velocity to avoid timing jitter in the readback signal.
Fortunately, low-frequency of positioning error is tolerable thanks to the timing-
recovery circuits. This makes using the thermal position sensor acceptable despite
its unreliability in measurement of errors at low frequencies. For y-direction control,
the main objective is to maintain the probes to be at the track centerline while
scanning in x-direction. In achieving this, both the low and high frequencies have
to be minimized. The thermal position sensors, therefore, cannot be fully trusted
because of the low-frequency noise, drift, and sensitivity variations. However, it is
advantageous to use the thermal position sensor due to its virtues of large range and
178 Z. Zhang et al.

great precision at high frequencies. To avoid the low-frequency noise induced by

using the thermal position sensors, the medium-derived PES can be utilized instead.
The two-sensor-based control strategy [32], also known as the frequency-
separated control [15, 30], is usually employed for the track-follow procedure. This
control scheme is to achieve best error measurement by using thermal position
sensor in high frequencies and the medium-derived PES in low frequencies.
Moreover, the medium-derived PES can be exploited to estimate drift and make
correction of using thermal position sensor at low frequencies.

5.6 Fabrication

Fabrication of macro-scale electro-magnetic nano-positioners is accomplished

based on traditional approaches, and in this section we only focus the fabrication
of meso- and micro-scale positioning systems. The fabrication of these positioning
systems can be divided into two main parts: fabrication of the micro coil and the
compliant guiding mechanisms. The compliant guiding mechanisms or flexure-
supported platform is usually fabricated using a standard silicon-on-insulator
MEMS (SOI-MEMS) micromachining technology. Figure 5.24 [33] shows the
fabrication process of a flexure-supported platform.
In fabrication of the flexure-supported platform, the most important considera-
tion is the thickness of the wafer. Since for planar scanning applications, out-of-
plane stiffness should be ensured. In this regard, the wafer should be fabricated
thicker. However, thicker wafer increases the etching time as a tradeoff. For out-of-
plane applications, the structure should be fabricated more but not too flexible to
both ensure out-of-plane motion and eliminate the gravity-induced errors.
Fabrication of micro coil is achieved based on two approaches, as in Fig. 5.25
[10]. One approach uses the patterned micro-mold to fabricate Cu coil with high
aspect ratio, shown in Fig. 5.25a. There are various mold materials available such
as SU-8, polyimide, photoresist, and silicon, etc. Fabrication procedure consists of
mainly three steps: mold patterning, electroplating, and removal of mold. However,
fabricating a high aspect ratio mold and removing the mold after electroplating are
difficult. Another approach is to fill Cu inside Si trench by electroplating and remove
the overflowed Cu using CMP (chemical mechanical polishing). The mold of high
aspect ratio Cu coil is easily fabricated using ICP-RIE (inductively coupled plasma
reactive ion etching) with high resolution, which ensures a lower resistance coil
fabrication. In addition, the flat surface thanks to the CMP makes it much easier for
other layers to assemble compared with the conventional approach. The procedure
can be divided into three steps: seed layer deposition, electroplating inside Si trench,
and CMP.
In constructing the whole system, one approach is to assemble the two chips
(one with the micro coil and the other one with the flexible-supported platform)
5 Electro-Magnetic Nano-Positioning 179

Fig. 5.24 Fabrication process for flexure-supported platform. (a) Ni/Cr layer are deposited using
liftoff, (b) platform shape is etched in the device layer using DRIE, (c) handle wafer is etched using
DRIE, and (d) platform is released by etching the SiO2 layer in HF and critical point drying [33]

together with the aid of microscope. To ensure high electro-magnetic force, the
distance between the coils and magnets should be minimized. Another approach
is to integrate the fabrications of micro coils and flexible platform together. The
choice of these two approaches depends on different applications.

5.7 Conclusions

Electro-magnetic nano-positioning technology is a relatively young field in nano

technologies. This chapter presents a general introduction and guidance on electro-
magnetic nano-positioning the aspects covered involve electro-magnetic actuation,
guiding mechanisms, system kinematics and dynamics, control, and fabrications.
These aspects focus mainly on the most common and general concepts and prin-
ciples of various electro-magnetic nano-positioning systems. Therefore, potential
180 Z. Zhang et al.

Fig. 5.25 Fabrication and process of Cu coil [10]

readers may benefit from it via an overview understanding of the nano-positioning

systems. The aim is to provide researchers with guidance on developing high-
performance nano-positioning systems for certain applications. In achieving this, a
variety of configurations of electro-magnetic nano-positioning systems are reviewed
and used as examples to illustrate the related methodologies and technologies.
Critical issues for each step in developing the systems are reviewed and summarized.
Not limited to current research, this chapter also presents some potential methods
such as the pseudo-rigid-body model that may be helpful in the design and analysis
of electro-magnetic nano-positioning systems.

References

1. D. Golda, M. Culpepper, Multi-axis electromagnetic moving-coil micro-actuator, in ASME

2006 International Mechanical Engineering Congress and Exposition, no. Imece2006-15191
(2006), pp. 57–63
2. G. Haugstad et al., Scanning Probe Microscopy (Wiley, 2013)
3. E. Eleftheriou et al., Millipede—a MEMS-based scanning-probe data-storage system. IEEE
Trans. Magn. 39(2), 938–945 (2003)
4. S. Naberhuis, Probe-based recording technology. J. Magn. Magn. Mater. 249, 447–451 (2002)
5. A. Malshe, K. Virwani, K. Rajurkar, D. Deshpande, Investigation of nanoscale electro
machining (nano-EM) in dielectric oil. CIRP Ann. Manuf. Technol. 54(1), 175–178 (2005)
6. M. Khatib, MEMS-based storage devices: integration in energy-constrained mobile systems.
Doctoral Dissertation, University of Twente, 2009
7. M. Khatib, E. Miller, P. Hartel, Workload-based configuration of MEMS-based storage devices
for mobile systems, in Proceedings of the 8th ACM & IEEE Conference on Embedded Software
(October 2008)
5 Electro-Magnetic Nano-Positioning 181

8. D. Golda, Design of high-speed, meso-scale nanopositioners driven by electromagnetic

actuators. Doctoral Dissertation, MIT, February 2008
9. S. Verma, W. Kim, J. Gu, Six-axis nanopositioning device with precision magnetic levitation
technology. IEEE/ASME Trans. Mechatronics 9(2), 384–391 (2004)
10. J. Choi, H. Park, K. Kim, J. Jeom, Electromagnetic micro x-y stage for probe-based data
storage. J. Semicond. Technol. Sci. 1(1), 84–93 (2001)
11. C. Jaejoon, P. Hong sik, K. Kim, J. Jeon, Electromagnetic micro x-y stage with very thick Cu
coil for probe-based mass data storage device. Proc. SPIE 4334, 363–371 (2001)
12. M. King, B. Zhu, S. Tang, High-precision control of a maglev linear actuator with nanoposi-
tioning capability. Mob. Robots 8(2), 520–531 (2001)
13. W. Kim, H. Maheshwari, High-precision magnetic levitation stage for photolithography, in
Proceedings of the American Control Conference (2002), pp. 4279–4284
14. W. Kim, D. Trumpert, High-precision planar magnetic levitation. Precis. Eng. 66–77 (1998)
15. A. Pantazi et al., Control of MEMS-based scanning-probe data-storage devices. IEEE Trans.
Control Syst. Technol. 15(5), 824–841 (2007)
16. M. King, B. Zhu, S. Tang, 6-Axis electromagnetically-actuated meso-scale nanopositioner.
IEEE Control Syst. Mag. 26–35 (2008)
17. L. Howell, Compliant Mechanisms (A Wiley-Interscience Publication, 2001)
18. Q. Xu, Design and development of a flexure-based dual-stage nanopositioning system with
minimum interface behavior. IEEE Trans. Autom. Sci. Eng. 9(3), 554–563 (2012)
19. Y. Yong, S. Moheimani, B. Kenton, K. Leang, Invited review article: high-speed flexure-guided
nanopositioning: mechanical design and control issues. Rev. Sci. Instrum. (83) (2012)
20. G. Anderson, A six degree of freedom flexural positioning stage, Masters Thesis, MIT, 1999
21. M. Culpepper, G. Anderson, P. Petri, HEXFLEX: a planar mechanism for six-axis manipula-
tion and alignment. Patent pending, Massachusetts Institute of Technology, Cambridge, 2001
22. M. Lantz et al., A vibration resistant nanopositioner for mobile parallel-probe storage
applications. J. Microelectromech. Syst. 16(1), 130–139 (2007)
23. S. Kota et al., Design of compliant mechanisms: applications to MEMS. Analog Integr. Circ.
Sig. Process. 29, 7–15 (2001)
24. S. Deepak, M. Dinesh, D. Sahu, G. Ananthasuresh, A comparative study of the formulations
and benchmark problems for the topology optimization of compliant mechanisms. J. Mech.
Robot. 1 (March 2008)
25. M. Bendsoe, O. Sigmund, Topology Optimization Theory (Springer, 2013)
26. J. Zhao, K. Zhou, Z. Feng, A theory of degrees of freedom for mechanisms. Mech. Mach.
Theory 39, 621–643 (2004)
27. C. DiBiasio, M. Culpepper, Design and characterization of a 6 degree-of-freedom meso-
scale nanopositioner with integrated strain sensing, in ASPE Proceedings, Design of Precision
Machines (2010)
28. S. Devasia, E. Eleftheriou, S. Moheimani, A survey of control issues in nanopositioning. IEEE
Trans. Control Syst. Technol. 15(5), 802–823 (2007)
29. M. Lantz, G. Binnig, M. Despont, U. Drechsler, A Micromechanical Thermal Displacement
Sensor with Nanometre Resolution, vol 15 (Institute of Physics Publishing, 2005)
30. A. Sebastian et al., Nanopositioning for probe-based data storage. IEEE Control Syst. Mag.
(2008)
31. A. Sebastian et al., Achieving subnanometer precision in a MEMS-based storage device during
self-servo write process. IEEE Trans. Nanotechnol. 7(5), 586–595 (2008)
32. A. Sebastian, A. Pantazi, Nanopositioning with multiple sensors: a case study in data storage.
IEEE Trans. Control Syst. Technol. 20(2), 382–394 (2012)
33. Y. Choi et al., A high-bandwidth electromagnetic MEMS motion stage for scanning applica-
tions. J. Micromech. Microeng. 22 (2012)
Chapter 6
Modeling of Piezoelectric-Actuated
Nanopositioning Stages Involving
with the Hysteresis

Guo-Ying Gu, Li-Min Zhu, Chun-Yi Su, and Han Ding

Abstract In this chapter, a general skeleton on modeling of piezoelectric-actuated

nanopositioning stages (PANSs) involving with the hysteresis is presented. Toward
this framework, a general model, including nonlinear electric behavior, voltage-
charge hysteresis, piezoelectric effect, and frequency response of the stage, is
developed to completely characterize dynamic behaviors of the PANS with both
electrical and mechanical components. For description of the hysteresis nonlinearity,
we firstly give a brief survey of the available hysteresis models in the literature.
Then, in order to validate the general model, the developed general model is divided
into two modules in terms of a three-order linear dynamic plant preceded by an
input hysteresis nonlinearity that is described by a modified Prandtl–Ishlinskii (MPI)
hysteresis model as an illustration. Subsequently, parameters of the linear part
and nonlinear part of the general model are identified, respectively. For the linear
part, an axiomatic-design-theory-based approach is adopted to identify the linear
plant. For the nonlinear part, a particle swarm optimization algorithm is utilized
for identification of the MPI model on basis of the identified parameters of the
linear part in the first step. Finally, the experimental and model simulation results of
a commercial PANS clearly verify the effectiveness of the proposed general model
and the corresponding identification method.

6.1 Introduction

Along with the rapid development of nanoscience and nanotechnology,

piezoelectric-actuated nanopositioning stages (PANSs) are becoming more and
more significant owing to their nanometer or sub-nanometer displacement

G.-Y. Gu () • L.-M. Zhu • H. Ding

State Key Laboratory of Mechanical System and Vibration, Shanghai Jiao Tong University,
Shanghai 200240, China
e-mail: [email protected]; [email protected]; [email protected]
C.-Y. Su
Concordia University, 1455 de Maisonneuve Blvd. W., Montreal, Quebec, Canada, H3G 1M8
e-mail: [email protected]

© Springer International Publishing Switzerland 2016 183

C. Ru et al. (eds.), Nanopositioning Technologies,
DOI 10.1007/978-3-319-23853-1_6
184 G.-Y. Gu et al.

resolution in many precision manufacturing equipments [13, 31, 65, 75], such
as micro/nano-manipulations, scanning probe microscopes, and ultra-precision
machine tools. A PANS system is generally composed of a flexure-hinge-
based mechanism, a piezoelectric actuator, a piezoelectric drive amplifier, and
a displacement sensor. Such a mechatronic system has attracted significant
attentions in the literature on its system modeling, control design, and experimental
applications due to the inherent complex hysteresis nonlinearity exhibited in the
piezoelectric actuators [25, 26, 29, 32, 33, 52, 76, 88].
Generally speaking, the modeling of the PANS is essential to account for
dynamic effects due to frequency response of the stage, voltage-charge hysteresis,
and nonlinear electric behavior, which is also related and important for control
design. However, in previous reported works [1, 19, 20, 22, 67], the attentions
focused on different aspects with different assumptions. For example, in [1, 20, 22,
67], the developed models mainly focused on the piezoelectric actuator itself, while
the electric behavior of the piezoelectric drive amplifier was simply ignored. On the
other hand, although Gao [19] proposed a linear modeling approach for the PANS
including the behavior of the piezoelectric drive amplifier, the hysteresis effect of
the piezoelectric actuator was totally missed.
It is obvious from the above discussions, a general skeleton on modeling of
the PANS is still missing. Based on the previous accumulated results, we propose
a general model for the united descriptions of the PANS with both electrical
and mechanical components. The proposed model reflects dynamic effects of
the PANS including the nonlinear electric behavior, voltage-charge hysteresis,
piezoelectric effect, and frequency response of the stage, so that any reported model
in the literature can be deduced from it, depending on the different applications.
To illustrate the validity of the general model, a modified Prandtl–Ishlinskii (MPI)
hysteresis model is utilized as one of the hysteresis models to describe the hysteresis
effect in the stage. Finally, identification and experimental results are conducted to
confirm such a model.
The remainder of this chapter is organized as follows. Section 6.2 starts with
description of the PANS. In Sect. 6.3, the proposed general model is detailed and
the available hysteresis models are surveyed in Sect. 6.4. In Sect. 6.5, the identified
method and model validation with experimental results are presented, and Sect. 6.6
concludes this chapter.

6.2 Piezoelectric-Actuated Nanopositioning Stage

6.2.1 System Description

Figure 6.1 shows an architecture of a PANS system. Therein, (1) the flexure-
hinge-guided mechanism is usually employed to provide motion by the elastic
deformations, which has the advantages of the monolithic structure and no sliding
6 Modeling of Piezoelectric-Actuated Nanopositioning Stages Involving: : : 185

Fig. 6.1 Architecture of a PANS system

parts, thereby avoiding undesired nonlinear effects such as backlash and friction
[48]; (2) the piezoelectric actuator is applied to realize the actuation by generating
force on the mechanism due to its excellent advantages of the large output force,
high displacement resolution, and fast response time [13, 31]; (3) the piezoelectric
driver amplifier (PDA) used to supply power for the piezoelectric ceramic actuator
(PCA) by either voltage control or charge control [59]; (4) a displacement sensor
[16], such as the inductive, piezoresistive, capacitive, or optical sensors, with high
resolution can be utilized to measure the real-time motion displacement of the stage
and the sensor module is used to convert the sensor’s signals to low-voltage signals
for the control systems; (5) the control algorithms are implemented in the real-time
control board to produce control commands for the piezoelectric actuator to achieve
the nanopositioning motion of the flexure-hinge-guided mechanism.

6.2.2 Working Principle

In general, the key enabling element of a PANS is the piezoelectric actuator, which
can manage small displacements in the range of sub-nanometers to several hundreds
micrometers. Therefore, the main advantages of the PANSs are due to the utilization
of piezoelectric actuators.
The piezoelectric actuator, made of piezoelectric materials, is a specific type of
smart material-based devices that can directly convert an electrical signal into a
physical displacement [31, 41, 62]. In the early time, the most applied piezoelectric
materials were piezoelectric crystals such as Quartz, tourmaline, and Rochelle
salt. The discovery of piezoelectric ceramic materials, such as barium titanate
and lead zirconate titanate (PZT) brought a major breakthrough owing to their
high Curie temperature and outstanding piezoelectric effect [41]. Compared to
conventional actuators, piezoelectric actuators can be made smaller and lighter
in a variety of forms, ranging from rectangular patches, thin disks, and tubes. In
applications, piezoelectric actuators are generally fabricated with three shapes [62],
i.e., piezoelectric bender actuators, piezoelectric tube actuators, and piezoelectric
stack actuators.
186 G.-Y. Gu et al.

Piezoelectric actuators work on the principle of piezoelectricity, which is an

electromechanical coupling phenomenon between electrical properties and mechan-
ical properties of piezoelectric materials. Mechanical stresses in the piezoelectric
materials produce measurable electric charge, which is referred to the direct
piezoelectric effect. Vice versa, mechanical strains are generated in response to an
applied electric field and this is called the inverse piezoelectric effect.
The most famous description of the piezoelectricity was published by a standards
committee of the IEEE Ultrasonics, Ferroelectrics, and Frequency Control Society
with the linearized constitutive equations as follows [13, 31, 41, 58]

Si D sEij Tj C dki Ek
(6.1)
Dm D dmj Tj C "Tmk Ek ;

where the subscript indexes i; j D 1; 2; : : : ; 6 and m; k D 1; 2; 3 refer to different

directions within the Cartesian coordinate system, S and T represent the strain
tensor and stress tensor, respectively, E and D are the electric field vector and
electric displacement vector, respectively, sE is the elastic compliance matrix when
subjected to a constant electrical field, d is the matrix of piezoelectric material
constants, "T is the dielectric constant matrix under condition of constant stress.
The first subscript index of the d-constant gives the “electrical” direction (i.e., field
or dielectric displacement) and the second one refers to the direction of mechanical
strain or stress.
The equations in (6.1) essentially state that the material stain and electrical
displacement are linearly affected by the subjected mechanical stress and electrical
field. However, these equations fail to explicitly describe the hysteresis nonlinearity
in piezoelectric actuators and the electromechanical dynamics that are observed in
PANSs. Therefore, lots of electromechanical models are reported in the literature to
describe the nonlinear dynamic behaviors of PANSs, which will be discussed in the
following.

6.3 Comprehensive Modeling Approach of PANSs

Comprehensive dynamic modeling of PANSs refers to develop an electromechanical

modeling method by completely representing the different relations (as shown in
Fig. 6.2) involved in the PANS.
This method was pioneered by Goldfarb et al. [22] and Adriaens et al. [1],
which was developed for the PCA. Moreover, Gao et al. [19] developed a linear
modeling approach for the PANS, but the hysteresis effect of the PCA was ignored.

Fig. 6.2 Schematic representation of different relations in PANSs

6 Modeling of Piezoelectric-Actuated Nanopositioning Stages Involving: : : 187

a b

Fig. 6.3 A general model of the PANS: (a) electrical aspect; (b) mechanical aspect

Following the accumulated results developed for the PCA in [1, 19, 22], we develop
a general schematic model of the PANS, which can be represented by Fig. 6.3. In the
following development, we shall use this structure as a base to completely represent
the characteristics of the PANS, assembling to the traditional DC motor, which will
be described from both electrical and mechanical aspects.

6.3.1 Electrical Modeling

From the electrical aspect, the PANS system can be modeled as an equivalent circuit
as shown in Fig. 6.3a, which is composed of a circuit of the PDA and a circuit of the
PCA. From Fig. 6.3a, the PDA is modeled as a fixed gain kamp plus an equivalent
internal resistance R0 of the driving circuit, vh is the generated voltage due to the
hysteresis effect H.q/, Tem represents the piezo effect, which is an electromechanical
transducer with transformer ratio, and CA represents the sum of the capacitances of
the total piezoelectric ceramics, which is electrically in parallel with the transformer.
The total charge in the PCA is q, and the resulting current flowing through the circuit
is qP . The charge stored in the linear capacitance CA is represented as qc . The charge
qp is the transduced charge from the mechanical side due to the piezoelectric effect,
the voltage vA is the transduced voltage.
Therefore, the complete electrical equations can be expressed as [27]

R0 qP .t/ C vh .t/ C vA .t/ D kamp vin .t/ (6.2)

vh .t/ D H.q/ (6.3)

188 G.-Y. Gu et al.

q.t/ D qc .t/ C qp .t/ (6.4)

vA .t/ D qc .t/=CA (6.5)

qp .t/ D Tem x.t/; (6.6)

where vin is the control input for the PDA, x is the output displacement of the
mechanical part, and kamp is the fixed gain of the voltage power amplifier.
By substituting (6.3)–(6.6) into (6.2), the complete electrical equations can be
further simplified as
 
H.q/
R0 CA qP .t/ C q.t/  Tem x.t/ D CA kamp vin .t/  : (6.7)
kamp

Remark 1. It is important to note it is R0 that makes a difference in the electric

part between the PANS and PCA in [1, 22]. Although the set of linear constitutive
equations in [1, 22] is fine for many applications, a more realistic model requires
the inclusion of R0 . Due to the existence of R0 in the PANS, the electrical part can
be expressed as a differential equation of the charge q. When R0 D 0, it is reduced
to the set of linear constitutive equations as discussed in [1, 22], which cannot lead
to (6.7). The role of (6.7) will be further explained in the following development.

6.3.2 Electromechanical Modeling

In mechanical aspect, the following dynamic electromechanical equations can be

obtained according to the piezoelectric effect and Newton’s Laws of Motion

FA D Tem vA .t/ (6.8)

mRx.t/ C bs xP .t/ C ks x.t/ D FA ; (6.9)

where FA is the transduced force from the electrical side; x is the output displace-
ment of the mechanical part; m, bs , and ks are the mass, damping coefficient, and
stiffness of the moving mechanism, respectively.
To present vA .t/ in (6.8) in terms of q.t/, it is obvious from (6.4)–(6.6) that

1 Tem
vA .t/ D q.t/  x.t/: (6.10)
CA CA
6 Modeling of Piezoelectric-Actuated Nanopositioning Stages Involving: : : 189

The dynamic electromechanical equation (6.9) can then be re-written as

Tem
mRx.t/ C bs xP .t/ C ks x.t/ D q.t/; (6.11)
CA
2
Tem
where ks D ks C CA
.

6.3.3 A General Dynamic Model

Combining the electrical model (6.7) and the electromechanical model (6.11), a
general dynamic model of the PANS is represented by the following set of equations
 
H.q/
R0 CA qP .t/ C q.t/  Tem x.t/ D CA kamp vin .t/  (6.12)
kamp

Tem
mRx.t/ C bs xP .t/ C ks x.t/ D q.t/: (6.13)
CA

These two equations represent an analytical model of the PANS, which surpris-
ingly assembles to the model of the traditional permanent magnet DC motor.
Remark 2.
(i) A complete comprehensive model described in (6.12) and (6.13) accounts
for both the dynamics of the PANS and the hysteresis inherent to the PCA.
Surprisingly, until now there is no such a complete description available in the
literature though a lot of papers have addressed this issue [5, 9, 51, 52, 77].
Possible reason is that the most of available results focused on a piezoelectric
actuator itself, not on the stage, i.e., [1, 20, 22, 67], where the dynamic models
consisting of a second-order linear plant with hysteresis nonlinearities, focused
on the PCA and ignored R0 . The proposed model can be reduced to this case
if the electric circuit of the PDA is neglected, that is R0 D 0 in (6.7). On the
other hand, Gao [19] proposed a linear modeling approach including both PDA
and PCA. However, the hysteresis effect of the PCA is ignored. The proposed
model can also be reduced to this case without considering the hysteresis effect
H.q/ in (6.12).
(ii) It is well known that dynamic model of a permanent magnet DC motor can be
described as:

i.t/ .t/
L C Ri.t/ C Kemf D vin .t/ (6.14)
dt dt
J R .t/ C BP .t/ D Kt i.t/; (6.15)
190 G.-Y. Gu et al.

where i is the armature current,  is the angular position, L, R, Kemf , and Kt are
motor’s inductance, resistance, back emf constant, and torque constant, J is the
inertia of the rotor and the equivalent mechanical load, and B is the damping
coefficient. From Eqs. (6.12)–(6.13), it is amused that the dynamic model of
the PANS is so similar to the traditional DC motor, except the hysteresis
nonlinearity H.q/. Therefore, the challenge for control of the PANS mainly
lies on existence of the non-smooth hysteresis nonlinearity H.q/, which usually
degrades the system performance in such manners as giving rise to undesirable
inaccuracies or oscillations, even leading to instability in the closed-loop
manner [83].
(iii) It is worth of mentioning that the charge control of PCA instead of the
voltage control can eliminate the effect of hysteresis from (6.11), which is also
demonstrated in [17]. This is exactly similar to directly control the armature
current of a permanent magnet DC motor. However, as similar to the DC
motor, the charge control has not been widely adopted due to the complicated
implementation of the power driver and the cost of such techniques [13].
Therefore, this paper discusses the related modeling and control problems
when the voltage control approach is used.

6.4 Hysteresis Modeling

As discussed in the previous section, it is the hysteresis inherent in the PCA that
makes the modeling and control of the PANS challenging. In order to describe the
hysteresis nonlinearity in (6.12), we will first give a brief survey of the available
hysteresis models in this section.

6.4.1 What Is Hysteresis?

The word “hysteresis” is of the Greek origin, which means “lagging behind” or
“coming behind.” The Scottish physicist, Alfred Ewing, introduced this word into
scientific vocabulary as follows [42]: “when there are two quantities M and N, such
that cyclic variations of N cause cyclic variation of M, then if the changes of M
lag behind those of N, we may say that there is hysteresis in the relation of M
and N.” As pointed out by Mayergoyz [57], this may be misleading and can create
the impression that the looping is the essence of hysteresis. However, it must not be
confused with “delay” or “phase lag,” which is not a hard nonlinearity and is present
in many linear systems.
Hysteresis in the PCA is a kind of non-smooth nonlinear phenomenon between
the applied voltage and the output displacement of the piezoelectric actuator [30].
This nonlinearity is multivalued and nonlocal memoryless, which results in the
major- and minor-loops behavior as shown in Fig. 6.4. It can be seen that, the
6 Modeling of Piezoelectric-Actuated Nanopositioning Stages Involving: : : 191

80

70

60
Displacement (μm)

50

40

30

20

10

0
0 50 100 150
Voltage (V)

Fig. 6.4 An illustration of the hysteresis nonlinearity

displacement of the output depends upon not only the current value but also the
previous dominant extrema of the input voltage. In general, the maximum error
caused by the hysteresis can be as much as 15 % of the travel range. With the
increase of frequencies of the input signals, the hysteresis caused errors may go
beyond 35 %.
In general, for treatment of the hysteresis, the first step is to develop mathematical
models that are sufficiently accurate, amenable to controller design for hysteresis
compensation, and efficient enough to use in real-time applications. As a conse-
quence, a number of hysteresis models are developed in the literature. The reader
may refer to [31] for a detailed survey.
Roughly speaking, the reported hysteresis models for PANSs can be classified
into two categories: physics-based models and phenomenological models.
Figure 6.5 depicts the categories of these different hysteresis models.

6.4.2 Physics-Based Models

Physics-based models are built on first principles of physics effects. The Jiles–
Atherton model [40, 71] is the most famous physics-based model, which was first
introduced for ferromagnetic hysteresis. On the basis of the Jiles–Atherton model,
Smith and Ounaies [78] proposed a domain wall model for hysteresis description
192 G.-Y. Gu et al.

Fig. 6.5 Category of hysteresis models for PANSs

of the PCA with an ordinary differential equation formulation that was amenable
to construct an inverse compensator for the linear controller design. The physics-
based model is generally very complicated, and the physics-based model developed
for one material may not be used for another kind of material.

6.4.3 Phenomenological Models

In contrast to physics-based models, phenomenological models are used to produce

behaviors similar to those of physical models without necessarily providing physical
insight into the modeling problems.
According to the differences of the mathematical structure, phenomenological
models for the hysteresis description in the piezo-actuated stages can be classified
as differential-equation-based hysteresis models, integral-equation-based hysteresis
models (generally called operator-based hysteresis models as well) and other
hysteresis models.
6 Modeling of Piezoelectric-Actuated Nanopositioning Stages Involving: : : 193

6.4.3.1 Differential-Equation-Based Models

Differential-equation-based models refer to a hysteresis description method in

the form of differential equations [84]. As shown in Fig. 6.5, the Duhem model,
Backlash-like model, and Bouc–Wen model are popular with differential equations
to describe the hysteresis in PANSs.
(1) Duhem model
From 1897, the Duhem model [55] has been used for hysteresis description.
Although it is originally developed to describe the magnetic hysteresis, the Duhem
model is also effective to represent the hysteresis of the PCA [54, 91].
The Duhem model has the properties that every state is an equilibrium under
constant inputs and the output can only change its character when the input changes
direction [55, 64]. For suitable functions f1 and f2 , the Duhem hysteresis model
between the input v and output w is expressed as [55]

P D f1 .w.t/; v.t// vP C .t/  f2 .w.t/; v.t// vP  .t/;

w.t/ (6.16)

where
jv.t/j
P ˙ v.t/
P
vP ˙ .t/ D : (6.17)
2
In applications, the model (6.17) can be simplified to a special case [11, 55, 63]
ˇ ˇ
dw ˇ dv ˇ dv
C a ˇˇ ˇˇ g.v; w/ D b ; (6.18)
dt dt dt

where a typical choice for g is g.v; w/ D w  c.v/.

(2) Backlash-like model
Based on the simplified Duhem model (6.18), Su et al. [81] developed a
Backlash-like model, which is described as
ˇ ˇ
dw ˇ dv ˇ dv
D ˛ ˇˇ ˇˇ Œcv  w C B1 ; (6.19)
dt dt dt

where ˛, c, and B1 are constant, satisfying c > B1 . Following the derivation in [81],
the solution of the model (6.19) is explicitly expressed as

w.t/ D cv.t/ C d.v.t// (6.20)

with

d.v.t// WD Œw0  cv0 e˛.vv0 /sgn.v/

P
˛vsgn.v/
P
Rv ˛sgn.v/
P (6.21)
Ce v0 ŒB1  ce d
194 G.-Y. Gu et al.

for vP constant where w0 D w.0/ and v0 D v.0/ are the initial condition, and sgn is
the well-known Sign function. The benefit for choosing the Backlash-like model is
that the hysteresis nonlinearity is expressed as a linear function of the input signal
plus a bounded disturbance. Thus, the conventional robust control approaches can be
utilized to deal with the hysteresis without constructing the inverse of the hysteresis
[27, 69, 81, 95].
(3) Bouc–Wen model
On the basis of the Duhem model (6.16), a hysteretic semi-physical model
was proposed initially by Bouc early in 1971 and further generalized by Wen
in 1976 [60]. Since then, the Bouc–Wen model has been known and utilized to
represent the hysteresis in the form

wP D AvP  ˇ vjwj
P n  ˛ jvj
P jwjn1 w; (6.22)

where the parameters n, A, ˇ, and ˛ are shape parameters of the hysteresis curves.
It can be seen from (6.22) that the input is vP rather that v although the hysteresis is
observed between v and w. The Bouc–Wen model has an interesting simplicity and
is able to represent a large class of hysteresis [60]. Because of the elastic structure
and material of piezoelectric ceramics, it is generally admitted that n D 1 to describe
the hysteresis of PANSs.
From the above discussions, the differential-equation-based models can be
essentially expressed as a first-order differential equation that relates the input to
the output in a hysteretic way. In such a way, it can be conveniently augmented by
introducing a state variable into the dynamic equations of the piezo-actuated system
for controller development [27, 51, 54, 68, 91]. However, it is difficult to obtain the
general solutions of differential-equation-based models. On the other hand, until
now, there is no method to construct the analytical inverses of the differential-
equation-based models for hysteresis compensation.

6.4.3.2 Operator-Based Hysteresis Models

Operator-based hysteresis models are kinds of models with an integral of weighed

elementary hysteresis operators. Based on the differences of elementary hysteresis
operators, operator-based hysteresis models for PANSs include the Preisach model,
Krasnosel’skii–Pokrovkii model, and Prandtl–Ishlinskii model as shown in Fig. 6.5.
(1) Preisach model
The elementary hysteresis operator in the Preisach model is called as Relay.
A geometric illustration of the Relay operator O˛ˇ Œv.t/ is shown in Fig. 6.6. With
the Relay operator, the Preisach model is expressed as [42]
Z Z
w.t/ D PŒv.t/ D .˛; ˇ/ O˛;ˇ Œv.t/d˛dˇ (6.23)
˛ˇ
6 Modeling of Piezoelectric-Actuated Nanopositioning Stages Involving: : : 195

Fig. 6.6 A geometric

illustration of the Relay
operator

Fig. 6.7 Block diagram of the Preisach model

with
8
ˆ
ˆ 0; v.t/  ˇ
ˆ
ˆ
ˆ
ˆ 1; v.t/  ˛
ˆ
ˆ
ˆ
< O˛ˇ Œv0 ; 8 2 Œt0 ; t; v./ 2 .ˇ; ˛/
O˛ˇ Œv.t/ D 0; if v. / 2 .ˇ; ˛/; 9t1 2 Œt0 ; t/; s:t: ; (6.24)
ˆ
ˆ
ˆ
ˆ v.t1 / D ˛ and 8 2 .t1 ; t v./ 2 .ˇ; ˛/
ˆ
ˆ
ˆ
ˆ 1; if v. / 2 .ˇ; ˛/; 9t1 2 Œt0 ; t/; s:t:
:̂
v.t1 / D ˇ and 8 2 .t1 ; t; v./ 2 .ˇ; ˛/

where .˛; ˇ/ is called as the density function, ˛ and ˇ are values of the switching
thresholds satisfying ˛  ˇ. The simple block diagram of the Preisach model (6.23)
can be approximately considered as parallel connections of a number of Relay
operators as shown in Fig. 6.7.
The Preisach model was first suggested almost in the 1930s [42, 55]. A thorough
discussion of the Preisach model may refer to the Mayergoyz’s monograph [6].
The Preisach model can be recognized as a fundamental toolkit of hysteresis
modeling for a wide range of hysteresis phenomena because of its general struc-
ture. However, due to the existence of the double integrals, identification and
196 G.-Y. Gu et al.

implementation of the Preisach model are usually very complicated [26]. Addition-
ally, the Preisach model is not analytically invertible. Thus, numerical methods are
generally adopted to obtain approximate inversions of the model in applications
[34, 35, 73, 74, 80, 85, 86].
(2) Krasnosel’skii–Pokrovkii model
The elementary hysteresis operator of the Krasnosel’skii–Pokrovkii (K–P) model
is called as a KP operator [18], which is a kind of generalized plays [84]. Like
the Preisach model, the K–P model is expressed using the double integrals as
follows [18]
Z Z
w.t/ D ƒŒv.t/ D . 2 ; 1 /kp Œv; . /.t/d 2 d 1 ; (6.25)
2 1

where . 2 ; 1 / denotes the density function, 1 and 2 are values of the switching
thresholds satisfying 2  1 . kp Œv; . /.t/ represents the KP operator with the
following form
8
< max 2P f0 . /; r.v; 2 /g; vP > 0
kp Œv; . /.t/ D min 2P f0 . /; r.v; 1 /g; vP < 0 (6.26)
:
kp ; vP D 0

with
8
< 1; v  x
r.v; x/ D 1 C ˛2 .v  x/; x < v < x C ˛ ; (6.27)
:
C1; v  x C ˛

where D . 1 ; 2 /, ˛ > 0, kp means the value of the KP operator does not

change from its previous value, 0 . / is the initial value at location satisfying
1  0 . /  1. Figure 6.8 shows a geometric illustration of the KP operator
kp Œv; . /.t/.

Fig. 6.8 Geometric

illustrations of the KP
operator
6 Modeling of Piezoelectric-Actuated Nanopositioning Stages Involving: : : 197

a b

Fig. 6.9 Geometric illustrations of (a) Play and (b) Stop operators

It is pointed out that the K–P model is mainly used for hysteresis modeling
and compensation in applications with shape memory alloy actuators [2, 3, 87].
Galinaities [18] investigated the properties of the K–P model and developed an
approximate inverse for hysteresis compensation of a PCA. Like the Preisach model,
the K–P model is also difficult to get an analytical inverse [18, 87].
(3) Prandtl–Ishlinskii model
The Prandtl–Ishlinskii (P–I) model was originally proposed as a model for
plasticity–elasticity [55]. Like the Preisach model, the P–I model is defined by the
elementary hysteresis operator called Play or Stop. Different from the discontinuous
Relay operator with two thresholds, the Play or Stop is a continuous operator with a
single threshold. Figure 6.9 shows the geometric illustrations of the Play (Fr Œv.t/)
and Stop (Er Œv.t/) operators. Considering the fact that Fr Œv.t/ C Er Œv.t/ D v.t/,
the P–I model defined by the play operator is detailed in this chapter as an
illustration.
With a threshold r for any piecewise monotone input function v.t/ 2 Cm Œ0; tE ,
the Play-based P–I model is expressed as [7, 43]
Z R
w.t/ D …Œv.t/ D p0 v.t/ C p.r/Fr Œv.t/dr (6.28)
0

with

w.0/ D Fr Œv.0/ D fr .v.0/; 0/

(6.29)
w.t/ D Fr Œv.t/ D fr .v.t/; w.ti //

for ti < t  tiC1 ; 0  i  N  1 with

fr .v; w/ D max.v  r; min.v C r; w//; (6.30)

198 G.-Y. Gu et al.

where p.r/ is a density function and p0 is a positive constant. The density function
p.r/ generally vanishes for large values of r, while the choice of R D 1 as the
upper limit of integration is widely used in the literature for the sake of convenience
[30, 44].
The P–I model (6.28) is closely linked with the Preisach model (6.23) even
if they seem very different at first glance. In fact, the P–I model falls into the
set of Preisach model and can be considered as a subclass of the Preisach model
[7, 85]. The main advantage of the P–I model over the Preisach model is that the
P–I model has the analytical inverse [38, 44], thus making it more efficient for
real-time applications. However, the P–I model essentially describes the symmetric
hysteresis effect. It fails to characterize the hysteresis effect with the asymmetric
behavior observed in the input-output relations of piezo-actuated stages. In an
attempt to overcome this limitation, several variations of P–I models for describing
the asymmetric hysteresis are reported. Kuhnen [45] proposed dead-zone operators
cascaded with the play operator. Alternatively, Gu et al. [30] proposed a modified
P–I model with a generalized input function without modifying the play operator.
Instead of using the play operator, the asymmetric backlash operator [14] was
adopted as the elementary operator to represent the asymmetric hysteresis behavior.
As another choice, two asymmetric operators [39] were utilized for the ascending
and descending branches of hysteresis loops, respectively. Janaideh et al. [37]
developed two nonlinear envelope function-based operators to form a generalized
P–I model for asymmetric hysteresis description. In current stage, interesting in
studying P–I model for hysteresis description and compensation in piezo-actuated
systems is increasing [15, 26, 28–30, 66, 70, 76].
(4) Rate-dependent model
It is pointed out that the operator-based hysteresis models are by nature devel-
oped to describe the hysteresis with the rate-independent memory effect. The
rate-independent memory effect of the hysteresis model means that the output of
the hysteresis model depends not on the variation rate of the input signals. In some
times, the rate-independent effect is also called as the quasi-static behavior, which
means the hysteresis output is not affected by the dynamics. In mathematics, the
definition of the rate independence can be described by Visone [85], Krejci [43],
and Brokate and Sprekels [7]: The operator @ is defined as rate-independent if and
only if the following condition holds

@Œx ı ' D @Œx ı ' (6.31)

with

' W t 2 Œ0; t ! '.t/ 2 Œ0; t; (6.32)

where the symbol ı is the composition operator, '.0/ D 0 and '.t/ D t. From
the definition of the Relay O˛ˇ (6.24) and Play Fr Œv.t/ (6.29), Preisach and P–I
hysteresis model are intrinsically rate-independent.
6 Modeling of Piezoelectric-Actuated Nanopositioning Stages Involving: : : 199

To overcome such limitation, Mayergoyz [56] developed a rate-dependent

Preisach model by introducing the speed of output signals in the density function.
Since then, a number of rate-dependent hysteresis models have employed this
dynamic density function for characterizing rate dependency of the input-output
relationships of piezo-actuated stages. Due to the noise problem for deriving the
output signals, Yu et al. [92] proposed an alternative method by introducing the
variation rate of the input signals to the density function of the Preisach model.
Mrad and Hu [61] introduced a time-dependent average rate of change of the input
voltage into the Preisach model to account for the rate-dependent hysteresis. With
the same idea in [61], Tan et al. [82] developed a rate-dependent P–I hysteresis
model. Furthermore, Janaideh et al. [36] developed a generalized rate-dependent P–I
model by modifying not only the density function but also the play operator with the
rate change of the input voltage. As reported, these models [36, 56, 61, 82, 92] can
well predict the rate-dependent behaviors of the hysteresis in piezo-actuated stages.
However, owing to the introductions of the derivatives of the input or output signals,
the density function and the elementary operators become more complicated, which
introduces difficulties in the model identification and real-time controller design for
hysteresis compensation.

6.4.3.3 Other Hysteresis Models

Besides the differential-based and operator-based models, there are some other
modeling approaches to describe the hysteresis in PANSs.
Intelligent modeling approaches such as neural networks [50, 79, 93], fuzzy
systems [49], and support vector machines [89] have been applied to predict
the hysteresis. Additionally, Cruz et al. [12] proposed a concept of the apparent
phase shift between input and output to describe the hysteresis. The advantage
of the concept is that a phaser can be conveniently used to develop hysteresis
compensator. Owing to the simplicity of the mathematical description, polynomial
function-based models [4, 10, 72] were developed to predict the rate-independent
hysteresis. Based on the experimental phenomena of the actuator’s hysteresis, Gu
and Zhu [23, 24] proposed an ellipse-based hysteresis model to characterize the
rate-dependent hysteresis of a PANS.
It should be noted that these hysteresis models can precisely and effectively
describe the hysteresis of PANSs in particular applications, although they are not
popular as the differential-based and operator-based models. On the other hand, it
demonstrates that, until now, there is no a general model to completely represent the
hysteresis behaviors of PANSs.
200 G.-Y. Gu et al.

6.5 Validation of the Proposed Model

To verify the proposed general dynamic model, a PANS (PZT mode P-753.31C
produced by Physik Instrumente GmbH & Co.) is used as a test bed. PZT mode P-
753.31C is driven by applied control input voltage in the range of 2 to 10 V and has
nominal 38 m expansion. The PDA is a voltage power amplifier with a fixed gain
of 10 to amplify the control input for driving the PCA. The unloaded resonance
frequency of the mechanical response is 2.9 kHz. The capacitive gap sensor is
integrated in the mechanical structure to measure the displacement of the stage.
The position servo-control module (PSCM) is used as a sensor module to transfer
the displacement to analogue voltage in the range of 0–10 V. The specifications of
the piezoelectric positioning stage P-753.31C are listed in Table 6.1.

6.5.1 Gain of the PSCM

Before model validation, the gain KPSCM of the PSCM used to transfer the
displacement to analogue voltage in the range of 0–10 V needs to be fixed, which is
calculated as

KPSCM D 10=.x10  x0 / .V=m/; (6.33)

where x10 is the real stage position when the sensor monitor output is 10 V, and x0 is
the real stage position when the sensor monitor output reaches to 0 V. For this stage,
x10 D 38 m and x0 D 0 m, thus one can get KPSCM D 10=38 V/m.

6.5.2 Identification of the Linear Dynamic Plant

It should be mentioned that, until now, there is no general approach to identify

the parameters of the analytical model (6.12) and (6.13) due to existence of the

Table 6.1 Properties of the PZT mode P-753.31C

Items Piezoelectric positioning stage Units
Driven voltage 2 to 10 V
Expansion displacement 0 to 38 m
Electrical capacitance 4.6 F
Stiffness in motion direction 16 N/m
Unloaded resonant frequency 2.9 kHz
Stage mass 0.25 kg
PSCM output 0–10 V
6 Modeling of Piezoelectric-Actuated Nanopositioning Stages Involving: : : 201

hysteresis nonlinearity. As a comprise, the commonly adopted approach in the

literature is to express the analytical model into a linear system preceded by a
hysteresis term. Then, the linear parameters and the hysteresis model will be
identified separately [21, 27, 53]. In the following development, we will follow this
line.
Since the PDA operates in the voltage control mode to supply charge for the
PCA to generate the piezoelectric expansion, the induced charge q.t/ in the circuit
should be represented as a function of control input vin .t/, that is, q.t/ D f .vin .t//.
.vin /
Without loss of generality, the term vin .t/  H.fkamp is re-defined as a new hysteresis
nonlinearity P.vin .t//

w.t/ D P.vin .t//; (6.34)

.vin /
where P.vin .t// D vin .t/ H.fkamp . In this case, the signal from P.vin .t// to the output
of the displacement x in (6.12) and (6.13) can be described by a third-order linear
system,

X.s/ K
G.s/ D D 3 2
(6.35)
W.s/ s C a0 s C a1 s C a2

with
8
ˆ
ˆ a0 D R01CA C bms
ˆ
< a1 D mRb0sCA C kms
; (6.36)
ˆ
ˆ a2 D mR0 CA
ks
:̂ kamp Tem
K D mR 0 CA

where K, ai ; i D f0; 1; 2g, are the parameters related to those parameters in (6.12)
and (6.13). Thus, the general model of the PANS can be represented by a linear
third-order plant (6.35) preceded by a hysteresis nonlinearity P.vin .t//. Figure 6.10
shows the block diagram of the general model.
Remark 3. We should mention that there are two approaches to handle the hystere-
sis nonlinearity H.q/ in literature. One is by assuming R0 D 0, the complete model
is described by

Fig. 6.10 Block diagram of the general model

202 G.-Y. Gu et al.

mRx.t/ C bs xP .t/ C ks x.t/ D Tem Œkamp vin .t/  H.q/: (6.37)

Then, the Bouc–Wen model is utilized to describe the hysteresis term H.q/, as in
.vin /
[51, 53]. The second approach is to treat the term vin .t/  H.fkamp as a new hysteresis
nonlinearity [5, 9, 94]. For different approaches, the treatment for the controller
designs is quite different because the hysteresis term in the first can be regarded as
a sum component while the second is as a preceding component.
From (6.35) and (6.36), the parameters for the linear dynamic part cannot be
obtained by simply using the specifications of the PANS, because the exact infor-
mation on the parameters bs , Tem , and R0 is generally not available. They are usually
identified based on the applied input voltage and corresponding displacement. In
this work, the axiomatic-design-theory-based approach (ADTBA) [27] is utilized
to identify parameters (K and ai ; i D 0; 1; 2) of the linear plant (6.35) using
experimental data of both frequency response and transient response. For model
identification of the linear dynamic part, the third-order linear dynamic model is
generally expressed as kp !n 2 =..s C 1/.s2 C 2& !n s C !n 2 //, where four parameters,
i.e., kp , !, & , and  , should be determined. According to the ADTBA, the frequency
response data is used to determine the parameters kp and !n , and the transient
response data is used to determine the parameters & and . It is worthy of mentioning
that small-amplitude input excitation signals should be applied to avoid distortion
from the preceded hysteresis nonlinearity as much as possible and the higher
sampling frequency (i.e., fs D 20 kHz) is set to capture the fast dynamic response
of the stage when the creep behavior can be conservatively neglected. The detailed
steps are described below:
(1) A small-amplitude band-limited white noise signal is used to excite the PCA.
The system identification toolbox of MATLAB is adopted to obtain the
frequency response of the PANS as shown in Fig. 6.11 indicated by the solid
blue line. The parameters kp and !n can be directly obtained through the
frequency response data.
(2) A small-amplitude step signal is used to excite the PCA. Figure 6.12 (indicated
by the solid blue line) shows the transient response of the PANS which is
captured during the first 6 ms. With the identified resonant frequency !n and
the gain kp in the first step, the transient response data is used to adjust time
constant  and damping coefficient & .
Based on the above two-step method, the identified parameters are compiled
in Table 6.2. Therefore, the linear dynamic plant G.s/ of the proposed model is
expressed as

4:767  106
G.s/ D : (6.38)
s3 C 7582s2 C 4:1  108 s C 1:677  1012
6 Modeling of Piezoelectric-Actuated Nanopositioning Stages Involving: : : 203

20

10
Magnitude (dB)

−10
Experimental data
−20 Model prediction

−30

−40
0

−720
Phase (deg)

−1440

−2160

−2880
0 1 2 3 4
10 10 10 10 10
Frequency (Hz)

Fig. 6.11 Comparison of frequency response of the PANS (blue solid-experimental data; red
dash-model simulation data)

6.5.3 Identification of the Hysteresis Nonlinearity

In order to identify the hysteresis nonlinearity in (6.34), the first step is to find
a hysteresis model, for instance, the differential-equation-based or operator-based
model (as detailed in Sect. 6.4), to describe the behaviors of the hysteresis. In this
chapter, the MPI model [30] is selected as an illustration for hysteresis description.
Certainly, other hysteresis models can also be selected. Without loss of generality,
in what follows we use the MPI hysteresis model to validate the proposed model.
Before introducing the MPI model, the one-side play operator For Œvin .t/ with a
threshold r  0 for any piecewise monotone input function vin .t/ 2 Cm Œ0; tE  is
defined as

u.0/ D For Œvin .0/ D for .vin .0/; 0/

(6.39)
u.t/ D For Œvin .t/ D for .vin .t/; u.ti //

for ti < t  tiC1 ; 0  i  N  1 with

for .vin ; u/ D max.vin  r; min.vin ; u// (6.40)

for 0 D t0 < t1 < : : : < tN D tE .

204 G.-Y. Gu et al.

2.5
Displacement (μm)

2
Experimental data
Model prediction
1.5

0.5

0
0 1 2 3 4 5 6
Time (ms)

Fig. 6.12 Comparison of 1-V step response of the PANS in the first 6 ms (blue solid-experimental
output response; red dash-model output response)

Table 6.2 Identified Parameters Value Units

parameters of the general
model for the PANS m 0.0515 kg
bs 114 Ns/m
ks 16.08 N/m
R0 40.5 
CA 4.6 F
Tem 4.575 N/V
kamp 10 –
a0 7582 –
a1 4:1  108 –
a2 1:677  1012 –
K 4:767  106 –

Then, with a finite number of the one-side play operators For Œvin .t/, the MPI
model can be expressed in the form of Gu et al. [30]

X
n
w.t/ D g.vin /.t/ C bi Fori Œvin .t/; (6.41)
iD1

3
where g.vin /.t/ D p1 vin .t/Cp2 vin .t/ is a function of the input vin .t/ with the constant
parameters p1 and p2 , n is the number of the adopted play operators for modeling,
and bi D p.ri /.ri  ri1 / is the weighted coefficient for the threshold ri , satisfying
6 Modeling of Piezoelectric-Actuated Nanopositioning Stages Involving: : : 205

i1
ri D jjvin .t/jj1 ; i D 1; 2; : : : ; n: (6.42)
n
For detailed descriptions of the MPI model, the reader may refer to [30].
With the MPI model for hysteresis description, it is ready to identify the
parameters in (6.41) on the basis of the previous identified parameters of the linear
dynamic plant (6.38). In fact, due to the highly nonlinear, high dimensional, and
multiple constraint characteristics, identification of the hysteresis models is also
a challenging task and many identification algorithms, such as the least square
method, the genetic algorithms, and particle swarm optimization (PSO), have been
reported to solve this problem [8, 46, 90]. PSO [90] has been shown to be superior
to its competitors for identification of the P–I model. In this work, the effective
informed adaptive PSO [47, 90] is borrowed to identify the MPI model.
From the MPI model described in (6.41), when ten one-side play operators are
utilized, i.e., n D 10, it can be seen that there are 12 parameters to be identified,
that is:

X D Œp1 ; p2 ; b1 ; b2 ; : : : ; b10 : (6.43)

As the parameters are difficult to find by trials, a minimize optimization is used to

solve the problem. The key point of the optimization is the selection of the fitness
function (or objective function), which has an important influence on the identified
results. In this work, the fitness function is chosen as:

1X a
N
F.X/ D .x  xi /2 (6.44)
N iD1 i

satisfying

«
x.t/ C a0 xR .t/ C a1 xP .t/ C a2 x.t/ D Kw.t/
3
P (6.45)
w.t/ D p1 vin .t/ C p2 vin .t/ C niD1 bi Fori Œvin .t/;

where N denotes the total number of the experimental data, xi and xia are the
experimental data and simulation data at the ith sampling time, respectively. When
the simulation data xia is very close to the experimental data xi , the identified result
X is thought to be a good result. To sum up, the flowchart of the PSO algorithm is
given in Fig. 6.13. Table 6.3 lists the identified parameters of the MPI model (6.41).
The detailed discussions on the PSO algorithms may refer to [47, 90].
Remark 4. It should be noted that the larger n is selected, it is more precision to
describe the hysteresis loops in theory. In real applications, it is verified that further
increase of the numbers of the play operators leads to no significant improvement
in the modeling accuracy. As in [30], there is a comparison of identification errors
with different values of n, where it can be seen that the identification errors are at the
same level with the selected values of n. The reader may refer to [30] for a detailed
discussion. Therefore, in this development, we select n D 10 for a case study.
206 G.-Y. Gu et al.

Fig. 6.13 The flowchart of the PSO identification algorithm

With the identified parameters of the linear dynamic plant and the nonlinear
MPI hysteresis model, we are now ready to validate the proposed general model.
Figure 6.14 shows the comparison of experimental data of the PANS and associated
prediction results of the identified general model. In Fig. 6.14, the waveforms of the
input voltage, displacement, prediction error, and hysteresis loops are shown.
From Fig. 6.14b, it can be obviously observed that the prediction results of
the developed general model agree well with the experimental measurements.
Figure 6.14c shows that the maximal prediction error is less than 1 % as a percentage
of the full displacement range. In addition, the comparison of the hysteresis loops is
given in Fig. 6.14d, where the asymmetric hysteresis nonlinearity of the PANS with
the major and minor loops are well described by the developed model.
6 Modeling of Piezoelectric-Actuated Nanopositioning Stages Involving: : : 207

Table 6.3 Identified Number ri bi pi

parameters of the MPI model
1 0 0:0136 0:1189
2 0:1 0:0164 0:0422
3 0:2 0:0362
4 0:3 0:0025
5 0:4 0:0178
6 0:5 0:0635
7 0:6 0:0514
8 0:7 0:0082
9 0:8 0:1189
10 0:9 0:0593

a 9 b 40
Experimental data
Model simulation data
8
35
Displacement (μm)

7
Voltage (V)

30
6

5
25

4
20
3

2 15
0 1 2 3 4 5 0 1 2 3 4 5
Time (s) Time (s)

c 0.4 d 40
Experimental data
Model simulation data
0.3
35
Displacement (μm)

0.2
Error (μm)

0.1 30

0
25

−0.1
20
−0.2

−0.3 15
0 1 2 3 4 5 2 3 4 5 6 7 8 9

Time (s) Voltage (V)

Fig. 6.14 Experimental validation of the proposed general model. (a) Input voltage. (b) Output
displacement. (c) Predicted error. (d) Hysteresis loops

Therefore, from Figs. 6.11, 6.12 and 6.14, the proposed model with the identified
parameters well characterize both the dynamics and hysteresis behaviors of the
PANS, which clearly validates our developed general model.
208 G.-Y. Gu et al.

6.6 Conclusion

PANSs are becoming promising in many precision manufacturing equipments.

However, they suffer from the inherent hysteresis effect of the PCAs. The hysteresis
effect exhibits complex nonlinear characteristics, which usually introduces unde-
sirable inaccuracies or oscillations and even instability. Interest in studying PANSs
with actuator hysteresis is also motivated by the fact that the hysteresis nonlinearity
is coupled with the system dynamics for which traditional control methods are
insufficient and thus require the comprehensive dynamic model of the PANS.
In this chapter, we first propose a general model to completely characterize
nonlinear electric behavior, voltage-charge hysteresis, piezoelectric effect, and
frequency response of the PANS in both electrical and mechanical aspects. After
a brief survey of the hysteresis models, a MPI hysteresis model is utilized as an
illustration to describe the hysteresis nonlinearity. To identify the parameters of the
general model, the adopted approach is to express the general model into a third-
order linear plant preceded by a MPI model. Then, the parameters of the linear plant
and the hysteresis term are identified separately, which is confirmed by experimental
results as well.

Acknowledgements This work was supported by the National Natural Science Foundation of
China under Grant No. 51405293.

References

1. H. Adriaens, W.L. de Koning, R. Banning, Modeling piezoelectric actuators. IEEE/ASME

Trans. Mechatron. 5(4), 331–341 (2000)
2. H.T. Banks, A.J. Kurdila, G. Webb, Identification of hysteretic control influence operators
representing smart actuators Part I: formulation. Math. Probl. Eng. 3(4), 287–328 (1997a)
3. H.T. Banks, A.J. Kurdila, G. Webb, Identification of hysteretic control influence operators
representing smart actuators Part II: convergent approximations. Math. Probl. Eng. 8(6),
536–550 (1997b)
4. S. Bashash, N. Jalili, A polynomial-based linear mapping strategy for feedforward compensa-
tion of hysteresis in piezoelectric actuators. J. Dyn. Syst. Meas. Control 130(3), 1–10 (2008)
5. S. Bashash, N. Jalili, Robust adaptive control of coupled parallel piezo-flexural nanoposition-
ing stages. IEEE/ASME Trans. Mechatron. 14(1), 11–20 (2009)
6. G. Bertotti, I.D. Mayergoyz, The Science of Hysteresis: Mathematical Modeling and Applica-
tions. (Academic, Oxford, 2006)
7. M. Brokate, J. Sprekels, Hysteresis and Phase Transitions (Springer, New York, 1996)
8. C. Chan, G. Liu, Hysteresis identification and compensation using a genetic algorithm with
adaptive search space. Mechatronics 17(7), 391–402 (2007)
9. X. Chen, T. Hisayam, Adaptive sliding-mode position control for piezo-actuated stage. IEEE
Trans. Ind. Electron. 55(11), 3927–3934 (2008)
10. S. Chonan, Z. Jiang, T. Yamamoto, Nonlinear hysteresis compensation of piezoelectric ceramic
actuators. J. Intell. Mater. Syst. Struct. 7(2),150–156 (1996)
11. B.D. Coleman, M.L. Hodgdon, On a class of constitutive relations for ferromagnetic hysteresis.
Arch. Ration. Mech. Anal. 99(4), 375–396 (1987)
6 Modeling of Piezoelectric-Actuated Nanopositioning Stages Involving: : : 209

12. J.M. Cruz-Hernandez, V. Hayward, Phase control approach to hysteresis reduction. IEEE
Trans. Control Syst. Technol. 9(1), 17–26 (2001)
13. S. Devasia, E. Eleftheriou, S.O.R. Moheimani, A survey of control issues in nanopositioning.
IEEE Trans. Control Syst. Technol. 15(5), 802–823 (2007)
14. R. Dong, Y. Tan, A modified Prandtl-Ishlinskii modeling method for hysteresis. Physica B
404(8–11), 1336–1342 (2009)
15. A. Esbrook, X. Tan, H.K. Khalil, Control of systems with hysteresis via servocompensation
and its application to nanopositioning. IEEE Trans. Control Syst. Technol. 21(3), 725–738
(2013)
16. A.J. Fleming, A review of nanometer resolution position sensors: operation and performance.
Sens. Actuators A Phys. 199, 106–126 (2013)
17. A.J. Fleming, K.K. Leang, Charge drives for scanning probe microscope positioning stages.
Ultramicroscopy 108(12), 1551–1557 (2008)
18. W.S. Galinaitis, Two methods for modeling scalar hysteresis and their use in controlling
actuators with hysteresis. Ph.D. thesis, Virginia Polytechnic Institute and State University
(1999)
19. Y.S. Gao, D.W. Zhang, C.W. Yu, Dynamic modeling of a novel workpiece table for active
surface grinding control. Int. J. Mach. Tools Manuf. 41(4), 609–624 (2001)
20. H. Georgiou, R.B. Mrad, Electromechanical modeling of piezoceramic actuators for dynamic
loading applications. J. Dyn. Syst. Meas. Contr. 128(3), 558–567 (2006)
21. F. Giri, Y. Rochdi, F.Z. Chaoui, A. Brouri, Identification of Hammerstein systems in presence
of hysteresis-backlash and hysteresis-relay nonlinearities. Automatica 44(3), 767–775 (2008)
22. M. Goldfarb, N. Celanovic, Modeling piezoelectric stack actuators for control of micromanip-
ulation. IEEE Control. Syst. Mag. 17(3), 69–79 (1997)
23. G.Y. Gu, L.M. Zhu, High-speed tracking control of piezoelectric actuators using an ellipse-
based hysteresis model. Rev. Sci. Instrum. 81(8), 085104 (2010)
24. G.Y. Gu, L.M. Zhu, Modeling of rate-dependent hysteresis in piezoelectric actuators using a
family of ellipses. Sens. Actuators A Phys. 165(2), 202–209 (2011)
25. G.Y. Gu, L.M. Zhu, Motion control of piezoceramic actuators with creep, hysteresis and
vibration compensation. Sens. Actuators A Phys. 197, 76–87 (2013)
26. G.Y. Gu, M.J. Yang, L.M. Zhu, Real-time inverse hysteresis compensation of piezoelectric
actuators with a modified Prandtl-Ishlinskii model. Rev. Sci. Instrum. 83(6), 065106 (2012)
27. G.Y. Gu, L.M. Zhu, C.Y. Su, H. Ding, Motion control of piezoelectric positioning stages:
modeling, controller design and experimental evaluation. IEEE/ASME Trans. Mechatron.
18(5), 1459–1471 (2013)
28. G.Y. Gu, L.M. Zhu, C.Y. Su, High-precision control of piezoelectric nanopositioning stages
using hysteresis compensator and disturbance observer. Smart Mater. Struct. 23(10), 105007
(2014a)
29. G.Y. Gu, L.M. Zhu, C.Y. Su, Integral resonant damping for high-bandwidth control of
piezoceramic stack actuators with asymmetric hysteresis nonlinearity. Mechatronics 24(4),
367–375 (2014b)
30. G.Y. Gu, L.M. Zhu, C.Y. Su, Modeling and compensation of asymmetric hysteresis
nonlinearity for piezoceramic actuators with a modified Prandtl-Ishlinskii model. IEEE Trans.
Ind. Electron. 61(3), 1583–1595 (2014c)
31. G.Y. Gu, L.M. Zhu, C.Y. Su, H. Ding, S. Fatikow, Modeling and control of piezo-
actuated nanopositioning stages: a survey. IEEE Trans. Autom. Sci. Eng. (2014d).
doi:10.1109/TASE.2014.2352364
32. G.Y. Gu, L.M. Zhu, C.Y. Su, H. Ding, S. Fatikow, Proxy-based sliding mode tracking control
of piezoelectric-actuated nanopositioning stages. IEEE/ASME Trans. Mechatron. 20(4), 1956–
1965 (2015)
33. H. Habibullah, H.R. Pota, I.R. Petersen, M.S. Rana, Tracking of triangular reference signals
using LQG controllers for lateral positioning of an AFM scanner stage. IEEE/ASME Trans.
Mechatron. 19(4), 1105–1114 (2014)
210 G.-Y. Gu et al.

34. H. Hu, H. Georgiou, R.B. Mrad, Enhancement of tracking ability in piezoceramic actuators
subject to dynamic excitation conditions. IEEE/ASME Trans. Mechatron. 10(2), 230–239
(2005)
35. R.V. Iyer, X. Tan, Control of hysteretic systems through inverse compensation. IEEE Control.
Syst. Mag. 29(1), 83–99 (2009)
36. M.A. Janaideh, S. Rakheja, C.Y. Su, Experimental characterization and modeling of rate-
dependent hysteresis of a piezoceramic actuator. Mechatronics 19(5), 656–670 (2009)
37. M.A. Janaideh, S. Rakheja, C.Y. Su, An analytical generalized Prandtl-Ishlinskii model
inversion for hysteresis compensation in micropositioning control. IEEE/ASME Trans.
Mechatron. 16(4), 734–744 (2011)
38. H. Janocha, K. Kuhnen, Real-time compensation of hysteresis and creep in piezoelectric
actuators. Sens. Actuators A Phys. 79(2), 83–89 (2000)
39. H. Jiang, H. Ji, J. Qiu, Y. Chen, A modified Prandtl-Ishlinskii model for modeling asymmetric
hysteresis of piezoelectric actuators. IEEE Trans. Ultrason. Ferroelectr. Freq. Control 57(5),
1200–1210 (2010)
40. D.C. Jiles, D.L. Atherton, Theory of ferromagnetic hysteresis. J. Magn. Magn. Mater. 61(1–2),
48–60 (1986)
41. T.L. Jordan, Z. Ounaies, Piezoelectric ceramics characterization. Technical report, ICASE,
NASA Langley Research Center, Hampton (2001)
42. M. Krasnosel’skii, P. Pokrovskii, Systems with Hysteresis (Springer, Berlin, 1989)
43. P. Krejci, Hysteresis Convexity and Dissipation in Hyperbolic Equations (Gakkotosho, Tokyo,
1996)
44. P. Krejci, K. Kuhnen, Inverse control of systems with hysteresis and creep. IEE Proc. Control
Theory Appl. 148(3), 185–192 (2001)
45. K. Kuhnen, Modeling, identification and compensation of complex hysteretic nonlinearities: a
modified Prandtl-Ishlinskii approach. Eur. J. Control 9(4), 407–418 (2003)
46. K. Kuhnen, P. Krejci, Compensation of complex hysteresis and creep effects in piezoelectri-
cally actuated systems-a new Preisach modeling approach. IEEE Trans. Autom. Control 54(3),
537–550 (2009)
47. Q. Li, W. Chen, Y. Wang, S. Liu, J. Jia, Parameter identification for PEM fuel-cell mechanism
model based on effective informed adaptive particle swarm optimization. IEEE Trans. Ind.
Electron. 58(6), 2410–2419 (2011)
48. C.X. Li, G.Y. Gu, M.J. Yang, L.M. Zhu, Design, analysis and testing of a parallel-kinematic
high-bandwidth XY nanopositioning stage. Rev. Sci. Instrum. 84(12), 125111 (2013a)
49. P. Li, F. Yan, C. Ge, X. Wang, L. Xu, J. Guo, P. Li, A simple fuzzy system for modelling
of both rate-independent and rate-dependent hysteresis in piezoelectric actuators. Mech. Syst.
Signal Process. 36(1), 182–192 (2013b)
50. W. Li, X. Chen, Compensation of hysteresis in piezoelectric actuators without dynamics
modeling. Sens. Actuators A Phys. 199, 89–97 (2013)
51. Y. Li, Q. Xu, Adaptive sliding mode control with perturbation estimation and PID sliding
surface for motion tracking of a piezo-driven micromanipulator. IEEE Trans. Control Syst.
Technol. 18(4), 798–810 (2010)
52. H.C. Liaw, B. Shirinzadeh, J. Smith, Enhanced sliding mode motion tracking control of
piezoelectric actuators. Sens. Actuators A Phys. 138(1), 194–202 (2007)
53. C.J. Lin, S.Y. Chen, Evolutionary algorithm based feedforward control for contouring of a
biaxial piezo-actuated stage. Mechatronics 19(6), 829–839 (2009)
54. C.J. Lin, P.T. Lin, Tracking control of a biaxial piezo-actuated positioning stage using
generalized Duhem model. Comput. Math. Appl. 64(5), 766–787 (2012)
55. J.W. Machi, P. Nistri, P. Zecca, Mathematical models for hysteresis. SIAM Rev. 35(1), 94–123
(1993)
56. I. Mayergoyz, Preisach models of hysteresis. IEEE Trans. Magn. 24(6), 2925–2927 (1988)
57. I. Mayergoyz, Mathematical Models of Hysteresis and Their Applications (Elsevier Academic
Press, Amsterdam, 2003)
6 Modeling of Piezoelectric-Actuated Nanopositioning Stages Involving: : : 211

58. A. Meitzler, H. Tiersten, A. Warner, D. Berlincourt, G. Couqin, F. Welsh III, IEEE standard on
piezoelectricity (1988)
59. J. Minase, T.F. Lu, B. Cazzolato, S. Grainger, A review, supported by experimental results, of
voltage, charge and capacitor insertion method for driving piezoelectric actuators. Precis. Eng.
34(4), 692–700 (2010)
60. M. Mohammed Ismail, F. Ikhouane, J. Rodellar, The hysteresis Bouc-Wen model, a survey.
Arch. Comput. Methods Eng. 16(2), 161–188 (2009)
61. R.B. Mrad, H. Hu, A model for voltage-to-displacement dynamics in piezoceramic actuators
subject to dynamic-voltage excitations. IEEE/ASME Trans. Mechatron. 7(4), 479–489 (2002)
62. C. Niezrecki, D. Brei, S. Balakrishnan, A. Moskalik, Piezoelectric actuation: state of the art.
Shock Vib. Dig. 33(4), 269–280 (2001)
63. J. Oh, D.S. Bernstein, Semilinear Duhem model for rate-independent and rate-dependent
hysteresis. IEEE Trans. Autom. Control 50(5), 631–645 (2005)
64. A. Padthe, B. Drincic, J. Oh, D. Rizos, S. Fassois, D. Bernstein, Duhem modeling of friction-
induced hysteresis. IEEE Control Syst. 28(5), 90–107 (2008)
65. G. Park, M.T. Bement, D.A. Hartman, R.E. Smith, C.R. Farrar, The use of active materials for
machining processes: a review. Int. J. Mach. Tools Manuf. 47(15), 2189–2206 (2007)
66. Y. Qin, B. Shirinzadeh, Y. Tian, D. Zhang, Design issues in a decoupled XY stage: static and
dynamics modeling, hysteresis compensation, and tracking control. Sens. Actuators A Phys.
194, 95–105 (2013)
67. M. Quant, H. Elizalde, A. Flores, R. Ramirez, P. Orta, G. Song, A comprehensive model for
piezoceramic actuators: modelling, validation and application. Smart Mater. Struct. 18(12),
125011 (2009)
68. M. Rakotondrabe, Bouc-Wen modeling and inverse multiplicative structure to compensate
hysteresis nonlinearity in piezoelectric actuators. IEEE Trans. Autom. Sci. Eng. 8(2), 428–431
(2011)
69. B. Ren, P. San, S. Ge, T.H. Lee, Adaptive dynamic surface control for a class of strict-feedback
nonlinear systems with unknown backlash-like hysteresis, in American Control Conference,
pp. 4482–4487 (2009)
70. L. Riccardi, D. Naso, B. Turchiano, H. Janocha, Design of linear feedback controllers for
dynamic systems with hysteresis. IEEE Trans. Control Syst. Technol. 22(4), 1268–1280 (2014)
71. S. Rosenbaum, M. Ruderman, T. Strohla, T. Bertram, Use of Jiles-Atherton and Preisach
hysteresis models for inverse feed-forward control. IEEE Trans. Magn. 46(12), 3984–3989
(2010)
72. C. Ru, L. Sun, Hysteresis and creep compensation for piezoelectric actuator in open-loop
operation. Sens. Actuators A Phys. 122(1), 124–130 (2005)
73. M. Ruderman, Direct recursive identification of the preisach hysteresis density function. J.
Magn. Magn. Mater. 348(12), 22–26 (2013)
74. M. Ruderman, T. Bertram, Discrete dynamic Preisach model for robust inverse control of
hysteresis systems, in 49th IEEE Conference on Decision and Control, pp. 3463–3468 (2010)
75. S.M. Salapaka, M.V. Salapaka, Scanning probe microscopy. IEEE Control Syst. Mag. 28(2),
65–83 (2008)
76. Y. Shan, K.K. Leang, Accounting for hysteresis in repetitive control design: nanopositioning
example. Automatica 48(4), 1751–1758 (2012)
77. J. Shen, W.Y. Jywe, H.K. Chiang, Y.L. Shu, Precision tracking control of a piezoelectric-
actuated system. Precis. Eng. 32(2), 71–78 (2008)
78. R. Smith, Z. Ounaie, A domain wall model for hysteresis in piezoelectric materials. J. Intell.
Mater. Syst. Struct. 11(1), 62–79 (2000)
79. D. Song, C.J. Li, Modeling of piezo actuator’s nonlinear and frequency dependent dynamics.
Mechatronics 9(4), 391–410 (1999)
80. G. Song, J.Q. Zhao, X.Q. Zhou, J.A.D. Abreu-Garcia, Tracking control of a piezoceramic
actuator with hysteresis compensation using inverse Preisach model. IEEE/ASME Trans.
Mechatron. 10(2), 198–209 (2005)
212 G.-Y. Gu et al.

81. C.Y. Su, Y. Stepanenko, J. Svoboda, T.P. Leung, Robust adaptive control of a class of
nonlinear systems with unknown backlash-like hysteresis. IEEE Trans. Autom. Control 45(12),
2427–2432 (2000)
82. U.X. Tan, T.L. Win, W.T. Ang, Modeling piezoelectric actuator hysteresis with singularity free
Prandtl-Ishlinskii model, in IEEE International Conference on Robotics and Biomimetics -
ROBIO2006, pp. 251–256 (2006)
83. G. Tao, F.L. Lewis, Adaptive Control of Nonsmooth Dynamic Systems (Springer, New York,
2001)
84. A. Visintin, Differential Models of Hysteresis (Springer, Berlin, 1994)
85. C. Visone, Hysteresis modelling and compensation for smart sensors and actuators. J. Phys.
Conf. Ser. 138(1), 012028 (2008)
86. S.R. Viswamurthy, R. Ganguli, Modeling and compensation of piezoceramic actuator
hysteresis for helicopter vibration control. Sens. Actuators A Phys. 135(2), 801–810 (2007)
87. G.V. Webb, D.C. Lagoudas, A.J. Kurdila, Hysteresis modeling of SMA actuators for control
applications. J. Intell. Mater. Syst. Struct. 9(6), 432–448 (1998)
88. Q. Xu, Y. Li, Micro-/nanopositioning using model predictive output integral discrete sliding
mode control. IEEE Trans. Ind. Electron. 59(2), 1161–1170 (2012)
89. Q. Xu, P.K. Wong, Hysteresis modeling and compensation of a piezostage using least squares
support vector machines. Mechatronics 21(7), 1239–1251 (2011)
90. M.J. Yang, G.Y. Gu, L.M. Zhu, Parameter identification of the generalized Prandtl-Ishlinskii
model for piezoelectric actuators using modified particle swarm optimization. Sens. Actuators
A Phys. 189, 254–265 (2013)
91. J. Yi, S. Chang, Y. Shen, Disturbance-observer-based hysteresis compensation for piezoelectric
actuators. IEEE/ASME Trans. Mechatron. 14(4), 456–464 (2009)
92. Y.H. Yu, Z.C. Xiao, N.G. Naganathan, R.V. Dukkipat, Dynamic Preisach modeling of
hysteresis for the piezoceramic actuator system. Mech. Mach. Theory 37(1), 75–89 (2002)
93. X. Zhao, Y. Tan, Modeling hysteresis and its inverse model using neural networks based on
expanded input space method. IEEE Trans. Control Syst. Technol. 16(3), 484–490 (2008)
94. J.H. Zhong, B. Yao, Adaptive robust precision motion control of a piezoelectric positioning
stage. IEEE Trans. Control Syst. Technol. 16(5), 1039–1046 (2008)
95. J. Zhou, C. Wen, C. Zhang, Adaptive backstepping control of a class of uncertain nonlinear
systems with unknown backlash-like hysteresis. Automatica 49(10), 1751–1759 (2004)
Chapter 7
Tracking Control for Nanopositioning Systems

Kam K. Leang and Andrew J. Fleming

Abstract The performance of nanopositioning systems is greatly affected by their

mechanical dynamics, and for piezo-actuated designs, induced structural vibration,
hysteresis, and creep can drastically limit positioning precision. Therefore, tracking
control, both feedback and feedforward control, plays an important role in achieving
high-performance operation, especially at high operating frequencies. This chapter
reviews popular feedback and feedforward control techniques for nanopositioning
systems. First, the effects of vibration, hysteresis, and creep are described, where
simple methods traditionally employed to avoid these effects are discussed. Second,
various models for nanopositioning systems for control system design, simulation,
and synthesis are presented. Finally, popular feedback and feedforward controllers
to handle vibration, hysteresis, and creep are presented, along with experimental
results.

7.1 Introduction

Nanopositioning systems rely on tracking controllers for optimum performance.

Extensive studies have been done to explore control techniques to compensate for
the effects of vibration, hysteresis, and creep in nanopositioning systems [1–4].
In general, these techniques can be categorized as either feedback control or
feedforward control, each with its advantages and disadvantages. In some cases,
integrated feedback and feedforward controllers are employed.
Feedback control schemes (see basic block diagram in Fig. 7.1a) such as
proportional-integral-derivative (PID) [5], state-feedback [6], and H1 control [7, 8]
have demonstrated substantial reduction of positioning errors due to creep and
hysteresis. The advantages of feedback control include: (1) the ability to handle
modeling errors and (2) robustness to parameter variation, such as those caused by

K.K. Leang ()

University of Utah, Salt Lake City, UT, USA
e-mail: [email protected]
A.J. Fleming
University of Newcastle, Callaghan, Australia
e-mail: [email protected]

© Springer International Publishing Switzerland 2016 213

C. Ru et al. (eds.), Nanopositioning Technologies,
DOI 10.1007/978-3-319-23853-1_7
214 K.K. Leang and A.J. Fleming

a Tracking Feedback
error input
Reference yref +∑ e Feedback u Nano- y
Output
output - controller positioner
y

Feedforward
b input
Desired yd Feedforward uff Nano- y
output controller positioner
Output

c
Desired yd Feedforward Feedforward input
output controller uopt
Tracking Feedback
error input
Reference yopt + e Feedback u + + Nano- y
∑ ∑ positioner Output
output - controller
y

Fig. 7.1 Control systems for nanopositioning: (a) feedback controller, (b) feedforward controller,
and (c) integrated feedback and feedforward controller

aging effects [9] and environmental changes (temperature changes [10]). Although
feedback control reduces positioning errors, the low gain margin in piezo positioners
tend to limit the achievable performance, i.e., high feedback gain tends to destabilize
the system [5, 11]. In particular, low structural damping in piezo positioners results
in high quality factor Q (i.e., sharp resonant peak) which gives rise to low gain
margin. In practice, a compromise is sought between performance and instability;
feedback gains (such as proportional, derivative, and integral terms) are adjusted
to improve performance without instability (e.g., in references [12, 13]). Moreover,
feedback-based approaches have had limited success in compensating for dynamics
effects [5].
To overcome some of the performance limitations inherent in feedback con-
trol (e.g., limited dynamic compensation due to low gain margin), model-based
feedforward control (see Fig. 7.1b) has been used to compensate for creep and
hysteresis effects [14–19]. The feedforward input is found by carefully model-
ing the complicated creep and hysteresis behaviors and inverting the models to
determine the control inputs. Many model-based approaches have been considered,
for instance, Smith et al. [20] developed a hysteresis model based on energy
methods and then used the model for the design of an inverse compensator. Croft
et al. [21, 22] and Stilson et al. [23] compensated for all three effects using an
inversion-based feedforward technique and they applied their technique to scanning
probe microscopy (SPM) imaging. However, a major disadvantage of using model-
based feedforward approaches to compensate for creep and hysteresis is that these
7 Tracking Control for Nanopositioning Systems 215

effects are difficult to model accurately and the approach is prone to errors [24].
Furthermore, model-based approaches can be computationally cumbersome to
implement.
Finally, integrated feedback and feedforward control schemes (see Fig. 7.1c)
have been developed [4, 25] to combine the advantages of both approaches. In
particular, feedback control minimizes the nonlinear effects such as hysteresis
and the operating speed is improved by the addition of feedforward input to
compensate for the vibrational dynamics. This chapter presents popular feedback
and feedforward controllers to handle vibration, hysteresis, and creep, along with
experimental results.

7.2 Challenges in Tracking Control of Nanopositioning

Systems

The performance of a nanopositioning system is affected by induced structural

vibration (vibrational dynamics), hysteresis, and creep in the piezoactuator, cross-
coupling behaviors in both the actuator and mechanical system, external dis-
turbances, and drift due to temperature variations [2, 4, 26, 27]. Cross-coupling
effects can be handled using multi-input-multi-output controllers and iterative
methods [28]. External disturbances are often dealt with using passive or active
vibration isolation systems. Below, the effects of vibrational dynamics, hysteresis,
and creep are discussed, as well as their impact on positioning performance.
Afterwards, tracking controllers to deal with these behaviors are presented.

7.2.1 Vibrational Dynamics

Vibration effects are caused by exciting the resonance modes of the positioning
system [22, 29] and in general vibration limits the operating bandwidth. For
example, the frequency response of a piezo-based positioner, such as the measured
response for a tube-shaped piezoactuator shown in Fig. 7.2, typically reveals sharp
resonant peaks. These peaks can easily be excited by certain command signals
like triangle inputs applied to control the positioner. Figure 7.3 shows a measured
displacement versus time response for a piezo-positioning system in which the three
effects of vibration, hysteresis, and creep are observable. In particular, the effect
of vibration leads to oscillations, which cause significant tracking error in the
displacement versus time response. The slow drift in the response is caused by creep,
and the curved appearance in the output response is due to hysteresis. Overall, such
effects cause distortion in nanopositioning applications, including image distortion
in atomic force microscope images (AFMs). In fact, vibrations cause ripples to
appear in AFM images during high-speed scanning [30].
216 K.K. Leang and A.J. Fleming

20
Magnitude (dB) 0
-20
-40
-60
-80
0 1 2 3
10 10 10 10
Frequency (Hz)

Fig. 7.2 Frequency response of a tube-shaped piezoactuator

120

100

80
Displacement (μm)

60

40

20

0.2 Hz (T = 5.0) 1.0 Hz (T = 1.0) 30 Hz (T = 0.033)

0
0 T 2T 3T
Time period (T)

Fig. 7.3 Measured piezoactuator response (scanning at 30 Hz) showing the combined effects of
vibration, hysteresis, and creep

Typically, scan rates (i.e., scan frequencies) are restricted to less than 1=100th
to 1=10th of the first resonant frequency to minimize the effects of vibration, thus
limiting the bandwidth of piezo-based systems because the achievable scan rate
is lower for increased resolution in positioning. However, higher operating speed
can be achieved by using stiffer piezoactuators with higher resonant frequencies
[31–33], for example, Ando et al. [34] used a stiff piezo with a resonant frequency
of 260 kHz in an AFM to image biological macromolecules in action. It is pointed
7 Tracking Control for Nanopositioning Systems 217

out that compact and stiff piezoactuators have shorter effective displacement ranges.
Therefore, the use of stiffer piezos to increase bandwidth also leads to reduction of
positioning range.

7.2.2 Hysteresis

Hysteresis, which is a nonlinear behavior between the applied electric field and the
mechanical displacement of a piezoelectric actuator, is believed to be caused by
irreversible losses that occur when similarly oriented electric dipoles interact upon
application of an electric field [35]. The effect of hysteresis on the displacement of a
piezoelectric actuator is more pronounced over large-range motion [5], as indicated
in Fig. 7.3.
The term hysteresis comes from the Greek word “to be late” or “come behind”
and it was first coined for application in 1881 by physicist Ewing when he was
studying magnetization. Interestingly, the year 1881 was when the Curie Brothers
were credited with the discovery of the piezoelectric effect. Hysteresis is often
referred to as a lag in the response. An interesting writing on the history of hysteresis
can be found in reference [36], which describes other systems which exhibit this
behavior.
The mechanism responsible for hysteresis in piezoelectric transducers is better
understood by considering the domain wall analogy for describing hysteresis
in magnetic materials [35, 37]. For example, magnetic materials consist of tiny
elementary magnetic dipoles. These particles align to an applied field. The analogy
to this in piezoelectric materials the unit cell of the piezoelectric crystal material
exhibits an electric dipole. The term domains of polarization refers to regions
of similarly-oriented dipoles, that is, a relatively large region of connected unit
cells having similarly-oriented net polarization. The imaginary boundary which
separates these regions are referred to as domain walls. These boundaries grow
or shrink depending on the nature of the applied field. For the simple case, an
isolated elementary dipole subjected to an applied field will orient itself to the
field instantaneously, and therefore displays no hysteresis. However, hysteresis
is said to arise due to “internal forces,” which cause the dipoles to exhibit a
preference for their orientation, and hence the motion of the domain walls are
retarded by such forces. These internal forces are attributed to material defects and
internal friction between dipoles and the domain walls. Even though the domain
wall analogy was conceived for magnetic materials, it can easily be extended to
materials which consist of elementary dipole-like particles, such as piezoelectric
materials. Additionally, hysteresis “remembers” the effect of the past, which further
complicates the problem in terms of precision control.
Hysteresis in piezo-based nanopositioning systems causes AFM image distor-
tion [5, 30], instability of the closed-loop [11], and loss in calibration [19, 38].
218 K.K. Leang and A.J. Fleming

By operating the nanopositioner over a relatively small range, say less than 10 %
of full range, hysteresis can be avoided. However, operating over such a narrow
range limits the ability of the actuator for long-range motion with sub-nanometer
resolution.

7.2.3 Creep

During slow and static positioning, creep and drift effects can cause over 30 % error
in positioning [39]. Creep is a low-frequency behavior, where its time constant
is typically a few minutes. Creep severely degrades the low-frequency and static
positioning ability of piezoelectric actuators [40–42].
In mechanics, creep is a rate-dependent deformation of the material when
subjected to a constant load or stress [43]. Similarly, creep in piezoelectric materials
is a rate-dependent deformation due to a constant electric field. Creep manifests
itself as the remnant polarization slowly increases after the onset of a constant field.
Figure 7.4 shows the effect of creep from an experimental piezoactuator. The
actuator is commanded to a reference position, say 25 m, but after a period of
15 min, the actuators’ position creeps to a new position of 33:41 m. As a result,
the error due to creep is 24:44 % of the total displacement range. To avoid creep,
the nanopositioner can be operated at a relatively fast rate (above 1 Hz scanning
frequency) [30]; however, such effort prevents the use of piezo positioners in slow
and static applications. For example, because of creep, it is difficult to precisely
fabricate nanofeatures using AFMs when the process time-scale is on the order of
minutes, e.g., see [40].
Methods to compensate for creep have been well studied in the past and some
examples include the use of feedback control, e.g., [5, 8, 13, 44], and model-based
feedforward control, e.g., [16, 19, 30, 42, 45, 46].
Displacement (μ m)

30

20

10
Desired
Measured
0
0 2 4 6 8 10 12 14 16
Time (min)

Fig. 7.4 The effects of creep in the output displacement measured over a period of 15 min
7 Tracking Control for Nanopositioning Systems 219

7.3 Modeling Vibration, Hysteresis, and Creep for Control

The displacement (stoke) response of a piezoelectric actuator employed in a

conventional nanopositioning system under an applied voltage where the effects of
temperature variations and self-heating are ignored consists of dynamics (vibration
and creep) and nonlinearities (such as hysteresis). An overview of modeling of
a piezo-based nanopositioner is considered. The main objective is to describe
a modeling approach which enables the application of feedforward as well as
feedback control for precision tracking control.
As illustrated in Fig. 7.3, the measured output response of a piezo-based nanopo-
sitioner consists of the combined effects of vibration, hysteresis, and creep. In
Fig. 7.3, the oscillations are caused by vibrational dynamics; the slow upward-drift
of the output over time is due to the creep effect; and finally, a noticeable curved
distortion in the output trajectory is due to hysteresis.
To effectively model these behaviors, one must consider the operating conditions
and when certain effects dominate. For instance, the amount of the dynamics on
the output response depends on the operating frequency. The operating frequency
is determined by the application in mind. As a result, the system’s dynamics, be it
low-frequency or high-frequency, are excited by the frequency of the input signal.
Take for example when the input frequency is close to a nanopositioning
system’s resonance frequencies. In this case, vibration becomes noticeably large.
This is not surprising when one examines the frequency response of a typical
piezoelectric actuator, where sharp resonances are common as shown in Fig. 7.2.
Piezoactuators tend to be highly resonant structures due to their high stiffness
and low structural damping. Therefore, input signals such as sawtooth signals can
excite the piezoactuator’s resonances, causing the output to oscillate or vibrate as
previously shown in Fig. 7.3.
At slow operating speeds, creep is a major source of positioning error. Creep in
piezoactuators is a low-frequency behavior, where the output drifts, especially when
the operation is offset from the center of the piezoactuator’s positioning range (see
Fig. 7.3).
Finally, hysteresis is significant over large-range displacements [5]. The opera-
tion of piezoactuators in their linear range helps avoid hysteresis. In general, the
linear range is within 5 to 10% of the maximal range of motion.
The hysteresis and dynamic effects are coupled [30]. For instance, when
the movement of the piezoactuator is large and slow, the piezoactuator exhibits
hysteresis and creep effects. As the input frequency increases, the piezoactuator’s
output response shows the addition of the vibrational dynamics. To model these
behaviors, the cascade model depicted in Fig. 7.5a is used. The range-dependent
hysteresis effect is treated as a rate-independent, input nonlinearity represented
by HŒu./. The vibrational dynamics and creep effects are captured by the linear
dynamics model G.s/. The cascade model structure is used extensively to model
piezoactuators and similar systems [30, 47]. Below, details on how to obtain models
for these behaviors are discussed.
220 K.K. Leang and A.J. Fleming

Piezoelectric actuator
Linear
Hysteresis dynamics
u v y
Input H [ u(.)] G(s) Output

Fig. 7.5 A cascade model structure for hysteresis, vibrational dynamics, and creep effects in
piezoactuators. Hysteresis, denoted by HŒu./, is modeled as an rate-independent, input nonlin-
earity that is output-range dependent. The input-frequency-dependent vibrational dynamics and
creep effects follow the hysteresis block, and they are captured by the linear dynamics model G.s/

7.3.1 Modeling Linear Dynamics: Vibration and Creep

The transfer function models for the vibrational dynamics and creep effects of a
piezoactuator can be obtained in a number of ways. One straightforward approach is
curve fitting the measured frequency and time responses over appropriate frequency
ranges. Another is developing a model using white noise data or a chirp signal. The
former approach is discussed below to model vibrational dynamics and creep.
To find the vibrational dynamics model, which relates the input u to the
displacement in the x, y, or z axis, a system identification algorithm is applied
to the measured frequency response. Consider a tube-shaped piezoactuator with
transverse range of approximately 100 m. First, the frequency response along the
x-axis is measured over a displacement range of less than 10 % of the maximal
range to avoid the hysteresis effect. To avoid creep, the response is measured over
a wide frequency range, in this case, 1 Hz to 2 kHz. Frequency responses can be
obtained using commercially available dynamic signal analyzers (such as Stanford
Research Systems SRT785) or other system identification approaches.
Consider the example measured frequency response for a piezoactuator shown in
Fig. 7.6, where the solid line is the measured response.
Using a system identification algorithm, such as the function “invfreqs” in
Matlab, a transfer function model is fitted to the measured response. The dash line
shown in Fig. 7.6 is the model given by

7:2  1013 s2 C 2:3  1016 s C 3:2  1021

Gv .s/ D :
s6 C 1:1  104 s5 C 9:5  107 s4 C 7:0  1011 s3 C 2:0  1015 s2 C 5:6  1018 s C 1:0  1022
(7.1)
At very low speed, the creep effect is significant. This effect can be captured
by the Kelvin-Voigt model, which consists of spring (ki ) and damper (ci ) elements
[16, 48]. The lumped-parameter model shown in Fig. 7.7 is linear, and its transfer
function is

x.s/ 1 X n
1
Gc .s/ D D C ; (7.2)
u.s/ k0 iD1
sci C ki
7 Tracking Control for Nanopositioning Systems 221

Magnitude of Gv (s) (dB)

20
Gain margin = -17.05 dB
0
-20
-40
-60 Measured piezoactuator response
-80 Linear model
0 1 2 3
10 10 10 10
Frequency (Hz)
Phase of Gv (s) (degrees)

-100

-200

-300 Measured piezosactuator response

Linear model
-400
0 1 2 3
10 10 10 10
Frequency (Hz)

Fig. 7.6 The frequency response of a piezoactuator for modeling the vibrational dynamics.
These plots show the measured frequency response (solid line), the magnitude and phase versus
frequency, of the piezoactuator over ˙2-m displacement range. The dash line is the linear
vibrational dynamics model

k1 k2 kn xn
k0
u

c1 c2 cn

Elastic Creep Model

deflection Kelvin-Voigt elements

Fig. 7.7 The spring-damper model for the creep effect. The parameters of the model are found by
curve fitting the measured step response of the piezoactuator

where x.s/ is the displacement of the piezoactuator and u.s/ is the applied input
voltage. In (7.2), k0 models the elastic behavior at dc, and the creep behavior is
captured by selecting an appropriate model order corresponding to the number of
spring-damper elements n. The parameters k0 , ki , and ci of (7.2) are often determined
by curve fitting the step response of the piezoactuator measured over a long period
of time.
222 K.K. Leang and A.J. Fleming

The creep effect can also be modeled by a logarithmic model of the form
  
t
x.t/ D x0 1 C log10 ; (7.3)
t0

where x.t/ is the actuator’s displacement, x0 is the nominal fast displacement to the
applied voltage, is the creep rate, and t0 is the settling time of the high-frequency
dynamics [19].

7.3.2 Modeling Hysteresis

Hysteresis is widely accepted as a nonlinear behavior characterized by a nonvanish-

ing input-output loop as the frequency decreases to zero. The behavior sometimes
corresponds to energy loss. Below are popular models for hysteresis for designing
and synthesizing controllers.

7.3.2.1 Simple Polynomial Model

The simplest model is a polynomial fit of the output versus input data, with the form,

y D an un C an1 un1 C    C a1 u C a0 ; (7.4)

where n is the order of the polynomial, y is the output, u is the input, and the ai ’s
coefficients can be found using a least-squares algorithm. It is pointed out that
this model ignores the branching (looping) behavior, that is, the ascending and
descending branches are the same. Such a model provides reasonable accuracy
for electrostrictive actuators as they tend to exhibit significantly less hysteresis
compared to piezoelectric actuators [49]. Inversion of the polynomial hysteresis
model for feedforward control is straightforward. A look-up table can be used in
this case.

7.3.2.2 Maxwell Slip Model

A more realistic representation is the Maxwell slip model, a lumped-parameter

approach [50]. In this model, the nonlinear input-output map and looping behavior
are caused by a parallel combination of elasto-slide elements as shown in Fig. 7.8.
An elasto-slide element consists of a massless linear spring and a massless block
that is subjected to Coulomb friction as shown in Fig. 7.8. The constitutive behavior
of the ith element is given by
7 Tracking Control for Nanopositioning Systems 223

N1
k1
Single elasto-slide element
μ
F
N2
k2
x
μ
F
x

Nn
kn
μ

Fig. 7.8 Maxwell slip model for hysteresis

(
ki .x  xbi / if jki .x  xbi /j < fi
Fi D (7.5)
fi sgn.Px/ and xbi D x  kfii sgn.Px/ else;

where x is the input displacement, Fi , ki , fi , and xbi are the output force, spring
stiffness, breakaway force, and block position, respectively, for the ith element. The
resultant output force is computed as

X
n
FD Fi : (7.6)
iD1

Starting in a relaxed state, the measured input-output map can be used to

determine the individual spring constants and break away forces (’s). The Maxwell
slip model has been applied to model and control a piezoelectric stack actuator
in [50].

7.3.2.3 Duhem Model

A finite-dimensional, differential model typically used to model ferromagnetically

soft materials [51] can be adapted to model the rate-independent hysteresis in the
piezoactuator. Such a model is given by

v.t/
P D ˛jPu.t/j ˇu.t/  v.t/ C uP .t/; (7.7)

where v.t/ is the output, u.t/ is the input, and ˛, ˇ, are positive constants.
The parameters of the model can obtained by a least-squares fit of the measured
224 K.K. Leang and A.J. Fleming

input-output data. For example, the input u.t/ D A sin.!t/, with ! D 2 rad/s
and A D 2 V, was applied to a piezoactuator and the output v.t/ was measured.
Assuming zero initial conditions, Eq. (7.7) was rewritten in the following matrix
form

v D Q‚; (7.8)

where v D Œv.t1 /; v.t2 /; v.t3 /; : : :T is the vector of measured outputs at specific
time instances; ‚ D Œ˛ˇ;  ˛; T is the vector of unknown parameters; and
2 3
q1 .t1 / q2 .t1 / q3 .t1 /
6q1 .t2 / q2 .t2 / q3 .t2 /7
6 7
Q D 6q .t / q .t / q .t /7 ; (7.9)
4 1 3 2 3 3 3 5
:: :: ::
: : :

where q3 .t/ D u.t/ and

Z Z
q1 .t/ D jPu.t/ju.t/dtI q2 .t/ D jPu.t/jv.t/dt: (7.10)

The identified parameters were ˛ D 0:3, ˇ D 1:2, and D 1:0. Figure 7.9a
compares the normalized measured and model outputs for u.t/ D 2 sin.2 t/ V.

Fig. 7.9 Comparison of measured and Duhem hysteresis model output for a piezoactuator: (a) The
normalized outputs versus time and (b) a hysteresis curve. Input was u.t/ D 2 sin.!t/ V, with
! D 2 rad/s
7 Tracking Control for Nanopositioning Systems 225

The hysteresis curves are compared in Fig. 7.9b. The results show good agreement
between the measured and model outputs, where the maximum steady-state error
was less than 2 % of the total range.

7.3.2.4 Preisach Model

A relatively accurate hysteresis model is the Preisach hysteresis model, which

was first developed in 1935 for magnetic materials [52]. This model has been
studied extensively to characterize the rate-independent hysteresis in piezoelectric
materials [53, 54], as well as many hysteretic systems, such as shape memory alloy
devices [18]. This model can be inverted for feedforward control [30].
The Preisach model is a phenomenological description, whereby the output of a
hysteretic system is the net effect of elementary relays, which represent the behavior
of individual domains within the material. These domains or relays can assume a
value of C1 or 1 depending on the current and future values of the input. The relay
operator R W R ! f1; C1g is defined as [54]:
8
< C1 u.t/ > ˛;
R˛;ˇ Œu.t/ D 1 u.t/ < ˇ; (7.11)
:
unchanged ˇ  u.t/  ˛;

where u.t/ is the input. The pair (˛; ˇ) in Eq. (7.11), such that ˛  ˇ, represents the
“up” and “down” switching values of the relay, respectively. Figure 7.10a shows an
example of an elementary relay (also called Preisach hysteron).
The output v.t/, which is an infinite sum of hysterons [Fig. 7.10b], is written as
“
v.t/ D HŒu.t/ D .˛; ˇ/R˛;ˇ Œu.t/d˛dˇ; (7.12)
˛ˇ

a Rα,β [u] b Hysteresis effect in a piezo, H [u]

+1 R α1, β1 [u](t) μ(α ,β)

1 1
. . .
. . .
. . .

β α u u(t) R α i, β i [u](t) μ(α ,β )

i i Σ v(t)
. . .
. . .
. . .
. . .
. . .
-1 . . .

Fig. 7.10 (a) The elementary relay (Preisach hysteron) and (b) the output v.t/ is the sum of
weighted hysterons
226 K.K. Leang and A.J. Fleming

where .˛; ˇ/ is called the Preisach weighting function, and .˛; ˇ/ belongs to the
Preisach plane P, defined as,

P , f.˛; ˇ/j˛  ˇI u  ˛; ˇ  uN g: (7.13)

The Preisach hysteresis model can be obtained experimentally from measured

output data, for instance, by applying an appropriate input voltage and measuring
the piezoactuator’s displacement response. In this case, based on the work of Banks
et al. [55], a discrete form of the output Eq. (7.12) can be approximated by

X
N
v.t/  R r Ar ; (7.14)
iD1

where Ai represents the area associated with the ith node, i is the average value of
the weighting surface over area Ai and R takes on value C1 or 1 depending on
the state of the node, or relay at the node.

7.3.2.5 Prandlt-Ishlinksii Model

Another operator-type hysteresis model which has recently been investigated for
piezoactuators is the Prandlt-Ishlinskii model, a phenomenological model that
is characterized by a play operator or a stop operator [56, 57]. This operator
is continuous and rate-independent, as shown in Eq. (7.15). If the input u is
continuous and monotone in subspace CŒti ; tiCn  of CŒ0; t, then the operator maps
function Hr W CŒ0; t ! CŒ0; t, that is,

Hr Œu.o/ D hr .u.0/; 0/ D 0;
Hr Œu.t/ D hr .u.t/; Hr Œu.t//: (7.15)

In the operator function, hr .u.t/; Hr Œu.ti // D max.ur; min.uCr; Hr Œu.ti1 ///.

The square bracket of the operator (Hr Œ) means that the operator is dependent on
input function. The operator maps an input function to an output function, which
represents the basic behavior of hysteresis, as illustrated in Fig. 7.11. The r in
the hr ./ is the threshold. The threshold r decides when the operator is turned on,
therefore with different r the operator maps the input and output curve in different
shapes as shown in Fig. 7.11.
With this operator the P-I model describes the input u and hysteresis output yh by
Z R
yh .t/ D ku C p.r/Hr Œu.t/dr; (7.16)
0

where k is the positive gain, and p.r/ is the density function, which affects the shape
and size of the hysteresis curve.
7 Tracking Control for Nanopositioning Systems 227

Fig. 7.11 The hysteresis play Pr

operator

-r r u

7.4 Feedback Control for Nanopositioning

A feedback controller works by comparing the reference behavior trajectory (such

as displacement) to the actual response (also displacement), see Fig. 7.1a. By
minimizing the positioning error, a feedback controller can compensate for all
forms of positioning errors that are within its effective bandwidth. Feedback
control is the most commonly used technique for eliminating positioning errors
in nanopositioning systems due to its simplicity and robustness. This section
provides an overview of feedback control techniques, including high-gain feedback
and damping-type controllers such integral resonant control. When the reference
trajectory is periodic, repetitive control can significantly improve the tracking
performance of a feedback loop. The repetitive control approach is briefly described
for nanopositioning systems.
Feedback control schemes such as traditional PID [5], state-feedback [6], gain
scheduling [58], and H1 control [3, 7] handle modeling errors and are robust
with respect to parameter variation. Although feedback control has been applied
extensively in nanopositioning, the low gain margin in piezo positioners tend to limit
the performance, i.e., high feedback gain tends to destabilize such systems [5, 11].
In practice, a compromise is sought between performance and instability; feedback
gains (such as proportional, derivative, and integral terms) are adjusted to improve
performance without instability [12, 13]. Additionally, the precision of feedback
controllers can be sensitive to sensor noise and can be bandwidth limited.
Feedback controllers such as standard proportional-integral (PI) controllers are
commonly used due to their robustness and high low-frequency gain, but PI con-
trollers offer limited closed-loop bandwidth. It was shown that for a nanopositioner
with second-order dynamics, the maximum closed-loop bandwidth !N BW is [59]

!N BW  2!n ; (7.17)
228 K.K. Leang and A.J. Fleming

where !n is the natural frequency and  is the damping ratio of the positioning
system. With this, low damping ratio severely limits the closed-loop bandwidth of a
PI-controlled system.
The simplest and most popular approach to suppress the sharp resonance
behavior is to use notch filters or inversion-based filters in the closed-loop [6, 25,
60]. By doing this, the gain margin and closed-loop bandwidth can be improved,
even up to the resonance frequency [60]. For example, the notch filter of the
following form can be used

.s  2 z1 /.s  2 z2 /
D.s/ D kD ; (7.18)
.s  2 p1 /.s  2 p2 /

where kD is a constant, z1 and z2 are the zeros of the filter that cancel the effect
of the sharp resonant peak of the nanopositioning system, and p1 and p2 are poles
to ensure a proper transfer function D.s/. Such notch filters can be realized using
analog op-amp circuits (e.g., [61], pp. 394–399).
Figure 7.12 compares the measured frequency response for a piezoscanner, a
notch filter, and the composite system. The measured gain margin of the original
system is 17:05 dB, whereas the gain margin of the composite system is 30:86 dB.
Thus, the notch filter significantly increase in the gain margin to enable high-gain
feedback control to compensate for effects such as hysteresis and creep, as well
as improve transient response [25]. It is pointed out that employing notch filters

20
Gain margin of composite system = 30.86 dB
Magnitude (dB)

0
-20
-40
Piezoscanner
-60 Notch filter
-80 Composite system
0 1 2 3
10 10 10 10
Frequency (Hz)
100
0
Phase (degrees)

-100
-200 Piezoscanner
Notch filter
-300
Composite system
-400 0 1 2 3
10 10 10 10
Frequency (Hz)

Fig. 7.12 Measured frequency response of the piezoscanner (dashed line), a notch filter (dotted
line), and a notch filter cascaded with the piezoscanner (solid line). The measured gain margin of
the original system is 17:05 dB, whereas the gain margin of the composite system is 30:86 dB
7 Tracking Control for Nanopositioning Systems 229

u ff
PI Regulator Plant
r + + y
ki +
Reference
Σ s Σ C(s) G(s)
Output
- -

Fig. 7.13 Advanced integral resonant control (IRC) with integral action and a feedforward input
uff [69]

requires knowledge of the resonance frequency. Even a small shift in the resonance
behavior as low as 10 %, say caused by changes in the payload mass, can lead to
instability of the closed-loop system.
Robust [8], H1 -based [62, 63], and loop-shaping [64] control techniques have
been considered for high-bandwidth operation. Other damping control techniques
include positive position feedback (PPF) control and polynomial-based controller
designs [65, 66]. In the PPF approach, an inner feedback loop is used to damp
the highly resonant mode of the positioner. Application of the PPF controller
on a commercial AFM system demonstrated damping as well as cross-coupling
compensation for a piezoelectric tube scanner [67]. Integral resonant control (IRC),
a method introduced as a means for augmenting the structural damping of resonant
systems with collocated sensors and actuators [68], damps the vibration and makes
the system robust to the unmodeled dynamics and resonance frequency variations
due to changes in the payload [69]. The block diagram of an advanced version
of the IRC approach which includes integral action and a feedforward input uff is
illustrated in Fig. 7.13, where G.s/ is the plant and the regulator C.s/ is

k
C.s/ D ; (7.19)
s  kDf

k is the constant, and Df is the feedthrough term [69]. Other active damping
techniques include receding horizon control [70, 71] as well as active and passive
shunt damping techniques for piezoelectric tube scanners [72, 73].
Recently, a feedback-based approach known as repetitive control (RC) was
exploited for tracking periodic trajectories in nanopositioning applications includ-
ing high-speed and metrological AFMs [74–76]. The RC provides high gain at
the harmonics of the reference trajectory by incorporating a signal generator
within the feedback loop [77, 78]. For piezo-based nanopositioning systems, the
RC can be designed for low tracking error in the presence of dynamic and
hysteresis effects [79, 80]. Some advantages of this approach include minimal
system modeling, robustness due to the feedback structure, and straightforward
digital implementation; however, a priori knowledge of the period of the reference
trajectory is required.
230 K.K. Leang and A.J. Fleming

Repetitive controller (RC)

+ B(z) -N A(z)
z Q(z) P1(z)
+
k P2(z) Go(z)
rc

R(z) + E(z) + C(z) Y(z)

Gc(z) G (z)
p
- +

Fig. 7.14 The repetitive control (RC) feedback system

To illustrate the concept of RC, consider an example of the discrete-time

RC closed-loop system shown in Fig. 7.14. The dynamics of the nanopositioner,
assumed to be linear, are represented by Gp .z/, where z D ej!Ts , ! 2 .0; =Ts /.
In the block diagram, Gc .z/ is a feedback controller, such as an existing PID
controller; Q.z/ is a low-pass filter for robustness; krc is the RC gain; and
P1 .z/ D zm1 and P2 .z/ D zm2 , where m1 ; m2 are non-negative integers, are positive
phase lead compensators to enhance the performance of the RC feedback system.
It is emphasized that the phase lead compensators zm1 and zm2 provide a linear phase
lead of (in units of radians)

1;2 .!/ D m1;2 Ts !; for ! 2 .0; =Ts /: (7.20)

The key component of the repetitive controller is the signal generator. To create
a signal generator with period Tp , the inner loop contains the pure delay zN , where
the positive integer N D Tp =Ts is the number of points per period Tp ; and Ts is
the sampling time. An analysis of the performance of the closed-loop system is
presented below, where the following assumptions are considered: (1) the reference
trajectory R.z/ is periodic and has period Tp and (2) the closed-loop system without
the RC loop is asymptotically stable, i.e., 1 C Gc .z/Gp .z/ D 0 has no roots outside
of the unit circle in the z-plane.
Assumptions 1 and 2 are easily met for many applications in nanopositioning,
including SPMs. For example in AFM imaging the lateral movements of the
piezoactuator are periodic, such as a triangle scanning signal. Also, most SPMs are
equipped with feedback controllers Gc .z/ to control the lateral positioning, which
can be tuned to be stable.
The transfer function of the signal generator (or RC block, Fig. 7.14) that relates
E.z/ to A.z/ is given by

A.z/ Q.z/P1 .z/zN Q.z/z.NCm1 /

D D : (7.21)
E.z/ 1  Q.z/P1 .z/zN 1  Q.z/z.NCm1 /
7 Tracking Control for Nanopositioning Systems 231

In the absence of both the low-pass filter Q.z/ and positive phase lead P1 .z/ D zm1 ,
the poles of the signal generator are 1  zN D 0; therefore, the frequency response
of the signal generator reveals infinite gain at the fundamental frequency and its
harmonics ! D 2n =Tp , where n D 1; 2; 3; : : :. The infinite gain at the harmonics
is what gives the RC its ability to track a periodic reference trajectory. As a result,
RC is a useful control method for applications such as SPM in which the scanning
motion is repetitive. Unfortunately, the RC also contributes phase lag which causes
instability. Therefore, the stability, robustness, and tracking performance of the RC
closed-loop system must be carefully considered.
For the RC system shown in Fig. 7.14, it has been shown that for stability the
following two conditions for the RC gain krc and linear phase lead 2 .!/ must be
met [74]:

2 cosŒT .!/ C 2 .!/

0 < krc < and (7.22)
A.!/
 =2 < ŒT .!/ C 2 .!/ < =2: (7.23)

In Eq. (7.23), the lead compensator P2 .z/ D zm2 accounts for the phase lag
of the closed-loop feedback system without RC. Enhanced designs of RC for
nanopositioning include dual-stage RC [79] and RC that accounts for hysteresis
behavior [80].
The tracking results comparing PID, regular RC, and the RC with the phase lead
compensators for ˙25-m scanning at 5, 10, and 25 Hz are presented in Fig. 7.15
and Table 7.1. The steady-state tracking errors, measured at the last two cycles, are
reported as a percentage of the range of motion (maximum error and root-mean-
squared error). Because the action of the repetitive controller is delayed by one scan
period, the tracking response for the first period are similar for the PID, RC, and
RC with phase lead compensation as shown in Fig. 7.15. However, after the first
period the RC begins to take action as illustrated by the reduction of the tracking
error from one cycle to the next. On the other hand, the tracking error of the PID
controller persists from one cycle to the next.
The 5 Hz scanning results shown in Fig. 7.15(a1) and (b1) and Table 7.1
demonstrate that the regular RC controller reduced maximum tracking error from
2.01 to 0.96 % compared to the PID controller, a 52 % reduction. By using RC
with the phase lead compensation, an additional 55 % improvement in tracking
performance was achieved. In this case, the maximum tracking error is 0.43 %.

Table 7.1 Tracking results for ˙25-m range

5 Hz 10 Hz 25 Hz
Controller emax .%/ erms .%/ emax .%/ erms .%/ emax .%/ erms .%/
PID 2:01 1:28 3:99 2:61 9:16 6:61
RC 0:96 0:21 2:74 0:79 8:86 3:69
RC + phase leads 0:43 0:08 0:46 0:10 1:78 0:57
232 K.K. Leang and A.J. Fleming

(a1) 30 5
4 (b1)
Displacement (μ m)

Tracking error (μ m)
20 3
10 2
1
0 0
-1
-10 -2
-20 -3
-4
-30 -5
0 0.5 1 1.5 2 0 0.5 1 1.5 2
Time (s) Time (s)
(a2) 30 5
4 (b2)
Displacement (μ m)

Tracking error (μ m)
20 3
10 2
1
0 0
-1
-10 -2
-20 -3
-4
-30 -5
0 0.2 0.4 0.6 0.8 1 0 0.2 0.4 0.6 0.8 1
Time (s) Time (s)
(a3) 30 10
(b3)
Displacement (μ m)

Tracking error (μ m)
20
5
10
0 0
-10
-5
-20
-30 -10
0 0.1 0.2 0.3 0.4 0.5 0 0.1 0.2 0.3 0.4 0.5
Time (s) Time (s)

Desired PID RC RC + phase leads

Fig. 7.15 Experimental tracking response and error for PID (dash-dot), RC (dashed line), and RC
with phase lead compensation [m1 D 6 and m2 D 0] (solid line) for 5 Hz (a1 and b1), 10 Hz (a2
and b2), and 25 Hz (a3 and b3) scanning

At 25 Hz, the tracking error of PID was unacceptable large at 9.16 %. In fact, for
AFM scanning operations the maximum tracking error should be less than a few
percent. The results in Table 7.1 show that the regular plug-in RC controller was
not able to improve the tracking performance at 25 Hz. However, the RC with phase
lead compensation gave lower maximum tracking error at 1.78 %. Therefore, the
RC with phase lead compensation enables precision tracking at higher scan rates.
Finally, scanning offset from the piezoactuator’s center position is demonstrated
as shown in Fig. 7.16. For this offset scanning operation, the PID controller
accounted for the low-frequency dynamics such as creep and the RC was used for
tracking the periodic trajectory. The tracking results in Fig. 7.16 show that the RC
was effective at minimizing the tracking error.

7.5 Feedforward Control for Nanopositioning Systems

Feedback controllers react to the measured tracking error; however, feedforward

controllers compensate or anticipate for deficit performance. As the operating speed
of a piezo-based nanopositioner increases, performance is limited by essentially
dynamic and hysteresis effects [1, 81]. Feedforward control is an open-loop
approach as depicted in Fig. 7.1b. As shown, an inverse model produces the
feedforward input uff that is applied to the positioning system. The accuracy of
7 Tracking Control for Nanopositioning Systems 233

15
Displ. (μm)
10

Desired PID RC RC + phase leads

0
0 0.1 0.2 0.3 0.4 0.5
Time (s)

5
Tracking error (μm)

PID RC RC + phase lead

-5
0 0.1 0.2 0.3 0.4 0.5
Time (s)

Fig. 7.16 Tracking results for offset triangle scan at 25 Hz

feedforward control, for example how close the actual output y matches the desired
output yd , depends on the quality of the inverse model and whether external distur-
bances are present. Being an open-loop approach, feedforward control is subject to
certain shortcomings, namely lack of robustness. However, for applications such as
vibration compensation with a reasonably accurate model of the system dynamics,
the advantages of feedforward control outweighs its disadvantages. In particular,
feedforward control can provide high-bandwidth positioning, exceeding that of
feedback-based methods. Also, feedforward control does not require continuous
sensor feedback, and thus sensor-noise related issues can be avoided entirely.
To improve robustness, feedforward can be integrated with feedback control, as
well to account for nonlinearity such as hysteresis [25]. Integrated feedback and
feedforward control also eliminates the need for modeling and inverting nonlinear
behaviors. Such a task can be difficult and computationally demanding. If iteration
is allowed, iterative feedforward techniques as discussed below provides good
performance with minimal modeling.

7.5.1 Simple DC-Gain Feedforward Control

At frequencies well below the dominant resonant peak, a simple feedforward input
uff can be computed by scaling the desired output trajectory yd .t/ by the inverse of
the DC-gain G.0/ of the system G.s/. For example,
234 K.K. Leang and A.J. Fleming

1
uff .t/ D yd .t/: (7.24)
G.0/

Simple DC-gain feedforward is often used when a plant model is difficult and/or
expensive to obtain. Also, if the frequency components of the desired trajectory
are well below the dominant resonances, the need for inverting high-frequency
dynamics for feedforward is less important. It is noted that simple DC-gain
feedforward offers good performance at low-frequency and over small displacement
ranges, and also under the assumption that the DG-gain remains stable.

7.5.2 Inversion-Based Feedforward Control for Linear

Dynamics

At higher frequencies, the effects of dynamics must be considered in the feed-

forward control input. In this case, the problem becomes inverting a dynamics
model G.s/ to find the feedforward input uff .t/ over a specific frequency range.
For piezoactuators, the dynamics include vibration and the creep effect [30].
Let G.s/ be a transfer function model that captures the linear dynamics of the
piezoactuator. In the frequency domain, the feedforward input that accounts for the
dynamics is given by

uff .j!/ D G1 .j!/yd .j!/; (7.25)

where G1 .j!/ is the inverse dynamics model.

In Eq. (7.25), the Fourier transform of the desired output trajectory yd .j!/ and
a plant model G.j!/ are needed to determine the feedforward input. If a measured
frequency response function G.j!/ is available, for example measured by a dynamic
signal analyzer, then the experimental data can be used directly to compute the
feedforward input uff .j!/. Then, the time-domain solution to Eq. (7.25) is found
by taking the inverse Fourier transform of uff .j!/.
One key feature of this feedforward technique is that it can be applied to
nonminimum-phase systems [82–84]. Although the dynamic effects are specifically
addressed in this section, the approach can be combined with alternative feedfor-
ward or feedback methods that compensate for hysteresis when the range of motion
becomes large [25].
The inversion-based method presented above may yield excessively large inputs
when the system has lightly damped system zeros. These large inputs can saturate
the voltage amplifiers that drive the piezoactuator, or, even worse, depole the
piezoactuator. Additionally, large model uncertainties around the resonant peaks or
lightly damped zeros can cause significant error in computing the feedforward input.
These model uncertainties thus lead to a lack of robustness when the inversion-based
feedforward method is used. The following optimal inversion approach is used to
7 Tracking Control for Nanopositioning Systems 235

account for these issues. Specifically, an optimal feedforward input is obtained by

minimizing the quadratic cost function [85]
Z 1 ˚ 
J.u/ D u .j!/R.j!/u.j!/ C Œx.j!/  xd .j!/ Q.j!/ Œx.j!/  xd .j!/ d!;
1
(7.26)
where ‘*’ denotes the conjugate transpose, and R.j!/ and Q.j!/ are non-negative,
frequency-dependent real-valued weights on the input energy and the tracking error,
respectively. The optimal feedforward input uff ;opt that minimizes (7.26) is
 
G .j!/Q.j!/
uff ;opt .j!/ D yd .j!/: (7.27)
R.j!/ C G .j!/Q.j!/G.j!/

By choosing the frequency-dependent weights R.j!/ and Q.j!/, it is possible

to systematically consider the effects of the input magnitude and the model
uncertainties. For instance, the input energy weight R.j!/ can be chosen to be
much larger than the tracking error weight Q.j!/ at frequencies where large model
uncertainties exist or around lightly damped zeros. For details and implementation
issues, see [86]. Other feedforward methods include input shaping techniques [87]
and shaped-triangle technique for high-speed AFM [88].
Both linear dynamics and hysteresis behaviors can be compensated for using
feedforward control by inverting the plant dynamics and nonlinearities (see
Fig. 7.17) [30]. Both the dynamic and hysteresis effect are often intertwined [89],
but can be modeled for feedforward control using the cascade structured shown
in Fig. 7.17. Specifically, the range-dependent hysteresis effect is treated as a rate-
independent, input nonlinearity represented by H. The vibrational dynamics (and
creep effect) are captured by the linear dynamics model G.s/. The cascade model
structure is used extensively to model piezoactuators and similar systems [30, 47].
It is pointed out that the model-based feedforward approach cannot correct for
positioning errors due to plant uncertainties and external perturbations [90]; but
feedback control can be used to minimize the uncertainties and then combined with
feedforward control [25, 91–94].

Feedforward Piezoactuator
Desired controller
output Hysteresis Dynamics Output
Inverse uff v
yd
system H [ .] G(s) y

Fig. 7.17 The feedforward control approach for piezoactuator system

236 K.K. Leang and A.J. Fleming

uk (t)
Positioning yk (t)
system - Σ+ yd (t)

FFT FFT

u k+1(t) uk+1 (ω) + ek (ω)

ILC
IFFT Σ+ algorithm

Fig. 7.18 Block diagram of frequency-domain ILC scheme

7.5.3 Iterative Feedforward Control

Many nanopositioning applications, for example SPM imaging, require tracking a

repeating (periodic) trajectory. The scanning motion in this case repeats from one
cycle to the next. Rather than model and invert the dynamics and nonlinearities of
a positioning system for feedforward control, for repeating scanning motion, the
feedforward input can be found using iterative techniques for both dynamics and
hysteresis compensation [39, 95, 96]. This approach is commonly referred to as
iterative learning control (ILC), where recently an approximate inverse of the plant
dynamics G was used in the input-update law [95]

O 1 .!/ yd .!/  yk .!/ ;

ukC1 .!/ D uk .!/ C .!/G (7.28)

to achieve faster convergence (k is the iteration number). A block diagram of

the frequency-domain implementation of the ILC control law (7.28) is shown in
Fig. 7.18. Particularly, the input uk .t/ and tracking error ek .t/ are Fourier trans-
formed, then the ILC algorithm is applied, producing the updated input ukC1 .!/.
The time-domain input ukC1 .t/ is obtained by inverse Fourier transform. Then the
input is applied to the system and the process is repeated [28, 97]. A model-less ILC
approach where the inverse G1 is obtained from the measured input-output data
from the previously iteration step was considered for AFM applications [98–100].
Piezo-based nanopositioners exhibit nonlinearity such as hysteresis. Although
the ILC approach is well suited for precise control of linear dynamics, the challenge
is developing a convergence criteria for hysteresis. The difficulty in proving
convergence of ILC algorithms for hysteretic systems arises due to two main
reasons: (1) branching effects and (2) nonlinearity of each branch [101].
The inability, to predict the direction in which the input needs to be changed for
reducing the output error in hysteretic systems, can be overcome if the input-output
behavior is restricted to belong on one single hysteresis branch. This observation
that the direction can be determined from the output error on a single branch was
used to prove the convergence for an ILC algorithm for hysteretic systems in [39].
First, the desired output trajectory is partitioned into monotonic sections (several
7 Tracking Control for Nanopositioning Systems 237

branches). The algorithm is applied to each section until a desired tracking precision
is achieved.
In particular, the ILC algorithm (ILCA) of the following form can be used to
compensate for hysteresis [39]:

ukC1 .t/ D uk .t/ C Œyd .t/  yk .t/; 8t 2 I: (7.29)

The ILCA Equation (7.29) converges if the desired trajectory vd .t/ is continuous
and monotonic over the finite time interval I. The iteration gain for convergence
is based on the parameters of the hysteresis model, such as the Preisach hysteresis
model. It is pointed out that a sufficiently small iteration gain can be found provided
the output different can be bounded above and below by the input difference [39].

7.5.4 Adding Feedback to Feedforward Control

Modeling and inverting the dynamic and hysteresis effects are effective methods
for precision positioning in piezoactuators in nanopositioning systems [30, 86].
However, because the approach exploits knowledge of the piezoactuator behavior,
the modeling process can be time-consuming, particularly when both the inverse
dynamics and inverse hysteresis are used. If a simpler method to account for
hysteresis is preferred over control performance, then high-gain feedback control
can be used to linearize the nonlinear behavior of the piezoactuator [25]. In this
case, the vibrational dynamics are modeled and inverted as described above, then the
feedforward input is combined with the feedback controller as shown in Fig. 7.1c.
However, when feedback is used, piezoactuators often exhibit low gain margin and
can cause instability. Thus, as mentioned earlier, notch filters can be used to suppress
the dominant resonant peaks.

7.5.5 Experimental Tracking Results

The results shown in Fig. 7.19 and Table 7.2 demonstrate the performance of
feedforward control and integrated feedforward and feedback control applied to a
nanopositioning system in an AFM. A high-gain feedback controller as described
in [25] was used to compensate for hysteresis and creep. The optimal inversion
feedforward controller was used to minimize the vibrational dynamics. The scan
range for the experiment is 50:00 m. The weightings of the optimal inversion
controller were chosen to give up tracking of all frequency components beyond
450 Hz (to avoid saturation). In particular, the weightings are Q D 1 and R D 0 for
!  450 Hz, and Q D 0 and R D 1 for ! > 450 Hz.
As shown by the results, the integrated feedback and optimal inversion feed-
forward approach substantially reduces positioning errors compared to just using
238 K.K. Leang and A.J. Fleming

Feedback Control Feedback/Feedforward Control

25 a 50 Hz 25 e 50 Hz
x (μm)

x (μm)
0 0

-25 -25
0 20 40 0 20 40
25 b 100 Hz 25 f 100 Hz
x (μm)

x (μm)
0 0

-25 -25
0 10 20 0 10 20
25 c 140 Hz 25 g 140 Hz
x (μm)

x (μm)
0 0

-25 -25
0 7 14 0 7 14
25 d 200 Hz 25 h 200 Hz
x (μm)

x (μm)

0 0

-25 -25
0 5 10 0 5 10
Time (ms) Time (ms)

Fig. 7.19 Experimental results: tracking of triangular reference trajectory. (a)–(d) Feedback
control. (e)–(h) Feedback and optimal inversion feedforward control. Optimal weightings were
chosen as Q D 1 and R D 0 for !  450 Hz, and Q D 0 and R D 1 for ! > 450 Hz. Solid line is
the measured response and dotted line is the desired trajectory

Table 7.2 Tracking Integrated

performance for triangular Feedback feedback/feedforward
reference trajectory using
Scan rate (Hz) emax (%) erms (%) emax (%) erms (%)
integrated feedback and
optimal inversion 50 4:37 3:58 1:54 0:74
feedforward control 100 9:49 6:79 2:95 1:39
140 12:11 9:33 2:94 1:42
200 18:15 13:69 2:03 1:18
Values reported as percentage of total output range (50:00 m)

feedback control for tracking a triangular trajectory as illustrated in Fig. 7.19 and
Table 7.2. Good tracking (i.e., emax  5 %) is achieved even at the 200 Hz scan
rate. Furthermore, the maximum and root-mean-square of the tracking error were
reduced by over 88 and 91 % at this scan rate compared to using only feedback
control.
7 Tracking Control for Nanopositioning Systems 239

7.6 Cross-Coupling Effects, Charge Control,

and Thermal Drift

Careful mechanical design can significantly minimize cross-coupling effects [76].

However, during high-speed operation such as fast raster scanning in video-rate
AFMs, the cross-coupling effect cannot be ignored as inertial effects and poor
boundary conditions can cause out-of-plane modes to be excited [27]. Control
methods to deal with cross-coupling issues include inverse-based feedforward
control [28], H1 -based control [27, 102, 103], iterative feedforward approach [104],
and adaptive control [105].
By controlling the charge (rather than voltage) applied to the piezoactuator,
hysteresis can be reduced. Early demonstration of the effectiveness of charge drives
have appeared in the literature [106–109]. More recently, advancements in the
design of charge drives for piezoelectric actuators have resulted in effective designs
for operating at low and high frequencies [110–112]. These charge drives have
also been integrated with inversion-based feedforward control for piezoelectric
actuators [113] and SPM [112].
Thermal expansion can severely limit the precision, repeatability, and overall per-
formance of a nanopositioning system. Specifically, the result includes temperature-
dependent drift in the motion and high thermal stresses which ultimately lead to
cracking, warping, or loosening of components. Unfortunately, for positioning at
the nanometer scale, thermal effects cannot be ignored. In a typical AFM operated
in ambient temperature, even a small 1ı change in temperature can cause as much
as 50 nm drift [114]. Feedback controllers have been used successfully to minimize
thermal drift [115]. For SPMs, an effective method to deal with drift is to measure
and compensating for the effect through consecutive SPM images [116–119].

7.7 Summary

This chapter focused on feedback and feedforward controllers for nanopositioning

systems. First, the effects of vibration, hysteresis, and creep were described.
Simple methods traditionally employed to avoid these effects were discussed.
Afterwards, the modeling of behavior of nanopositioners was presented and these
models are typically used in control system design and simulation. Finally, popular
feedback and feedforward controllers to handle vibration, hysteresis, and creep were
presented. The controllers included IRC, repetitive control, model-based inversion
feedforward control, and iterative feedforward control methods.
240 K.K. Leang and A.J. Fleming

References

1. G.M. Clayton, S. Tien, K.K. Leang, Q. Zou, S. Devasia, A review of feedforward control
approaches in nanopositioning for high speed SPM. ASME J. Dyn. Syst. Meas. Control.
131(6), 061101 (19 pages) (2009)
2. S. Devasia, E. Eleftheriou, S.O.R. Moheimani, A survey of control issues in nanopositioning.
IEEE Trans. Control Syst. Technol. 15(5), 802–823 (2007)
3. S.M. Salapaka, M.V. Salapaka, Scanning probe microscopy. IEEE Control. Syst. Mag. 28(2),
65–83 (2008)
4. A.J. Fleming, K.K. Leang, Design, Modeling and Control of Nanopositioning Systems
(Springer, New York, 2014)
5. R.C. Barrett, C.F. Quate, Optical scan-correction system applied to atomic force microscopy.
Rev. Sci. Instrum. 62(6), 1393–1399 (1991)
6. Y. Okazaki, A micro-positioning tool post using a piezoelectric actuator for diamond turning
machines. Precis. Eng. 12(3), 151–156 (1990)
7. S. Korson, A.J. Helmicki, An h1 based controller for a gas turbine clearance control system,
in Proceedings of the IEEE Conference on Control Applications (1995), pp. 1154–1159
8. S. Salapaka, A. Sebastin, J.P. Cleveland, M.V. Salapaka, High bandwidth nano-positioner: a
robust control approach. Rev. Sci. Instrum. 73(9), 3232–3241 (2002)
9. F. Lowrie, M. Cain, M. Stewart, M. Gee, Time dependent behaviour of piezo-electric
materials. Technical Report 151, National Physical Laboratory, 1999
10. H.-J. Lee, D.A. Saravanos, The effect of temperature dependent material properties on the
response of piezoelectric composite materials. J. Intell. Mater. Syst. Struct. 9(7), 503–508
(1998)
11. J.A. Main, E. Garcia, Piezoelectric stack actuators and control system design: strategies and
pitfalls. AIAA J. Guid. Control. Dyn. 20(3), 479–485 (1997)
12. F. Braet, R.D. Zanger, S. Kammer, E. Wisse, Noncontact versus contact imaging: an atomic
force microscope study on hepatic endothelial cells in vitro. Int. J. Imaging Syst. Technol.
8(2), 162–167 (1997)
13. G. Schitter, P. Menold, H.F. Knapp, F. Allgöwer, A. Stemmer, High performance feedback
for fast scanning atomic force microscopes. Rev. Sci. Instrum. 72(8), 3320–3327 (2001)
14. J. Schafer, H. Janocha, Compensation of hysteresis in solid-state actuators. Sensors Actuators
A 49, 97–102 (1995)
15. P. Ge, M. Jouaneh, Tracking control of a piezoceramic actuator. IEEE Trans. Control Syst.
Technol. 4(3), 209–216 (1996)
16. H. Janocha, K. Kuhnen, Real-time compensation of hysteresis and creep in piezoelectric
actuators. Sensors Actuators A 79, 83–89 (2000)
17. S. Mittal, C.-H. Menq, Hysteresis compensation in electromagnetic actuators through
preisach model inversion. IEEE/ASME Trans. Mechatro. 5(4), 394–409 (2000)
18. S. Majima, K. Kodama, T. Hasegawa, Modeling of shape memory alloy actuator and tracking
control system with the model. IEEE Trans. Control Syst. Technol. 9(1), 54–59 (2001)
19. O.M.E. Rifai, K. Youcef-Toumi, Creep in piezoelectric scanners of atomic force microscopes,
in Proceeding of the American Control Conference (2002), pp. 3777–3782
20. R.C. Smith, M.V. Salapaka, A. Hatch, J. Smith, T. De, Model development and inverse
compensator design for high speed nanopositioning, in Proceeding of the 41st IEEE
Conference on Decision and Control (2002), pp. 3652–3657
21. D. Croft, D. McAllister, S. Devasia, High-speed scanning of piezo-probes for nano-
fabrication. ASME J. Manuf. Sci. Eng. 120(3), 617–622 (1998)
22. D. Croft, S. Devasia, Vibration compensation for high speed scanning tunneling microscopy.
Rev. Sci. Instrum. 70(12), 4600–4605 (1999)
23. S. Stilson, A. McClellan, S. Devasia, High-speed solution switching using piezo-based micro-
positioning stages. in Proceedings of the American Control Conference (2001),pp. 2238–2243
7 Tracking Control for Nanopositioning Systems 241

24. S. Devasia, Should model-based inverse inputs be used as feedforward under plant uncer-
tainty? IEEE Trans. Autom. Control 47(11), 1865–1871 (2002)
25. K.K. Leang, S. Devasia, Feedback-linearized inverse feedforward for creep, hysteresis, and
vibration compensation in AFM piezoactuators. IEEE Trans. Control Syst. Technol. 15(5),
927–935 (2007)
26. T.R. Hicks, P.D. Atherton, The Nanopositioning Book (Queensgate Instruments Ltd., Torquay,
1997)
27. Y. Yong, S.O.R. Moheimani, B.J. Kenton, K.K. Leang, Invited review: High-speed flexure-
guided nanopositioning: mechanical design and control issues. Rev. Sci. Instrum. 83(12),
121101 (2012)
28. S. Tien, Q. Zou, S. Devasia, Iterative control of dynamics-coupling-caused errors in
piezoscanners during high-speed AFM operation. IEEE Trans. Control Syst. Technol. 13(6),
921–931 (2005)
29. A.E. Holman, P.M.L.O. Scholte, W.C. Heerens, F. Tuinstra, Analysis of piezo actuators in
translation construction. Rev. Sci. Instrum. 66(5), 3208–3215 (1995)
30. D. Croft, G. Shed, S. Devasia, Creep, hysteresis, and vibration compensation for piezoactua-
tors: atomic force microscopy application. ASME J. Dyn. Syst. Meas. Control. 123(1), 35–43
(2001)
31. T. Sulchek, R. Hsieh, J.D. Adams, S.C. Minne, C.F. Quate, High-speed atomic force
microscopy in liquid. Rev. Sci. Instrum. 71(5), 2097–2099 (2000)
32. G. Schitter, K.J. Astrom, B.E. DeMartini, P.J. Thurner, K.L. Turner, P.K. Hansma, Design and
modeling of a high-speed AFM-scanner. IEEE Trans. Control Syst. Technol. 15(5), 906–915
(2007)
33. K.K. Leang, A.J. Fleming, High-speed serial-kinematic AFM scanner: design and drive
considerations, in American Control Conference, Invited Session on Modeling and Control of
SPM (2008), pp. 3188–3193
34. T. Ando, N. Kodera, D. Maruyama, E. Takai, K. Saito, A. Toda, A high-speed atomic force
microscope for studying biological macromolecules in action. Jpn. J. Appl. Phys., Part 1
41(7B), 4851–4856 (2002)
35. D.C. Jiles, D.L. Atherton, Theory of ferromagnetic hysteresis. J. Magn. Magn. Mater. 61,
48–60 (1986)
36. R. Cross, Unemployment, Hysteresis, and the Natural Rate Hypothesis (Basil Blackwell Ltd.,
New York, 1988)
37. H. Cao, A.G. Evans, Nonlinear deformation of ferroelectric ceramics. J. Am. Ceram. Soc.
76, 890–896 (1993)
38. R.S. Robinson, Interactive computer correction of piezoelectric creep in scanning tunneling
microscopy images. J. Comput.-Assist. Microsc. 2(1), 53–58 (1996)
39. K.K. Leang, S. Devasia, Design of hysteresis-compensating iterative learning control for
piezo positioners: application to atomic force microscopes. Mechatronics 16(3–4), 141–158
(2006)
40. S.M. Hues, C.F. Draper, K.P. Lee, R.J. Colton, Effect of PZT and PMN actuator hysteresis
and creep on nanoindentation measurements using force microscopy. Rev. Sci. Instrum. 65(5),
1561–1565 (1994)
41. K.R. Koops, P.M.L.O. Scholte, W.L. de Koning, Observation of zero creep in piezoelectric
actuators. Appl. Phys. A 68, 691–697 (1999)
42. H. Jung, D.-G. Gweon, Creep characteristics of piezoelectric actuators. Rev. Sci. Instrum.
71(4), 1896–1900 (2000)
43. W.D. Callister, Materials Science and Engineering: An Introduction (Wiley, New York, 1994)
44. G. Schitter, A. Stemmer, Fast closed loop control of piezoelectric transducers. J. Vac. Sci.
Technol. B 20(1), 350–352 (2002)
45. H. Jung, J.Y. Shim, D Gweon, New open-loop actuating method of piezoelectric actuators for
removing hysteresis and creep. Rev. Sci. Instrum. 71(9), 3436–3440 (2000)
46. P. Krejci, K. Kuhnen, Inverse control of systems with hysteresi and creep. IEE Proc. Control
Theory Appl. 148(3), 185–192 (2001)
242 K.K. Leang and A.J. Fleming

47. X. Tan, J.S. Baras, Adaptive identification and control of hysteresis in smart materials. IEEE
Trans. Autom. Control 50(6), 827–839 (2005)
48. L.E. Malvern, Introduction to the Mechanics of a Continuous Medium (Prentice-Hall,
Englewood Cliffs, 1969)
49. M. Hu, H. Du, S.-F. Ling, Z. Zhou, Y. Li, Motion control of an electrostrictive actuator.
Mechatronics 14(2), 153–161 (2004)
50. M. Goldfarb, N. Celanovic, Modeling piezoelectric stack actuators for control of microma-
nipulation. IEEE Control. Syst. Mag. 17(3), 69–79 (1997)
51. B.D. Coleman, M.L. Hodgdon, A constitutive relation for rate-independent hysteresis in
ferromagnetically soft materials. Int. J. Eng. Sci. 24(6), 897–919 (1986)
52. F. Preisach, Uber die magnetische nachwirkung. Z. Phys. 94, 277–302 (1935)
53. P. Ge, M. Jouaneh, Modeling hysteresis in piezoceramic actuators. Precis. Eng. 17(3), 211–
221 (1995)
54. I.D. Mayergoyz, Mathematical Models of Hysteresis (Springer, New York, 1991)
55. H.T. Banks, A.J. Kurdila, G. Webb, Identification of hysteretic confluence operators
representing smart actuators: convergent approximations. Technical Report, North Carolina
State University CRSC, April 1997
56. M.A. Janaideh, C.-Y. Su, S. Rakheja, Development of the rate-dependent Prandtl-Ishlinskii
model for smart actuators. Smart Mater. Struct. 17, 035026 (11pp) (2008)
57. K. Kuhnen, Modeling, identification and compensation of complex hysteretic nonlinearities:
a modified prandtl-ishlinskii approach. Eur. J. Control. 9(4), 407–418 (2003)
58. R.J.E. Merry, N.C.T. de Kleijn, M.J.G. van de Molengraft, M. Steinbuch, Using a walking
piezo actuator to drive and control a high precision stage. IEEE/ASME Trans. Mechatro.
14(1), 21–31 (2009)
59. A.J. Fleming, Nanopositioning system with force feedback for high-performance tracking
and vibration control. IEEE Trans. Mechatro. 15(3), 433–447 (2010)
60. D.Y. Abramovitch, S. Hoen, R. Workman, Semi-automatic tuning of PID gains for atomic
force microscopes, in American Control Conference (2008), pp. 2684–2689
61. H.Y-F. Lam, Analog and Digital Filters: Design and Realization (Prentice-Hall, New York,
1979)
62. A. Sebastian, S.M. Salapaka, Design methodologies for robust nanopositioning. IEEE Trans.
Control Syst. Technol. 13(6), 868–876 (2005)
63. A. Stemmer, G. Schitter, J.M. Rieber, F. Allgöwer, Control strategies towards faster
quantitative imaging in atomic force microscopy. Eur. J. Control. 11(4–5), 384–395 (2005)
64. A. Daniele, S. Salapaka, M.V. Salapaka, M. Dahleh, Piezoelectric scanners for atomic
force microscopes: design of lateral sensors, identification and control, in Proceeding of
the American Control Conference (1999), pp. 253–257
65. B. Bhikkaji, M. Ratnam, A.J. Fleming, S.O.R. Moheimani, High-performance control of
piezoelectric tube scanners. IEEE Trans. Control Sys. Technol. 15(5), 853–866 (2007)
66. B. Bhikkaji, M. Ratnam, S.O.R. Moheimani, PVPF control of piezoelectric tube scanners.
Sensors Actuators A Phys. 135(2), 700–712 (2007)
67. I.A. Mahmood, S.O.R. Moheimani, Making a commercial AFM more accurate and faster
using positive position feedback control. Rev. Sci. Instrum. 80(6), 063705 (8 pages) (2009)
68. S.S. Aphale, A.J. Fleming, S.O.R. Moheimani, Integral resonant control of collocated smart
structures. Smart Mater. Struct. 16, 439–446 (2007)
69. A.J. Fleming, S.S. Aphale, S.O.R. Moheimani, A new method for robust damping and
tracking control of scanning probe microscope positioning stages. IEEE Trans. Nanotechnol.
9(4), 438–448 (2010)
70. N. Kodera, H. Yamashita, T. Ando, Active damping of the scanner for high-speed atomic
force microscopy. Rev. Sci. Instrum. 76(5), 05378–1–053708–5 (2005)
71. A.J. Fleming, A.G. Wills, S.O.R. Moheimani, Sensor fusion for improved control of
piezoelectric tube scanners. IEEE Trans. Control Sys. Technol. 15(6), 1265–6536 (2008)
72. A.J. Fleming, S.O.R. Moheimani, Sensorless vibration suppression and scan compensation
for piezoelectric tube nanopositioners. IEEE Control Sys. Technol. 14(1), 33–44 (2006)
7 Tracking Control for Nanopositioning Systems 243

73. S.S. Aphale, A.J. Fleming, S.O.R. Moheimani, High speed nano-scale positioning using a
piezoelectric tube actuator with active shunt control. Micro Nano Lett. 2(1), 9–12 (2007)
74. U. Aridogan, Y. Shan, K.K. Leang, Design and analysis of discrete-time repetitive control
for scanning probe microscopes. ASME J. Dyn. Syst. Meas. Control. 131, 061103 (12 pages)
(2009)
75. R.J.E. Merry, M.J.C. Ronde, R. van de Molengraft, K.R. Koops, M. Steinbuch, Directional
repetitive control of a metrological AFM. IEEE Trans. Control Sys. Technol. 19(6),
1622–1629 (2011)
76. B.J. Kenton, K.K. Leang, Flexure design using metal matrix composite materials: nanopo-
sitioning example, in IEEE International Conference on Robotics and Automation (ICRA)
(2012)
77. K.K. Chew, M. Tomizuka, Digital control of repetitive errors in disk drive systems. IEEE
Control Syst. Mag. 10(1), 16–20 (1990)
78. M. Steinbuch, S. Weiland, T. Singh, Design of noise and period-time robust high-order
repetitive control, with application to optical storage. Automatica 43(12), 2086–2095 (2007)
79. Y. Shan, K.K. Leang, Dual-stage repetitive control with Prandtl-Ishlinskii hysteresis inversion
for piezo-based nanopositioning. Mechatronics 22, 271–281 (2012)
80. Y. Shan, K.K. Leang, Accounting for hysteresis in repetitive control design: nanopositioning
example. Automatica 48(8), 1751–1758 (2012)
81. K.K. Leang, Q. Zou, S. Devasia, Feedforward control of piezoactuators in atomic force
microscope systems: inversion-based compensation for dynamics and hysteresis. IEEE
Control Syst. Mag. 29(1), 70–82 (2009, Special issue on Hysteresis)
82. E. Bayo, A finite-element approach to control the end-point motion of a single-link flexible
robot. J. Robot. Syst. 4, 63–75 (1987)
83. S. Devasia, D. Chen, B. Paden, Nonlinear inversion-based output tracking. IEEE Trans.
Autom. Control 41(7), 930–942 (1996)
84. Q. Zou, S. Devasia, Preview-based stable-inversion for output tracking. ASME J. Dyn. Syst.
Meas. Control 121(4), 625–630 (1999)
85. J.S. Dewey, K.K. Leang, S. Devasia, Experimental and theoretical results in output-trajectory
redesign for flexible structures. ASME J. Dyn. Syst., Meas. Control 120(4), 456–461 (1998)
86. Q. Zou, S. Devasia, Preview-based optimal inversion for output tracking: application to
scanning tunneling microscopy. IEEE Control Sys. Technol. 12(3), 375–386 (2004)
87. A.J. Fleming, A.G. Wills Optimal periodic trajectories for band-limited systems. IEEE
Control Sys. Technol. 17(3), 552–562 (2009)
88. G. Schitter, P.J. Thurner, P.K. Hansma, Design and input-shaping control of a novel scanner
for high-speed atomic force microscopy. Mechatronics 18(5–6), 282–288 (2008)
89. R.J.E. Merry, R. van de Molengraft, M. Steinbuch, Modeling of a walking piezo actuator.
Sensors Actuators A Phys. 162, 51–60 (2010)
90. Y. Zhao, S. Jayasuriya, Feedforward controllers and tracking accuracy in the presence of plant
uncertainty. ASME J. Dyn. Syst. Meas. Control 117, 490–495 (1995)
91. G. Schitter, A. Stemmer, F. Allgöwer, Robust 2DOF-control of a piezoelectric tube
scanner for high-speed atomic force microscopy, in American Control Conference (2003),
pp. 3720–3725
92. Y. Li, J. Bechhoefer, Feedforward control of closed-loop piezoelectric translation stage for
atomic force microscope. Rev. Sci. Instrum. 78(1), 013702 (2007)
93. S.S. Aphale, S. Devasia, S.O.R. Moheimani, High-bandwidth control of a piezoelectric
nanopositioning stage in the presence of plant uncertainties. Nanotechnology 19(12), 125503
(2008)
94. Y. Wu, Q. Zou, Robust inversion-based 2-DOF control design for output tracking:
piezoelectric-actuator example. IEEE Control Sys. Technol. 17(5), 1069–1082 (2009)
95. Y. Wu, Q. Zou, Iterative control approach to compensate for both the hysteresis and the
dynamics effects of piezo actuators. IEEE Control Sys. Technol. 15(5), 936–944 (2007)
96. D.A. Bristow, J. Dong, A.G. Alleyne, P. Ferriera, S. Salapaka, High bandwidth control of
precision motion instrumentation. Rev. Sci. Instrum. 79, 103704 (2008)
244 K.K. Leang and A.J. Fleming

97. J. Ghosh, B. Paden, A pseudoinverse-based iterative learning control. IEEE Trans. Autom.
Control 47(5), 831–837 (2002)
98. Y. Li, J. Bechhoefer, Feedforward control of a piezoelectric flexure stage for AFM, in
American Control Conference (2008), pp. 2703–2709
99. G.M. Clayton, S. Devasia, Image-based compensation of dynamic effects in scanning
tunneling microscopes. Nanotechnology 16, 809–818 (2005)
100. K.S. Kim, Q. Zou, Model-less inversion-based iterative control for output tracking: piezo
actuator example. in American Control Conference (2008), pp. 2710–2715
101. M. Brokate, J. Sprekels, Hysteresis and Phase Transitions (Springer, New York, 1996)
102. A. Pantazi, A. Sebastian, H. Pozidis, E. Eleftheriou, Two-sensor-based H1 control for
nanopositioning in probe storage, in Proceedings of the 44th IEEE Conference on Decision
and Control, and the European Control Conference (2005), pp. 1174–1179
103. J. Dong, S.M. Salapaka, P.M. Ferreira, Robust mimo control of a parallel kinematics
nano-positioner for high resolution high bandwidth tracking and repetitive tasks, in IEEE
Conference on Decision and Control (2007), pp. 4495–4500
104. Y. Wu, J. Shi, C. Su, Q. Zou, A control approach to cross-coupling compensation of
piezotube scanners in tapping-mode atomic force microscope imaging. Rev. Sci. Instrum.
80(4), 0433709 (2009)
105. S. Bashash, N. Jalili, Robust adaptive control of coupled parallel piezo-flexural nanoposition-
ing stages. IEEE/ASME Trans. Mechatron. 14(1), 11–20 (2009)
106. R.H. Comstock, Charge control of piezoelectric actuators to reduce hysteresis effects. US
Patent 4,263,527, 1979
107. H. Kaizuka, B. Siu, A simple way to reduce hysteresis and creep when using piezoelectric
actuators. Jpn. J. Appl. Phys. 27(5), 773–776 (1988)
108. J.A. Main, E. Garcia, D.V. Netwon, Precision position control of piezoelectric actuators using
charge feedback. AIAA J. Guid. Control. Dyn. 18(5), 1068–1073 (1995)
109. K. Furutani, M. Urushibata, N. Mohri, Displacement control of piezoelectric element by
feedback of induced charge. Nanotechnology 9, 93–98 (1998)
110. A.J. Fleming, S.O.R. Moheimani, Precision current and charge amplifiers for driving highly
capacitive piezoelectric loads. IEE Electron. Lett. 39(3), 282–284 (2003)
111. A.J. Fleming, S.O.R. Moheimani, A grounded-load charge amplifier for reducing hysteresis
in piezoelectric tube scanners. Rev. Sci. Instrum. 76, 073707–1 (2005)
112. A.J. Fleming, K.K. Leang, Evaluation of charge drives for scanning probe microscope
positioning stages, in American Control Conference, Invited session on Advanced Mechanism
Design, Modeling, and Control of SPMs (2008), pp. 2028–2033
113. G.M. Clayton, T. Tien, A.J. Fleming, S.O.R. Moheimani, S. Devasia, Inverse-feedforward of
charge controlled piezopositioners. Mechatronics 18, 273–281 (2008)
114. B. Mokaberi, A.A.G. Requicha, Drift compensation for automatic nanomanipulation with
scanning probe microscopes. IEEE Trans. Autom. Sci. Eng. 3(3), 199–207 (2004)
115. Q. Yang, S. Jaqannathan, E.W. Bohannan, Atomic force microscope-based nanomanipulation
with drift compensation. Int. J. Nanotechnol. 3(4), 527–544 (2006)
116. V.Y. Yurov, A.N. Klimov, Scanning tunneling microscope calibration and reconstruction of
real image: drift and slope elimination. Rev. Sci. Instrum. 65(5), 1551–1557 (1994)
117. R. Staub, D. Alliata, C.Nicolini, Drift elimination in the calibration of scanning probe
microscopes. Rev. Sci. Instrum. 66(3), 2513–2516 (1995)
118. J.T. Woodward, D.K. Schwartz, Removing drift from scanning probe microscope images of
periodic samples. J. Vac. Sci. Technol. B 16(1), 51–53 (1998)
119. S.H. Huerth, H.D. Hallen, Quantitative method of image analysis when drift is present in a
scanning probe microscope. J. Vac. Sci. Technol. B 21(2), 714–718 (2000)
Chapter 8
Position Sensors for Nanopositioning

Andrew J. Fleming and Kam K. Leang

Abstract Position sensors with nanometer resolution are a key component of many
precision imaging and fabrication machines. Since the sensor characteristics can
define the linearity, resolution, and speed of the machine, the sensor performance
is a foremost consideration. The first goal of this article is to define concise
performance metrics and to provide exact and approximate expressions for error
sources including nonlinearity, drift, and noise. The second goal is to review current
position sensor technologies and to compare their performance. The sensors con-
sidered include: resistive, piezoelectric, and piezoresistive strain sensors; capacitive
sensors; electrothermal sensors; eddy-current sensors; linear variable displacement
transformers; interferometers; and linear encoders.

8.1 Introduction

The sensor requirements of a nanopositioning system are among the most demand-
ing of any control system. The sensors must be compact, high-speed, immune to
environmental variation, and able to resolve position down to the atomic scale.
In many applications, such as atomic force microscopy [1, 2] or nanofabrication
[3, 4], the performance of the machine or process is primarily dependent on
the performance of the position sensor, thus, sensor optimization is a foremost
consideration.
In order to define the performance of a position sensor, it is necessary to
have strict definitions for the characteristics of interest. At present, terms such as
accuracy, precision, nonlinearity, and resolution are defined loosely and often vary
between manufacturers and researchers. The lack of a universal standard makes it

A.J. Fleming ()

University of Newcastle, Callaghan, NSW, Australia
e-mail: [email protected]
K.K. Leang
University of Utah, Salt Lake City, UT, USA
e-mail: [email protected]

© Springer International Publishing Switzerland 2016 245

C. Ru et al. (eds.), Nanopositioning Technologies,
DOI 10.1007/978-3-319-23853-1_8
246 A.J. Fleming and K.K. Leang

difficult to predict the performance of a particular sensor from a set of specifications.

Furthermore, specifications may not be in a form that permits the prediction of
closed-loop performance.
This article provides concise definitions for the linearity, drift, bandwidth, and
resolution of position sensors. The measurement errors resulting from each source
are then quantified and bounded to permit a straightforward comparison between
sensors. An emphasis is placed on specifications that allow the prediction of closed-
loop performance as a function of the controller bandwidth.
Although there are presently no international standards for the measurement or
reporting of position sensor performance, this article is aligned with the definitions
and methods reported in the ISO/IEC 98:1993 Guide to the Expression of Uncer-
tainty in Measurement [5], and the ISO 5723 Standard on Accuracy (Trueness and
Precision) of Measurement Methods and Results [6].
The noise and resolution of a position sensor is potentially one of the most
misreported sensor characteristics. The resolution is commonly reported without
mention of the bandwidth or statistical definition and thus has little practical
value. To improve the understanding of this issue, the relevant theory of stochastic
processes is reviewed in Sect. 8.2. The variance is then utilized to define a concise
statistical description of the resolution, which is a straightforward function of the
noise density, bandwidth, and 1=f corner frequency.
The second goal of this article is to provide a tutorial introduction and compari-
son of sensor technologies suitable for nanopositioning applications. To be eligible
for inclusion, a sensor must be capable of a 6 -resolution better than 10 nm with
a bandwidth greater than 10 Hz. The sensor can not introduce friction or contact
forces between the reference and moving target, or exhibit hysteresis or other
characteristics that limit repeatability.
The simplest sensor considered is the metal foil strain gauge discussed in
Sect. 8.3.1. These devices are often used for closed-loop control of piezoelectric
actuators but are limited by temperature dependence and low sensitivity [7].
Piezoresistive and piezoelectric strain sensors provide improved sensitivity but at
the cost of stability and DC performance.
The most commonly used sensors in nanopositioning systems [8] are the capac-
itive and eddy-current sensors discussed in Sects. 8.3.4 and 8.3.6. Capacitive and
eddy-current sensors are more complex than strain sensors but can be designed with
sub-nanometer resolution, albeit with comparably small range and low bandwidth.
They are used extensively in applications such as atomic force microscopy [2, 9–11]
and nanofabrication [4, 12]. The linear variable displacement transformer (LVDT)
described in Sect. 8.3.7 is a similar technology that is intrinsically linear. However,
this type of sensor is larger than a capacitive sensor and due to the larger range, is
not as sensitive.
To achieve high absolute accuracy over a large range, the reference standard is the
laser heterodyne interferometer discussed in Sect. 8.3.8. Although bulky and costly,
the interferometer has been the sensor of choice for applications such as IC wafer
steppers [13, 14] and metrological systems [15]. New fiber interferometers are also
discussed that are extremely compact and ideal for extreme environments.
8 Position Sensors for Nanopositioning 247

Aside from the cost and size, the foremost difficulties associated with an interfer-
ometer are the susceptibility to beam interference, variation in the optical medium,
and alignment error. Since an interferometer is an incremental position sensor, if the
beam is broken or the maximum traversing speed is exceeded, the system must be
returned to a known reference before continuing. These difficulties are somewhat
alleviated by the absolute position encoders described in Sect. 8.3.9. A position
encoder has a read-head that is sensitive to a geometric pattern encoded on a
reference scale. Reference scales operating on the principle of optical interference
can have periods of 128 nm and a resolution of a few nanometers.
Other sensor technologies that were considered but did not fully satisfy the
eligibility criteria include optical triangulation sensors [16], Hall effect sensors, and
magnetoresistive sensors. In general, optical triangulation sensors are available in
ranges from 0.5 mm to 1 m with a maximum resolution of approximately 100 nm.
Hall effect sensors are sensitive to magnetic field strength and hence the distance
from a known magnetic source. These sensors have a high resolution, large range,
and wide bandwidth but are sensitive to external magnetic fields and exhibit hys-
teresis of up to 0.5 % which degrades the repeatability. The magnetoresistive sensor
is similar except that the resistance, rather than the induced voltage, is sensitive
to magnetic field. Although typical anisotropic magnetoresistive (AMR) sensors
offer similar characteristics to the Hall effect sensor, recent advances stimulated
by the hard disk industry have provided major improvements [17]. In particular, the
giant magnetoresistive effect (GMR) can exhibit two orders of magnitude greater
sensitivity than the AMR effect which equates to a resistance change of up to
70 % at saturation. Such devices can also be miniaturized and are compatible with
lithographic processes. Packaged GMR sensors in a full-bridge configuration are
now available from NVE Corporation, NXP Semiconductor, Siemens, and Sony.
Aside from the inherent nonlinearities associated with the magnetic field, the major
remaining drawback is the hysteresis of up to 4 % which can severely impact the
performance in nanopositioning applications. Despite this, miniature GMR sensors
have shown promise in nanopositioning applications by keeping the changes in
magnetic field small [18, 19]. However, to date, the linearity and hysteresis of this
approach has not been reported.

8.2 Sensor Characteristics

8.2.1 Calibration and Nonlinearity

Position sensors are designed to produce an output that is directly proportional to

the measured position. However, in reality, all position sensors have an unknown
offset, sensitivity, and nonlinearity. These effects must be measured and accounted
for in order to minimize the uncertainty in position.
248 A.J. Fleming and K.K. Leang

Fig. 8.1 The actual position x

versus the output voltage of a
position sensor. The 50
calibration function fcal .v/ is
Actual Output

Actual Position ( μ m)
an approximation of the
25 x = fa (v)
sensor mapping function
fa .v/ where v is the voltage
resulting from a displacement 0
x. em .v/ is the residual error
−25 Calibration Curve
x = fcal (v)
−50
v
−10 5 0 10 10

Output Voltage (V)

5
Error em(v)

0 v
−10 5 0 10 10
−5
Output Voltage (V)

The typical output voltage curve for a capacitive position sensor is illustrated
in Fig. 8.1. A nonlinear function fa .v/ maps the output voltage v to the actual
position x. The calibration process involves finding a curve fcal .v/ that minimizes
the mean-square error, known as the least-squares fit, defined by

X
N
  D arg min Œxi  fcal .; vi /2 ; (8.1)
iD1

where vi and xi are the data points and   is the vector of optimal parameters for
fcal .; v/. The simplest calibration curve, as shown in Fig. 8.1, is a straight line of
best fit,

fcal .v/ D 0 C 1 v: (8.2)

In the above equation, the sensor offset is 0 and the sensitivity is 1 m/V. More
complex mapping functions are also commonly used, including the higher order
polynomials

fcal .v/ D 0 C 1 v C 2 v 2 C 3 v 3    : (8.3)

Once the calibration function fcal .v/ is determined, the actual position can be
estimated from the measured sensor voltage. Since the calibration function does not
perfectly describe the actual mapping function fa .v/, a mapping error results. The
mapping error em .v/ is the residual of (8.1), that is
8 Position Sensors for Nanopositioning 249

em .v/ D fa .v/  fcal .  ; v/: (8.4)

If em .v/ is positive, the true position is greater than the estimated value and vice
versa. Although the mapping error has previously been defined as the peak-to-peak
variation of em .v/ [20], this may underestimate the positioning error if em .v/ is not
symmetric. A more conservative definition of the mapping error (em ) is

em D ˙ max jem .v/j : (8.5)

It is also possible to specify an unsymmetrical mapping error such as C max em .v/,

 min em .v/ however this is more complicated. For the sake of comparison, the
maximum mapping error (nonlinearity) is often quoted as a percentage of the full-
scale range (FSR), for example

max jem .v/j

Mapping Error.%/ D ˙100 : (8.6)
FSR
Since there is no exact consensus on the reporting of nonlinearity, it is important
to know how the mapping error is defined when evaluating the specifications
of a position sensor. A less conservative definition than that stated above may
exaggerate the accuracy of a sensor and lead to unexplainable position errors. It may
also be necessary to consider other types of nonlinearity such as hysteresis [21].
However, sensors that exhibit hysteresis have poor repeatability and are generally
not considered for precision sensing applications.

8.2.2 Drift and Stability

In addition to the nonlinearity error discussed above, the accuracy of a positioning

sensor can also be severely affected by changes in the mapping function fa .v/.
The parameters of fa .v/ may drift over time, or be dependent on environmental
conditions such as temperature, humidity, dust, or gas composition. Although, the
actual parametric changes in fa .v/ can be complicated, it is possible to bound the
variations by an uncertainty in the sensitivity and offset. That is,

fa .v/ D .1 C ks /fa .v/ C ko ; (8.7)

where ks is the sensitivity variation usually expressed as a percentage, ko is the offset

variation, and fa .v/ is the nominal mapping function at the time of calibration. With
the inclusion of sensitivity variation and offset drift, the mapping error is

ed .v/ D .1 C ks /fa .v/ C ko  fcal .v/: (8.8)

250 A.J. Fleming and K.K. Leang

Fig. 8.2 The worst-case x

range of a linear mapping
function fa .v/ for a given 50
error in sensitivity and offset.
Actual Output

Actual Position ( μ m)
In this example the greatest
25 x = fa (v)
error occurs at the maximum
and minimum of the range
0

−25 Calibration Curve

x = fcal (v)
−50
v
−10 −5 0 5 10

Output Voltage (V)

10
Error ed(v)

0 v
−10 −5 0 5 10
−10
Output Voltage (V)

Equations (8.7) and (8.8) are illustrated graphically in Fig. 8.2. If the nominal
mapping error is assumed to be small, the expression for error can be simplified to

ed .v/ D ks fcal .v/ C ko : (8.9)

That is, the maximum error due to drift is

ed D ˙ .ks max jfcal .v/j C ko / : (8.10)

Alternatively, if the nominal calibration can not be neglected or if the shape of the
mapping function actually varies with time, the maximum error due to drift must be
evaluated by finding the worst-case mapping error defined in (8.5).

8.2.3 Bandwidth

The bandwidth of a position sensor is the frequency at which the magnitude of the
transfer function v.s/=x.s/ drops by 3 dB. Although the bandwidth specification
is useful for predicting the resolution of a sensor, it reveals very little about the
measurement errors caused by sensor dynamics. For example, a sensor phase-lag of
only 12 degrees causes a measurement error of 10 % FSR.
8 Position Sensors for Nanopositioning 251

If the sensitivity and offset have been accounted for, the frequency domain
position error is

ebw .s/ D x.s/  v.s/; (8.11)

which is equal to

ebw .s/ D x.s/ .1  P.s// ; (8.12)

where P.s/ is the sensor transfer function and .1  P.s// is the multiplicative error.
If the actual position is a sine wave of peak amplitude A, the maximum error is

ebw D ˙A j1  P.s/j : (8.13)

The worst-case error occurs when A D FSR/2, in this case,

FSR
ebw D ˙ j1  P.s/j : (8.14)
2
The error resulting from a Butterworth response is plotted against normalized
frequency in Fig. 8.3. Counter to intuition, the higher order filters produce more
error, which is surprising because these filters have faster roll-off, however, they
also contribute more phase-lag. If the poles of the filter are assumed to be equal to
the cut-off frequency, the low-frequency magnitude of j1  P.s/j is approximately

f
j1  P.s/j  n ; (8.15)
fc

Fig. 8.3 The magnitude of 1

10
error caused by the sensor
dynamics P.s/. The
0
frequency axis is normalized 10
to the sensor 3 dB bandwidth.
Lower order sensor dynamics
−1
result in lower error but 10
|1 − P(s)|

n=8
typically result in
significantly lesser 4
−2 2
bandwidths. In this example 10
the dynamics are assumed to n=1
be nth order Butterworth
−3
10

−4
10
−4 −3 −2 −1 0 1
10 10 10 10 10 10
Normalized Frequency
252 A.J. Fleming and K.K. Leang

where n is the filter order and fc is the bandwidth. The resulting error is approxi-
mately

f
ebw  ˙A n : (8.16)
fc

That is, the error is proportional to the magnitude of the signal, filter order, and
normalized frequency. This is significant because the sensor bandwidth must be
significantly higher than the operating frequency if dynamic errors are to be avoided.
For example, if an absolute accuracy of 10 nm is required when measuring a signal
with an amplitude of 100 m, the sensor bandwidth must be ten-thousand times
greater than the signal frequency.
In the above derivation, the position signal was assumed to be sinusoidal, for
different trajectories, the maximum error must be found by simulating Eq. (8.12).
Although the RMS error can be found analytically by applying Parseval’s equality,
there is no straightforward method for determining the peak error, aside from
numerical simulation. In general, signals that contain high-frequency components
such as square and triangle waves cause the greatest peak error.

8.2.4 Noise

In addition to the actual position signal, all sensors produce some additive measure-
ment noise. In many types of sensor, the majority of noise arises from the thermal
noise in resistors and the voltage and current noise in conditioning circuit transistors.
As these noise processes can be approximated by Gaussian random processes, the
total measurement noise can also be approximated by a Gaussian random process.
A Gaussian random process produces a signal with normally distributed values
that are correlated between instances of time. We also assume that the noise process
is zero-mean and that the statistical properties do not change with time, that is, the
noise process is stationary. A Gaussian noise process can be described by either the
autocorrelation function or the power spectral density. The autocorrelation function
of a random process X is

RX . / D E ŒX .t/X .t C / ; (8.17)

where E is the expected value operator. The autocorrelation function describes the
correlation between two samples separated in time by . Of special interest is RX .0/
which is the variance of the process. The variance of a signal is the expected value
of the varying part squared. That is,

Var X D E .X  E ŒX /2 : (8.18)

8 Position Sensors for Nanopositioning 253

Another term used to quantify the dispersion of a random process is the standard
deviation  which is the square-root of variance,
p
X D standard deviation of X D Var X : (8.19)

The standard deviation is also the root-mean-square (RMS) value of a zero-mean

random process.
The power spectral density SX .f / of a random process represents the distribution
of power or variance across frequency f . For example, if the random process under
consideration was measured in Volts, the power spectral density would have the
units of V2 /Hz. The power spectral density can be found by either the averaged
periodogram technique, or from the autocorrelation function. The periodogram
technique involves averaging a large number of Fourier transforms of a random
process,
 
1
2E jF fXT .t/gj2 ) SX .f / asT ) 1: (8.20)
T

This approximation becomes more accurate as T becomes larger and more records
are used to compute the expectation. In practice, SX .f / is best measured using
a spectrum or network analyzer, these devices compute the approximation pro-
gressively so that large time records are not required. The power spectral density
can also be computed from the autocorrelation function. The relationship between
the autocorrelation function and power spectral density is known as the Wiener–
Khinchin relations, given by

Z1
SX .f / D 2F fRX . /g D 2 RX . /ej2 f
d; and (8.21)
1

Z1
1 1
RX . / D F 1 fSX .f /g D SX .f /ej2 f
df : (8.22)
2 2
1

If the power spectral density is known, the variance of the generating process can
be found from the area under the curve, that is

Z1
2 2
X D E X .t/ D RX .0/ D SX .f / df : (8.23)
0

Rather than plotting the frequency distribution of power or variance, it is often

convenient to plot the frequency distribution of the standard deviation, which is
referred to as the spectral density. It is related to the standard power spectral density
function by a square-root, that is,
254 A.J. Fleming and K.K. Leang

p
Spectral Density D SX .f /: (8.24)
p p
The units of SX .f / are units= Hz rather than units2 =Hz. The spectral density is
preferred in the electronics literature as the RMS value of a noise process can be
determined directly from the noise density
p and effective bandwidth. For example, if
the noise density is a constant c V= Hz and the process is perfectly band limited p
to fc Hz, the RMS value or standard deviation of the resulting signal is c fc .
To distinguish between ppower spectral density and noise density, A is used for power
spectral density and A is used for noise density. An advantage of the spectral
density is that a gain k applied to a signal u.t/ also scales the spectral density by k.
This differs from the standard power spectral density function that must be scaled
by k2 .
Since the noise in position sensors is primarily due to thermal noise and 1=f
(flicker) noise, the power spectral density can be approximated by

fnc
S.f / D A C A; (8.25)
jf j

where A is the power spectral density and fnc is the noise corner frequency illustrated
in Fig. 8.4. The variance of this process can be found by evaluating Eq. (8.23).
That is,
Z fh
fnc
2 D A C A df : (8.26)
fl jf j

where fl and fh define the bandwidth of interest. Extremely low-frequency noise

components are considered to be drift. In positioning applications, fl is typically
chosen between 0.01 and 0.1 Hz. By solving Eq. (8.26), the variance is

fh
 2 D Afnc ln C A.fh  fl /: (8.27)
fl

Fig. 8.4 A constant power S(f ) (dB)

spectral density that exhibits
1=f noise at low frequencies. 1= f
The dashed lines indicate the
asymptotes

log f
fnc
8 Position Sensors for Nanopositioning 255

If the upper-frequency limit is due to a linear filter and fh >> fl , the variance can be
modified to account for the finite roll-off of the filter, that is

fh
 2 D Afnc ln C Ake fh : (8.28)
fl

where ke is a correction factor that accounts for the finite roll-off. For a first,
second, third, and fourth order response ke is equal to 1.57, 1.11, 1.05, and 1.03,
respectively [22].

8.2.5 Resolution

The random noise of a position sensor causes an uncertainty in the measured

position. If the distance between two measured locations is smaller than the
uncertainty, it is possible to mistake one point for the other. In fabrication and
imaging applications, this can cause manufacturing faults or imaging artifacts. To
avoid these eventualities, it is critical to know the minimum distance between two
adjacent but unique locations.
Since the random noise of a position sensor has a potentially large dispersion, it
is impractically conservative to specify a resolution where adjacent locations never
overlap. Instead, it is preferable to state the probability that the measured value
is within a certain error bound. Consider the plot of three noisy measurements in
Fig. 8.5 where the resolution ıy is shaded in gray. The majority of sample points

y(t) y

y3

y2 δy

y1

t fy

Fig. 8.5 The time-domain recording y.t/ of a position sensor at three discrete positions y1 , y2 , and
y3 . The large shaded area represents the resolution of the sensor and the approximate peak-to-peak
noise of the sensor. The probability density function fy of each signal is shown on the right
256 A.J. Fleming and K.K. Leang

in y2 fall within the bound y2 ˙ ıy =2. However, not all of the samples of y2 lie
within the resolution bound, as illustrated by the overlap of the probability density
functions. To find the maximum measurement error, the resolution is added to other
error sources as described in Sect. 8.2.6.
If the measurement noise is approximately Gaussian distributed, the resolution
can be quantified by the standard deviation  (RMS value) of the noise. The
empirical rule [23] states that there is a 99.7 % probability that a sample of a
Gaussian random process lie within ˙3 . Thus, if we define the resolution as
ı D 6 there is only a 0.3 % probability that a sample lies outside of the specified
range. To be precise, this definition of resolution is referred to as the 6 -resolution.
Beneficially, no statistical measurements are required to obtain the 6 -resolution if
the noise is Gaussian distributed.
In other applications where more or less overlap between points is tolerable,
another definition of resolution may be more appropriate. For example, the 4 -
resolution would result in an overlap 4.5 % of the time, while the 10 -resolution
would almost eliminate the probability of an overlap. Thus, it is not the exact
definition that is important; rather, it is the necessity of quoting the resolution
together with its statistical definition.
Although there is no international standard for the measurement or reporting of
resolution in a positioning system, the ISO 5725 Standard on Accuracy (Trueness
and Precision) of Measurement Methods and Results [6] defines precision as the
standard deviation (RMS Value) of a measurement. Thus, the 6 -resolution is
equivalent to six times the ISO definition for precision.
If the noise is not Gaussian distributed, the resolution can be measured by obtain-
ing the 99.7 percentile bound directly from a time-domain recording. To obtain a
statistically valid estimate of the resolution, the recommended recording length is
100 s with a sampling rate 15 times the sensor bandwidth [24]. An anti-aliasing
filter is required with a cut-off frequency 7.5 times the bandwidth. Since the signal
is likely to have a small amplitude and large offset, an AC coupled preamplifier is
required with a high-pass cut-off of 0.03 Hz or lower [25].
Another important parameter that must be specified when quoting resolution is
the sensor bandwidth. In Eq. (8.28), the variance of a noise process is shown to
be approximately proportional to the bandwidth fh . By combining Eq. (8.28) with
the above definition of resolution,p the 6 -resolution can be found as a function of
the bandwidth fh , noise density A, and 1=f corner frequency fnc ,
s
p fh
6 resolution D 6 A fnc ln C ke fh : (8.29)
fl

From Eq. (8.29), it can be observed that the resolution is approximately proportional
to the square-root of bandwidth when fh >> fnc . It is also clear that the 1=f corner
frequency limits the improvement that can be achieved by reducing the bandwidth.
Note that Eq. (8.29) relies on a noise spectrum of the form (8.25) which may
not adequately represent some sensors. The resolution of sensors with irregular
8 Position Sensors for Nanopositioning 257

Fig. 8.6 The resolution

6 6.0 nm
versus bandwidth of a
60 ppm
position sensor with
p a noise
density of 10 pm/ Hz and a 5
1=f corner frequency of 10 Hz

Resolution (nm)
(fl D 0:01 Hz and ke D 1). At 4
low frequencies, the noise is
dominated by 1=f noise; 3 2.1 nm
however, at high frequencies, 21 ppm
the noise increases by a factor
2 0.83 nm
of 3.16 for every decade of 0.53 nm
0.41 nm 8.3 ppm
bandwidth 4.1 ppm 5.4 ppm
1

0
0 1 2 3 4
10 10 10 10 10
Bandwidth fh (Hz)

spectrum’s can be found by solving (8.23) numerically. Alternatively, the resolution

can evaluated from time-domain data, as discussed above.
The trade-off between resolution and bandwidth can be illustrated by considering
p
a typical position sensor with a range of 100 m, a noise density of 10 pm/ Hz,
and a 1=f corner frequency of 10 Hz. The resolution is plotted against bandwidth
in Fig. 8.6. When the bandwidth is below 100 Hz, the resolution is dominated by
1=f noise. For example, the resolution is only improved by a factor of two when the
bandwidth is reduced by a factor of 100. Above 1 kHz, the resolution is dominated
by the flat part of the power spectral density,
pthus a ten times increase in bandwidth
from 1 to 10 kHz causes an approximately 10 reduction in resolution.
Many types of position sensors have a limited full-scale range (FSR); examples
include strain sensors, capacitive sensors, and inductive sensors. In this class of
sensor, sensors of the same type and construction tend to have an approximately
proportional relationship between the resolution and range. As a result, it is
convenient to consider the ratio of resolution to the full-scale range, or equivalently,
the dynamic range (DNR). This figure can be used to quickly estimate the resolution
from a given range, or conversely, to determine the maximum range given a certain
resolution. A convenient method for reporting this ratio is in parts-per-million
(ppm), that is

6 resolution
DNRppm D 106 : (8.30)
Full scale range

This measure is equivalent to the resolution in nanometers of a sensor with a range

of 1 mm. In Fig. 8.6 the resolution is reported in terms of both absolute distance and
the dynamic range in ppm. The dynamic range can also be stated in decibels,

Full scale range

DNRdb D 20 log10 : (8.31)
6 resolution
258 A.J. Fleming and K.K. Leang

Table 8.1 Summary of the exact and simplified worst-case measurement errors
Error source Exact Simplified bound
Mapping error em fa .v/  fcal .  ; v/ ˙ max jem .v/j
Drift ed .1 C ks /fa .v/ C ko  fcal .v/ ˙ .ks max jfcal .v/j C ko /
F 1 fx.s/.1  P.s//g
Anf
Bandwidth ebw ˙ fc (sine-wave)
p q f
Noise ı NA 6 A fnc ln fhl C ke fh

Due to the strong dependence of resolution and dynamic range on the bandwidth
of interest, it is clear that these parameters cannot be reported without the frequency
limits fl and fh , to do so would be meaningless. Even if the resolution is reported
correctly, it is only relevant for a single operating condition. A better alternative is
to report the noise density and 1=f corner frequency, which allows the resolution
and dynamic range to be calculated for any operating condition. These parameters
are also sufficient to predict the closed-loop noise of a positioning system that
incorporates the sensor [26]. If the sensor noise is not approximately Gaussian or the
spectrum is irregular, the resolution is measured using the process described above
for a range of logarithmically spaced bandwidths.

8.2.6 Combining Errors

The exact and worst-case errors described in Sect. 8.2 are summarized in Table 8.1.
In many circumstances, it is not practical to consider the exact error as this is
dependent on the position. Rather, it is preferable to consider only the simplified
worst-case error. An exception to the use of worst-case error is the drift-error ed . In
this case, it may be unnecessarily conservative to consider the maximum error since
the exact error is easily related to the sensor output by the uncertainty in sensitivity
and offset.
To calculate the worst-case error et , the individual worst-case errors are summed,
that is

et D em C ed C ebw C ı=2; (8.32)

where em , ed , ebw , and ı=2 are the mapping error, the drift error, the error due to finite
bandwidth, and the error due to noise whose maximum is half the resolution ı. The
sum of the mapping and drift error can be referred to as the static trueness error
es which is the maximum error in a static position measurement when the noise
is effectively eliminated by a slow averaging filter. The total error and the static
trueness error are illustrated graphically in Fig. 8.7.
8 Position Sensors for Nanopositioning 259

Fig. 8.7 The total y

uncertainty of a
two-dimensional position average measured
measurement is illustrated by position
the dashed box. The total δ
uncertainty et is due to both
the static trueness error es and
the noise ı

y measurement
et es

actual
position

total uncertainty

x measurement

8.2.7 Metrological Traceability

The error of a position sensor has been evaluated with respect to the true position.
However, in practice, the “true” position is obtained from a reference sensor that
may also be subject to calibration errors, nonlinearity, and drift. If the tolerance of
the calibration instrument is significant, this error must be included when evaluating
the position sensor accuracy. However, such consideration is usually unnecessary
as the tolerance of the calibration instrument is typically negligible compared to
the position sensor being calibrated. To quantify the tolerance of a calibration
instrument, it must be compared to a metrological reference for distance. Once the
tolerance is known, measurements produced by the instrument can then be related
directly to the reference, such measurements are said to be metrologically traceable.
Metrological traceability is defined as “the property of a measurement result
whereby the result can be related to a reference through a documented unbroken
chain of calibrations, each contributing to the measurement uncertainty” [27]. The
reference for a distance measurement is the meter standard, defined by the distance
traveled by light in vacuum over 1/299 792 458 s. Laser interferometers are readily
calibrated to this standard since the laser frequency can be compared to the time
standard which is known to an even higher accuracy than the speed of light.
Metrological traceability has little meaning by itself and must be quoted with
an associated uncertainty to be valid [27]. If a position sensor is calibrated by an
instrument that is metrologically traceable, subsequent measurements made by
the position sensor are also metrologically traceable to within the bounds of the
uncertainty for a specified operating environment [5].
260 A.J. Fleming and K.K. Leang

To obtain metrologically traceable measurements with the least uncertainty, an

instrument should be linked to the reference standard through the least number of
intervening instruments or measurements. All countries have a national organization
that maintains reference standards for the calibration instruments. It should be noted
that these organizations have individual policies for the reporting of traceability
if their name is quoted. For example, to report that a measurement is NIST
Traceable, the policy of the National Institute of Standards and Technology (USA),
must be adhered to. Examples of measurement standards organizations include:
• National Measurement Institute (NIM), Australia
• Bureau International des Poids et Mesures (BIPM), France
• Physikalisch-Technische Bundesanstalt (PTB), Germany
• National Metrology Institute of Japan (NMIJ), Japan
• British Standards Institution (BMI), United Kingdom
• National Institute of Standards and Technology (NIST), USA.

8.3 Nanometer Position Sensors

8.3.1 Resistive Strain Sensors

Due to their simplicity and low-cost, resistive strain gauges are widely used for
position control of piezoelectric actuators. Resistive strain gauges can be integrated
into the actuator or bonded to the actuator surface. An example of a piezoelectric
actuator and resistive strain gauge is pictured in Fig. 8.14a. Other application
examples can be found in references [28–31].
Resistive strain gauges are constructed from a thin layer of conducting foil
laminated between two insulating layers. With a zig-zag conductor pattern, strain
gauges can be designed for high sensitivity in only one direction, for example, elon-
gation. When a strain gauge is elongated, the resistance increases proportionally.
The change in resistance per unit strain is known as the gauge factor GF defined by

R=RG
GF D ; (8.33)

where R is the change in resistance from the nominal value RG for a strain .
As the gauge factor is typically in the order of 1 or 2, the change is resistance is
similar in magnitude to the percentage of strain. For a piezoelectric transducer with
a maximum strain of approximately 0.1 %, the change in resistance is around 0.1 %.
This small variation requires a bridge circuit for accurate measurement.
In Fig. 8.14a, a 10 mm Noliac SCMAP07 piezoelectric actuator is pictured with
a strain gauge bonded to each of the two non-electrode sides. The strain gauges
are Omega SGD-3/350-LY13 gauges, with a nominal resistance of 350 Ohms
and package dimensions of 74 mm. The electrical wiring of the strain gauges
8 Position Sensors for Nanopositioning 261

Fig. 8.8 A two-varying- +5V

element bridge circuit that
contains two fixed resistors
and two strain dependent
resistors. All of the nominal
resistance values are equal. V− V+
A simultaneous change in the 2000 Vs
two-varying-elements
produces a differential
voltage across the bridge

is illustrated in Fig. 8.8. The two-varying-element bridge circuit is completed by

two dummy 350 Ohm wire wound resistors and excited by a 5 Volt DC source.
The differential bridge voltage (V C  V  ) is acquired and amplified by a Vishay
Micro-Measurements 2120B strain gauge amplifier. The developed voltage from a
two-varying-element bridge is
 
Av Vb R
Vs D ; (8.34)
2 RG C R=2

where Av D 2000 is the differential gain and Vb D 5 V is the excitation voltage.

By substituting (8.33) into (8.34) and neglecting the small bridge nonlinearity,1 the
measured voltage is proportional to the strain  and displacement d by

Vs D 12 Av Vb GF (8.35)
1
Vs D A V GFd;
2L v b
(8.36)

where L is the actuator length. With a gauge factor of 1, the position sensitivity of
the amplified strain sensor is predicted to be 0.5 V/m which implies a full-scale
voltage of 5 V from a displacement of 10 m. The actual sensitivity was found to be
0.3633 V/m [31].
The bridge configuration shown in Fig. 8.8 is known as the two-varying-element
bridge. It has twice the sensitivity of a single-element bridge but is also slightly
nonlinear and sensitive to temperature variations between the gauge and bridge
resistances. A detailed review of bridge circuits and their associated instrumentation
can be found in reference [32]. The best configuration is the four-varying-element
differential bridge. This arrangement requires four strain gauges, two of which
experience negative strain and another two that experience positive strain. Since the

1
In a two-varying-element bridge circuit, the nonlinearity due to R=2 in Eq. (8.34) is 0.5 %
nonlinearity per percent of strain [32]. Since the maximum strain of a piezoelectric actuator
is 0.1 %, the maximum nonlinearity is only 0.05 % and can be neglected. If this magnitude of
nonlinearity is not tolerable, compensating circuits are available [32].
262 A.J. Fleming and K.K. Leang

Fig. 8.9 The noise density of 2

10
the strain sensor and
instrumentation. The
spectrum can be

pm/ Hz
approximated by a constant
1
spectral density and 1=f noise

√
10

0
10
0 1 2
10 10 10
f (Hz)

bridge is made entirely from the same elements, the four-varying-element bridge
is insensitive to temperature variation. The bridge nonlinearity is also eliminated.
In applications where regions of positive and negative strain are not available, the
two-varying-element bridge is used.
Compared to other position sensors, strain gauges are compact, low-cost, precise,
and highly stable, particularly in a full-bridge configuration [30, 32]. However, a
major disadvantage is the high measurement noise that arises from the resistive
thermal noise and the low sensitivity. The power spectral density of the resistive
thermal noise is

S.f / D 4kTR V2 =Hz; (8.37)

where k is the Boltzmann constant (1:38  1023 ), T is the room temperature in

Kelvin (300ı ), and R is the resistance of each element in the bridge. In addition to
the thermal noise, the current through the bridge also causes 1=f .
The strain gauge pictured inpFig. 8.14a has a resistance of 350 Ohms, hence
the spectral density is 2.4 nV/ Hz. p Since the sensitivity is 0.3633 V/m, the
predicted spectral density is 13 pm/ Hz. This figure agrees with the experimentally
measured spectral density
p plotted in Fig. 8.9. The sensor exhibits a noise density of
approximately 15 pm/ Hz and a 1=f noise corner frequency of around 5 Hz. This
compares poorly with the noise p density of a typical inductive or capacitive sensor
which is on the order of 1 pm/ Hz for a range of 10 m. Hence, strain gauges
are rarely used in systems designed for high resolution. If they are utilized in such
systems, the closed-loop bandwidth must be severely restrained.
As an example of strain gauge resolution, we consider a typical two-varying-
element strain gauge with an excitation of 5 V and a gauge factor of 1. The full-
scale voltage is predicted to be 2.5 mV for a 0.1 % strain. If we assume a 1=f noise
corner frequency of 5 Hz, fl D 0:01 Hz, and a first-order bandwidth of 1 kHz (ke D
1:57). The resolution predicted by Eq. (8.29) is 580 nV or 230 ppm. In other words,
if the full-scale range was 100 m, the resolution would be 23 nm, which is not
competitive.
8 Position Sensors for Nanopositioning 263

8.3.2 Piezoresistive Strain Sensors

In 1954, a visiting researcher at Bell Laboratories, C. S. Smith, demonstrated that

“exceptionally large” resistance changes occur in silicon and germanium when
subjected to external strain [33]. This discovery was the foundation for today’s
semiconductor piezoresistive sensors that are now ubiquitous in applications such
as integrated pressure sensors and accelerometers [34].
Compared to metal foil strain gauges that respond only to changes in geometry,
piezoresistive sensors exhibit up to two orders-of-magnitude greater sensitivity.
In addition to their high strain sensitivity, piezoresistive sensors are also easily
integrated into standard integrated circuit and MEMS fabrication processes which is
highly advantageous for both size and cost. The foremost disadvantages associated
with piezoresistive sensors are the low strain range (0.1 %), high temperature
sensitivity, poor long-term stability, and slight nonlinearity (1 %) [34]. The elim-
ination of these artifacts requires a more complicated conditioning circuit than
metal foil strain gauges; however, integrated circuits are now available that partially
compensate for nonlinearity, offset, and temperature dependence, for example, the
Maxim MAX1450.
As shown in Fig. 8.10, a typical integrated piezoresistive strain sensor consists of
a planar n-doped resistor with heavily doped contacts. When the sensor is elongated
in the x-axis, the average electron mobility increases in that direction, reducing
resistance [34]. The effect is reverse during compression, or if the resistor is p-type.
Since the piezoresistive effect is due to changes in the crystal lattice, the effect
is highly dependent on the crystal orientation. The change in resistance can be
expressed as,

R D RG L xx C T yy C zz ; (8.38)

where R is the change in resistance; RG is the nominal resistance; xx , yy , and zz
are the tensile stress components in each axis; and L and T are the longitudinal
and transverse piezoresistive coefficients which are determined from the crystal
orientation [34].
Due to the temperature dependence and low strain range, piezoresistive sensors
are primarily used in microfabricated devices where the difficulties are offset by the
high sensitivity and ease of fabrication, for example, meso-scale nanopositioners

Fig. 8.10 A cross-section of metal n+ contacts

a piezoresistive strain sensor.
Deformation of the
semiconductor crystal causes z
a resistance change n-well
one-hundred times that of a
resistive strain gauge x p substrate
y
264 A.J. Fleming and K.K. Leang

[35] and MEMs devices [36]. Discrete piezoresistive sensors are also available
for standard macro-scale nanopositioning applications, for example, Micron Instru-
ments SS-095-060-350PU. Discrete piezoresistive strain sensors are significantly
smaller than metal foil gauges, for example, the Micron Instruments SS-095-060-
350PU is 2.4 mm0.4 mm. The sensitivity is typically specified in the same way as
a metal foil sensor, by the gauge factor defined in Eq. (8.33). While the gauge factor
of a metal foil sensor is between 1 and 2, the gauge factor of the Micron Instruments
SS-095-060-350PU is 120.
Due to the temperature dependence of piezoresistive strain sensors, practical
application requires a closely collocated half- or full-bridge configuration, similar
to a metal foil gauge. The required signal conditioning is also similar to the
metal foil gauges. If an accuracy of better than 1 % is required, or if large
changes in temperature are expected, the piezoresistive elements must be closely
matched and the signal conditioning circuit must be compensated for temperature
and nonlinearity. Two fully integrated bridge conditioning circuits include the
MAX1450 and MAX1452 from Maxim Integrated Products, USA.
Alike metal foil strain gauges, the noise in piezoresistive sensors is predom-
inantly thermal and 1=f noise [34]. However, since piezoresistive sensors are
semiconductors, the 1=f noise can be substantially worse [34]. Consider the Micron
Instruments SS-095-060-350PU piezoresistive sensor which has a gauge factor of
120 and a resistance of 350 . In a two-varying-element bridge with 2-V excitation,
Eq. (8.36) predicts that a full-scale strain
p of 0.1 % develops 120 mV. The thermal
noise due to the resistance is 2.4 nV/ Hz. If the 1=f noise corner frequency is
assumed to be 10 Hz, the resolution with a first-order bandwidth of 1000 Hz is
130 nV which implies a 6 -resolution of 590 nV or 4.9 ppm. Restated, if the full-
scale displacement was 100 m, the resolution would be 0.49 nm.
Although the majority of piezoresistive sensors are integrated directly into
MEMS devices, discrete piezoresistive strain sensors are available from: Kulite
Semiconductor Products Inc., USA; and Micron Instruments, USA.

8.3.3 Piezoelectric Strain Sensors

In addition to their actuating role, piezoelectric transducers are also widely utilized
as high sensitivity strain sensors [37–43].
The basic operation of a piezoelectric strain sensor is illustrated in Fig. 8.11a. In
this case the applied force F and resulting strain h/h is aligned in the same axis
as the polarization vector. The polarization vector points in the same direction as
the internal dipoles which is opposite in direction to the applied electric field. Thus,
compression of the actuator results in a voltage of the same polarity as the voltage
applied during polarization. From the stress-voltage form of the piezoelectric
constituent equations, the developed electric field E is
8 Position Sensors for Nanopositioning 265

Fig. 8.11 A piezoelectric a F

stack and plate strain sensor.
The polarization vector is
shown as a downward arrow.
(a) Axial sensors are typically
used to measure dynamic
forces while (b) flexional
sensors are used to measure Δh
changes in strain or curvature
Vs

Axial sensor
b

F Vs

ΔL

Flexional sensor

h
E D q33 ; (8.39)
h
where h is the change in thickness, h is the thickness, and q33 is the piezoelectric
coupling coefficient for the stress-voltage form. The constant q33 is related to the
piezoelectric strain constant d33 by

d33
q33 D ; (8.40)
 T sD

where  T is the permittivity under constant stress (in Farad/m) and sD is the elastic
compliance under constant electric displacement (in m2 /N). If the piezoelectric
voltage constant g33 is known instead of q33 or d33 , q33 can also be derived from
q33 D g33 =sD . By multiplying (8.40) by the thickness h, the measured voltage can
be written as

Vs D q33 h; (8.41)

If there are multiple layers, the voltage is

q33
Vs D h; (8.42)
n
266 A.J. Fleming and K.K. Leang

where n is the number of layers. The developed voltage can also be related to the
applied force [31].

nd33 d33 h
Vs D F; or Vs D T F; (8.43)
C n A

where C is the transducer capacitance defined by C=n2  T A=h and A is the area.
The voltage developed by the flexional sensor in Fig. 8.11b is similar to the axial
sensor except for the change of piezoelectric constant. In a flexional sensor, the
applied force and resulting strain are perpendicular to the polarization vector. Hence,
the g31 constant is used in place of the g33 constant. Assuming that the length L is
much larger than the width and thickness, the developed voltage is
g31
Vs D F; (8.44)
L
which can be rewritten in terms of the stiffness k and strain,

L
Vs D g31 k (8.45)
L
g31 A L
Vs D D ; (8.46)
s L L
where A is the cross-sectional area equal to width  thickness.
When mounted on a host structure, flexional sensors can be used to detect the
underlying stress or strain as well as the curvature or moment [37, 44, 45]. In
nanopositioning applications, the electrodes of a piezoelectric tube act as a plate
sensor and can be used to detect the strain and hence displacement [39, 40, 42]. This
application is illustrated in Fig. 8.12.
Due to the high mechanical stiffness of piezoelectric sensors, thermal or Boltz-
mann noise is negligible compared to the electrical noise arising from interface
electronics. As piezoelectric sensors have a capacitive source impedance, the
noise density NVs .!/ of the sensor voltage Vs is due primarily to the current
noise in generated by the interface electronics. The equivalent electrical circuit of
a piezoelectric sensor and high-impedance buffer is shown in Fig. 8.13. Neglecting
the leakage resistance R, the noise density of the sensor voltage is

1
NVs .!/ D in ; (8.47)
C!
where NVs and in are the noise densities
p of the sensor voltage and current noise,
measured in Volts and Amps per Hz, respectively.
The experimentally measured and predicted noise density of a piezoelectric
sensor is plotted in Fig. 8.14. The sensor is a 2-mm Noliac CMAP06 stack mounted
on top of 10-mm long actuator, the assembly is mounted in the nanopositioning
stage pictured in Fig. 8.15. The sensor has a capacitance of 30 nF and the voltage
8 Position Sensors for Nanopositioning 267

Cp triax cable

Rp Vs
Vp

Fig. 8.12 A piezoelectric tube actuator with one electrode utilized as a strain sensor. The electrical
equivalent circuit consists of the induced piezoelectric voltage Vp in series with the transducer
capacitance. The dielectric leakage and input impedance of the buffer circuit are modeled by the
parallel resistance Rp . An effective method for shielding the signal is to use a triaxial cable with the
intermediate shield driven at the same potential as the measured voltage. (Tube drawing courtesy
K. K. Leang)

Fig. 8.13 The electrical

model of a piezoelectric force Vs
sensor. The open-circuit
C in
voltage Vp is high-pass
filtered by the transducer
capacitance C and leakage R
resistance R. The current Vp
source in represents the
current noise of a
high-impedance buffer

p
buffer (OPA606) has a noise density of 2 fA/ Hz: Further details on the behavior
of piezoelectric force sensors can be found in [31].
In Fig. 8.14b the noise density of the piezoelectric sensor is observed to be more
than two orders of magnitude less than the strain and inductive sensors at 100 Hz.
The noise density also continues to reduce at higher frequencies. However, at low
frequencies the noise of the piezoelectric force sensor eventually surpass the other
sensors. As the noise density is equivalent to an integrator excited by white noise,
the measured voltage drifts significantly at low frequencies. A time record that
illustrates this behavior is plotted in Fig. 8.16. The large drift amplitude is evident.
Thus, although the piezoelectric force sensor generates less noise than the strain
and inductive sensors at frequencies in the Hz range and above, it is inferior at
frequencies below approximately 0.1 Hz.
In addition to noise, piezoelectric force sensors are also limited by dielectric
leakage and finite buffer impedance at low frequencies. The induced voltage Vp
shown in Fig. 8.13 is high-pass filtered by the internal transducer capacitance C and
the leakage resistance R. The cut-off frequency is
268 A.J. Fleming and K.K. Leang

Fig. 8.14 (a) A piezoelectric a

stack actuator with an
integrated force sensor and
two resistive strain gauges
bonded to the top and bottom
surface (the bottom gauge is
not visible). In (b), the noise
density of the piezoelectric
sensor is compared to the
resistive strain gauge and a
Actuator with integrated force sensor
Kaman SMU9000-15N
and strain gauge
inductive sensor, pall signals
are scaled to nm/ Hz. The b −1
simulated noise of the 10
piezoelectric force sensor is Inductive
also plotted as a dashed line
−2
10 Strain

−3
10
Piezo
−4
10

Simulated
−5
10
0 1 2
10 10 10
f (Hz)
√
Noise power spectral density (nm/ Hz)

1
fhp D Hz: (8.48)
2 RC
The buffer circuit used in the results above has an input impedance of 100 M, this
results in a low-frequency cut-off of 0.05 Hz. To avoid a phase lead of more than
6 degrees, the piezoelectric force sensor cannot be used to measure frequencies of
less than 0.5 Hz.
Piezoelectric actuators and sensors are commercially available from: American
Piezo (APC International, Ltd.), USA; CeramTec GmbH, Germany; Noliac A/S,
Denmark; Physik Instrumente (PI), Germany; Piezo Systems Inc., USA; and Sensor
Technology Ltd., Canada.
8 Position Sensors for Nanopositioning 269

Fig. 8.15 A nanopositioning

platform with a
two-varying-element strain
gauge fitted to the y-axis
actuator [31]. The
nanopositioner is driven by
two piezoelectric stack
actuators that deflect the
sample platform by a
maximum of 10 m in the x
and y lateral axes

Fig. 8.16 Low-frequency

noise of the piezoelectric 20

sensor pictured in Fig. 8.14a,

scaled to nanometers. The 10
peak-to-peak noise over 220 s
d (nm)

is 38 nm or 26 mV 0

−10

−20

0 50 100 150 200

t (s)

8.3.4 Capacitive Sensors

Capacitive sensors are the most commonly used sensors in short-range nanoposi-
tioning applications. They are relatively low-cost and can provide excellent linearity,
resolution, and bandwidth [46]. However, due to the electronics required for
measuring the capacitance and deriving position, capacitive sensors are inherently
more complex than sensors such as resistive strain gauges. Larger ranges can be
achieved with the use of an encoder-style electrode array [47].
All capacitive sensors work on the principle that displacement is proportional
to the change in capacitance between two conducting surfaces. If fringe effects are
neglected, the capacitance C between two parallel surfaces is

 0 r A
CD ; (8.49)
h
where 0 is the permittivity of free space, r is the relative permittivity of the
dielectric (or dielectric constant), A is the area between the surfaces, and h is the
distance between the surfaces.
270 A.J. Fleming and K.K. Leang

d d

Lateral moving plate Moving dielectric Axial moving plate

Fig. 8.17 Types of capacitive sensor. The axial moving plate produces the highest sensitivity but
the smallest practical travel range. Lateral moving plate and moving dielectric sensors are most
useful in long-range applications

Three types of capacitive sensor are illustrated in Fig. 8.17. The lateral moving
plate design is used for long-range measurements where the plate spacing can be
held constant. This is often achieved with two concentric cylinders mounted on
the same axis. In this configuration, the change in capacitance is proportional to
the change in area and hence position. A similar arrangement can be found in the
moving dielectric sensor where the area and distance are constant but the dielectric is
variable. This approach is not commonly used because a solid dielectric is required
that causes friction and mechanical loading.
The axial moving plate, or parallel plate capacitive sensor is the most common
type used in nanopositioning applications. Although the useful range is smaller than
other configurations, the sensitivity is proportionally greater. The capacitance of a
moving plate sensor is

 0 r A
CD ; (8.50)
d
hence, the sensitivity is

dC C0
D F=m; (8.51)
dd d0

where C0 and d0 are the nominal capacitance and distance. Thus, for a sensor with a
nominal capacitance of 10 pF and spacing of 100 m, the sensitivity is 100 fF/m.
The sensitivity of different capacitive sensor types is compared in [20].
A practical parallel plate capacitive sensor is illustrated in Fig. 8.18. In addition
to the probe electrode, a guard electrode is also used to shield the probe from
nearby electric fields and to improve linearity. The guard electrode is driven at the
same potential as the probe but is not included in the capacitance measurement. As
the fringing effect in the electric field is only present at the outside electrode, the
nonlinearity in the capacitance measurement and distance calculation is reduced.
A summary of correction terms for different guard electrode geometries can be
found in references [20] and [46].
8 Position Sensors for Nanopositioning 271

Target
Probe Electrode

Gaurd Electrode

Fig. 8.18 A capacitive sensor probe and electrode configuration. The guard electrode is driven at
the same potential as the probe in order to linearize the electric field and reduce fringing effects

To measure the capacitance and thus derive the position, a wide variety of circuits
are available [21, 46]. The simplest circuits are timing circuits where the timing
capacitor is replaced by the sensor capacitance. Examples include the ubiquitous
555 timer in the one-shot or free-running oscillator modes. The output of a one-
shot circuit is a pulse delay proportional to the capacitance. Likewise, the output
of the oscillator is a square-wave whose frequency is proportional to capacitance.
Although these techniques are not optimal for nanopositioning applications, they
are simple, low-cost, and can be directly connected to a microcontroller with no
analog-to-digital converters.
A direct measurement of the capacitance can be obtained by applying an AC
voltage V to the probe electrode and grounding the target. The resulting current I is
determined by Ohms law,

I D j!VC; (8.52)

where ! is the excitation frequency in rad/s. Since the current is proportional to

capacitance, this method is useful for the lateral moving plate and moving dielectric
configurations where the displacement is also proportional to capacitance. For the
axial moving plate configuration, where the displacement is inversely proportional
to capacitance, it is more convenient to apply a current and measure the voltage. In
this case, the measured voltage in response to an applied current is

I
VD ; (8.53)
j!C

which is inversely proportional to capacitance and thus proportional to displace-

ment.
Regardless of whether the current or voltage is the measured variable, it is
necessary to compute the AC magnitude of the signal. The simplest circuit that
achieves this is the single-diode demodulator or envelope detector shown in
Fig. 8.19a. Although simple, the linearity and offset voltage of this circuit are
dependent on the diode characteristics which are highly influenced by temperature.
272 A.J. Fleming and K.K. Leang

Sync. Demodulator
a b
ref
Cr Vs Cr mag
input
Vs

C C

Fig. 8.19 Demodulation circuits for measuring capacitance. The linearity, temperature sensitivity,
and noise performance of the synchronous detector are significantly better than the single-diode
envelope detector. (a) single-diode demodulator and (b) Synchronous demodulator

A better option is the synchronous demodulator with balanced excitation shown

in Fig. 8.19b. A synchronous demodulator can be constructed from a filter and
voltage controlled switch [21, 46]. Integrated circuit demodulators such as the
Analog Devices AD630 are also available. Synchronous demodulators provide
greatly improved linearity and stability compared to single-diode detectors.
The balanced excitation in Fig. 8.19b eliminates the large DC offset produced
by single ended demodulators, such as Fig. 8.19a. The balanced configuration also
eliminates the offset sensitivity to changes in the supply voltage, which greatly
improves the stability. Although single ended excitation can be improved with a
full-bridge configuration, this requires a high common-mode rejection ratio, which
is difficult to obtain at high frequencies.
In general, capacitive sensors with guard electrodes can provide excellent
linearity in ideal conditions (10 ppm or 0.001 %); however, practical limitations
can significantly degrade this performance. A detailed analysis of capacitive sensor
nonlinearity in [20] concluded that the worst sources of nonlinearity are tilting and
bowing. Tilting is the angle between the two parallel plates and bowing is the depth
of concavity or convexity.
A summary of the error analysis performed in [20] is contained in Table 8.2.
Considering that the linearity of an capacitive sensor in ideal conditions can be
0.001 %, the effect of tilting and bowing severely degrades the performance. These
errors can be reduced by careful attention to the mounting of capacitance sensors.
It is recommended that capacitive sensors be fixed with a spring washer rather than a
screw. This can significantly reduce mounting stress on the host structure and sensor.
In addition to deformation, excessive mounting forces can slowly relieve over time
causing major drifts in offset, linearity, and sensitivity.
The magnitude of error due to tilting and bowing can be reduced by increasing
the nominal separation of the two plates, this also increases the range. However, if
the area of the sensor is not increased, the capacitance drops, which increases noise.
The noise developed by a capacitive sensor is due primarily to the thermal and
shot-noise of the instrumentation electronics. Due to the demodulation process,
the noise spectral density is relatively flat and does not contain a significant
8 Position Sensors for Nanopositioning 273

Table 8.2 A summary of Errors due to tilting

error sources in a parallel
plate capacitive sensor Tilt angle (mrad) 2 5
studied in [20] Nonlinearity (%) 0.08 0.6
Offset (%) 0.35 2.4
Scale error (%) 0.8 5.4
Errors due to bowing
Bow depth (m) 10 30
Nonlinearity (%) 0.025 0.33
Offset (%) 5 18
Scale error (%) 3 11
The sensor has a gap of 100 m,
a radius of 6 mm, and a nominal
capacitance of 10 pF

1=f component. Although the electronic noise remains constant with different
sensor configurations, the effective position noise is proportional to the inverse
of sensitivity. As the sensitivity is C0 =d0 (8.51), if the capacitance is doubled by
increasing the area, the position noise density is reduced by half. However, if the
nominal gap d0 is doubled to improve the linearity, the capacitance also halves,
which reduces the sensitivity and increases the noise density by a factor of four. The
position noise density is minimized by using the smallest possible plate separation
and the largest area.
A typical commercial capacitive sensor
p with a range of 100 m has a noise
density of approximately of 20 pm/ Hz [40]. The 1=f corner frequency of a
capacitive sensor is typically very low, around 10 Hz. With a first-order bandwidth
of 1 kHz, the resolution predicted by Eq. (8.29) is 2.4 nm or 24 ppm. This can be
reduced to 0.55 nm or 5.5 ppm by restricting the bandwidth to 10 Hz.
Capacitive position sensors are commercially available from: Capacitec, USA;
Lion Precision, USA; Micro-Epsilon, Germany; MicroSense, USA; Physik Instru-
mente (PI), Germany; and Queensgate Instruments, UK. Two commercially avail-
able devices are pictured in Fig. 8.20.

8.3.5 MEMs Capacitive and Thermal Sensors

MEMs capacitive sensors operate on a similar principles to their macro-scale

counterpart discussed in the previous section. However, due to their small size, a
more complicated geometry is required to achieve a practical value of capacitance.
The comb-type sensor illustrated in Fig. 8.21a is a common variety found in a
number of nanopositioning applications, for example, [48, 49]. In this configuration,
the total capacitance is approximately proportional to the overlap area of each
electrode array.
274 A.J. Fleming and K.K. Leang

Fig. 8.20 An example of two commercially available capacitive sensors. Photos courtesy of
(a) Queensgate Instruments, UK and (b) Micro-Epsilon, Germany

a b c
1 1 1

2 2 3
2 3

Fig. 8.21 Three examples of MEMs capacitive sensor geometries. (a) Standard comb sensor;
(b) Differential comb sensor; and (c) Incremental capacitive encoder

The basic comb sensor can be improved by employing a differential detection

method as illustrated in Fig. 8.21b. Here, two sets of excitation electrodes (terminals
2 and 3) are driven 180 degrees out of phase. Thus, at the central position, the
potential at terminal 1 is zero. This configuration provides a higher sensitivity than
the basic comb sensor and is used extensively in devices such as accelerometers and
gyroscopes [46, 50].
To increase the range of motion beyond a single inter-electrode spacing, the con-
figuration in Fig. 8.21c uses withdrawn electrodes to form a capacitive incremental
encoder [51–53]. The slider can now move freely in either direction, limited only
by the length of the excitation array. As the slider moves horizontally, the induced
voltage at terminal 1 alternates between the phase of terminals 2 and 3. A second
array is typically used to create a quadrature signal for ascertaining the direction
of travel. This approach can provide a large travel range with high resolution but
the decoding electronics is more complicated and the performance is sensitive to
the separation between the arrays. If the two arrays can be overlain vertically, the
capacitance can be increased while the difficulties with array separation are reduced
[54, 55].
8 Position Sensors for Nanopositioning 275

Vout

R1 R2

Vs

Fig. 8.22 An electrothermal position sensor. The two stationary microheaters are driven by a
constant voltage source Vs. The rate of heat transfer and the resulting temperature is proportional
to the overlap between the heater and the heatsink. The position of the heatsink can be estimated
by measuring the current difference between the two microheaters which indicates the difference
in resistance and temperature

Electrothermal sensors are an alternate class of position sensors first utilized in

nanopositioning applications by IBM in 2005 [56]. An example of a differential
electrothermal position sensor is illustrated in Fig. 8.22. Two microheaters are
driven by a DC voltage source resulting in a temperature increase. Due to the heat
transfer between the microheater and moving heatsink, the temperature of each
microheater becomes a function of the overlap area and hence position. The heatsink
position is estimated by measuring the difference in current which is related to the
resistance and temperature.
An advantage of electrothermal sensors over capacitive sensors is the compact
size which has made them appealing in applications such as data storage [57–59]
and nanopositioning [60, 61]. The noise performance of electrothermal sensors can
be similar or superior to capacitive sensors under certain conditions. However, due
to the elevated temperature, electrothermal sensors are known to exhibit a significant
amplitude of low-frequency noise [49].
With a range of 100 m, a thermal
p position sensing scheme achieved a noise
density of approximately 10 pm/ Hz with a 1/f corner frequency of approximately
3 kHz [60]. This resulted in a resolution of 10 nm over a bandwidth of 4 kHz. As a
result of the low-frequency noise and drift, an auxiliary position sensor was utilized
at frequencies below 24 Hz [60].

8.3.6 Eddy-Current Sensors

Eddy-current, or inductive proximity sensors, operate on the principle of elec-

tromagnetic induction [62, 63]. As illustrated in Fig. 8.23, an eddy-current probe
consists of a coil facing an electrically conductive target. When the coil is excited
by an AC current, the resulting magnetic field passes through the conductive target
and induces a current according to Lenz’s law. The current flows at right angles to
276 A.J. Fleming and K.K. Leang

Fig. 8.23 The operating Conducting Target

principle of an eddy-current
sensor. An alternating current Magnetic Field
in the coil induces eddy
currents in the target.
Increasing the distance Coil
between the probe and target
reduces the eddy currents and
also the effective resistance of
the coil Probe Body
Eddy
Current

a coil b shield c reference coil

field
core

Fig. 8.24 Types of eddy-current sensor. The unshielded type has the greatest range but is affected
by nearby fields and conductors. A shield makes the magnetic field more directional but reduces
the range. A reference coil can be used to reduce the sensitivity to temperature.(a) Unshielded, (b)
Shielded, and (c) Balanced

the applied magnetic field and develops an opposing field. The eddy currents and
opposing field become stronger as the probe approaches the target.
The distance between probe and target is detected by measuring the AC resis-
tance of the excitation coil which depends on the magnitude of the opposing field
and eddy current. The required electronics are similar to that of a capacitive sensor
and include an oscillator and demodulator to derive the resistance [21, 62, 64].
Three common types of eddy-current sensor are depicted in Fig. 8.24. The
unshielded sensor has a large magnetic field that provides the greatest range;
however, it also requires the largest target area and is sensitive to nearby conductors.
Shielded sensors have a core of permeable material such as Permalloy which reduces
the sensitivity to nearby conductors and requires less target area; however, they also
have less range. The balanced type has a second shielded or non-inductive coil that
is used to null the effect of temperature variation [65]. The second coil is used in a
divider or bridge configuration such as that illustrated in Fig. 8.25.
8 Position Sensors for Nanopositioning 277

Fig. 8.25 Synchronous Sync. Demodulator

demodulation circuit for a
balanced eddy-current sensor. ref
Lr mag
Lr and Rr are the inductance Vs
and resistance of the input
Rr
reference coil

Another type of position sensor similar to an eddy-current sensor is the inductive

proximity sensor, also referred to as a differential reluctance transducer if a
reference coil is present. Rather than a conductive target, an inductive proximity
sensor requires a ferromagnetic target. Since the reluctance of the magnetic path
is proportional to the distance between the probe and target, the displacement
can be derived from the coil inductance. Inductive proximity sensors have the
same construction and electronics requirement as an eddy-current sensor. Their
main drawback compared to eddy-current sensors is the temperature dependent
permeability of the target material and the presence of magnetic hysteresis.
Eddy-current sensors are not as widely used as capacitive sensors in nanopo-
sitioning applications due to the temperature sensitivity and range concerns. The
temperature sensitivity arises from the need of an electrical coil in the sensor
head and the varying resistance of the target. The minimum range of an eddy-
current sensor is limited by the minimum physical size of the coil, which imposes
a minimum practical range of between 100 and 500 m. In contrast, capacitive
sensors are available with a range of 10 m which can provide significantly higher
resolution in applications with small travel ranges.
The major advantage of eddy-current and inductive sensors is the insensitivity to
dust and pollutants in the air-gap and on the surface of the sensor. This gives them a
significant advantage over capacitive sensors in industrial applications.
The noise performance of an eddy-current sensor can be similar to that of a
capacitive sensor. For example, the noise density of the Kaman SMU9000-15N
which has a range of 500 m is plotted in Fig. 8.14b. The 1=f corner frequency p
is approximately 20 Hz and the constant density is approximately 20 pm/ Hz.
Equation (8.29) predicts a resolution of 5 nm or 10 ppm with a bandwidth of 1 kHz.
Due to the physical size of the coils, smaller ranges and higher resolution is difficult
to achieve.
Eddy-current position sensors are commercially available with ranges of approxi-
mately 100 m to 80 mm. Manufacturers include: Micro-Epsilon, Germany; Kaman
Sensors, USA; MicroStrain, USA; Keyence, USA; Lion Precision, USA; and Ixthus
Instrumentation, UK. Two commercially available devices are pictured in Fig. 8.26.
278 A.J. Fleming and K.K. Leang

a b

Lion Precision, USA Micro-Epsilon, Germany

Fig. 8.26 Two commercially available eddy-current sensors. (a) Photos courtesy of Lion Preci-
sion, USA and (b) Micro-Epsilon, Germany

Magnetic Field

Core
Push Rod

Coil A
Driving
Coil Coil B

Fig. 8.27 The operating principle of a linear variable displacement transducer (LVDT). Changes
in the core position produce a linear differential change in the coupling between the driving coil
and the pick-up coils

8.3.7 Linear Variable Displacement Transformers

Linear variable displacement transformers (LVDTs) are used extensively for dis-
placement measurement with ranges of 1 mm to over 50 cm. They were originally
described in a patent by G. B. Hoadley in 1940 (US Patent 2,196,809) and became
popular in military and industrial applications due to their ruggedness and high
resolution [21].
The operating principle of an LVDT is illustrated in Fig. 8.27. The stationary part
of the sensor consists of a single driving coil and two sensing coils wound onto a
thermally stable bobbin. The movable component of the transducer is a permeable
material such as Nickel–Iron (Permalloy), and is placed inside the bobbin. The core
is long enough to fully cover the length of at least two coils. Thus, at either extreme,
the central coil always has a complete core at its center.
Since the central coil always has a complete core, all of the magnetic flux is
concentrated in the core. As the core moves, the amount of flux passing through
each sensor coil is proportional to the length of core contained within. Hence, the
8 Position Sensors for Nanopositioning 279

Coil A
Driving
Coil
Vs

Coil B

Fig. 8.28 The relationship between the sensor coil voltage and core position in an LVDT. The coil
voltage is proportional to the amount or core it contains

Fig. 8.29 An LVDT

conditioning circuit with a
synchronous demodulator and
differential amplifier [21]
Coil A
Vs

Coil B

displacement of the core is proportional to the difference in voltage induced in the

sensor coils. This principle is illustrated in Fig. 8.28.
In addition the components in Fig. 8.27, a bearing is required to guide the motion
of the core through the bobbin. An external case is also required that can be
constructed from a permeable material to provide magnetic shielding of the coils.
It is important that the push-rod be constructed from a non-magnetic material such
as aluminum or plastic, otherwise it contributes erroneously to the coupling between
the coils.
The electronics required by an LVDT are similar to that required for a capacitive
or inductive sensor. An oscillator excites the driving coil with a frequency of around
1 kHz. Although higher frequencies increase the sensor bandwidth they also induce
eddy currents in the core that are detrimental to performance [21]. Alike a capacitive
or eddy-current sensor, a demodulator is required to determine the AC magnitude of
the voltage induced in each coil. A simple synchronous demodulator circuit for this
purpose is shown in Fig. 8.29 [21]. The square-wave oscillator is replaced by a sine-
wave oscillator if the electronics and LVDT are not physically collocated. Other
demodulation circuits include the single-diode demodulator in Fig. 8.19a and the
AD630-based demodulator in Fig. 8.19b.
280 A.J. Fleming and K.K. Leang

The greatest advantages of LVDTs are the infinitesimal resolution, large range,
simplicity, and ruggedness. Very low levels of electrical noise can be achieved due
to the low-impedance of the sensing coils. Nonlinearity is also below 1 % without
the need for field calibration or mapping functions. The major drawbacks of LVDTs
include the limited bandwidth and sensitivity to lateral motion. Due to eddy currents
and the inter winding capacitance, the excitation frequency is limited to a few tens of
kHz, which limits the bandwidth to between 100 Hz and 1 kHz. Although classified
as a non-contact sensor, bearings are required to guide the core linearly through the
bobbin. This can be a significant disadvantage in nanopositioning applications if the
sensor adds both friction and mass to the moving platform. However, if the platform
is already flexure-guided, additional bearings may not be required. LVDTs are most
suited to one-degree-of-freedom applications with relatively large displacement
ranges of approximately 1 mm or greater. A range of less than 0.5 mm is difficult
to achieve due to the small physical size of the coils. A notable exception is the air
core LVDT coils used to detect position in the Asylum Research (USA) atomic force
microscopes [66]. The air core eliminates eddy-current losses and Barkhausen noise
caused by the high permeability materials. An RMS noise of 0.19 nm was reported
for a range of 16 m which equates to a resolution of approximately 1.14 nm and a
dynamic range of 71 ppm [66].
The theoretical resolution of LVDT sensors is limited primarily by the Johnson
noise of the coils and Barkhausen noise in the magnetic materials [66]. However,
standard conditioning circuits like the Analog Devices AD598 produce electronic
noise on the order of 50 Vp-p with a bandwidth of 1 kHz. This imposes a
resolution of approximately 10 ppm when using a driving amplitude of 5 Vp-p.
Since the smallest commercially available range is 0.5 mm, the maximum resolution
is approximately 5 nm with a 1 kHz bandwidth.
Due to their popularity, LVDTs and the associated conditioning electronics are
widely available. Some manufacturers of devices that may be suitable in micro- and
nanopositioning applications include: Macro Sensors, USA; Monitran, UK; Singer
Instruments, Israel; MicroStrain, USA; Micro-Epsilon, USA; and Honeywell, USA.
Two commercially available LVDTs are pictured in Fig. 8.30.

b
a

Singer Instruments, Israel Macro Sensors, USA

Fig. 8.30 Commercially available LVDT sensors. Photos courtesy of (a) Singer Instruments,
Israel and (b) Macro Sensors, USA
8 Position Sensors for Nanopositioning 281

8.3.8 Laser Interferometers

Since 1960, the meter length standard has been defined by optical means. This
change arose after Michelson invented the interferometer which improved the
accuracy of length measurement from a few parts in 107 , to a few parts in 109 [67].
Thus, in 1960, the meter was redefined in terms of the orange line from a 86 Kr
discharge lamp.
In 1983, the meter was redefined as the length traveled by light in a vacuum
during a time interval of 1/299 792 458 s [67]. This definition was chosen because
the speed of light is now fixed and the primary time standard, based on the 133 Cs
clock, is known to an accuracy of a few parts in 1011 [67]. Length measurements
are performed by interferometry using lasers with a frequency measured against
the time standard. With a known frequency and speed, the laser wavelength can be
found to an extremely high accuracy. Stabilized lasers are now available with pre-
cisely calibrated wavelengths for metrological purposes. Metrological traceability
is described further in Sect. 8.2.7.
The operating principle of a Michelson interferometer is described in Fig. 8.31.
A laser beam is split into two paths, one that is reflected by a moving mirror and
another reflected by a stationary mirror. The movement of the mirror is measurable
by observing the fringe pattern and intensity at the detector. If the distance between
the paths is an integer number of wavelengths, constructive interference occurs.
The displacement of the moving mirror, in wavelengths, is measured by counting
the number of interference events that occur. The phase of the interference, and
hence the displacement between interference events, can also be derived from the
detector intensity.

moving mirror
fixed mirror
beam splitter

laser

detector

Fig. 8.31 The operation of a Michelson interferometer. The laser light is split into two paths, one
that encounters a moving mirror and another that is fixed. The two beams are recombined and
interfere at the detector. If the distance between the paths is an integer number of wavelengths,
constructive interference occurs
282 A.J. Fleming and K.K. Leang

Although simple, the Michelson interferometer is rarely used directly for dis-
placement metrology. Due to the reference path, the Michelson interferometer is
sensitive to changes or movement in the reference mirror and the beam splitter.
Differences between the optical medium in the reference and measurement path are
also problematic. Furthermore, the Michelson interferometer is not ideal for sub-
wavelength displacement measurements as the phase sensitivity is a function of the
path length. For example, at the peaks of constructive and destructive interference,
the phase sensitivity is zero.
Modern displacement interferometers are based on the heterodyne interferometer
by Duke and Gordon from Hewlett-Packard in 1970 [68]. Although similar in
principle to a Michelson interferometer, the heterodyne interferometer, overcomes
many of the problems associated with the Michelson design. Most importantly, the
phase sensitivity remains constant regardless of the path length.
Since the original work in 1970, a wide variety of improvements have been made
to the basic heterodyne interferometer, for example, [69]. All of these devices work
on the heterodyne principle, where the displacement is proportional to the phase
(or frequency) difference between two laser beams. In heterodyne interferometers,
the displacement signal is shifted up in frequency which avoids 1=f noise and
provides immunity from low-frequency light-source intensity variations.
In the original design, the two frequencies were obtained from a He–Ne laser
forced to oscillate at two frequencies separated by 2 MHz. However, later designs
utilize acousto-optic frequency shifters to achieve a similar result. An example
application of a heterodyne interferometer is pictured in Fig. 8.32. Here, the angle
and displacement of a linear positioning stage is measured using two interferometers
and a single laser source.
A drawback of conventional interferometers is the large physical size and
sensitivity to environmental variations which preclude their use in extreme envi-
ronments such as within a cryostat or high magnetic field. To allow measurement
in such environments, the miniature fiber interferometer, pictured in Fig. 8.33a, was
developed [70]. The measuring head contains a single-mode optical fiber with a
9 m core diameter coupled to a collimator lens. Approximately 4 % of the applied
light is immediately reflected off the fiber termination and is returned down the fiber,
forming the reference beam. The transmitted light passes through the collimator lens
and is reflected off the slightly angled target mirror back towards the fiber surface

Fig. 8.32 A ZMI™two-axis

heterodyne interferometer
with a single laser source for
measuring the angle and
displacement of a positioning
stage. Courtesy of Zygo, USA
8 Position Sensors for Nanopositioning 283

Fig. 8.33 The operating principle of an Attocube FPS miniature fiber interferometer [70], courtesy
of Attocube, Germany. In (a) the transmitted light is reflected from the mirror, the fiber surface,
the mirror again, and is then focused onto the fiber core. The interferogram plotted in (b) shows
the direct reflected power (black) and the quadrature reflected power (red) versus displacement.
The quadrature signal is obtained by modulating the laser wavelength and demodulating at the
receiver. By plotting the power of the direct and quadrature signals (c), the direction of travel and
sub-wavelength displacement can be resolved

but away from the core. As the fiber surface is a poor reflector, only 4 % of the
incident light is reflected from the fiber surface. This reflected light travels back
through the lens, is reflected off the mirror and is coupled directly to the fiber core,
thus forming a Fabry-Perot interferometer with a cavity length equal to twice the
distance between the fiber and mirror.
As the cavity length changes, the two beams interfere so that the reflected power
is modulated periodically by the distance as illustrated in Fig. 8.33b. A problem
with the basic interferogram is the lack of directional information. To resolve the
direction of travel, the light-source wavelength is modulated at a high-frequency and
demodulated at the receiver to provide an auxiliary interferogram in quadrature with
the original. By considering both the directly reflected power and the demodulated
reflected power, the direction of travel and can be deduced from the phase angle
shown in Fig. 8.33c.
Since the miniature fiber interferometer is physically separated from the laser and
receiver electronics it is both physically small and robust to extreme environments
such as high vacuum, cryogenic temperatures, and magnetic fields. Due to the
284 A.J. Fleming and K.K. Leang

secondary reflection from the fiber surface, the fiber interferometer is also less
sensitive to mirror misalignment compared to some other interferometers.
In general, laser interferometers are the most expensive displacement sensors
due to the required optical, laser, and electronic components. However, unlike
other sensors, laser interferometers have an essentially unlimited range even though
the resolution can exceed 1 nm. Furthermore, the accuracy, stability, and linearity
exceed all other sensors. For these reasons, laser interferometers are widely used
in applications such as semiconductor wafer steppers and display manufacturing
processes. They are also used in some speciality nanopositioning applications
that require metrological precision, for example, the metrological AFM described
in [15].
Aside from the cost, the main drawback of laser interferometers is the suscepti-
bility of the beam to interference. If the beam is broken, the position is lost and the
system has to be restarted from a known reference. The position can also be lost if
the velocity of the object exceeds the maximum velocity imposed by the electronics.
The maximum velocity is typically a few centimeters per second and is not usually
a restriction; however, if the object is subject to shock loads, maximum velocity can
become an issue.
The noise of laser interferometers is strongly dependent on the instrument
type and operating environment. As an example, the Fabry-Perot interferometer
discussed in reference [70] has a 1=f noise cornerp frequency of approximately 10 Hz
and a noise density of approximately 2 pm/ Hz. This results in a resolution of
approximately 1.6 nm with a 12 kHz bandwidth. Equation (8.29) predicts a resolu-
tion of 0.49 nm with a 1 kHz bandwidth. Although the resolution of interferometers
is excellent, small range sensors such as capacitive or piezoresistive sensors can
provide higher resolution. However, the comparison is hardly fair considering that
interferometers have a range in the meters while small range sensors may be
restricted to 10 m or less.
Some manufacturers of interferometers designed for stage metrology and posi-
tion control include: Agilent, USA; Attocube, Germany (Fibre Interferometer);
Keyence, Japan (Fibre Interferometer); Renishaw, UK; Sios, Germany; and Zygo,
USA. Instruments from these manufacturers are pictured in Figs. 8.33a and 8.34.

Fig. 8.34 Two commercially b

available laser
interferometers. Photos
courtesy of (a) Agilent, USA
a
and (b) Sios, Germany
8 Position Sensors for Nanopositioning 285

8.3.9 Linear Encoders

A linear encoder consists of two components, the reference scale and the read-head.
The read-head is sensitive to an encoded pattern on the reference scale and produces
a signal that is proportional to position. Either the scale or the read-head can be free
to move, however, the scale is typically fixed since the read-head is usually lighter.
The earliest form of linear encoder consisted of a bar with a conductive metal
pattern, read by a series of metal brushes [21]. Although simple, the constant contact
between the brush and scale meant a very limited life and poor reliability.
In the 1950s optical linear encoders became available for machine tools. The
reference scales were glass with a photochemically etched pattern. The photolitho-
graphic method used to produce the scale resulted in the highest resolution and
accuracy at the time.
Although today’s optical encoders still produce the highest resolution, other
technologies have also become available. Magnetic or inductive linear encoders can
not match the absolute accuracy or resolution of an optical scale encoder, however,
they are cheaper and more tolerant of dust and contamination. The most common
type of encoder is possibly the capacitive encoder found in digital callipers. These
devices use a series of conductive lines on the slider and scale to produce a variable
capacitor.
The operation of a simple reflective optical encoder is illustrated in Fig. 8.35.
Light from a laser diode is selectively reflected from the scale onto a photodetector.
As the read-head is moved relative to the scale, the peaks in received power
correspond the distance between the reflective bars. In between the peaks, the
position can be estimated from the received power. Rather than partial reflection,
other gratings contain height profiles that modulate the proximity and thus received
power [71].

Light Source
Photo detector
Power

Position

Glass Scale

Fig. 8.35 The operation of a simple reflective optical encoder. The peaks in the received power
correspond to the distance between reflective bars
286 A.J. Fleming and K.K. Leang

Fig. 8.36 The image

scanning technique is used
for reference scales with a
grating pitch of between 10
and 200 m. Image courtesy
of Heidenhain, Germany

There are two major difficulties with the design illustrated in Fig. 8.35. First, the
received power is highly sensitive to any dust or contamination on the scale. Second,
it is difficult to determine the direction of motion, particularly at the peaks where
the sensitivity approaches zero.
To provide immunity to dust and contamination, commercial optical encoders
use a large number of parallel measurements to effectively average out errors.
This principle relies on the Moire phenomenon [72] and is illustrated by the
image scanning technique shown in Fig. 8.36. In Fig. 8.36 a parallel beam of
light is projected onto a reflective scale through a scanning reticle. The reflected
Moire pattern is essentially the binary product of the scanning reticle and the
scale and is detected by an array of photodetectors. Aside from the immunity
to contamination, this technique also provides a quadrature signal that provides
directional information.
Optical reference scales are encoded with a geometric pattern that describes
either the absolute position or the incremental position. Absolute scales contain
additional information that can make them physically larger than incremental scales.
Compared to an incremental encoder, an absolute encoder is also typically more
sensitive to alignment errors, lower in resolution, slower, and more costly. The
benefit of an absolute scale is that the read-head does not need to return to a known
reference point after a power failure or read error.
The noise of high resolution optical encoders is described as “jitter” and is
typically on the order of 1 nm RMS, or 6 nm peak-to-peak. The overall accuracy
is around 5 m/m [73], however, accuracies as high as 0.5 m/m are possible with
ranges up to 270 mm [74].
The highest resolution optical encoders operate on the principle of interference
[74, 75]. The technique involves light that is diffracted through a transparent phase
grating in the read-head and reflected from a step grating on the scale [74]. Since this
technique operates on the principle of diffraction, extremely small signal periods of
down to 128 nm are possible with a resolution on the order of a few nanometers.
8 Position Sensors for Nanopositioning 287

a
b

Fig. 8.37 Two commercially available optical linear encoders. Photos courtesy of (a) Heidenhain,
Germany and (b) Renishaw, UK

Table 8.3 Summary of position sensor characteristics

Sensor type Range DNR Resolution (nm) Max. BW (kHz) Accuracy
Metal foil 10 – 500 m 230 ppm 23 1–10 1 % FSR
Piezoresistive 1 – 500 m 4.9 ppm 0.49 > 100 1 % FSR
Capacitive 10 m – 10 mm 24 ppm 2.4 100 0.1 % FSR
Electrothermal 10 m – 1 mm 100 ppm 10 10 1 % FSR
Eddy current 100 m – 80 mm 10 ppm 1 40 0.1 % FSR
LVDT 0.5 – 500 mm 10 ppm 5 1 0.25 % FSR
Interferometer m 0.49 > 100 1 ppm FSR
Encoder m 6 > 100 5 ppm FSR
The dynamic range (DNR) and resolution are approximations based on a full-scale range of
100 m and a first-order bandwidth of 1 kHz

Other encoder technologies include techniques where the position information is

actually encoded into the medium being scanned. Examples of this approach include
hard disk drives [76] and MEMS mass storage devices [58].
Companies that produce linear encoders suitable for nanometer scale metrology
include: Heidenhain, Germany; MicroE Systems, USA; and Renishaw, UK. Two
instruments from these manufacturers are pictured in Fig. 8.37.

8.4 Comparison and Summary

Due to the extreme breadth of position sensor technologies and the wide range of
applications, it is extremely difficult to make direct performance comparisons. In
many applications, characteristics such as the physical size and cost play a greater
role than performance. Nevertheless, it is informative to compare some aspects of
performance.
In Table 8.3 the specifications under consideration are the range, the dynamic
range, the 6 -resolution, the maximum bandwidth, and the typical accuracy.
Consider the following notes when interpreting the results in Table 8.3:
288 A.J. Fleming and K.K. Leang

• The quoted figures are representative of commercially available devices and do

not imply any theoretical limits.
• The dynamic range and 6 -resolution is an approximation based on a full-scale
range of 100 m and a first-order bandwidth of 1 kHz. The low-frequency limit
is assumed to be fl D 0:01 Hz.
• The quoted accuracy is the typical static trueness error defined in Sect. 8.2.6.
Metal foil strain gauges are the simplest and lowest cost sensor considered in
this study. Due to their size (a few mm2 ) strain gauges are suitable for mounting
directly on to actuators or stages with a range from 10 to 500 m. The parameters
in Table 8.3 pertain to the example of a two-varying-element bridge discussed in
Sect. 8.3.1. Although strain gauges can be calibrated to achieve higher accuracy, it is
reasonable to consider an error of 1 % FSR due to drift and the indirect relationship
between the measured strain and actual displacement.
Piezoresistive sensors are smaller than metal foil strain gauges and can be bonded
to actuators that are only 1 mm long with a range of up to 1 m. Although the
resolution of piezoresistive sensors is very high, the absolute accuracy is limited
by nonlinearity, temperature sensitivity, and inexact matching. An error budget of
1 % FSR is typical. Although strain sensors require contact with the actuator or
flexural components, they do not introduce forces between the reference and moving
platforms, thus, in this sense, they are considered to be non-contact.
Capacitive sensors are relatively simple in construction, provide the highest
resolution over short ranges, are insensitive to temperature, and can be calibrated
to an accuracy of 0.01 % FSR. However, in general purpose applications where the
sensor is not calibrated after installation, alignment errors may limit the accuracy
to 1 % FSR. The capacitive sensor parameters under consideration are described in
Sect. 8.3.4.
Eddy-current sensors can provide excellent resolution for travel ranges greater
than 100 m. They are more sensitive to temperature than capacitive sensors but are
less sensitive to dust and pollutants which is important in industrial environments.
The quoted noise and resolution is calculated from the example discussed in
Sect. 8.3.6.
LVDT sensors are among the most popular in industrial applications requiring a
range from a few millimeter to tens of centimeter. They are simple, have a high
intrinsic linearity and can be magnetically shielded. However, they also have a
low bandwidth and can load the motion with inertia and friction. The maximum
resolution is limited by the physical construction of the transducer which is
generally suited to ranges of greater than 1 mm. The bandwidth of LVDT sensors is
limited by the need to avoid eddy currents in the core. With an excitation frequency
of 10 kHz, the maximum bandwidth is approximately 1 kHz.
Compared to other sensor technologies, laser interferometers provide an unprece-
dented level of accuracy. Stabilized interferometers can achieve an absolute accu-
racy exceeding 1 ppm, or in other words, better than 1 um/m. Nonlinearity is also on
the order of a few nanometers. Due to the low-noise and extreme range, the dynamic
8 Position Sensors for Nanopositioning 289

range of an interferometer can be as high as a few parts per billion, or upwards

of 180 dB. The quoted resolution in Table 8.3 is associated with the Fabry-Perot
interferometer discussed in Sect. 8.3.8.
Linear encoders are used in similar applications to interferometers where abso-
lute accuracy is the primary concern. Over large ranges, absolute accuracies of up
to 5 ppm or 5 m/m are possible. Even greater accuracies are possible with linear
encoders working on the principle of diffraction. The accuracy of these sensors can
exceed 1 ppm over ranges of up to 270 mm, which is equivalent to the best laser
interferometers.

8.5 Outlook and Future Requirements

One of the foremost challenges of position sensing is to achieve high resolution

and accuracy over a large range. For example, semiconductor wafer stages require
a repeatability and resolution in the nanometers while operating over a range in
the tens of centimeters [13, 14]. Such applications typically use interferometers or
high resolution optical encoders which can provide the required performance but
can impose a significant cost. Long-range sensors are also becoming necessary
in standard nanopositioning applications due to the development of dual-stage
actuators [77–80] and stepping mechanisms [81, 82]. Capacitive sensors can be
adapted for this purpose by using a periodic array of electrodes [55]. Such
techniques can also be applied to magnetic or inductive sensing principles. Due to
the increasing availability of long-range nanopositioning mechanisms, an increased
focus on the development of cost effective long-range sensors is required.
A need is also emerging for position sensors capable of measuring position
at frequencies up to 100 kHz. Applications include: high-speed surface inspection
[83, 84]; nanofabrication [3, 4, 12, 85], and imaging of fast biological and physical
processes [10, 86–90]. Although, many sensor technologies can provide a band-
width of 100 kHz, this figure is the 3 dB bandwidth where phase and time delay
render the signal essentially useless in a feedback loop. High-speed position sensors
are required with a bandwidth in the MHz that can provide accurate measurements
at 100 kHz with negligible phase shift or time delay. Due to the operating principle
of modulated sensors such as capacitive and inductive sensors, this level of
performance is difficult to achieve due to the impractically high carrier frequency
requirement. Applications requiring a very high sensor bandwidth typically use an
auxiliary sensor for high bandwidth tasks, for example, a piezoelectric sensor can
be used for active resonance damping [41, 43]. Technologies such as piezoresistive
sensors [91] have also shown promise in high-speed applications since a carrier
frequency is not required. Magnetoresistive sensors are also suitable for high-
frequency applications if the changes in field strength can be kept small enough
to mitigate hysteresis [18, 19].
Due to the lack of cost effective sensors that provide both high-resolution
and wide bandwidth, recent research has also considered the collaborative use of
290 A.J. Fleming and K.K. Leang

multiple sensors. For example, in reference [40] a piezoelectric strain sensor and
capacitive sensor were combined. The feedback loop utilized the capacitive sensor
at low frequencies and the piezoelectric sensor at high frequencies. This approach
retains the low-frequency accuracy of the capacitive sensor and the wide bandwidth
of the piezo sensor while avoiding the drift from the piezo sensor and wide-band
noise from the capacitive sensor. The closed-loop noise was reduced from 5 nm with
the capacitive sensor to 0.34 nm with both sensors. Piezoelectric force sensors have
also been used for high-frequency damping control while a capacitive, inductive, or
strain is used for tracking control [31, 41].
Data storage systems are an example application that requires both long range but
extreme resolution and increasingly wide bandwidth. In these applications, a media
derived position error signal (PES) can provide the requisite range and resolution
but not the bandwidth. In reference [58] a MEMs storage device successfully
combined the accuracy of a media derived position signal with the speed of
an electrothermal sensor. Electrothermal sensors have also been combined with
capacitive sensors to reduce the inherent 1/f noise [49]. Multiple sensors can be
combined by complementary filters [41] or by an optimal technique in the time-
domain [40] or frequency domain [60]. Given the successful applications to date, it
seems likely that the trend of multiple sensors will continue, possibly to the point
where multiple sensors are packaged and calibrated as a single unit.

References

1. D.Y. Abramovitch, S.B. Andersson, L.Y. Pao, G. Schitter, A tutorial on the mechanisms,
dynamics, and control of atomic force microscopes, in Proceedings of American Control
Conference, pp. 3488–3502, New York City, NY, 2007
2. S.M. Salapaka, M.V. Salapaka, Scanning probe microscopy. IEEE Control Syst. 28(2), 65–83
(2008)
3. A.A. Tseng (ed.), Nanofabrication: Fundamentals and Applications (World Scientific, Singa-
pore, 2008)
4. J.A. Vicary, M.J. Miles, Pushing the boundaries of local oxidation nanolithography: short
timescales and high speeds. Ultramicroscopy 108(10), 1120–1123 (2008)
5. ISO/IEC Guide 98:1993 guide to the expression of uncertainty in measurement, international
organization for standardization, 1994
6. ISO 5725 - accuracy (trueness and precision) of measurement methods and results, interna-
tional organization for standardization, 1994
7. G. Schitter, R.W. Stark, A. Stemmer, Sensors for closed-loop piezo control: strain gauges
versus optical sensors. Meas. Sci. Technol. 13, N47–N48 (2002)
8. S. Devasia, E. Eleftheriou, S.O.R. Moheimani, A survey of control issues in nanopositioning,
IEEE Trans. Control Syst. Technol. 15(5), 802–823 (2007)
9. K.K. Leang, Q. Zou, S. Devasia, Feedforward control of piezoactuators in atomic force
microscope systems. IEEE Control Syst. 29(1), 70–82 (2009)
10. A.J. Fleming, B.J. Kenton, K.K. Leang, Bridging the gap between conventional and
video-speed scanning probe microscopes. Ultramicroscopy 110(9), 1205–1214 (2010).
doi:10.1016/j.ultramic.2010.04.016
11. A.J. Fleming, S.S. Aphale, S.O.R. Moheimani, A new method for robust damping and tracking
control of scanning probe microscope positioning stages. IEEE Trans. Nanotechnol. 9(4),
438–448 (2010). doi:10.1109/TNANO.2009.2032418
8 Position Sensors for Nanopositioning 291

12. A.A. Tseng, S. Jou, A. Notargiacomo, T.P. Chen, Recent developments in tip-based nanofabri-
cation and its roadmap. J. Nanosci. Nanotechnol. 8(5), 2167–2186 (2008)
13. H. Butler, Position control in lithographic equipment. IEEE Control Syst. 31(5), 28–47 (2011)
14. S. Mishra, J. Coaplen, M. Tomizuka, Precision positioning of wafer scanners. segmented
iterative learning control for nonrepetitive disturbances. IEEE Control Syst. 27(4), 20–25
(2007). doi:10.1109/MCS.2007.384130
15. R. Merry, M. Uyanik, R. van de Molengraft, R. Koops, M. van Veghel, M. Steinbuch,
Identification, control and hysteresis compensation of a 3 DOF metrological AFM. Asian J.
Control 11(2), 130–143 (2009)
16. Y. Shan, J. Speich, K. Leang, Low-cost IR reflective sensors for submicrolevel posi-
tion measurement and control. IEEE/ASME Trans. Mechatron. 13(6), 700–709 (2008).
doi:10.1109/TMECH.2008.2005407
17. S. Parkin, X. Jiang, C. Kaiser, A. Panchula, K. Roche, M. Samant, Magneti-
cally engineered spintronic sensors and memory. Proc. IEEE 91(5), 661–680 (2003).
doi:10.1109/JPROC.2003.811807
18. D.R. Sahoo, A. Sebastian, W. Häberle, H. Pozidis, E. Eleftheriou, Scanning probe microscopy
based on magnetoresistive sensing. Nanotechnology 22(14), 145501 (2011)
19. V. Kartik, A. Sebastian, T. Tuma, A. Pantazi, H. Pozidis, D.R. Sahoo, High-bandwidth
nanopositioner with magnetoresistance based position sensing. Mechatronics 22(3), 295–301
(2012)
20. T.R. Hicks, P.D. Atherton, Y. Xu, M. McConnell, The Nanopositioning Book (Queensgate
Instruments Ltd, Berkshire, UK, 1997)
21. D.S. Nyce, Linear Position Sensors. Theory and Application (Wiley, Hoboken, 2004)
22. W.C. van Etten, Introduction to Noise and Random Processes (Wiley, West Sussex, England,
2005)
23. R.G. Brown, P.Y.C. Hwang, Introduction to Random Signals and Applied Kalman Filtering
(Wiley, New York, 1997)
24. A.J. Fleming, A method for measuring the resolution of nanopositioning systems. Rev. Sci.
Instrum. 83(8), 086101 (2012)
25. A.J. Fleming, Measuring picometer nanopositioner resolution, in Proceedings of Actuator,
Bremen, 2012
26. A.J. Fleming, Estimating the resolution of nanopositioning systems from frequency domain
data, in Proceedings of IEEE International Conference on Robotics and Automation, St. Paul,
MN, pp. 4786–4791, 2012
27. JCGM 200:2008 international vocabulary of metrology - basic and general concepts and
associated terms (VIM), 3rd edn. (Bureau International des Poids et Mesures, Paris, 2008)
28. T.-F. Lu, D. Handley, Y.K. Yong, Position control of a 3 DOF compliant micro-motion stage,
in Proceedings of Control, Automation, Robotics and Vision Conference, vol. 2, pp. 1274–278,
2004 . doi:10.1109/ICARCV.2004.1469029
29. W. Dong, L.N. Sun, Z.J. Du, Design of a precision compliant parallel positioner driven by dual
piezoelectric actuators. Sensors Actuators A 135(1), 250–256 (2007)
30. G. Schitter, P.J. Thurner, P.K. Hansma, Design and input-shaping control of a novel scanner for
high-speed atomic force microscopy. Mechatronics 18(5–6), 282–288 (2008)
31. A.J. Fleming, K.K. Leang, Integrated strain and force feedback for high performance
control of piezoelectric actuators. Sensors Actuators A 161(1–2), 256–265 (2010).
doi:10.1016/j.sna.2010.04.008
32. W. Kester, J. Bryant, W. Jung, S. Wurcer, C. Kitchin, Chapter 4. Sensor signal conditioning, in
Op Amp Applications Handbook, ed. by W. Jung. Analog Devices (Newnes, Oxford, 2002)
33. C.S. Smith, Piezoresistance effect in germanium and silicon. Phys. Rev. 94(1), 42–49 (1954).
doi:10.1103/PhysRev.94.42
34. A. Barlian, W.-T. Park, J. Mallon, A. Rastegar, B. Pruitt, Review: semiconductor piezore-
sistance for microsystems. Proc. IEEE 97(3), 513–552 (2009). doi:10.1109/JPROC.2009.
2013612
292 A.J. Fleming and K.K. Leang

35. C.M. DiBiasio, M.L. Culpepper, Design of a meso-scale six-axis nanopositioner with
integrated position sensing, in Proceedings of Fifth Annual International Symposium on
Nanomanufacturing, Singapore, 2008
36. R. Messenger, Q. Aten, T. McLain, L. Howell, Piezoresistive feedback control
of a mems thermal actuator. J. Microelectromech. Syst. 18(6), 1267–1278 (2009).
doi:10.1109/JMEMS.2009.2035370
37. J. Sirohi, I. Chopra, Fundamental understanding of piezoelectric strain sensors. J. Intell. Mater.
Syst. Struct. 11, 246–257 (2000)
38. A.J. Fleming, S.O.R. Moheimani, Control oriented synthesis of high performance piezoelectric
shunt impedances for structural vibration control. IEEE Trans. Control Syst. Technol. 13(1),
98–112 (2005)
39. J. Maess, A.J. Fleming, F. Allgöwer, Simulation of dynamics-coupling in piezoelectric tube
scanners by reduced order finite element models. Rev. Sci. Instrum. 79(1–9), 015105 (2008).
doi:10.1063/1.2826428
40. A.J. Fleming, A.G. Wills, S.O.R. Moheimani, Sensor fusion for improved control of
piezoelectric tube scanners. IEEE Trans. Control Syst. Technol. 15(6), 1265–6536 (2008).
doi:10.1016/j.mechatronics.2007.07.006
41. A.J. Fleming, Nanopositioning system with force feedback for high-performance tracking and
vibration control. IEEE Trans. Mechatron. 15(3), 433–447 (2010)
42. Y.K. Yong, B. Ahmed, S.O.R. Moheimani, Atomic force microscopy with a 12-electrode
piezoelectric tube scanner. Rev. Sci. Instrum. 81(1–10), 033701 (2010)
43. Y.K. Yong, A.J. Fleming, S.O.R. Moheimani, A novel piezoelectric strain sensor for simultane-
ous damping and tracking control of a high-speed nanopositioner. IEEE/ASME Trans. Mecha-
tron. (doi: 10.1109/TMECH.2012.2193895, in press). doi:10.1109/TMECH.2012.2193895
44. S.O.R. Moheimani, A.J. Fleming, Piezoelectric Transducers for Vibration Control and Damp-
ing (Springer, London, 2006)
45. A. Preumont, Mechatronics, Dynamics of Electromechanical and Piezoelectric Systems
(Springer, Berlin, 2006)
46. L.K. Baxter, Capacitive Sensors: Design and Applications (IEEE Press, Piscataway, NJ, 1997)
47. M. Kim, W. Moon, E. Yoon, K.-R. Lee, A new capacitive displacement sensor with
high accuracy and long-range. Sensors Actuators A Phys. 130–131(14), 135–141 (2006).
doi:10.1016/j.sna.2005.12.012
48. L.L. Chu, Y.B. Gianchandani, A micromachined 2d positioner with electrothermal actuation
and sub-nanometer capacitive sensing. J. Micromech. Microeng. 13(2), 279–285 (2003)
49. Y.K. Zhu, S.O.R. Moheimani, M.R. Yuce, Simultaneous capacitive and electrothermal position
sensing in a micromachined nanopositioner. Electron Device Lett. 32(8), 1146–1148 (2011)
50. G.T.A. Kovacs, Micromachined Transducers Sourcebook (McGraw Hill, Boston, 1998)
51. A.A. Kuijpers, G.J.M. Krijnen, R.J. Wiegerink, T.S.J. Lammerink, M. Elwenspoek, 2d-finite-
element simulations for long-range capacitive position sensor. J. Micromech. Microeng. 13(4),
S183–S189 (2003)
52. A.A. Kuijpers, G.J.M. Krijnen, R.J. Wiegerink, T.S.J. Lammerink, M. Elwenspoek, A
micromachined capacitive incremental position sensor: part 1. analysis and simulations. J.
Micromech. Microeng. 16(6), S116–S124 (2006)
53. A.A. Kuijpers, G.J.M. Krijnen, R.J. Wiegerink, T.S.J. Lammerink, M. Elwenspoek, A
micromachined capacitive incremental position sensor: part 2. experimental assessment. J.
Micromech. Microeng. 16(6), S125–S134 (2006)
54. J.-I. Lee, X. Huang, P. Chu, Nanoprecision MEMS capacitive sensor for lin-
ear and rotational positioning. J. Microelectromech. Syst. 18(3), 660–670 (2009).
doi:10.1109/JMEMS.2009.2016275
55. S.-C. Lee, R.D. Peters, Nanoposition sensors with superior linear response to position and
unlimited travel ranges. Rev. Sci. Instrum. 80(4), 045109 (2009). doi:10.1063/1.3123407
56. M.A. Lantz, G.K. Binnig, M. Despont, U. Drechsler, A micromechanical thermal displacement
sensor with nanometre resolution. Nanotechnology 16(8), 1089–1094 (2005)
8 Position Sensors for Nanopositioning 293

57. A. Pantazi, A. Sebastian, G. Cherubini, M. Lantz, H. Pozidis, H. Rothuizen, E. Eleftheriou,

Control of mems-based scanning-probe data-storage devices, IEEE Trans. Control Syst.
Technol. 15(5), 824–841 (2007). doi:10.1109/TCST.2006.890286
58. A. Sebastian, A. Pantazi, H. Pozidis, E. Elefthriou, Nanopositioning for probe-based data
storage. IEEE Control Syst. 28(4), 26–35 (2008)
59. A. Sebastian, D. Wiesmann, Modeling and experimental identification of silicon micro-
heater dynamics: a systems approach. J. Microelectromech. Syst. 17(4), 911–920 (2008).
doi:10.1109/JMEMS.2008.926980
60. A. Sebastian, A. Pantazi, Nanopositioning with multiple sensors: A case study in data storage.
IEEE Trans. Control Syst. Technol. 20(2), 382–394 (2012). doi:10.1109/TCST.2011.2177982
61. Y. Zhu, A. Bazaei, S.O.R. Moheimani, M. Yuce, Design, modeling and control of a micro-
machined nanopositioner with integrated electrothermal actuation and sensing, IEEE/ASME J.
Microelectromech. Syst. 20(3), 711–719 (2011)
62. J. Fraden, Handboook of Modern Sensors: Physics, Designs, and Applications (Springer, New
York, 2004)
63. S. Fericean, R. Droxler, New noncontacting inductive analog proximity and inductive linear
displacement sensors for industrial automation. IEEE Sensors J. 7(11), 1538–1545 (2007)
64. S.D. Roach, Designing and building an eddy current position sensor, Sensors J. Sens. Technol.
15(9), 56–74 (1998)
65. Q. Li, F. Ding, Novel displacement eddy current sensor with temperature compen-
sation for electrohydraulic valves. Sensors Actuators A Phys. 122(1), 83–87 (2005).
doi:10.1016/j.sna.2005.04.008
66. R. Proksch, J. Cleveland, D. Bocek, Linear variable differential transformers for high precision
position measurements, US Patent 7,262,592, 2007
67. P. Hariharan, Basics of Interferometry, 2nd edn. (Academic, New York, 2007)
68. J.N. Dukes, G.B. Gordon, A two-hundred-foot yardstick with graduations every microinch.
Hewlett-Packard J. 21(2), 2–8 (1970)
69. G.E. Sommargren, A new laser measurement system for precision metrology, in Proceedings
of Precision Engineering Conference, Dallas, Texas, 1986
70. K. Karrai, P. Braun, Miniature long-range laser displacement sensor, in Proceedings of
Actuator, Bremen, Germany, pp. 285–288, 2010
71. A. Khiat, F. Lamarque, C. Prelle, P. Pouille, M. Leester-Schädel, S. Büttgenbach, Two-
dimension fiber optic sensor for high-resolution and long-range linear measurements. Sensors
Actuators A Phys. 158(1), 43–50 (2010). doi:10.1016/j.sna.2009.12.029
72. R.S. Sirohi, Optical Methods of Measurement - Wholefield Techniques (CRC Press, Taylor and
Francis Group, Boca Raton, FL, 2009)
73. FASTRACK high-accuracy linear encoder scale system. data sheet l-9517-9356-01-b (2009).
www.renishaw.com
74. Heidenhain exposed linear encoders (2011). www.heidenhain.com
75. J.-Y. Lee, H.-Y. Chen, C.-C. Hsu, C.-C. Wu, Optical heterodyne grating interferometry for
displacement measurement with subnanometric resolution. Sensors Actuators A Phys. 137(1),
185–191 (2007). doi:10.1016/j.sna.2007.02.017
76. B.M. Chen, T.H. Lee, K. Peng, V. Venkatarmanan, Hard Disk Drive Servo System (Springer,
London, 2006)
77. Y. Michellod, P. Mullhaupt, D. Gillet, Strategy for the control of a dual-stage nano-positioning
system with a single metrology, in Proceedings of Robotics, Automation and Mechatronics,
pp. 1–8, 2006. doi:10.1109/RAMECH.2006.252649
78. L. Chassagne, M. Wakim, S. Xu, S. Topçu, P. Ruaux, P. Juncar, Y. Alayli, A 2d nano-
positioning system with sub-nanometric repeatability over the millimetre displacement range.
Meas. Sci. Technol. 18(11), 3267–3272 (2007)
79. A.J. Fleming, Dual-stage vertical feedback for high speed-scanning probe microscopy. IEEE
Trans. Control Syst. Technol. 19(1), 156–165 (2011). doi:10.1109/TCST.2010.2040282
80. J. Zheng, A. Salton, M. Fu, Design and control of a rotary dual-stage actuator positioning
system. Mechatronics 21(6), 1003–1012 (2011). doi:10.1016/j.mechatronics.2011.03.011
294 A.J. Fleming and K.K. Leang

81. C.-L. Chu, S.-H. Fan, A novel long-travel piezoelectric-driven linear nanopositioning stage,
Precis. Eng. 30(1), 85–95 (2006). doi:10.1016/j.precisioneng.2005.05.002
82. R. Merry, M. Maassen, M. van de Molengraft, N. van de Wouw, M. Steinbuch, Modeling and
waveform optimization of a nano-motion piezo stage. IEEE/ASME Trans. Mechatron. 16(4),
615–626 (2011). doi:10.1109/TMECH.2010.2050209
83. G. Borionetti, A. Bazzalia, R. Orizio, Atomic force microscopy: a powerful tool for surface
defect and morphology inspection in semiconductor industry. Eur. Phys. J. Appl. Phys. 27(1–3),
101–106 (2004)
84. A. Humphris, M. McConnell, D. Catto, A high-speed atomic force microscope capable of
video-rate imaging. Microsc. Anal. SPM Suppl. 20, 29–31 (2006)
85. A. Ferreira, C. Mavroidis, Virtual reality and haptics for nanorobotics. IEEE Robot. Autom.
Mag. 13(3), 78–92 (2006)
86. G.E. Fantner, G. Schitter, J.H. Kindt, T. Ivanov, K. Ivanova, R. Patel, N. Holten-Andersen,
J. Adams, P.J. Thurner, I.W. Rangelow, P.K. Hansma, Components for high speed atomic force
microscopy. Ultramicroscopy 106(2–3), 881–887 (2006)
87. M. Kobayashi, K. Sumitomo, K. Torimitsu, Real-time imaging of DNA-streptavidin complex
formation in solution using a high-speed atomic force microscope. Ultramicroscopy 107(2–3),
184–190 (2007)
88. G. Schitter, K.J. Åström, B.E. DeMartini, P.J. Thurner, K.L. Turner, P.K. Hansma, Design and
modeling of a high-speed AFM-scanner. IEEE Trans. Control Syst. Technol. 15(5), 906–915
(2007)
89. L.M. Picco, L. Bozec, A. Ulcinas, D.J. Engledew, M. Antognozzi, M. Horton, M.J. Miles,
Breaking the speed limit with atomic force microscopy. Nanotechnology 18(4), 044030(1–4)
(2007)
90. T. Ando, T. Uchihashi, T. Fukuma, High-speed atomic force microscopy for nano-visualization
of dynamic biomolecular processes. Prog. Surf. Sci. 83(7–9), 337–437 (2008)
91. E. Guliyev, T. Michels, B. Volland, T. Ivanov, M. Hofer, I. Rangelow, High speed quasi-
monolithic silicon/piezostack spm scanning stage. Microelectron. Eng. 98(0), 520–523 (2012).
doi:10.1016/j.mee.2012.07.059
Chapter 9
MEMS Nanopositioners

Jason J. Gorman

9.1 Introduction

Nanopositioning mechanisms, or nanopositioners, have evolved quickly over the

last few decades due to a growing need for nanoscale precision motion control
in applications including microscopy, lithography, nanomanufacturing, and optics.
The critical differentiator between nanopositioners and other motion stages is that
nanopositioners are capable of nanoscale positioning resolution, which is generally
interpreted as a few nanometers and below. Motion stages with mechanical elements
that have friction, such as screws, linear bearing slides, and rotational bearings,
are not capable of nanoscale resolution. The backlash caused by friction limits the
precision in these mechanisms to tens of nanometers, and complex control systems
are often required to reach this level. As a result, nanopositioner designs typically
use a flexure mechanism to guide the motion of the stage, which exhibit no friction
since the motion is based on structural bending, thereby providing continuous
smooth movements.
The simplest nanopositioner is composed of a flexure mechanism and an actuator
that drives the mechanism. In most high-precision applications, nanopositioners also
include sensors for measuring the motion, which are used for feedback control and
other error compensation methods. Among the many nanopositioner designs in the
literature, the combination of piezoelectric actuators, a metal flexure mechanism,
and capacitive or optical sensors is by far the most common. This combination can
be seen in numerous linear, multi-axis, and rotational nanopositioners [1–3] and
has been the basis of the majority of commercially available nanopositioners [4–6].

J.J. Gorman ()

National Institute of Standards and Technology, 100 Bureau Drive,
Stop 8212, Gaithersburg, MD 20899-8212, USA
e-mail: [email protected]

© Springer International Publishing Switzerland 2016 295

C. Ru et al. (eds.), Nanopositioning Technologies,
DOI 10.1007/978-3-319-23853-1_9
296 J.J. Gorman

These motion stages have radically improved the performance of many precision
instruments, including scanning probe microscopes [7] and optically lithography
tools for semiconductor electronics manufacturing [8].
In tandem with the evolution of nanopositioners over the last few decades,
microelectromechanical systems (MEMS) have emerged to create a major industry
and have enabled numerous technology advances including automotive airbag
deployment, inertial navigation for unmanned vehicles, and smartphones. Most
MEMS are composed of combinations of actuators, flexure mechanisms, and
sensors, as seen in accelerometers, gyroscopes, and resonators. Furthermore, these
devices have been shown to be capable of exceptional precision compared to their
macroscale counterparts. As a result, MEMS are an obvious technology platform for
extending the capabilities of nanopositioners. This opportunity was realized early on
in the development of both nanopositioners and MEMS by Akamine et al. [9] who
developed a microscale piezoelectric cantilever for scanning tunneling microscopy
that can move with three degrees of freedom. Since then, MEMS nanopositioners
have progressed rapidly due to exciting applications that are not possible with
macroscale nanopositioners, including multi-tip scanning probe microscopy (SPM)
for high-bandwidth imaging, lithography, and data storage.
This chapter will provide a review of the state of the art for MEMS nanoposition-
ers. The next section discusses the motivation for using MEMS in nanopositioning,
including the components within a nanopositioner, the performance requirements
for these systems, the advantages in using MEMS to create high-performance
motion stages, and the applications that are driving the development of this
technology. This is followed by a review of the actuators and sensors that have
been demonstrated in MEMS nanopositioners, with an emphasis on the advantages
and disadvantages related to precision motion control. An assessment of the
performance of MEMS nanopositioners to date in terms of resolution, range,
bandwidth, and size is then presented. Finally, the chapter concludes with thoughts
on the current limitations and future directions for MEMS nanopositioners.

9.2 Motivation for MEMS Nanopositioners

Nanopositioners are composed of actuators, a flexure mechanism, sensors, and a

control system. The block diagram in Fig. 9.1 describes the interactions between
these components. For a single-axis system, a control voltage, u, is amplified and
applied to the actuator. This causes the actuator to exert a force, F, on the flexure
mechanism, thereby causing a displacement, x. The displacement is then measured
with the sensor and the sensing signal is processed by the readout electronics to
produce a voltage, Vx , which is proportional to the displacement. The sensor voltage
is then sent to the control system, which generates the control voltage, u, based on
the selected algorithm. Open-loop control is also possible but it typically yields
significantly reduced precision.
9 MEMS Nanopositioners 297

Electronics

Electronics
F x

Amplifier

Readout
Flexure
Actuators Sensors
Mechanism

u Control Vx
System

Fig. 9.1 Anatomy of a nanopositioner. Block diagram showing the components within a nanopo-
sitioner and the interactions between the components. u D control voltages, F D forces applied by
actuators, x D displacements of MEMS nanopositioner, Vx D measured displacements in voltage

Actuators for controlled motion and sensors for displacement measurement

are widely used in MEMS including those found in accelerometers, gyroscopes,
microphones, and pressure sensors. As a result, MEMS technologies provide a
compelling approach to implementing the device anatomy shown in Fig. 9.1.
The main question is whether MEMS nanopositioners can match the performance
obtained by macroscale nanopositioners. While there is considerable variation in the
performance attained with macroscale nanopositioners, the motion resolution can
be as good as 0.05 nm, the range can be as high as a few hundred micrometers, and
the dynamic range is typically above 105 . The achievable motion bandwidth ranges
from tens of hertz to several kilohertz and is inversely related to the motion range.
There are hard constraints between these four metrics (resolution, range, dynamic
range, and bandwidth) but they can be optimized for a given application. For
example, scanning tunneling microscopy (STM) usually requires greater resolution
than atomic force microscopy (AFM). Therefore, a nanopositioner for STM may
have motion resolution on the order of tens of picometers but with a range of only
a few micrometers, whereas a nanopositioner for AFM may have sub-nanometer
resolution and a range around 100 m.
The motivation for the development of MEMS nanopositioners is a combination
of advantages related to performance, functionality, and new applications that are
only accessible with small mechanisms. Scaling down in size presents one set
of advantages. First, MEMS have high resonant frequencies compared to larger
systems, resulting in greater motion bandwidth. Second, MEMS can be integrated
on a silicon chip with other technologies, such as electronics and photonics,
thereby providing a path to low-power embedded systems with high functionality.
Finally, arrays of nanopositioners can be fabricated on a single chip and controlled
independently. The main disadvantages in scaling down in size are the resulting
payload size and mass limits.
298 J.J. Gorman

Piezoelectric Actuators

Capacitance Probes

Fig. 9.2 An XY nanopositioner with piezoelectric actuators, a nested flexure mechanism with
motion amplification, and capacitance probes (J.J. Gorman, S. Bergna, N.G. Dagalakis, unpub-
lished)

Another set of advantages results from the integration and economy of scale
inherent in MEMS fabrication processes. A representative macroscale XY nanoposi-
tioner is shown in Fig. 9.2. It includes two piezoelectric stack actuators, a nested XY
flexure mechanism that amplifies the actuator motion, and two capacitive position
sensors. The flexure mechanism is fabricated using wire electrodischarge machining
(wire EDM), which is a slow, expensive, and serial process. The actuators and
sensors are assembled into the mechanism by hand, adding additional time and
cost to fabrication. Additionally, misalignment between components is inevitable,
causing off-axis forces and erroneous position measurements, thereby reducing the
accuracy of motion. In comparison, MEMS fabrication uses parallel processes, the
mechanisms are typically monolithic, and assembly is not required. Therefore, the
cost is significantly reduced and motion errors due to misalignment of sensors and
actuators are all but eliminated.
Figure 9.3 shows an XY MEMS nanopositioner with electrothermal actuators and
a nested flexure mechanism that amplifies motion, similar to the mechanism shown
in Fig. 9.2. Comparing Figs. 9.2 and 9.3, the advantages of MEMS fabrication
are evident. The MEMS nanopositioner is fabricated in a single layer of single-
crystal silicon. Complex flexure mechanisms, such as the multi-lever arm design
with circular notch flexure joints can be fabricated with ease using deep reactive ion
etching (DRIE). The actuators and flexure mechanism are directly integrated and
the only misalignments possible between the two are caused by lithography errors,
9 MEMS Nanopositioners 299

Fig. 9.3 An XY MEMS nanopositioner with electrothermal actuators and a nested flexure
mechanism with motion amplification, similar to that shown in Fig. 9.2 (J.J. Gorman, S. Bergna,
N.G. Dagalakis, unpublished)

which are typically negligible. This is in stark contrast to the complex couplings
necessary to assemble the actuators and sensors in the macroscale nanopositioner in
Fig. 9.2.
The materials available in MEMS fabrication also provide significant advantages
over those common in macroscale nanopositioners. Single-crystal silicon is the
most widely used material for MEMS nanopositioners due to several factors. It
is broadly available with widely varying specifications, its material properties are
well known, and there are many well-established etch processes available, making
it a relatively easy material to use in MEMS fabrication. Silicon has a much higher
Young’s modulus and a lower coefficient of thermal expansion than the materials
used in macroscale nanopositioners (i.e., metals), resulting in stiffer structures
and improved thermal stable. Monolithic integration in silicon further improves
thermal stability by avoiding mismatched material properties as found in assembled
nanopositioners.
There are several applications driving the development of MEMS nanoposition-
ers with SPM being the most prominent and well explored. By definition, SPM [10],
including STM and AFM, requires a nanopositioner to scan the probe or sample
with exceptional precision. There is a continual push to increase the scan rate in
SPM in order to reduce drift in surface images, maximize data collection, and
observe dynamic nanoscale phenomena. As described above, MEMS are capable
of high motion bandwidth making them an obvious solution to the SPM scanning
problem. Akamine et al. [9] were the first to demonstrate a MEMS multi-axis
300 J.J. Gorman

Fig. 9.4 Active AFM cantilever array. (a) Schematic of cantilever array showing actuation
and sensing elements, (b) embedded readout electronics, and (c) fabricated cantilever array (S.
Hafizovic et al. © PNAS [18])

scanner for STM, achieving atomic resolution with a piezoelectric cantilever. This
work was followed by the development of multi-axis electrostatic nanopositioners
with integrated tunneling tips that were designed for multi-tip STM measurements
[11–13].
Most of the research on MEMS scanners has focused on AFM due to the broader
applications compared to STM and the simpler imaging mechanism (i.e., tip force
vs. tunneling current). Due to their small size and high level of device integration,
MEMS have provided a path for active AFM cantilever arrays with independent
force sensing and Z axis actuation [14–18] (see Fig. 9.4). In addition to the increased
motion bandwidth due to scaling down in size, cantilever arrays provide multiplexed
measurements that increase the image size and speed of image collection by
increasing the number of cantilevers. While multiple tips can be controlled in
the vertical direction (e.g., 10 cantilevers [14]), scanning of the cantilever array
along the X and Y axes is performed with a macroscale nanopositioner. Arrays
of independent AFM scanners (XYZ) would be capable of significantly higher
throughput due to their higher motion bandwidth. As a step in this direction, an XY
MEMS nanopositioner has been used to image metal patterns on the scanner with
resolution and bandwidth that is competitive with traditional AFM [19–22]. Most
recently, a fully integrated on-chip AFM with three axes of motion and embedded
force sensing has been shown to be capable of imaging nanoscale step heights with
9 MEMS Nanopositioners 301

Fig. 9.5 On-chip AFM. (a) Schematic showing three motion axes with electrothermal actuation
and cantilever with piezoresistive strain sensing and embedded tip, (b) fabricated on-chip AFM
(N. Sarkar et al. © IEEE [24])

3 nm pk-pk vertical resolution [23, 24] (see on-chip AFM in Fig. 9.5). Applications
for on-chip AFM beyond nanoscale imaging include probe-based data storage [25]
and nanolithography [14].
Another active application area for MEMS nanopositioners is the manipulation
of nanostructures for nanomanufacturing and material testing. A number of nanoma-
nipulator concepts have been explored in which probes with nanoscale sharpness are
used to push and tweeze nanostructures [26–29]. Nanopositioners have also been
used to measure the mechanical properties of nanowires and carbon nanotubes by
attaching them to the motion stage and pulling on them while observing the event in
a scanning electron microscope (SEM) or transmission electron microscope (TEM)
[30–32]. By scaling down the material testing experiment, the accuracy of the results
can be improved due to a significant reduction in the distance between sensors
and actuators and because they can more readily fit in small sample chambers as
found in TEMs. Many other applications for MEMS nanopositioners are expected
in the areas of adaptive optics (e.g., tunable lens arrays), photonics (e.g., laser cavity
stabilization), and physics research (e.g., optical and atomic traps), among others.
Returning to Fig. 9.1, it is clear that for MEMS nanopositioners to be successful
in the applications above, each of the components (actuators, flexure mechanism,
sensors, controller) must perform at a level comparable to their macroscale coun-
terparts. There has been considerable research on actuators and sensors for MEMS
nanopositioners, which will be discussed in the following sections. Control system
design for MEMS nanopositioners varies greatly depending on the actuator, flexure
mechanism, and sensor selections, and the application of interest. As a result, this
subject is discussed in detail in the following chapter. The final component in a
MEMS nanopositioner, the flexure mechanism, is critical to its performance and
requires careful design to achieve stable motion over the desired range. However,
flexure mechanism design at the microscale is largely the same as at the macroscale
(e.g., see [33, 34]), so this topic is not covered here. However, it is worth noting the
variety of mechanism designs that have been used to date.
302 J.J. Gorman

Fig. 9.6 The HexFlex nanopositioner. This mechanism uses electrothermal actuators and a two-
level flexure mechanism to generate six-axis motion. (a) Fabricated mechanism, (b) exploded
diagram showing two levels of thermal actuators. (S.-C. Chen and M.L. Culpepper © Elsevier
[47])

Early mechanisms relied heavily on cantilevered beams [9, 14–18] and clamped
beams [35]. Double parallelogram flexures were then found to constrain the motion
better than simple beam structures, resulting in pure translational motion along the
degree of freedom of interest with minimal off-axis motion [35–38]. Multi-axis
MEMS nanopositioners require more complex mechanisms to properly constrain
the motion. A number of different designs have been used for XY stages, including
beam structures [12, 19, 39, 40], a four-bar flexure [41], and nested structures with
circular notch flexure hinges [28, 42] (see Fig. 9.3). Other multi-axis mechanisms
of interest include XYZ stages [43–45] (see Fig. 9.8), an XY stage [46], and even a
six-axis parallel mechanism [47], as shown in Fig. 9.6. Almost all of these flexure
mechanism designs work equally well at the microscale as at the macroscale. One of
the biggest challenges in developing an adequate flexure mechanism for MEMS is
finding a design that is compatible with the fabrication process. In many cases, this
limits the mechanism to two-dimensional geometry in the plane of the substrate,
which can make it difficult to properly constrain both in-plane and out-of-plane
motions simultaneously. Other important design challenges include the trade-off
between range and bandwidth (i.e., fundamental resonant frequency), cross-talk
between motion axes, and achieving linearity over the entire workspace.

9.3 Actuation

Actuators for nanopositioning must meet several stringent performance require-

ments. Smooth continuous motion with position fluctuations on the order of 1 nm or
below is required. Sufficient force must be generated to move the flexure mechanism
over a range from 1 to 100 m, depending on the application. In general, a motion
bandwidth of at least several hundred hertz is necessary but significantly higher
bandwidth is typically expected since this is a driving motivation for using MEMS
nanopositioners. Depending on the level of integration required for the application,
this must all be done within a volume from 0.1 to around 5 mm3 . Actuation in
9 MEMS Nanopositioners 303

the plane of the supporting substrate and out of the plane are both required for
many applications of MEMS nanopositioners, such as SPM, since they depend
on multi-axis motion. There are numerous books [48, 49] and review articles
[50–52] that describe the design and modeling of MEMS actuators and sensors
in detail. Therefore, the goal of this section and the following section on sensors
is not to provide an in-depth description on how various actuators and sensors
operate. Rather, the focus is on highlighting the approaches that have been used in
MEMS nanopositioners and discussing their advantages and disadvantages within
this context. Among the many intriguing approaches for actuation in MEMS, almost
all of the MEMS nanopositioners to date rely on electrostatic, electrothermal, or
piezoelectric actuation, which are discussed below. The most obvious omission
from this discussion is electromagnetic actuation, which can be very effective at
the macroscale, particularly linear Lorentz force actuators. However, due to poor
scaling and difficulties in fabricating microscale rare-earth magnets, electromag-
netic actuation has received little attention with respect to MEMS nanopositioners,
with a few exceptions (e.g., see [53] and its reference list).

9.3.1 Electrostatic Actuators

Electrostatic actuation is the most common approach for MEMS nanopositioners,

as well as for MEMS in general. The design and modeling of electrostatic actuators
are well understood, making them easy to integrate into mechanisms and achieve a
predictable level of performance. An electrostatic actuator is a capacitor in which
one side of the capacitor can move because it is attached to a flexure mechanism.
When a voltage is applied across the capacitor a charge is generated, resulting in
an electrostatic force that is proportional to the voltage squared. The three most
common types of electrostatic actuators are the comb actuator, the parallel-plate
actuator, and the parallel-plate actuator with comb structure, as shown in Fig. 9.7.
The electrostatic comb actuator (Fig. 9.7a) uses a pair of interdigitated combs
to generate a force in the direction parallel to the comb fingers (i.e., the x-axis).
Numerous MEMS nanopositioners have used this actuator due to its simplicity of
design, well-characterized force model, and considerable motion range [12, 19,
35–41, 43, 44, 46]. An example of an XYZ MEMS nanopositioner with comb
actuators for the X and Y axes is shown in Fig. 9.8. The force can be optimized
by minimizing the gap between the fingers, g, and maximizing the thickness
thickness, t. Fabrication on silicon-on-insulator wafers using DRIE is the best
method for achieving these high aspect ratio structures. Although the force is
proportional to the voltage squared, the actuator response can easily be linearized by
canceling the nonlinearity with a square root function, either with an analog circuit
or digital signal processor. The motion range of a comb actuator can be several
hundred micrometers through careful design of the connected flexure mechanism
[54, 55], which is exceptionally large for MEMS nanopositioners. Most comb
actuators are designed for in-plane motion, but out-of-plane motion (z axis) can also
304 J.J. Gorman

Fig. 9.7 Common a

electrostatic actuator designs:
(a) Comb actuator, (b) g -
parallel-plate actuator, and (c)
parallel-plate actuator with a
comb structure
t + x

z
g -
+

c
y
g -

t +
L

be achieved by using the levitation effect [13] or an offset in the vertical position of
the two combs [56, 57]. This results in a much smaller motion range, typically on
the order of a few micrometers.
The electrostatic parallel-plate actuator (Fig. 9.7b) has been used at many scales
and its application in MEMS predates the use of comb actuators. Several actuator
geometries for in-plane [11] and out-of-plane motion [43, 58–61] have been used
in MEMS nanopositioners. An example is shown in Fig. 9.8, where the parallel-
plate actuator drives the Z axis of an XYZ MEMS nanopositioner. The design of
this actuator is simple and easy to fabricate. However, the force generated by a
parallel-plate actuator is nonlinearly related to the displacement of the moving plate,
resulting in a pull-in instability, which is generally not an issue for comb actuators.
The instability causes the moving plate to snap in to the stationary plate when the
displacement due to an applied voltage exceeds one third of the nominal gap, g.
The motion range before pull-in can be improved through mechanical design (e.g.,
see [62]), but it is still small compared to comb actuators. When g is made larger to
accommodate a greater range, the required input voltage necessary to achieve a long
9 MEMS Nanopositioners 305

Fig. 9.8 An XYZ MEMS nanopositioner that uses electrostatic comb actuators for the X and Y
axes and an electrostatic parallel-plate actuator for the Z axis (X. Liu, K. Kim, and Y Sun © IOP
[43])

Fig. 9.9 An XY MEMS nanopositioner that uses electrostatic parallel-plate actuators with a comb
structure (Y. Sun et al. © Elsevier [64])

range increases and becomes impractical, thereby setting the motion limit to only
a few micrometers. The parallel-plate approach is typically used for out-of-plane
actuation but it can be implemented for in-plane motion by using a comb structure
with many parallel-plate actuators working in parallel, as shown in Figs. 9.7c and
9.9 [63, 64].
Among the many positive attributes for electrostatic actuation, possibly the most
attractive is that it has a flat dynamic response (i.e., no resonances). The dynamics of
the nanopositioner are defined solely by the flexure mechanism, making it possible
to achieve the desired motion bandwidth through straightforward modeling of the
electrostatic actuator and the structural dynamics of the flexure mechanism. The
greatest challenge in using an electrostatic actuator is that the ratio of the actuation
force over the in-plane actuator area is low in comparison to other actuators. As a
result, large actuators are necessary to generate an adequate motion range in most
cases. This limits the electrostatic actuation to applications in which the deeply
scaled integration of MEMS nanopositioners arrays is not required.
306 J.J. Gorman

9.3.2 Electrothermal Actuators

Electrothermal actuators are the second most popular type of actuator for MEMS
nanopositioners. In these actuators, a current is passed through an electromechanical
structure with an appropriate resistance (typically between 100  and 1 k). This
results in Joule heating causing the structure to expand due to the increase in tem-
perature. If designed correctly, the thermal expansion results in linear motion that
is proportional to the voltage squared since the temperature increase is proportional
to the power dissipation. By properly constraining the electromechanical structure,
the actuator motion can be significantly larger than that achieved through linear
expansion alone. Thermal expansion provides continuous smooth motion that can
easily be controlled through the actuation voltage.
Almost all electrothermal actuators can be defined as being one of following two
types: (1) Joule heating through the bulk of a single-material actuator or (2) Joule
heating of stacked, thermally mismatched materials (i.e., bimorph actuation). In the
first type, current flows through the bulk of the structure and the actuator uniformly
expands and contracts across its cross-section. The two most common geometries
for this type are the chevron actuator and the U-shape actuator, as shown in Fig. 9.10.
The chevron actuator [65] has a symmetric geometry with angled beams connected
to a center shuttle that transmits the force. When current flows through the beams,
they heat up and their length increases, causing the center shuttle to move. The
resulting output motion is related to the drive voltage, the geometry of the beams,

Fig. 9.10 Common a

electrothermal actuator + x
designs. (a) chevron actuator,
(b) U-shape actuator, and (c)
q
thermal bimorph actuator
-

b y
hot

+ cold
-

c heater
z
+
- material A
material B
9 MEMS Nanopositioners 307

and the beam angle. A smaller beam angle provides larger motion for the same drive
voltage but also reduces the actuator stiffness, thereby reducing the force that can be
transmitted. Numerous MEMS nanopositioners have relied on this actuator [65–72]
due to its simplicity of design and because is generates highly linear motion.
The U-shape actuator [73] is composed of a narrow beam (hot arm) and a
wide beam (cold arm) that are connected in a continuous loop. When a current
flows through the loop, the hot arm heats up and expands, while the cold arm
experiences a much smaller increase in temperature due to its lower resistance.
This mismatch in expansion causes motion at the tip of the actuator along an arc.
Nanopositioning typically requires linear motion so the arc motion of the U-shape
actuator is unacceptable in most cases. This has limited it use in comparison to the
chevron actuator. Linear motion can be achieved though by coupling two actuators
together as shown in [47] (see Fig. 9.6). The geometry shown in Fig. 9.10b and in
[47, 73] is for in-plane motion. However, the principle of using a hot and cold arm
has been demonstrated for out-of-plane motion as well [74].
Bimorph actuators that use Joule heating are composed of two or more materials
that have different coefficients of thermal expansion and a heater, which is a
conductive structure that is driven electrically (see Figs. 9.10c and 9.11). When a
current is applied to the heater, the actuator heats up and moves due to internal
stress caused by the mismatch in thermal expansion and the thermal gradient. This
approach is most commonly used with cantilevers [16–18, 75, 76] because large
motion can be generated at the cantilever tip due to small strain at the base from
thermal expansion. Furthermore, it is relatively easy to stack the materials necessary
to achieve a large mismatch in their expansion. As with the U-shape actuator,

Fig. 9.11 Thermal bimorph

cantilever. (a) Section ii is the
thermal bimorph actuator and
section i contains a
piezoresistive sensor, (b)
schematic of cantilever
cross-section (T. Akiyama
et al. © AIP [75])
308 J.J. Gorman

bimorph cantilevers generate motion along an arc making them less attractive for
nanopositioning. However, the thermal bimorph concept has been implemented
to generate in-plane linear motion by using a CMOS (complementary metal-
oxide semiconductor) fabrication process to create actuators composed of dielectric
structures and metal heaters [23, 24, 77]. These in-plane actuators use beam bending
resulting from bimorph actuation but the heaters are strategically placed within the
beams and multiple bimorph beams are coupled together. This design cancels out
the rotational motion typically found in bimorph actuators, thereby providing linear
output motion. As a result, this design can be used in applications similar to that of
the chevron actuator.
While there is considerable variation in the motion range of thermal actuators,
the maximum displacement is around 20 m. The reduced range compared to
electrostatic actuators is frequently compensated using a flexure mechanism with
motion amplification [28, 42]. Similar to electrostatic actuators, the displacement
of electrothermal actuators is proportional to the square of the applied voltage.
However, when this input nonlinearity is canceled using an analog circuit or digital
signal processor, their dynamic response is nearly linear with the exception of
temperature-dependent material properties. The chevron and U-shape actuators are
straightforward to fabricate since they only require a single device layer, making
these devices attractive for silicon-on-insulator processes.
In addition to ease of fabrication, electrothermal actuators are robust mechanisms
that can withstand a large number of heating cycles, are generally unaffected by
surface contamination, and are less prone to electrical breakdown compared to
electrostatic actuators. The biggest disadvantage of electrothermal actuators is their
bandwidth, which is limited by their thermal time response. Typically, a bandwidth
in the range of 100 Hz can be achieved, although this is dependent on size, heat
sinking, and whether the actuator is operating in air or vacuum. The bandwidth
can be extended to some degree through open-loop control [75, 78]. Even with
open-loop control, the bandwidth is limited by the maximum drive voltage that
can be sustained without damage and the rate at which the actuator can lose heat.
This drawback of electrothermal actuation must be strongly considered for a given
application.

9.3.3 Piezoelectric Actuators

Multi-layer piezoelectric actuators, such as those shown in Fig. 9.2, are by far
the most commonly used actuators in macroscale nanopositioners. Interestingly,
piezoelectric actuators have found very limited use in MEMS nanopositioners. This
is largely due to the small piezoelectric coefficients that can be achieved (in this
case, the d33 coefficient), thereby requiring stacked actuators with thick layers.
For example, typical multi-layer piezoelectric actuators made with lead zirconate
9 MEMS Nanopositioners 309

titanate (PZT) can achieve approximately 1 m of displacement at 120 V for

a 1 mm thick material layer. Clearly, this approach does not scale well since a
1 m thick layer would result in only 1 nm of displacement. As a result, linear
actuators in which the motion has a one-to-one correspondence to the expansion
of a piezoelectric material are generally not found in MEMS with the exception of
resonators, which can operate with exceptionally small motion.
One approach for achieving adequate motion range with a MEMS piezoelectric
actuator is to use a bimorph cantilever design [9, 14, 15]. Expansion of the
piezoelectric material causes the cantilever to bend, resulting in vertical motion
of the cantilever tip. Displacements of tens of micrometers can be achieved with
this approach. However, linear motion cannot be generated with this configuration,
limiting its application. An intriguing approach to solving this problem uses
multiple bimorph actuators to generate in-plane linear motion through out-of-
plane bending, resulting in a range of a few micrometers [79]. In addition to
the challenge of generating linear motion, the deposition of high-quality thin-film
piezoelectric materials with properties comparable to those found in macroscale
materials remains an open area of research. As a result, piezoelectric actuators are
not currently practical for MEMS nanopositioners but this may change in the near
future.

9.4 Sensing

The most commonly used sensors in macroscale nanopositioners are capacitive

sensors and strain gauge sensors, including both metal and piezoresistive types.
Laser interferometry is also used when exceptional accuracy and stability are
required and cost and size are not a concern. These displacement sensors must meet
a set of stringent performance requirements similar to those described for actuators
in the previous section: resolution of 1 nm or below, range from 1 to 100 m, and
bandwidth of a few kilohertz. In addition, linearity and thermal stability are critical
since they directly affect the motion accuracy. However, it is more difficult to define
these metrics and their values are largely unreported for MEMS in the literature.
Among the MEMS sensors for linear displacement measurement, either capac-
itive, thermal, or piezoresistive sensors have been used in almost all MEMS
nanopositioners to date, as described below. The most intriguing option that has
yet to receive adequate attention is optical sensing, including interferometric [80]
and diffractive optics [81] sensors. With the continued integration of photonics
into MEMS, optical sensors in MEMS nanopositioners are expected over the next
decade.
310 J.J. Gorman

9.4.1 Capacitive Sensors

Capacitive sensing has been integral to a number of MEMS sensor technologies

including accelerometers, gyroscopes, pressure sensors, and force sensors, and their
mechanical and electronic design is well developed. This is because they can achieve
exceptional displacement resolution and are easy to fabricate in standard MEMS
processes. As a result, capacitive sensors are the most common displacement sensors
used in MEMS nanopositioners. Capacitive sensors convert a displacement into a
change in capacitance, which is measured using readout electronics. The variable
capacitor designs that are typically used are the same as the electrostatic actuators
shown in Fig. 9.7: the comb sensor [29, 66, 82–84], parallel-plate sensor [58], and
parallel-plate sensor with comb structure [63, 64]. Similar to electrostatic actuators,
the comb sensor has a linear relationship between capacitance and displacement,
whereas parallel-plate designs are nonlinear. Since sensor linearity is critical in
nanopositioning, the comb sensor is highly preferred. Sub-nanometer resolution has
been achieved using both single-ended comb sensors, with [29, 66] and without [83]
motion amplification, and with differential comb sensors [84]. Comb sensors have
many of the same advantages as comb actuators, including large range, flat dynamic
response, high bandwidth, and ease of fabrication.
The readout electronics that measure the change in capacitance when the
nanopositioner moves are just as important as the structure of the sensor. The most
sensitive readout electronics for capacitive sensors are AC bridge and switched
capacitor circuits [85]. These circuits can achieve a noise floor better than 0.1 aF/rt-
Hz, making it possible to reach sub-nanometer resolution, and a dynamic range
above 105 , which are ideal characteristics for nanopositioning. Furthermore, these
circuits are compatible with CMOS electronics [85, 86], making it possible to
integrate the MEMS and electronics on a single chip or co-integrate in a single
package.
As with electrostatic actuators, the main disadvantage of comb sensors is the
area required to attain a sufficiently sensitive device. An important recent innovation
for capacitive sensing in nanopositioners is the use of the same comb structure for
both actuation and sensing [87, 88]. This reduces the area required for actuation
and sensing by half, thereby increasing its bandwidth by reducing the mass of the
nanopositioner. In one configuration, the actuation signal and the AC signal used to
measure the comb capacitance are separated in the frequency domain and filtering is
used to lock in on the sensor signal [87]. The frequency of the AC signal is selected
to be well above the mechanical resonances of the structure so that it does not impart
a measurable motion. Alternatively, the comb has been used as a tunable capacitor
within a resonant LC circuit [88]. The resonant frequency of the circuit is modulated
by the motion of the nanopositioner, which can be calibrated to attain a sensitivity
in Hz/nm. The change in frequency of the resonator can be measured using a phased
locked loop, providing a high bandwidth sensor output. Actuation and sensing on
the same comb has yet to be compared to the performance of independent actuation
and sensing channels and warrants further investigation.
9 MEMS Nanopositioners 311

9.4.2 Thermal Sensors

Over the last decade, thermal sensors have emerged as a viable alternative to
capacitive sensors for displacement measurement in MEMS [71, 72, 83, 89–92]. In
its simplest form, a thermal sensor uses heat conduction between an electrothermal
heater and a moving heat sink, typically the motion stage, as a measure of position.
The heater is most commonly made of doped silicon and is shaped to evenly
distribute the heat over the area of the heater that interacts with the motion stage.
The heater and the motion stage interact through a small gap (<5 m) that conducts
heat away from the heater. As the overlapping area between the motion stage
and heater changes due to displacement, the temperature of the heater changes,
resulting in a change in the resistance of the heater. The change in resistance is
measured with readout electronics and the output voltage is calibrated using an
accurate measurement of the motion of the nanopositioner. The sensor and motion
stage can be stacked vertically [72, 89, 91] or they can interact in the plane of
motion [71, 83, 90]. In most cases, the thermal sensor and actuator moving the
motion stage are separated to avoid coupling of their dynamics. However, it is
possible to intentionally couple an electrothermal actuator to a thermal sensor,
which significantly reduces the area consumed by the mechanism [91].
Thermal sensors are typically measured in pairs in which the overlapping area
increases for one heater and decreases for the other for a given motion of the stage.
Assuming a constant voltage across the heaters, the current flowing through one
heater increases and the other decreases. By measuring the difference between the
two signals, sensor drift due to environmental temperature fluctuations and sensor
aging can be canceled out. This differential configuration of thermal sensors for
measuring the motion of a moving stage is shown in Fig. 9.12. Trans-impedance
amplifiers are used to convert the sensor currents to voltage, which are then
subtracted from each other to attain the drift-free sensor output. Interestingly, when
directly compared to capacitive sensing, thermal sensors yielded a lower noise floor
above 4 Hz in one implementation [83], demonstrating the applicability of this
approach to nanopositioning. Thermal sensors, similar to the piezoresistive sensors
discussed in the next subsection, have high 1/f noise compared to capacitive sensors
due to fluctuations in the conductivity of the heaters. This issue has been addressed
through AC modulation of the heater voltage and measuring the output voltage with
a lock-in amplifier at the modulation frequency, which has been shown to improve
the noise floor at low frequency by a factor of 2 [92].
Among the many promising features of thermal sensors, they have been shown
to be capable of a resolution of 0.5 nm, a motion range of tens of micrometers,
and 10 kHz bandwidth [89]. Additionally, the majority of the sensors to date are
no bigger than 400 m  200 m, which is considerably smaller than capacitive
sensors. The fabrication of most existing thermal sensor designs is straightforward,
particularly the in-plane design shown in Fig. 9.12 since it uses a standard silicon-
on-insulator process. Sensor linearity is design dependent, with some demonstrating
high linearity [89] and others with strong nonlinearity [71]. The largest drawback of
312 J.J. Gorman

Fig. 9.12 Thermal displacement sensor with differential readout electronics. (a) Schematic of
approach for differential readout of two thermal sensors, (b) scanning electron micrographs of
the thermal sensors, motion stage, and the electrothermal actuator that moves the stage (Zhu et al.
© IEEE [90])

thermal sensors is that they act as heat sources that can cause significant asymmetric
thermal expansion in the mechanism. MEMS nanopositioners require excellent
motion accuracy and stability and thermal expansion could degrade performance
in these areas. This issue requires further study to determine if it will impact MEMS
nanopositioner applications.

9.4.3 Piezoresistive Sensors

Piezoresistive sensing was one of the first sensing mechanisms to be used in MEMS
and has been integral to many of the most common sensor products including
9 MEMS Nanopositioners 313

pressure sensors, accelerometers, gyroscopes, and microphones [93]. Piezoresistors

are similar to metal strain gauges in that their resistance changes due stress.
However, metal strain gauges change resistance only due to geometric changes
caused by the stress. Piezoresistors, which are fabricated from semiconducting
materials, most commonly silicon, also change resistance due to a stress-induced
change in the bandgap of the material, which is typically a much larger effect than
the change due to geometric effects. Piezoresistive sensors are frequently used in
low-precision macroscale nanopositioners because they are compact, allowing them
to be directly bonded to piezoelectric stack actuators. Unfortunately, their resolution
is typically one to two orders of magnitude worse than capacitive sensors. The
first MEMS nanopositioners to use piezoresistive sensors were cantilever arrays
designed for SPM [15–18]. One example is shown in Fig. 9.4, in which four
piezoresistors are located at the base of each AFM cantilever within the array. The
piezoresistors are embedded in the cantilever through selective doping of silicon.
This approach measures the induced strain due to the out-of-plane motion of the
cantilever tip rather than measuring the tip motion directly.
Measurement of in-plane motion using piezoresistive sensors has been achieved
by using suspended silicon structures that extend or compress when the nanopo-
sitioner moves. For example, long slender beams were used as piezoresistors
to measure the displacement of a motion stage over a 12 m range [94]. The
piezoresistors were fabricated in the same silicon layer as the nanopositioner, as
shown in Fig. 9.13, greatly simplifying fabrication since actuation and sensing are
formed in the same etch step and selective doping is not required. Similar to the
simultaneous actuation and sensing approach used with electrostatic comb actuators
[87, 88], piezoresistive sensing can be combined with electrothermal actuation
[70, 95]. The beams in a chevron electrothermal actuator experience a change in
resistance during actuation due to stress and a change in resistivity due to the
change in temperature. As a result, the change in resistance can be measured during
actuation to attain the actuator displacement. However, due to coupling between
actuation and sensing, the response is nonlinear and it is not straightforward to
separate these coupling effects.
Another approach for integrating piezoresistors into flexural structures uses a
CMOS–MEMS fabrication process that is also compatible with making independent
electrothermal actuators [23, 24], as described in the previous section. In this
case, the piezoresistors are composed of polysilicon and are embedded in a silicon
dioxide structural layer. Both in-plane and out-of-plane motion can be measured by
strategically placing the piezoresistors to measure the strain related to the desired
degree of freedom.
There are only a handful of cases where piezoresistive sensing has been applied
successfully to MEMS nanopositioners so it is difficult to quantify the expected
resolution and range. Hafizovic et al. [18] demonstrated a 1 nm resolution over a
1 m range, while Bazaei et al. [94] achieved 5 nm resolution over a range of 12 m.
While the resolution is promising, the dynamic range reported to date is quite low,
with the best result among the included references being 8000. The resolution is
generally limited by the high level of 1/f noise and Johnson noise that is universal
314 J.J. Gorman

Fig. 9.13 MEMS nanopositioner with piezoresistive sensors. (a) Scanning electron micrograph
showing thin silicon piezoresistors (Rp1 and Rp2 ) connected to the motion stage, (b) Schematic
describing the strain in each piezoresistor due to stage motion (Bazaei et al. © IEEE [94])

to all resistors. The use of AC readout electronics can eliminate the contribution
from the 1/f noise, but the Johnson noise is still a major contributor to the noise
floor.
The biggest advantages of piezoresistive sensors are their small size, thereby
minimizing the overall area consumed by a mechanism, and their high bandwidth,
which is typically limited by the structural dynamics of the flexure mechanism
and the readout electronics. Attaining piezoresistors with a high gauge factor (i.e.,
high sensitivity) typically requires selective ion implantation or diffusion doping,
complicating the fabrication process. However, adequate displacement resolution
9 MEMS Nanopositioners 315

has been achieved without selective doping [94], so a high gauge factor may not
be required in some applications. Thermal drift is another well-known problem
with piezoresistors that could be a perceived issue for MEMS nanopositioners.
However, this problem is easily mitigated by using a full Wheatstone bridge to
readout the resistance change and by including reference piezoresistors in the design
that do not experience stress. In general, piezoresistive sensors are an intriguing
approach for displacement measurement in MEMS nanopositioners that requires
more research to determine whether sub-nanometer resolution can be achieved over
tens of micrometers of motion.

9.5 Performance Assessment

A number of different MEMS nanopositioners have been described in the previous

sections. An open question is whether these mechanisms can perform as well as
macroscale nanopositioners. Although this is a difficult question to answer since
many of the performance characteristics of interest are rarely reported, this section
provides an assessment of the performance of existing MEMS nanopositioners
and compares the results with the typical performance achieved at the macroscale.
The focus of this comparison is on the aggregate rather than highlighting the top
performing mechanisms and does not examine the effect of actuator and sensor
selection on performance. The following subsection discusses the methodology for
the assessment and is followed by the presentation and discussion of the results.

9.5.1 Methodology

Four performance metrics have been selected for this assessment: range, resolution,
bandwidth, and in-plane area. Data was cataloged from the relevant references
previously cited in this chapter and [96–103], with 59 references providing usable
data points. A minimum of the in-plane area and one other metric were required for a
reference to be included in this analysis. The range is defined as either a stated value
or the maximum displacement presented in the reference. Resolution is defined as
either the stated RMS resolution or is calculated from the displacement noise floor
in m/Hz1/2 using the bandwidth. Due to 1/f noise and other noise sources, there
can be a considerable difference between the RMS noise and that calculated using
the noise floor and bandwidth. However, this is unavoidable for this assessment
since resolution is so rarely stated explicitly. The bandwidth is taken as the lowest
value among the stated bandwidth, the half-power point of the amplitude frequency
response, and the first resonant frequency. Finally, the in-plane area is typically
extracted from provided images and is defined as the width multiplied by the length.
For mechanisms with multiple degrees of freedom, the width and length are divided
by the degrees of freedom to provide a fair comparison with single axis devices.
316 J.J. Gorman

Fig. 9.14 Range as a function of in-plane mechanism area for MEMS nanopositioners

9.5.2 Results

Among the four metrics of interest, range is the most widely reported. Comparing
motion range versus in-plane area, as shown in Fig. 9.14, there is a wide distribution
in both of these metrics. Most MEMS nanopositioners have a range below 30 m
with some exceptions going to 100 m and beyond (some outliers were not included
in the figures). This is shorter than the average range for macroscale nanopositioners
but is expected for MEMS and 30 m is sufficient for many of the applications
of interest. One clear trend is that larger area is required for the longest-range
mechanisms, which is obvious since more force and larger flexure mechanisms are
required for large range. More interesting is that below 10 m the same range is
achieved by mechanisms ranging in area from 104 to 107 m2 . This is due in part to
the trade-offs in actuator selection and the common use of flexure mechanisms that
are not optimized for the expected range, resulting in excessively large dimensions.
Resolution is the single most important performance metric since it determines
whether a motion stage can truly achieve nanoscale motion precision. Unfortunately,
achieving a clear assessment of resolution is challenging since multiple definitions
are used including the RMS value of the displacement noise and the displacement
noise floor. Furthermore, it is commonly not addressed in publications; less than half
of the references providing range data also include resolution. Finally, independent
confirmations of resolution, as done with a laser interferometer at the macroscale,
are rare. Even with these challenges, there are some useful insights attained by
looking at resolution as a function of in-plane area, as shown in Fig. 9.15. Similar
to the range data, there is a wide distribution for resolution, spanning more than two
orders of magnitude. Most of the available data is for larger mechanisms (1 mm2
or larger) but there is no obvious trend indicating a relationship between scale and
9 MEMS Nanopositioners 317

Fig. 9.15 Resolution as a function of in-plane mechanism area for MEMS nanopositioners

resolution. In terms of magnitude, there are a handful of papers reporting sub-

nanometer resolution but the majority of mechanisms are above 2 nm, making it
difficult to define them as nanopositioners.
As previously discussed, the dynamic range (DR) (i.e., range/resolution) for
macroscale nanopositioners is typically greater than 105 . Assuming that the range is
below 30 m as described in the previous paragraph and a DR D 105 , the resolution
should be less than 0.3 nm as indicated by the dashed line in Fig. 9.15. It is clear
that only a few reported designs can perform at this level. Taking another look at
this issue, range versus resolution is plotted in Fig. 9.16 along with contour lines
for dynamic range. The majority of reported designs perform below DR D 104 and
no MEMS nanopositioner considered in this analysis has achieved a DR  105 . This
is the single greatest limitation in the state of the art. The resolution needs to be
improved to meet that achieved at the macroscale while also increasing the dynamic
range beyond 105 (i.e., improve resolution while not decreasing range).
As expected, bandwidth is the metric in which MEMS nanopositioners show
exceptional performance, as shown in Fig. 9.17. Most reported mechanisms have
bandwidth greater than 100 Hz and more than half have demonstrated bandwidth
greater than 1 kHz. This bandwidth range is sufficient for many applications and
can clearly be further optimized considering some examples above 10 kHz have
been demonstrated. The data in Fig. 9.17 shows that, as expected, bandwidth is
linked to in-plane area with the greatest bandwidth occurring on the low end of
the spectrum (200 kHz at 6  104 m2 ) and the smallest bandwidth at the high end
(20 Hz at 7  106 m2 ). Based on Figs. 9.14, 9.15, and 9.17, the majority of MEMS
nanopositioners to date have an area greater than 2.5  105 m2 (i.e., equivalent to
a 500 m square). Reduction in this area would improve bandwidth and facilitate
large-scale integration of arrays for scanning probe applications.
318 J.J. Gorman

Fig. 9.16 Range as a function of resolution for MEMS nanopositioners. DR dynamic range

Fig. 9.17 Bandwidth as a function of in-plane mechanism area for MEMS nanopositioners

While the presented data and analysis is clearly not complete it does provide
some important insights into the state of the art. The range and bandwidth of existing
mechanisms are adequate for known applications and in some cases the bandwidth is
exceptional compared to non-MEMS approaches. Unfortunately, the resolution and
dynamic range found in this analysis are both worse than that found in macroscale
nanopositioners by an order of magnitude or more in most cases. This provides a
clear direction for future research with a goal of achieving resolution <0.1 nm RMS
and DR > 105 .
9 MEMS Nanopositioners 319

9.6 Conclusion

MEMS nanopositioners have emerged over the last two decades as a highly capable
class of mechanisms that merges precision engineering design principles with
MEMS sensors and actuators. They are already being used successfully in SPM and
probe-based data storage, with products expected within the next few years, and
hold promise in many other applications (e.g., nanomanufacturing, nanomaterial
testing, biomedical, photonics). This chapter has provided an overview of the critical
components within these mechanisms, provided discussion on the advantages and
disadvantages of the most common actuators and sensors, and presented a meta-
analysis of their performance to date.
Among the possible actuator technologies of interest, electrostatic, electrother-
mal, and piezoelectric actuators were identified as the most compatible and widely
used with MEMS nanopositioners. Piezoelectric actuators still require substan-
tial research to reach the current performance levels achieved by the other two
approaches, particularly with respect to piezoelectric material deposition and
actuator design. Selection between electrostatic and electrothermal actuation can
be determined almost solely on bandwidth requirements. Bandwidth above a few
hundred hertz requires electrostatic actuation in most cases because the thermal
response of electrothermal actuators is typically too slow. Otherwise, electrothermal
actuators are preferred over electrostatic actuators when their low bandwidth can be
tolerated because they require lower actuation voltage, consume less area, and are
more reliable.
The most commonly used sensing approaches for MEMS nanopositioners are
capacitive, thermal, and piezoresistive sensors. While capacitive sensing has been
most commonly used in precision MEMS applications, thermal and piezoresistive
sensing have been shown to achieve resolution and range performance that is
approaching that of capacitive sensing in some situations. Furthermore, unlike the
actuators discussed above, all three approaches provide adequate bandwidth for
almost all nanopositioning applications. As a result, sensor selection must be based
on available in-plane area (capacitive sensing takes the most area), the intended
fabrication process (piezoresistive sensing is maximized by local ion implantation
or diffusion doping), and the operating environment (thermal sensors are less
sensitive in vacuum), among other factors. A detailed study of the three sensor types
that compares their performance limits under equivalent constraints and operating
conditions would be highly beneficial to future nanopositioner designs.
The presented performance assessment of existing MEMS nanopositioners
revealed two deficiencies in the state of the art. First, the achievable resolution in
the majority of devices reported in the study is worse by an order of magnitude or
more in comparison to macroscale nanopositioners. A resolution of 0.1 nm RMS is
commonly achieved in macroscale nanopositioners but most MEMS nanoposition-
ers are above 1 nm. Second, the dynamic range among the studied mechanisms was
found to be an order of magnitude worse than their macroscale counterparts (>105 )
in most cases. Somewhat less critical, it was found that that the majority of MEMS
320 J.J. Gorman

nanopositioners have an in-plane area greater than 2.5  105 m2 , or a 500 m
square. This size limitation should be improved because it currently restricts the
capabilities of nanopositioner arrays due to poor fill factor. The other performance
metrics of interest, range, and bandwidth have been shown to be adequate in many
of the existing mechanisms.
Looking to the future, research should be focused on improving resolution and
dynamic range since these metrics are the best indicators of usability in nanoposi-
tioning applications and the state of the art is inadequate. If 0.1 nm RMS resolution
and a dynamic range above 105 can be achieved, MEMS nanopositioners will be
capable of replacing macroscale nanopositioners in scanning probe applications.
Another area that warrants more attention is the integration of scanner arrays
with independent XYZ motion. Previous work on parallel operation has focused
on cantilever arrays with a single degree of freedom in each cantilever. Recent
developments in on-chip AFMs [24] have opened the door for independent scanning
probe arrays that will be significantly faster and more fault tolerant than existing
cantilever arrays. Embedded integration of the mechanisms and electronics will be
critical to the success of such systems. Other important areas of research include
analysis of positioning stability, cross talk between axes, and the limitations of
actuation and readout electronics. With sustained efforts in these research areas, it is
expected that MEMS nanopositioners will outperform macroscale nanopositioners
within the following decade.

References

1. J.W. Ryu, D.-G. Gweon, K.S. Moon, Optimal design of a flexure hinge based XY® wafer
stage. Precis. Eng. 21, 18–28 (1997)
2. P. Gao, S.-M. Swei, Z. Yuan, A new piezodriven precision micropositioning stage utilizing
flexure hinges. Nanotechnology 10, 394–398 (1999)
3. A. Elmustafa, M.G. Lagally, Flexural-hinge guided motion nanopositioner stage for precision
machining: finite element simulations. Precis. Eng. 25, 77–81 (2001)
4. http://www.physikinstrumente.com
5. http://www.npoint.com
6. http://www.aerotech.com
7. S. Salapaka, A. Sebastian, J.P. Cleveland, M.V. Salapaka, High bandwidth nano-positioner: a
robust control approach. Rev. Sci. Instrum. 73, 3232–3241 (2002)
8. C.-W. Lee, S.-W. Kim, An ultraprecision stage for alignment of wafers in advanced
microlithography. Precis. Eng. 21, 113–122 (1997)
9. S. Akamine, T.R. Albrecht, M.J. Zdeblick, C.F. Quate, Microfabricated scanning tunneling
microscope. IEEE Electron Device Lett. 10, 490–492 (1989)
10. E. Meyer, H.J. Hug, R. Bennwitz, Scanning Probe Microscopy: The Lab on a Tip (Springer,
New York, 2004)
11. J.J. Yao, S.C. Arney, N.C. MacDonald, Fabrication of high frequency two-dimensional
nanoactuators for scanned probe devices. J. Microelectromech. Syst. 1, 14–22 (1992)
12. Y. Xu, N.C. MacDonald, S.A. Miller, Integrated micro-scanning tunneling microscope. Appl.
Phys. Lett. 67, 2305–2307 (1995)
9 MEMS Nanopositioners 321

13. S.A. Miller, K.L. Turner, N.C. MacDonald, Microelectromechanical scanning probe instru-
ments for array architectures. Rev. Sci. Instrum. 68, 4155–4162 (1997)
14. S.C. Minne, J.D. Adams, G. Yaralioglu, S.R. Manalis, A. Atalar, C.F. Quate, Centimeter scale
atomic force microscope imaging and lithography. Appl. Phys. Lett. 73, 1742–1744 (1998)
15. S.C. Minne, G. Yaralioglu, S.R. Manalis, J.D. Adams, J. Zesch, A. Atalar, C.F. Quate,
Automated parallel high-speed atomic force microscopy. Appl. Phys. Lett. 72, 2340–2342
(1998)
16. D. Lange, T. Akiyama, C. Hagleitner, A. Tonin, H.R. Hidber, P. Niedermann, U. Staufer,
N.F. de Rooij, O. Brand, H. Baltes, Parallel scanning AFM with on-chip circuitry in CMOS
technology, in Proceedings of IEEE MEMS (IEEE, New York, 1999), pp. 447–452
17. T. Volden, M. Zimmermann, D. Lange, O. Brand, H. Baltes, Dynamics of CMOS-based
thermally actuated cantilever arrays for force microscopy. Sensors Actuators A 115, 516–522
(2004)
18. S. Hafizovic, D. Barrettino, T. Volden, J. Sedivy, K.U. Kirstein, O. Brand, A. Hierlemann,
Single chip mechatronic microsystem for surface imaging and force response studies. Proc.
Natl. Acad. Sci. 101, 17011–17015 (2004)
19. A.G. Fowler, A.N. Laskovski, A.C. Hammond, S.O.R. Moheimani, A 2-DOF electrostatically
actuated MEMS nanopositioner for on-chip AFM. J. Microelectromech. Syst. 21, 771–773
(2012)
20. A. Mohammadi, A.G. Fowler, Y.K. Yong, S.O.R. Moheimani, A feedback controlled MEMS
nanopositioner for on-chip high-speed AFM. J. Microelectromech. Syst. 23, 610–619 (2014)
21. M. Maroufi, A. Bazaei, S.O.R. Moheimani, A high-bandwidth MEMS nanopositioner for
on-chip AFM: design, characterization, and control. IEEE Trans. Control Syst. Technol. 23,
504–512 (2015)
22. M. Maroufi, A.G. Fowler, A. Bazaei, S.O.R. Moheimani, High-stroke silicon-on-insulator
MEMS nanopositioner: control design for non-raster scan atomic force microscopy. Rev. Sci.
Instrum. 86, 023705 (2015)
23. N. Sarkar, G. Lee, R.R. Mansour, CMOS-MEMS dynamic FM atomic force microscope, in
Proceedings of Transducers (IEEE, New York, 2013), pp. 916–919
24. N. Sarkar, D. Strathearn, G. Lee, M. Olfat, R.R. Mansour, A 0.25 mm3 atomic
force microscope on-a-chip, in Proceedings of IEEE MEMS (IEEE, New York, 2015),
pp. 732–735
25. E. Eleftheriou et al., Millipede – a MEMS-based scanning-probe data-storage system. IEEE
Trans. Magn. 39, 938–945 (2003)
26. Y. Ahn, T. Ono, M. Esahi, Si multiprobes integrated with lateral actuators for independent
scanning probe applications. J. Micromech. Microeng. 15, 1224–1229 (2005)
27. N. Sarkar, C. Baur, E. Stach, Z. Jandric, R. Stallcup, M. Ellis et al., Modular MEMS
experimental platform for transmission electron microscopy, in Proceedings of IEEE MEMS
(IEEE, New York, 2006), pp. 146–149
28. J.J. Gorman, Y.-S. Kim, A.E. Vladar, N.G. Dagalakis, Design of an on-chip microscale
nanoassembly system. Int. J. Nanomanuf. 1, 710–721 (2007)
29. X. Liu, J. Tong, Y. Sun, A millimeter-sized nanomanipulator with sub-nanometer positioning
resolution and large force output. Smart Mater. Struct. 16, 1742–1750 (2007)
30. S. Lu, D.A. Dikin, S. Zhang, F.T. Fisher, J. Lee, R.S. Ruoff, Realization of nanoscale
resolution with a micromachined thermally actuated testing stage. Rev. Sci. Instrum. 75,
2154–2162 (2004)
31. Y. Zhu, H.D. Espinosa, An electromechanical material testing system for in situ electron
microscopy and applications. Proc. Natl. Acad. Sci. 102, 14503–14508 (2005)
32. B. Pant, B.L. Allen, T. Zhu, K. Gall, O.N. Pierron, A versatile microelectromechanical system
for nanomechanical testing. Appl. Phys. Lett. 98, 053506 (2011)
33. S.T. Smith, D.G. Chetwynd, Foundations of Ultra-Precision Mechanism Design (CRC Press,
Boca Raton, 1992)
34. S.T. Smith, Flexures: Elements of Elastic Mechanisms (Taylor & Francis, London, 2000)
322 J.J. Gorman

35. R. Legtenberg, A.W. Groeneveld, M. Elwenspoek, Comb-drive actuators for large displace-
ments. J. Micromech. Microeng. 6, 320–329 (1996)
36. V.P. Jaecklin, C. Linder, N.F. de Rooij, J.M. Moret, R. Bischof, F. Rudolf, Novel polysilicon
comb actuators for XY-stages, in Proceedings of IEEE MEMS (IEEE, New York, 1992),
pp. 147–149
37. D. Kobayashi, T. Hirano, T. Furuhata, H. Fujita, An integrated lateral tunneling unit, in
Proceedings of IEEE MEMS (IEEE, New York, 1992), pp. 214–219
38. M.I. Lutwyche, Y. Wada, Manufacture of micromechanical scanning tunnelling microscopes
for observation of the tip apex in a transmission electron microscope. Sensors Actuators A
48, 127–136 (1995)
39. C.S.B. Lee, S. Han, N.C. MacDonald, Single crystal silicon (SCS) XY-stage fabricated by
DRIE and IR alignment, in Proceedings of IEEE MEMS (IEEE, New York, 2000), pp. 28–33
40. C.-H. Kim, Y.-K. Kim, Micro XY-stage using silicon on a glass substrate. J. Micromech.
Microeng. 12, 103–107 (2002)
41. J. Dong, D. Mukhopadhyay, P.M. Ferreira, Design, fabrication and testing of a silicon-
on-insulator (SOI) MEMS parallel kinematics XY stage. J. Micromech. Microeng. 17,
1154–1161 (2007)
42. Y.-S. Kim, J.-M. Yoo, S.H. Yang, Y.-M. Choi, N.G. Dagalakis, S.K. Gupta, Design,
fabrication and testing of a serial kinematic MEMS XY stage for multifinger manipulation. J.
Micromech. Microeng. 22, 085029 (2012)
43. X. Liu, K. Kim, Y. Sun, A MEMS stage for 3-axis nanopositioning. J. Micromech. Microeng.
17, 1796–1802 (2007)
44. J. Dong, P.M. Ferreira, Electrostatically actuated cantilever with SOI-MEMS parallel kine-
matic XY stage. J. Microelectromech. Syst. 18, 641–651 (2009)
45. Y.S. Kim, N.G. Dagalakis, S.K. Gupta, Design of MEMS based three-axis motion stage by
incorporating a nested structure. J. Micromech. Microeng. 24, 075009 (2014)
46. D. Mukhopadhyay, J. Dong, E. Pengwang, P. Ferreira, A SOI-MEMS-based 3-DOF planar
parallel-kinematics nanopositioning stage. Sens. Actuators A 147, 340–351 (2008)
47. S.-C. Chen, M.L. Culpepper, Design of a six-axis micro-scale nanopositioner—HexFlex.
Precis. Eng. 30, 314–324 (2006)
48. S.D. Senturia, Microsystem Design (Springer, New York, 2001)
49. C. Liu, Foundations of MEMS (Prentice Hall, Upper Saddle River, 2011)
50. D.J. Bell, T.J. Lu, N.A. Fleck, S.M. Spearing, MEMS actuators and sensors: observations on
their performance and selection for purpose. J. Micromech. Microeng. 15, S153–S164 (2005)
51. N.B. Hubbard, M.L. Culpepper, L.L. Howell, Actuators for micropositioners and nanoposi-
tioners. Appl. Mech. Rev. 59, 324–334 (2006)
52. R.C. Batra, M. Porfiri, D. Spinello, Review of modeling electrostatically actuated microelec-
tromechanical systems. Smart Mater. Struct. 16, R23–R31 (2007)
53. Y.-M. Choi, J.J. Gorman, N.G. Dagalakis, S.H. Yang, Y.-S. Kim, J.-M. Yoo, A high-bandwidth
electromagnetic MEMS motion stage for scanning applications. J. Micromech. Microeng. 22,
105012 (2012)
54. J.D. Grade, H. Jerman, T.W. Kenny, Design of large deflection electrostatic actuators. J.
Microelectromech. Syst. 12, 335–343 (2003)
55. M. Olfatnia, S. Sood, J.J. Gorman, S. Awtar, Large stroke electrostatic comb-drive actuators
enabled by a novel flexure mechanism. J. Microelectromech. Syst. 22, 483–494 (2013)
56. T. Akiyama, U. Staufer, N.F. de Rooij, Atomic force microscopy using an integrated comb-
shape electrostatic actuator for high-speed feedback motion. Appl. Phys. Lett. 76, 3139–3141
(2000)
57. J.L.A. Yeh, C.-Y. Hui, N.C. Tien, Electrostatic model for an asymmetric combdrive. J.
Microelectromech. Syst. 9, 126–135 (2000)
58. M.S.C. Lu, G.K. Fedder, Position control of parallel-plate microactuators for probe-based
data storage. J. Microelectromech. Syst. 13, 759–769 (2004)
59. A.G. Onaran, M. Balantekin, W. Lee, W.L. Hughes, B.A. Buchine, R.O. Guldiken et al.,
A new atomic force microscope probe with force sensing integrated readout and active tip.
Rev. Sci. Instrum. 77, 023501 (2006)
9 MEMS Nanopositioners 323

60. E.C.M. Disseldorp, F.C. Tabak, A.J. Katan, M.B.S. Hesselberth, T.H. Oosterkamp, J.W.M.
Frenken et al., MEMS-based high speed scanning probe microscopy. Rev. Sci. Instrum. 81,
043702 (2010)
61. F.C. Tabak, E.C.M. Disseldorp, G.H. Wortel, A.J. Katan, M.B.S. Hesselberth, T.H. Oost-
erkamp et al., MEMS-based fast scanning probe microscopes. Ultramicroscopy 110, 599–604
(2010)
62. E.S. Hung, S.D. Senturia, Extending the travel range of analog-tuned electrostatic actuators.
J. Microelectromech. Syst. 8(4), 497–505 (1999)
63. D.A. Horsley, N. Wongkomet, R. Horowitz, A.P. Pisano, Precision positioning using a
microfabricated electrostatic actuator. IEEE Trans. Magn. 35, 993–999 (1999)
64. Y. Sun, D. Piyabongkarn, A. Sezen, B.J. Nelson, R. Rajamani, A high-aspect-ratio two-axis
electrostatic microactuator with extended travel range. Sens. Actuators A 102, 49–60 (2002)
65. R. Cragun, L.L. Howell, Linear thermomechanical microactuators, in Proceedings of ASME
IMECE (ASME, New York, 1999), pp. 181–188
66. L.L. Chu, Y.B. Gianchandani, A micromachined 2D positioner with electrothermal actuation
and sub-nanometer capacitive sensing. J. Micromech. Microeng. 13, 279–285 (2003)
67. N.B. Hubbard, L.L. Howell, Design and characterization of a dual-stage, thermally actuated
nanopositioner. J. Micromech. Microeng. 15, 1482–1493 (2005)
68. S. ergna, J.J. Gorman, N.G. Dagalakis, Design and modeling of thermally actuated MEMS
nanopositioners, in Proceedings of ASME International Mechanical Engineering Congress
and Exposition (ASME, New York, 2005), pp. 561–568
69. J.J. Gorman, Y.-S. Kim, N.G. Dagalakis, Control of MEMS nanopositioners with nano-scale
resolution, in Proceedings of ASME International Mechanical Engineering Congress and
Exposition (ASME, New York, 2006), pp. 151–159
70. R.K. Messenger, Q.T. Aten, T.W. McLain, L.L. Howell, Piezoresistive feedback control of a
MEMS thermal actuator. J. Microelectromech. Syst. 18, 1267–1278 (2009)
71. Y. Zhu, A. Bazaei, S.O.R. Moheimani, M.R. Yuce, Design, modeling, and control of
a micromachined nanopositioner with integrated electrothermal actuation and sensing. J.
Microelectromech. Syst. 20, 711–719 (2011)
72. M. Rakotondrabe, A.G. Fowler, S.O.R. Moheimani, Control of a novel 2-DoF MEMS
nanopositioner with electrothermal actuation and sensing. IEEE Trans. Control Syst. Technol.
22, 1486–1497 (2014)
73. J.H. Comtois, V.M. Bright, M.W. Phipps, Thermal microactuators for surface-
micromachining processes, in Proceedings of SPIE 2642, Micromachined Devices and
Components (SPIE, Bellingham, 1995), pp. 10–21
74. D. Yan, A. Khajepour, R. Mansour, Design and modeling of a MEMS bidirectional vertical
thermal actuator. J. Micromech. Microeng. 14, 841–850 (2004)
75. T. Akiyama, U. Staufer, N.F. de Rooij, Fast driving technique for integrated thermal bimorph
actuator toward high-throughput atomic-force microscopy. Rev. Sci. Instrum. 73, 2643–2646
(2002)
76. G. Vitellaro, G. L’Episcopo, C. Trigona, B. Ando, S. Baglio, A compliant MEMS device
for out-of-plane displacements with thermo-electric actuation. J. Microelectromech. Syst. 23,
661–671 (2014)
77. P.J. Gilgunn, L. Jingwei, N. Sarkar, G.K. Fedder, CMOS-MEMS lateral electrothermal
actuators. J. Microelectromech. Syst. 17, 103–114 (2008)
78. D.O. Popa, B.H. Kang, J.T. Wen, H.E. Stephanou, G. Skidmore, A. Geisberger, Dynamic
modeling and input shaping of thermal bimorph MEMS actuators, in Proceedings of IEEE
ICRA (IEEE, New York, 2003), pp. 1470–1475
79. K.R. Oldham, J.S. Pulskamp, R.G. Polcawich, M. Dubey, Thin-Film PZT lateral actuators
with extended stroke. J. Microelectromech. Syst. 17, 890–899 (2008)
80. M.W. Pruessner, T.H. Stievater, W.S. Rabinovich, In-plane microelectromechanical resonator
with integrated Fabry–Pérot cavity. Appl. Phys. Lett. 92, 081101 (2008)
81. U. Krishnamoorthy, R.H. Olsson III, G.R. Bogart, M.S. Baker, D.W. Carr, T.P. Swiler, P.J.
Clews, In-plane MEMS-based nano-g accelerometer with sub-wavelength optical resonant
sensor. Sens. Actuator A 145–146, 283–290 (2008)
324 J.J. Gorman

82. L. Ji, Y. Zhu, S.O.R. Moheimani, M.R. Yuce, A micromachined 2DOF nanopositioner with
integrated capacitive displacement sensor, in Proceedings of IEEE Sensors (IEEE, New York,
2010), pp. 1464–1467
83. Y. Zhu, S.O.R. Moheimani, M.R. Yuce, Simultaneous capacitive and electrothermal position
sensing in a micromachined nanopositioner. IEEE Electron Device Lett. 32, 1146–1148
(2011)
84. B. Koo, X. Zhang, J. Dong, S.M. Salapaka, P.M. Ferreira, A 2 degree-of-freedom SOI-MEMS
translation stage with closed-loop positioning. J. Microelectromech. Syst. 21, 13–22 (2012)
85. N. Yazdi, H. Kulah, K. Najafi, Precision readout circuits for capacitive microaccelerometers,
in Proceedings of IEEE Sensors (IEEE, New York, 2004), pp. 28–31
86. J. Wu, G.K. Fedder, L.R. Carley, A low-noise low-offset capacitive sensing amplifier for a 50
m/rt-Hz monolithic CMOS MEMS accelerometer. IEEE J. Solid-State Circuits 39, 722–730
(2004)
87. J. Dong, P.M. Ferreira, Simultaneous actuation and displacement sensing for electrostatic
drives. J. Micromech. Microeng. 18, 035011 (2008)
88. S.I. Moore, S.O.R. Moheimani, Displacement measurement with a self-sensing MEMS
electrostatic drive. J. Microelectromech. Syst. 23, 511–513 (2014)
89. M.A. Lantz, G.K. Binnig, M. Despont, U. Drechsler, A micromechanical thermal displace-
ment sensor with nanometre resolution. Nanotechnology 16, 1089–1094 (2005)
90. Y. Zhu, A. Bazaei, S.O.R. Moheimani, M.R. Yuce, A micromachined nanopositioner with on-
chip electrothermal actuation and sensing. IEEE Electron Device Lett. 31, 1161–1163 (2010)
91. J. Chow, Y. Lai, Displacement sensing of a micro-electro-thermal actuator using a monolithi-
cally integrated thermal sensor. Sens. Actuators A 150, 137–143 (2009)
92. A. Mohammadi, M.R. Yuce, S.O.R. Moheimani, A low-flicker-noise MEMS electrothermal
displacement sensing technique. J. Microelectromech. Syst. 21, 1279–1281 (2012)
93. A.A. Barlian, W.-T. Park, J.R. Mallon Jr., A.J. Rastegar, B.L. Pruitt, Review: semiconductor
piezoresistance for microsystems. Proc. IEEE 97, 513–552 (2009)
94. A. Bazaei, M. Maroufi, A. Mohammadi, S.O.R. Moheimani, Displacement sensing with
silicon flexures in MEMS nanopositioners. J. Microelectromech. Syst. 23, 502–504 (2014)
95. J. Ouyang, Y. Zhu, Z-shaped MEMS thermal actuators: piezoresistive self-sensing and
preliminary results for feedback control. J. Microelectromech. Syst. 21, 596–604 (2012)
96. Y. Ando, T. Ikehara, S. Matsumoto, Design, fabrication and testing of new comb actuators
realizing three-dimensional continuous motions. Sens. Actuators A 97–98, 579–586 (2002)
97. C.-H. Kim, H.-M. Jeong, J.-U. Jeon, Y.-K. Kim, Silicon micro XY-stage with a large area
shuttle and no-etching holes for SPM-based data storage. J. Microelectromech. Syst. 12,
470–478 (2003)
98. Y. Ando, Development of three-dimensional electrostatic stages for scanning probe micro-
scope. Sensors Actuators A 114, 285–291 (2004)
99. L. Gu, X. Li, H. Bao, B. Liu, Y. Wang, M. Liu et al., Single-wafer-processed nano-positioning
XY-stages with trench-sidewall micromachining technology. J. Micromech. Microeng. 16,
1349–1357 (2006)
100. X. Chen, D.-W. Lee, Integrated microactuation scanning probe microscopy system. J. Vac.
Sci. Technol. B 27, 1408–1412 (2009)
101. Y.-S. Kim, N.G. Dagalakis, S.K. Gupta, Creating large out-of-plane displacement electrother-
mal motion stage by incorporating beams with step features. J. Micromech. Microeng. 23,
055008 (2013)
102. A.G. Fowler, A. Bazaei, S.O.R. Moheimani, Design and analysis of nonuniformly shaped
heaters for improved MEMS-based electrothermal displacement sensing. J. Microelec-
tromech. Syst. 22, 687–694 (2013)
103. X. Zhang, B. Koo, S.M. Salapaka, J. Dong, P.M. Ferreira, Robust control of a MEMS probing
device. IEEE/ASME Trans. Mechatron. 19, 100–108 (2014)
Chapter 10
Control Issues of MEMS Nanopositioning
Devices

Y. Zhu, S.O.R. Moheimani, M.R. Yuce, and A. Bazaei

Abstract In this chapter, the control issues of microelectromechanical system

(MEMS) nanopositioning devices are introduced and discussed. The real-time
feedback control of a novel micro-machined 1-degree-of-freedom (1-DoF) thermal
nanopositioner with on-chip electrothermal position sensors is presented. The
actuation works based on the thermal expansion of V-shaped silicon beams. The
sensing mechanism works based on measuring the resistance difference between
two electrically biased identical silicon beams. The resistance difference varies
with displacement. The heat conductance of the sensor beams varies oppositely
with the position of the movable stage, resulting in different beam temperatures
and resistances. A pair of position sensors are operated in differential mode to
reduce low-frequency drift. The micro-machined nanopositioner has a nonlinear
static input–output characteristic. The electrothermal actuator has a dynamic range
of 14.4 m and the electrothermal sensor has a low drift of 8.9 nm over 2000 s. An
open-loop controller is first designed and implemented. It is experimentally shown
that uncertainties result in unacceptable positioning performance. Hence, feedback
control is required for accurate positioning. The on-chip displacement sensor is able
to provide high-resolution displacement control. Therefore, a real-time closed-loop
feedback control system is designed using a proportional-integral (PI) controller
together with the nonlinear compensator used for the open-loop control system. The
closed-loop system provides acceptable and robust tracking resolution for a wide
range of set point values. The step response results show a positioning resolution of
7.9 nm and a time constant of 1.6 ms in a 10 m stroke. For triangular reference
tracking, which is required in raster-scanned scanning probe microscopy (SPM), the
steady-state tracking error has a standard deviation of 20 nm within a wide range of
10 m.

Y. Zhu () • S.O.R. Moheimani • M.R. Yuce • A. Bazaei

Griffith University, Nathan, QLD, Australia
e-mail: [email protected]; [email protected]

© Springer International Publishing Switzerland 2016 325

C. Ru et al. (eds.), Nanopositioning Technologies,
DOI 10.1007/978-3-319-23853-1_10
326 Y. Zhu et al.

10.1 Introduction

High precision nanopositioners have been extensively used in many applications

such as scanning tunneling microscopy (STM) [1], atomic force microscopy (AFM)
[2], and emerging ultrahigh density probe storage systems [3–7]. These nanoposi-
tioners typically have high positioning accuracy with a large dynamic range and a
wide bandwidth, enabling fast and robust closed-loop position control. Macro-scale
nanopositioners can achieve nanometre-scale positioning resolution and accuracy,
however they are relatively bulky and costly [8–11]. Microelectromechanical system
(MEMS) nanopositioners have attracted significant interest recently because of their
small size, low cost, fast dynamics, and the emergence of applications such as
probe-based data storage [12, 13], and scanning probe microscopy [14–16]. Closed-
loop feedback control of these positioners is highly desirable if a high degree of
displacement precision is required, and such a control system needs an accurate
source of position information [17–20]. However, many of the MEMS nanoposi-
tioners reported in the literature are not equipped with on-chip sensors due to the
restrictions associated with micro-fabrication processes [21–23]. Thus, the in-plane
movements are often measured by laser reflectance microscopes [24, 25] or optical
microscopes [26], making the footprint of the entire system fairly large. There are
several exceptions in the literature, e.g., embedded on-chip capacitive and piezo-
resistive displacement sensors were integrated in thermally and electrostatically
actuated positioners in [27–30]. Nevertheless, only open-loop results were obtained
and complex fabrication processes were required for electrical insulation between
electrical heating and sensing circuits in [27] and for sidewall implementation of
the piezo-resistors in [29]. In addition, limited sensing resolutions were achieved in
[28] and [29]. Recently, a thermal sensing scheme was used in a probe-based storage
device [31, 32]. Micro-heaters were used to measure the motion of a MEMS micro-
scanner with resolution of less than 1 nm. Compared to comb capacitive sensors,
thermal sensors are much more compact and can be easily integrated with actuators
in a MEMS device [33–35]. In [36, 37], off-chip electromagnetic coil actuators were
adopted for scanner actuation along with the thermal sensing approach reported
in [31, 32], where a complex mass-balanced structure was designed for vibration
resistance purposes.
In this chapter, a novel electrothermal position sensor is integrated with an
electrothermal actuator in the same MEMS chip without the need for inclusion of
extra electrical insulation fabrication process as reported in [27], or assembling two
chips as reported in [36, 37]. Although the electrothermal actuators consume more
power compared with other MEMS actuators, for example, electrostatic actuator and
piezoelectric actuators, they do offer certain advantages. They can operate under low
voltages, generate large forces, and enjoy a high degree of vibration resistance due
to their stiff structures [38].
This chapter introduces a MEMS nanopositioner with on-chip electrothermal
actuation and electrothermal sensing, which was micro-fabricated in a low-cost
bulk silicon process. Measurement results show that the positioner has a dynamic
10 Control Issues of MEMS Nanopositioning Devices 327

range of 14.4 m and an open-loop bandwidth of 101 Hz. To reduce the low-
frequency thermal drift, the sensors are operated in a pair and measured by a
differential circuitry. The sensor was measured to have a small low-frequency drift
of 8.9 nm over 2000 s. The on-chip displacement sensing enables a feedback control
capability. A model of the positioner is derived and a proportional-integral (PI)
feedback controller is implemented digitally in a dSPACE rapid prototyping system
to investigate the closed-loop performance of the positioner. Open-loop and closed-
loop performance of the system in tracking a step displacement were investigated.
The closed-loop step response results show a steady-state RMS positioning error of
8.6 nm, while the corresponding value of the open-loop seek operation can reach
up to 1180 nm. High tracking performance was also achieved for a 10 Hz triangular
reference by using a 2-DoF feedback control system consisting of a PI controller
and a prefilter.
The chapter is organized as follows: In Sect. 10.2, the electrothermal positioning
scheme and its modeling are introduced. Micro-fabrication process and design
parameters are described in Sect. 10.3. Section 10.4 presents the experimental
setup and system characterization of the positioner. Open-loop static and dynamic
responses are presented to verify that the positioner behaves as predicted. Sec-
tion 10.5 presents the details of the open-loop and closed-loop control design. The
tracking performances of step and triangle references are presented in this section.
Section 10.6 concludes this chapter.

10.2 Design and Modeling of Micro-Machined

Nanopositioner

10.2.1 Design

The proposed nanopositioner consists of electrothermal actuator and sensor on a

single silicon chip. On-chip actuation and sensing enable a compact positioning sys-
tem with small footprint and high resolution. Figure 10.1 illustrates the conceptual
schematic view of the nanopositioner. The positioner stage is actuated by a chevron
electrothermal actuator. To move stage, dc current is applied across the heaters
to heat up the silicon beams. Compared to other actuation methods, for example,
electrostatic actuation and piezoelectric actuation, the electrothermal actuators are
simple to implement with low actuation voltage and large force. Thanks to the high
stiffness of the structures, complex mass-balanced structures as proposed in [37]
are not required to resist vibration. As shown in Fig. 10.1, a pair of thin hot arms
are designed to have an angle of 0.57ı with respect to each other in the V-shaped
electrothermal actuator. The operating voltage range is typically from 0 to 15 V
in chevron actuators, which depends on the beam geometries. The displacement is
proportional to the square of actuation voltage and the largest stroke of the actuator
is limited by the buckling of the hot arms at high temperatures (>700 ı C) [25].
328 Y. Zhu et al.

Fig. 10.1 Schematic diagram of the nanopositioner with electrothermal actuator, electrothermal
position sensor, and differential amplifier: (a) before actuation; (b) after actuation [39]

The position sensors are based on electrothermal principle. Two beam-shaped

resistive heaters R1 and R2 are made from n-doped silicon (2 m wide, 100 m
long, and 25 m thick). The positioner stage (rectangular plate) is placed close to
the beam heaters with a 2 m air gap. Initially, the positioner stage is at the rest
position before applying voltage across the actuator, as illustrated in Fig. 10.1a. At
this position, the two edges of the sink plate are precisely aligned to the center of
the two thermal resistive sensors R1 and R2 . To heat up the sensors, a dc voltage
source Vdc is applied to both sensors (heaters). Consequently, the heat generated in
the resistive heaters is conducted through the 2 m air gap to the heat sink plate
(positioner stage). At the rest position in Fig. 10.1a, the ambient condition for both
sensors is identical; therefore, the dissipated heat fluxed out of the sensors is same.
Thus, it equalizes the temperature and resistance of the two sensors.
In Fig. 10.1b, a voltage is applied on the actuator beams to move the positioner
stage towards left. As a consequence, the ambient condition is changed to be differ-
ent. The heat fluxes dissipated out of the both sensors become different, resulting in
a corresponding difference in the temperatures of both sensors. Also, the resistance
is the function of temperature in silicon material. Therefore the resistance of sensor
R1 and sensor R2 will be different. Thus, the positioner stage’s displacement can be
10 Control Issues of MEMS Nanopositioning Devices 329

sensed by measuring the resistance changes between the two sensors. With a fixed
dc voltage Vdc applied to both sensors, the differential changes of the resistance
result in current variations in the beam resistors because of Ohm’s law. The currents
are then converted to an output voltage using a pair of trans-impedance amplifiers
and an instrumentation amplifier. To suppress the common-mode noise, the gains
of these two trans-impedance amplifiers must be well matched by adjusting the
feedback resistance of the trans-impedance amplifiers. Employing the differential
topology allows the sensor output to be immune from undesirable drift effects due
to changes in ambient temperature or aging effects.

10.2.2 Modeling

This section presents a large-signal analysis for the anticipation of temperature

and resistance values of the heaters in terms of a pre-specified displacement and
bias voltage. We adopt a lumped parameter approach and static (time-invariant)
conditions to simplify the analysis [40]. Each sensing resistor is regarded as
an infinitesimal solid element with no internal thermal convection. With this
assumption, the resistor temperature is presented by a scalar variable rather than
a spatial distribution, which would require solving a complicated boundary value
PDE problem. In addition, using this assumption, we can combine a number of heat
transfer coefficients between the object and various parts in its surrounding such as
heat sink, air gap, and conducting pads, into an overall thermal resistance coefficient.
Applying Newton’s law of cooling, the transfer of the heat from the object to air is
mainly governed by thermal conduction in a thin air layer beside the object instead
of thermal convection. Although thermal convection can participate in the transfer
of heat outside the foregoing thin layer, its overall affect in stationary conditions can
be modeled as an equivalent thermal resistance between the object and air. In this
way, each sensing resistor has a uniform temperature distribution. We now proceed
to a lumped parameter analysis for temperature prediction of heaters.
Assuming a constant temperature coefficient ˛(1/ı C), the electrical resistance
R of an infinitesimal resistor at temperature T is described by Ro Œ1 C ˛ .T  To /,
where To and Ro () describe the ambient temperature and the resistance value
at room temperature, respectively. Let the overall thermal resistance between the
resistor and the surrounding area, including the thermal conduction through its
electrical wiring to a voltage source, be K1 (ı C/W). Assuming a constant bias
voltage V across the resistor and a stationary condition, the electrical power
delivered to the resistor should be equal to the heat power delivered by the resistor
to the ambient area:

V2
D K .T  To / (10.1)
Ro Œ1 C ˛ .T  To /
330 Y. Zhu et al.

Fig. 10.2 Schematic diagram

of the biased electrothermal
position sensors [39]

Thanks to the simplifying assumption that constant ˛ does not vary with the
temperature, we can obtain a closed-form solution for the temperature T from Eq.
(10.1). Hence, the temperature and electrical resistance of the resistor at bias voltage
V are determined by the following expressions:
q
4˛V 2
1C 1C Ro K
R D Ro (10.2)
2
q
4˛V 2
1C Ro K
1
T D To C (10.3)
2˛

Before applying Eqs. (10.2) and (10.3) to the heaters in Fig. 10.2, we need to
introduce individual heat resistances K1 1 and K2 1 to the ambient area for each
electric resistors R1 and R2 , respectively. A horizontal displacement of the heat
sink plate in the left direction will increase the heat conductance of the left heater,
whilst reducing that of the right heater. These monotonic variations of the heat
conductance values with displacement can roughly be modeled by the following
linear relationships:

K1 .Qx/ D Kmin C .Kmax  Kmin / xQ ;

(10.4)
K2 .Qx/ D Kmin C .Kmax  Kmin / .1  xQ /

where xQ  x=L describes a dimensionless position variable, L is the heater length,

and Kmin 1 and Kmax 1 represent the maximum and minimum overall heat resistance
of each heater, respectively. We can now use Eqs. (10.2) and (10.3) to predict the
stationary values of temperature and resistance for each heater:
10 Control Issues of MEMS Nanopositioning Devices 331

Table 10.1 Electrothermal Parameters Values

sensor parameters and values
for simulation ˛(K1 ) 3.7822  104
Ro () 390
To (ı C) 27
Kmin (W/K) 8  105
Kmax (W/K) 16  105
V(V) 6

q
4˛V 2
1C 1C Ro Ki .Qx/
Ri .Qx; V/ D Ro ;
2
q (10.5)
2
1 C R4˛V
o Ki .Q
x/
1
Ti .Qx; V/ D To C ; i 2 f1; 2g
2˛
where Ro is value of unbiased heater electrical resistance at room temperature. When
the heat sink displacement is zero.Qx D 0:5/; we have assumed an identical heat
conductance K1 D K2 D .Kmax C Kmin / =2 for the heaters. Hence, the sensing
resistors are also identical when the displacement is zero. In this way, the output
voltage in Fig. 10.1 is proportional to R1 1
1  R 2 . This result is used in the following
simulation to convert the amount of heat sink displacement into a dimensionless
variable Ro R1 1
1  R2 , which is proportional to the output voltage in Fig. 10.1.
Considering the values reported in Table 10.1 for model parameters and applying
Eqs. (10.4) and (10.5), the stationary temperature, resistance, and sensor output
values as a function of heat sink displacement are obtained for each sensing heater
as depicted in Fig. 10.3a–c, respectively. In spite of the nonlinear variation of the
individual resistance values with displacement, the sensor output demonstrates a
highly linear relationship with the displacement.

10.3 Microfabrication

The device was fabricated in a commercial silicon-on-insulator (SOI) Micromachin-

ing process in a MEMS foundry (MEMSCAP) [41]. This process allows fabrication
of silicon structures with a fixed 25 m of thickness and a minimum 2 m of width
and gap. The final device structure is illustrated in Fig. 10.4, and the fabrication
process is briefly described as follows:
1. Microfabrication starts with a highly doped n-type 25 m silicon device layer,
on which metal pads (20 nm Cr and 500 nm Au) are deposited by e-beam
evaporation and patterned through lift-off process. The metal pads are to provide
ohmic contact for probes and/or bonding wires.
2. Silicon is lithographically patterned to define the structures. Then deep reactive
ion etch (DRIE) is applied from the front side of the wafer to define both the
anchored and movable features of the structure.
332 Y. Zhu et al.

a b Resistances c Sensor output

850 1.32
0.06
T1 R1 / Ro Ro(R-1
1
-R-1
2
)
800 1.3
R /R 0.04
T2 2 o
Temperature (Centigrade)

750 1.28
0.02
700 1.26
0
650 1.24
-0.02
600 1.22
-0.04
550 1.2

-0.06
500 1.18
0.2 0.4 0.6 0.8 0.2 0.4 0.6 0.8 0.2 0.4 0.6 0.8
x/L x/L x/L

Fig. 10.3 Simulated steady-state values versus displacement for (a) resistor temperatures,
(b) resistance values, and (c) normalized sensor output [39]

Fig. 10.4 SOIMUMPs

microfabrication structures
from MEMSCAP [41]

3. A protective polyimide layer is applied to the top surface of the silicon layer to
protect the silicon structures during the following etching process.
4. Another DRIE is applied from the back side of the wafer to etch completely
through the substrate in defined areas, stopping on the buried oxide layer. Then a
trench underneath the movable structures is created.
5. The exposed buried oxide is removed using a wet oxide etch process (HF
etching).
6. The polyimide coat on the front side is removed by oxygen plasma, thereby
allowing the movable structure to be fully released. The front side oxide layer is
removed using a vapor HF process. Then a large contact metal pad (i.e. electrical
grounded contact) is patterned on the substrate to reduce parasitics.
The scanning electron microscope (SEM) image of the whole device and a close-
up section of it are provided in Fig. 10.5.
To reduce the thermal coupling effects from the electro-thermal actuators to
thermal sensors, a number of holes were embedded in the center shuttle between
10 Control Issues of MEMS Nanopositioning Devices 333

Fig. 10.5 SEM images of the micro-machined nanopositioner. The close-up section shows the
details of a pair of electrothermal sensors, and positioner stage as a heat sink for the sensors [39]

Table 10.2 Design parameters and values

Parameters Values
Thermal actuator beam width (m) 5
Thermal actuator beam length (m) 2  400
Thermal actuator beam angle (degree) 0.57
Thermal actuator beam quantity (pair) 10
Suspension beam width (m) 5
Suspension beam length (m) 400
Suspension beam angle (degree) 0
Suspension beam quantity (pair) 3
Thermal sensor beam width (m) 2
Thermal sensor beam length (m) 100
Gap between thermal sensor and heat sink plate (m) 2
Heat sink plate dimension (m  m) 200  200
Device thickness (m) 25
Measured resistance of thermal sensor R1 at room temperature () 388.9
Measured resistance of thermal sensor R2 at room temperature () 391.6
Measured resistance of electro-thermal actuator at room temperature () 116

actuators and the heat sink plate. The holes are expected to improve the thermal
convection, thereby thermally insulating the heat sink plate from actuators. The
suspension beams help the thermal insulation as well, due to the heat transfers to
the substrate through the suspension beams and its anchors. Table 10.2 summarizes
the device design parameters and measured beam resistance.
334 Y. Zhu et al.

10.4 Experimental Characterization

The displacement of the nanopositioner was calibrated using a PolytecTM Planar

Motion Analyzer (PMA-400, Polytec GmbH). To accurately characterize the in-
plane displacement of the positioner stage, digital image capturing and analysis
methods were utilized. The PMA has an ordinary speed camera with a limited
exposure time of 82 ms, however it can measure periodic processes at frequencies
as high as 1 MHz by using the stroboscopic principle [42]. Step response is
essential to characterize the dynamic performance of the positioner. Time constant
and displacement can be obtained through step response measurement. Therefore,
step voltages were applied to the actuator periodically, and the PMA recorded the
displacement time response in each period, as shown in Fig. 10.6. The step time
response of the actuator has been characterized using different actuation voltages
of 8.5 and 9.0 V to exam the step displacement and time constant. As illustrated in
Fig. 10.7, after 20 ms of actuation, the positioner stage reached the stable positions
of 13.6 and 14.4 m with 8.5 and 9.0 V step actuation voltages, respectively. The
time constant is approximately 1.6 ms for both cases.
To control the positioner stage with high resolution, the static characterization
of the actuator and the sensor must be implemented accurately. Particularly, two
calibration curves are required: displacement versus actuation voltage curve for
the actuator and output voltage versus displacement curve for the sensor. Accurate
displacement measurement is crucial for the above two calibration curves. There-
fore, PMA is utilized to measure the in-plane displacement. The static displacement
versus actuation voltage measured data with a fitted fifth order polynomial is
illustrated in Fig. 10.8a. With the actuation voltage of 9 V, the electrothermal

Fig. 10.6 Displacement measurement using PolytecTM Planar Motion Analyzer (PMA-400,
Polytec GmbH)
10 Control Issues of MEMS Nanopositioning Devices 335

Fig. 10.7 Time response of the electrothermal actuator with step actuation voltages of 8.5 and
9.0 V

Fig. 10.8 Experimental calibration curves: (a) displacement versus actuation voltage for actuator;
(b) output voltage versus displacement for sensor

actuator can achieve a maximum displacement of 14.4 m, which is matched well
with the step response measurement in Fig. 10.7. Due to the nature of the thermal
expansion, the actuator can only move forward in one direction after application of
a voltage. However, the actuator can be biased to an initial position by a dc offset
voltage, such that it can move forward or backward by increasing or decreasing the
applied voltage. For instance, for a 10 m travel range with bidirectional movement,
the actuator can be actuated by a 4.5 V dc voltage to the middle position (5 m) of
the desired range.
To calibrate the position sensor, the instrumentation amplifier outputs were
recorded at every actuation voltage, as illustrated in Fig 10.8b. The sensors were
336 Y. Zhu et al.

Fig. 10.9 Optical

micrograph of the position
sensor’s illumination with an
applied voltage of 10 V to the
left silicon beam. No voltage
was applied to the right
sensor for comparison
purpose [39]

Fig. 10.10 Output voltage

versus displacement with
different sensor bias voltages
of 6 and 7 V [39]

biased with 6 V and the instrumentation amplifier gain was set at 90.3 V/V. With
the bias voltage of 6 V, a power of 60 mW is consumed in each sensor and the
sensitivity is 0.27 mV/nm with a displacement up to 6 m. As shown in Fig. 10.8b,
the sensitivity drops when the displacement is more than 6 m. A possible reason is
that the temperature distribution is not uniform in the thermal sensor beams. Because
of the heat loss at the anchors of the sensor beams, the center of the beam is hotter
than its two ends, as described in [43].
To visualize the temperature distribution, a 10 V dc voltage was applied to the left
sensor in Fig. 10.9. Visible light was observed in the center of the silicon beam while
the two ends of the sensor beams have no light. The right sensor is dark because no
voltage was applied to it for the comparison purpose. It was observed that 10 V is
safe for the silicon beam without damage. However, the bias voltage should be far
less than 10 V for the sensor’s reliability reason. Therefore, the sensor is calibrated
with 6 V bias voltage.
It will be of interest to investigate the effect on the sensor sensitivity from the
bias voltage. The relationships between positioner displacement and sensor output
at various sensor biases are illustrated in Fig. 10.10. The sensitivity can be increased
by increasing the sensor bias voltage. However, the nonlinear nature of the response
could not be eliminated by increasing the bias voltage.
10 Control Issues of MEMS Nanopositioning Devices 337

Fig. 10.11 Measured

resistance versus voltage with
no bias voltage applied to the
sensor pair

As actuator and sensor are on a single silicon chip, we cannot assume that the
displacement does not have any effect on the temperature of the area around
the sensor resistors. To investigate the thermal coupling from actuator to sensor,
the sensor resistances were measured using multimeter without applying an external
bias voltage. As shown in Fig. 10.11, it is observed that the un-biased sensor
resistances increase by about 6.4 % for a 12 m displacement, just due to transfer
of heat from actuator to sensor. The difference between the two curves in Fig. 10.11
is because of the variations and errors in beam dimensions during microfabrication.
Dynamic characterization is crucial for control implementation. HP35670A
spectrum analyzer was utilized to characterize the positioner’s dynamic perfor-
mance. To characterize the frequency response, the nanopositioner can be regarded
as an input–output system, which is connected to the spectrum analyzer. The input
actuation voltage versus sensor output voltage curve can be obtained from the
measured data in Fig. 10.8. The fitted fifth order polynomial calibration curve is
illustrated in Fig. 10.12. To drive the actuator at high frequency, a dc voltage of
4.5 V superimposed with an ac voltage of 1 V was applied to the electrothermal
actuator. Swept sinusoidal measurements were carried out from 1 Hz to 51.2 kHz
by applying the input voltage to the actuator and sensing from the on-chip thermal
sensors. The frequency response, illustrated in Fig. 10.13, shows that the open-loop
bandwidth of the positioner is 101 Hz, which is relatively low compared to the
thermal structures fabricated by PolyMUMPs [44]. The reason is because of the
fact that the heat transfer to surrounding air in SOIMUMPs is much slower than that
to the substrate in PolyMUMPs. However, the bandwidth of 101 Hz is satisfactory
for some positioning applications.
Figure 10.14 shows the open-loop sensor drift when the positioner is at rest,
which was measured at the output of the instrumentation amplifier over a period of
2000 s under normal laboratory conditions. Low-frequency drift can be calculated
338 Y. Zhu et al.

Fig. 10.12 Experimental calibration curve: actuation voltage versus sensor output voltage for the
overall system including actuator and sensor

Fig. 10.13 Measured

open-loop frequency response
of the MEMS nanopositioner
using HP35670A spectrum
analyzer

by the moving average of the measured data. Thanks to the differential sensing of
the sensor pair, the open-loop amplifier output has a low drift of 2.4 mV over 2000 s,
which corresponds to 8.9 nm displacement.

10.5 Control Implementation

Having known the static and dynamic parameters of electrothermal actuators and
sensors, proper control strategies can be implemented to achieve high positioning
resolution and stability for the MEMS nanopositioner. In this section, open-loop and
closed-loop position control strategies are investigated to control the MEMS posi-
tioner to have a positioning resolution in nanometre range. A dSPACE-1103 rapid
10 Control Issues of MEMS Nanopositioning Devices 339

Fig. 10.14 Experimental results of sensor drift (open-loop) over 2000 s

Fig. 10.15 Block diagram of the open-loop control system including input and output look up
tables, reference, electrothermal actuator and sensor [39]

prototyping system was used for real-time implementation of the controllers and
®
data acquisition. The dSPACE system is fully programmable from the Simulink
block diagram in Matlab environment.

10.5.1 Open-Loop Control

The block diagram of the open-loop control system is shown in Fig. 10.15. Accurate
look-up-tables are required for fast computation of static nonlinear input–output
mappings. For a specific input value to the block, it generates an output value
based on interpolation or extrapolation among the recorded data values. The look-
up-table at the input of the electro-thermal actuator converts the reference signal
in micrometers to an appropriate actuation voltage, as illustrated in Fig. 10.16a,
which was obtained using PMA as described in Sect. 10.5. However, the number
and precision of measured data points are not enough to provide a smooth curve.
Therefore, a fifth order polynomial was fitted to the measured data in Fig. 10.16a to
generate the data for the input look-up-table. In a similar way, the look-up-table at
the output of the electrothermal actuator was obtained to convert the sensor output
voltage to displacement in micrometers, as shown in Fig. 10.16b. For a sensor bias
voltage of 7 V, a third order polynomial was fitted to the measured data points in
Fig. 10.10 to generate smooth and monotonic data for the output look-up-table.
340 Y. Zhu et al.

Fig. 10.16 The data used in input and output look-up-tables [39]

To avoid the infinite slope of the curve around zero displacement in Fig. 10.16a, the
origin for displacement was shifted to C1 micrometer. Then both input and output
look-up-tables were input to Simulink to control the electrothermal actuator.
To evaluate the control capability of the open-loop control method, a staircase
reference signal is utilized as shown in Fig. 10.17. The reference voltage increased
from 0 to 10 V with 2.5 V steps and 50 ms time intervals, and then decreased to
0 V in same manner. Unfortunately, it can be seen in Fig. 10.17 that the open-
loop controller cannot provide an acceptable tracking performance for staircase
displacement. A maximum displacement error of 0.96 m (9.6 % error) was
observed at 10 m stroke. This is due to the plant uncertainties and the drift in
the sensor and electronic circuits. Plant herein refers to the combination of the
electrothermal MEMS device and electronic circuits for actuation and sensing.
Therefore, feedback control strategies have to be considered to compensate the
uncertainties and errors for accurate positioning.
10 Control Issues of MEMS Nanopositioning Devices 341

Fig. 10.17 Tracking performance of open-loop control method for a staircase reference. Large
over-shoot errors are observed

Fig. 10.18 Block diagram of

one-degree-of-freedom
(1DoF) PI control system
including input and output
look up tables, reference,
electrothermal actuator and
sensor, PI controller [39]

10.5.2 Closed-Loop Control

For closed-loop control, we incorporated a proportional-integral controller (PI

controller) in addition to the nonlinear mappings, as illustrated in Fig. 10.18. The
integration part in the PI controller provides a closed-loop unity low frequency gain
form reference to displacement for set point tracking and robustness to uncertainties
and disturbances. The proportional part in the PI controller is utilized to reduce the
overshoot in step response. The integral gain and proportional gain were determined
by experimental measurements. With an integral gain of ki D 700 and a proportional
gain of kp D 1.3, the positioning performance is significantly improved compared to
open-loop case, as shown in Fig. 10.19. Based on this control scheme, a controllable
desired response of 2.5 m steps over a 10 m range was obtained with a maximum
time constant of 1.4 ms. As illustrated in Fig. 10.19, the overshoot issue in open-loop
control was also eliminated.
342 Y. Zhu et al.

Fig. 10.19 Tracking

performance of 1DoF
closed-loop control method
for a staircase reference

7.55
Displacement (µm)

7.5

7.45
0.12 0.125 0.13 0.135 0.14 0.145 0.15
Time (sec)

Fig. 10.20 Close-up view of the 7.5 m closed-loop seek operation in Fig.10.19. RMS positioning
error of 8.6 nm is observed [39]

Figure 10.20 is the close-up view of the 7.5 m closed-loop seek operation in
Fig. 10.19, which indicates a steady-state RMS positioning error of 8.6 nm. As
a comparison, the steady-state RMS positioning error of a similar open-loop seek
operation can reach up to 1180 nm.
To further investigate the tracking capability of the nanopositioner, a 10 Hz
triangular reference covering a 10 m range was applied. The one-degree-of-
freedom (1DoF) PI control system in Fig. 10.18 yielded an acceptable control
performance, as shown in Fig. 10.21 (labeled “No prefilter”). With an integral gain
of ki D 2000 and a proportional gain of kp D 4, the steady-state tracking error has a
standard deviation of 90 nm, which is less than 1 % of the 10 m full range.
The control performance can be further improved with a 2DoF control system
consisting of a PI controller and a prefilter. The look-up-tables provide the nonlinear
mappings described earlier. To provide better stability margins we used lower
10 Control Issues of MEMS Nanopositioning Devices 343

Fig. 10.21 Performance of 1DoF (no prefilter) and 2DoF (with prefilter) control systems in
tracking of a triangular reference. RMS positioning error of 90 nm is observed [39]

gains of ki D 1100 and kp D 2 for the PI controller. This also reduces the effect
of measurement noise on the controlled displacement output due to feedback. The
tracking performance for a 10 Hz triangular reference is shown in Fig. 10.21 (labeled
“With prefilter”). It can be observed that displacement output closely follows a
desired triangular reference within a wide range of 10 m with a standard deviation
of 20 nm.

10.6 Conclusion

The control issues of MEMS nanopositioning devices are introduced and discussed.
A novel micro-machined silicon nanopositioner with on-chip electrothermal actua-
tors and electrothermal sensors has been presented with nanometre resolution and
low drift. The MEMS positioner was embedded in a feedback loop to realize a
precise position control. Due to the nonlinear nature of the electrothermal actuator,
a nonlinear inversion block was added to the feedback loop to linearize the plant.
The experimental results showed that the positioner with a PI controller achieved a
high degree of positioning accuracy with high robustness.
344 Y. Zhu et al.

The electrothermal actuator positions the heat-sink plate, which is a suitable

platform for samples located on it as a load. This one-dimensional positioner is
a preliminary experimental device for the future two-dimensional positioners [45],
which can be used as a nanopositioner stage for scanning probe microscopy. As
the forces induced by thermal expansion of the arms are large, it is expected that
the performance of the positioner would not significantly change if small probes
or samples are attached to the platform. Due to the limitation of the SOIMUMPs
fabrication process, it is not possible to design a two-dimensional thermal-based
positioner in a single device layer. However, the thermal sensors and actuators can
be placed in different layers, which can be fabricated in METALMUMPs process
through MEMSCAP.

Acknowledgment This research is funded by Australian Research Council (ARC) discovery

grant- DP0774287 and the Griffith University Research Infrastructure Program fund from Griffith
University, Australia.

References

1. G. Binnig, H. Rohrer, The scanning tunneling microscope. Sci. Am. 253, 50–56 (1986)
2. G. Binnig, C. Quate, C. Gerber, Atomic force microscope. Phys. Rev. Lett. 56(9), 930–933
(1986)
3. A. Pantazi, A. Sebastian, G. Cherubini, M. Lantz, H. Pozidis, H. Rothuizen, E. Eleftheriou,
Control of MEMS-based scanning-probe data-storage devices. IEEE Trans. Nanotechnol.
15(5), 824–841 (2007)
4. A. Pantazi, M.A. Lantz, G. Cherubini, H. Pozidis, E. Eleftheriou, A servomechanism for a
micro-electro-mechanical-system-based scanning-probe data storage device. Nanotechonology
15(10), 612–621 (2004)
5. G. Binnig, M. Despont, U. Drechsler, W. Häberle, M. Lutwyche, P. Vettiger, H.J. Mamin,
B.W. Chui, T.W. Kenny, Ultrahigh-density atomic force microscopy data storage with erase
capability. Appl. Phys. Lett. 74(9), 1329–1331 (1999)
6. M. Despont, J. Brugger, U. Drechsler, U. Dürig, W. Häberle, VLSI-NEMS chip for parallel
AFM data storage. Sensors Actuators A Phys. 80(2), 100–107 (2000)
7. P. Vettiger, G. Cross, M. Despont, U. Drechsler, U. Dürig, B. Gotsmann, W. Häberle, M. Lantz,
H. Rothuizen, R. Stutz, G. Binnig, The ‘millipede’-nanotechnology entering data storage. IEEE
Trans. Nanotechnol. 1(1), 39–55 (2002)
8. Y.K. Yong, S.S. Aphale, S.O. Moheimani, Design, identification, and control of a flexure-based
XY stage for fast nanoscale positioning. IEEE Trans. Nanotechnol. 8(1), 46–54 (2009)
9. Y.K. Yong, S.S. Aphale, S.O. Moheimani, Atomic force microscopy with a 12-electrode
piezoelectric tube scanner. Rev. Sci. Instrum. 81(3), 033701 (2010)
10. B.J. Kenton, K.K. Leang, Design and control of a three-axis serial-kinematic high-bandwidth
nanopositioner. IEEE/ASME Trans. Mechatron. 17(2), 356–369 (2012)
11. S. Schonhardt, J.G. Korvink, J. Mohr, U. Hollenbach, U. Wallrabe, Optimization of an
electromagnetic comb drive actuator. Sens. Actuators A Phys. 154(2), 212–217 (2009)
12. C.H. Kim, Y.K. Kim, Micro XY-stage using silicon on a glass substrate. J. Micromech.
Microeng. 12(2), 103–107 (2002)
13. C.K. Pang, Y. Lu, J. Chen, H. Zhu, J. Yang, J. Mou, J. Guo, B.M. Chen, T.H. Lee, Design,
fabrication, sensor fusion, and control of a micro X-Y stage media platform for probe-based
storage systems. Mechatronics 19, 1158–1168 (2009)
10 Control Issues of MEMS Nanopositioning Devices 345

14. N. Sarkar, R. Mansour, A CMOS-MEMS scanning probe microscope with integrated position
sensors, in 1st Microsystems and Nanoelectronics Research Conference, 2008. MNRC 2008,
pp. 77–80, 15 October 2008. doi:10.1109/MNRC.2008.4683382
15. N. Sarkar, R.R. Mansour, O. Patange, K. Trainor, CMOS-MEMS atomic force microscope, in
2011 16th International Solid-State Sensors, Actuators and Microsystems Conference (TRANS-
DUCERS), pp. 2610–2613, 5–9 June 2011. doi:10.1109/TRANSDUCERS.2011.5969836
16. D. Barrettino, S. Hafizovic, T. Volden, J. Sedivy, K. Kirstein, A. Hierlemann, H. Baltes, CMOS
monolithic atomic force microscope, in 2004 Symposium on VLSI Circuits. Digest of Technical
Papers, pp. 306–309, 17–19 June 2004. doi:10.1109/VLSIC.2004.1346597
17. S. Devasia, E. Eleftheriou, S.O. Moheimani, A survey of control issues in nanopositioning.
IEEE Trans. Control Syst. Technol. 15(5), 802–823 (2007)
18. A. Sebastian, D. Wiesmann, Modeling and experimental identification of silicon microheater
dynamics: a systems approach. J. Microelectromech. Syst. 17(4), 911–920 (2008)
19. Y.C. Chen, R.T. M’Closkey, T.A. Tran, B. Blaes, A control and signal processing integrated
circuit for the JPL-Boeing micromachined gyroscopes. IEEE Trans. Control Syst. Technol.
13(2), 286–300 (2005)
20. V. Todorov, V. Stavrov, J. Kreuter, Sub nm-resolution static measurement with MEMS
displacement sensors. Procedia Eng. 25, 591–594 (2011)
21. L. Gu, X. Li, H. Bao, B. Liu, Y. Wang, M. Liu, Z. Yong, B. Cheng, Single wafer processed
nanopositioning XY-stage with trench-sidewall micromachining technology. J. Micromech.
Microeng. 16(7), 1349–1357 (2006)
22. C.S.B. Lee, S. Han, N.C. MacDonald, Single crystal silicon (SCS) XY-stage fabricated
by DRIE and IR alignment, in The Thirteenth Annual International Conference on Micro
Electro Mechanical Systems, 2000. MEMS 2000, pp. 28–33, 23–27 January 2000. doi:10.1109/
MEMSYS.2000.838485
23. A.G. Fowler, A.N. Laskovski, A.C. Hammond, S.O. Moheimani, A 2-DoF electrostatically
actuated MEMS nanopositioner for on-chip AFM. J. Microelectromech. Syst. 21(4), 771–773
(2012)
24. J.J. Gorman, Y. Kim, N.G. Dagalakis, Control of MEMS nanopositioners with nano-scale
resolution, in Proceeding of IMECE2006, Chicago, 2006
25. R. Hickey, D. Sameoto, T. Hubbard, M. Kujath, Time and frequency response of two-arm
micromachined thermal actuators. J. Micromech. Microeng. 13(1), 40–46 (2003)
26. Y. Sun, M.A. Greminger, D.P. Potasek, B.J. Nelson, A visually servoed MEMS manipulator,
in Experimental Robotics VIII (Springer, Berlin/Heidelberg, 2003), pp. 255–264
27. L.L. Chu, Y.B. Gianchandani, A micromachined 2D positioner with electrothermal actuation
and sub-nanometer capacitive sensing. J. Micromech. Microeng. 13(2), 279–285 (2003)
28. Y.S. Choi, Y. Zhang, D.W. Lee, A thermal-driven silicon micro xy-stage integrated with
piezoresistive sensors for nanopositioning. J. Micromech. Microeng. 22(5), 055002 (2012)
29. L. Sun, J. Wang, W. Rong, X. Li, H. Bao, A silicon integrated micro nano-positioning XY-stage
for nano-manipulation. J. Micromech. Microeng. 18(12), 125004 (2008)
30. L. Ji, Y. Zhu, S.O. Moheimani, M.R. Yuce, A micromachined 2DOF nanopositioner with
integrated capacitive displacement sensor, in Sensors, 2010 IEEE, Hawai, 2010
31. M.A. Lantz, G.K. Binnig, M. Despont, U. Drechsler, A micromechanical thermal displacement
sensor with nanometre resolution. Nanotechnology 16(8), 1089–1094 (2005)
32. G.K. Binnig, M. Despont, M.A. Lantz, P. Vettiger, Thermal movement sensor. International
Patent WO 2004/020328 A1
33. N.B. Hubbard, M.L. Culpeper, L.L. Howell, Actuators for micropositioners and nanoposition-
ers. Appl. Mech. Rev. Trans. ASME 59(6), 324–334 (2006)
34. Y. Zhu, S.O. Moheimani, M.R. Yuce, Simultaneous capacitive and electrothermal position
sensing in a micromachined nanopositioner. IEEE Electron Device Lett. 32(8), 1146–1148
(2011)
35. Y. Zhu, A. Bazaei, S.O. Moheimani, M.R. Yuce, A micromachined nanopositioner with on-chip
electrothermal actuation and sensing. IEEE Electron Device Lett. 31(10), 1161–1163 (2010)
346 Y. Zhu et al.

36. A. Sebastian, A. Pantazi, S.R. Moheimani, H. Pozidis, E. Eleftheriou, Achieving subnanometer

precision in a MEMS-based storage device during self-servo write process. IEEE Trans.
Nanotechnol. 7(5), 586–595 (2008)
37. M.A. Lantz, H.E. Rothuizen, U. Drechsler, W. Haberle, M. Despont, A vibration resistant
nanopositioner for mobile parallel-probe storage applications. J. Microelectromech. Syst.
16(1), 130–139 (2007)
38. V. Kaajakari, Practical MEMS (NV: Small Gear Publishing, Las Vegas, 2009)
39. Y. Zhu, A. Bazaei, S.R. Moheimani, M.R. Yuce, Design, modeling, and control of a
micromachined nanopositioner with integrated electrothermal actuation and sensing. J. Micro-
electromech. Syst. 20(3), 711–719 (2011)
40. A. Bazaei, Y. Zhu, R. Moheimani, M.R. Yuce, Analysis of nonlinear phenomena in a thermal
micro-actuator with a built-in thermal position sensor. IEEE Sensors J. 12(6), 1772–1784
(2012)
41. [Online]. Available: http://www.memscap.com/en_mumps.html
42. Software Manual – Planar Motion Analyzer Software 2.3, Polytec. [Online]. Available: www.
polytec.com
43. U. Durig, Fundamentals of micromechanical thermoelectric sensors. J. Appl. Phys. 98, 044906
(2005)
44. C.D. Lott, T.W. McLain, J.N. Harb, L.L. Howell, Modeling the thermal behavior of a surface-
micromachined linear-displacement thermomechanical microactuator. Sensors Actuators A
101, 239–250 (2002)
45. M. Rakotondrabe, A.G. Fowler, S.R. Moheimani, Control of a novel 2-DoF MEMS nanopo-
sitioner with electrothermal actuation and sensing. IEEE Trans. Control Syst. Technol. 22(4),
1486–1497 (2014)
Chapter 11
A Review of Nanomanipulation in Scanning
Electron Microscopes

Devin K. Luu, Chaoyang Shi, and Yu Sun

Abstract Nanomanipulation under scanning electron microscope (SEM) imaging

has enabled the characterization of nanomaterials and nanostructures and the proto-
typing/assembly of nanoscale devices. In this chapter, techniques for nanorobotic
manipulation in SEMs are reviewed. Nanomanipulation platforms, nanomanipu-
lation tools, gas injection systems, imaging techniques for automation, and appli-
cations in nanomaterial characterization and nanodevice assembly are discussed.
Many micro-tools, changeable toolboxes, and automation techniques are emerging.
Nanoscale laboratories built around SEMs are becoming a powerful platform to
enable flexible prototyping of nanomaterial-based devices in the nanotechnology,
biotechnology, and nanoelectronics sectors.

11.1 Introduction

Nanomanipulation is a technology for manipulating objects at the nanometer scale

which addresses the demand to conduct fundamental investigations in areas such
as material science, biology, and electronics and to improve industrial procedures
to allow rapid prototyping and inspection of nanoscale devices. For example,
mechanical and electrical properties of nanowires can be measured, and they can be
precisely maneuvered and placed to assemble transistors. Multi-walled nanotubes
can be used as linear or rotational bearings on nanoscale mechanical structures.
Nanomanipulation technology was first demonstrated in 1990, when Donald
Eigler and Erhard Schweizer of IBM used a scanning tunneling microscope (STM)
to position 35 xenon atoms on a metal substrate to form the letters “IBM” [1].
Although individual atoms had been moved before, this marked the first time atoms
were placed deliberately with such precision. The STM used by IBM is in a class
of microscopes called scanning probe microscopes, which physically move a sharp

D.K. Luu • C. Shi • Y. Sun ()

Advanced Micro and Nanosystems Lab, University of Toronto, Toronto, ON, Canada
e-mail: [email protected]

© Springer International Publishing Switzerland 2016 347

C. Ru et al. (eds.), Nanopositioning Technologies,
DOI 10.1007/978-3-319-23853-1_11
348 D.K. Luu et al.

tip over a surface to detect a surface profile. Many studies of nanomanipulation

have been performed using another type of scanning probe microscopes, called
atomic force microscopes (AFM) [2–4]. The advantage of the AFM is that it is
not only capable of atomic scale resolution, but it is also much more compact.
However, the disadvantage of using scanning probe microscopes alone is their
limited motion range and slow image feedback. It can take several minutes to
acquire an image. Even today, performing simultaneous manipulation and imaging
using AFM remains a challenge.
Two other microscope technologies used to image manipulate materials at
the nanometer level are transmission electron microscopes (TEM) and scanning
electron microscopes (SEM). The TEM is capable of real-time, sub-nanometer
resolution imaging, allowing high precision characterization and measurements [5].
However, the specimen chamber and scanning/observation workspace are small,
demanding special sample preparation techniques [6]. The SEM provides real-time
imaging with up to 1 nm resolution and boasts a larger specimen chamber and
scanning area, allowing integration and development of robotic manipulation sys-
tems. These advantages allow simultaneous high resolution imaging, manipulation,
and direct interactions with samples [7]. A comparison of high resolution imaging
technologies is shown in Table 11.1.
Nanomanipulation platforms with multiple degrees of freedom piezoelectric
actuator-based manipulators capable of manipulating nanoscale objects in nanome-
ter resolution have been built specifically inside SEM to perform tasks ranging from
electrical probing [8] to cell nanosurgery [9]. Electron beam induced deposition
(EBID) systems can be combined with nanomanipulation platforms to perform
pick-and-place and assembly tasks [10] or to directly deposit nanostructures. AFM
systems have been integrated into SEM for simultaneous imaging and nanoma-
nipulation, combining the advantages of both instruments to allow electrical and
mechanical characterization of materials, SEM-guided topography imaging as well
as crystallography, and chemistry and surface morphology analysis [11]. Recent
progress also involves the integration of AFMs and focused ion beam (FIB) systems
to form a nanoscale laboratory capable of simultaneous imaging, fabrication,
and nanomanipulation [12]. This chapter reviews existing SEM nanomanipulation
techniques.

Table 11.1 A comparison of high resolution imaging technologies

Typical imaging area Resolution Image acquisition speed
AFM 100  100 m Lateral: depends on tip geometry Minutes
Vertical: 0.1 nm
TEM 3 mm diameter ring 0.1 nm in-plane Milliseconds
SEM 70  50 mm 1 nm in-plane Milliseconds
11 A Review of Nanomanipulation in Scanning Electron Microscopes 349

11.2 Scanning Electron Microscopy Imaging for Automation

The SEM is capable of providing 1 nm resolution visual feedback at up to

25 Hz with a depth of field on the order of micrometers. Not only are these
properties useful for teleoperation of nanomanipulation systems, but they also
make the SEM a promising candidate as a vision sensor for automation. However,
many nanomanipulation systems still operate in open loop without any feedback
mechanisms. Further developments in SEM imaging techniques can enable efficient
and robust automation in SEM nanomanipulation. This section introduces basic
SEM imaging principles and summarizes the methods for imaging de-noising, drift
compensation, depth sensing for automation, and 3D sample reconstruction.

11.2.1 SEM Imaging Basics

The electron’s short wavelength allows it to probe at scales much smaller than
optical microscopes (OM). Where the wavelength of visible light limits OM
resolution to about 200 nm, the resolution of SEM imaging can be as fine as 1 nm.
The depth of field in the SEM is also much higher, as shown in Fig. 11.1 [13]. In
an SEM, an electron gun generates a beam of electrons towards the sample. The
condenser and objective lenses focus the beam to a point on the sample. A beam
deflector controls the beam to scan across the sample. Electrons hitting the surface of
the sample cause secondary or backscatter electrons to be emitted and then detected
by an electron detector, which converts the incoming electrons to signals, sending
them to a video screen as shown in Fig. 11.2.

Fig. 11.1 The OM image, showing only a small cross section of the specimen, has a smaller depth
of field than the SEM image, which shows the entire specimen and the background with clarity [13]
350 D.K. Luu et al.

Fig. 11.2 Operation of the SEM. An electron gun fires an electron beam, which is focused by two
lenses, and is maneuvered by the beam deflector to scan the sample. Electrons are emitted from the
sample and detected by an electron detector, sending a signal to a video screen

Another important aspect of the SEM is its vacuum environment because air
in the SEM chamber can interfere with electron paths. SEMs typically use high
vacuum in the imaging/specimen chamber and use a specimen exchange port to
preserve the vacuum within the SEM specimen chamber when a sample is brought
into or taken out of the specimen chamber. Because the working principle is based
on detecting electrons emitted from the sample, it is important to control specimen
charging. Non-conductive specimens are typically sputter coated with gold to ensure
no charge is built up.

11.2.2 Imaging Modes

The two main imaging modes are secondary electron (SE) mode, showing topo-
graphical information, and backscattered electron (BSE) mode, showing composi-
tion information [14]. SEs are produced when an incoming electron with sufficient
energy collides with the sample to induce emission of electrons. Sharp edges on the
sample emit more SEs while flatter surfaces emit fewer. BSEs are produced when
an electron is emitted towards a sample, and the emitted electron’s path is altered
by the charge of the nuclei in the sample’s atoms, similar to how the path of a comet
is altered by the gravity of a star. Atoms with higher atomic numbers cause more
BSEs. The difference between SE and BSE images is depicted in Fig. 11.3.
11 A Review of Nanomanipulation in Scanning Electron Microscopes 351

Fig. 11.3 The SE image on the left shows topographical information. The BSE image on the right
shows composition of the sample, represented as different colors in the image [14]

Table 11.2 Parameters affecting SEM imaging

Low High
Acceleration voltage • Wider spot size • Narrower spot size
• Lower sample penetration • Higher sample penetration
• Risk of sample damage
Aperture size • Narrower spot size • Wider spot size
• Lower signal-to-noise ratio • Higher signal-to-noise ratio
Beam current • Fewer electrons hit sample • More electrons hit sample
• Lower signal-to-noise ratio • Higher signal-to-noise ratio

11.2.3 Parameters Affecting SEM Imaging

The three main parameters affecting imaging quality are acceleration voltage,
aperture size, and beam current. These settings affect both the spot size of the
electron beam and the number of electrons striking the sample. There is no all-
around best setting, and the operator must tune these parameters in order to obtain
the best possible image for each specific circumstance. A summary of the effects of
changing the above parameters is summarized in Table 11.2.
In general, a smaller spot size gives higher resolution, and more electrons striking
the sample gives a higher signal-to-noise ratio. However, when more electrons
strike the sample, it becomes difficult to focus the electron beam, which can
sacrifice imaging resolution. Although acceleration voltage can be increased to
narrow the spot size, increasing it too much can increase the risk of sample damage,
and too much sample penetration can introduce artifacts into the image via the
352 D.K. Luu et al.

emission of secondary electrons from inside the sample. Therefore, there is always
a trade-off between resolution and noise in SEM imaging. Accelerating voltage and
beam current must also be controlled to limit charging of the sample when low
conductivity samples are imaged in order to limit image drift.

11.2.4 In-plane Imaging

To use SEM images as feedback for visual servoing (i.e., vision-based control),
the image must be stable and clear. There are two main difficulties preventing
accurately locating in-plane features: noise and drift. Noise in SEM images obscures
details in images and is caused by random factors in the ambient environment.
Noise is especially prevalent when imaging delicate samples when the accelerating
voltage and beam current cannot be too high so there is a low signal-to-noise ratio.
Image drift occurs when the entire image shifts, which is caused either by the
specimen physically shifting or the electron beam shifting. Image drift can also
occur if the specimen does not conduct charge away from its surface sufficiently
quickly, or if there are interfering magnetic fields. Low acceleration voltages at high
magnification and short irradiation times can also cause SEM image drift.
Methods for de-noising images such as Gaussian smoothing [15], neighborhood
filter [16], and anisotropic filter [17] often obscure relevant image details. Total
variation minimization [18, 19] and non-local (NL) means de-noising can smooth
out noise and are also effective in preserving sharp edges and details [20, 21]. The
result of the NL-means de-noising algorithm applied on an SEM image is shown in
Fig. 11.4. To compensate for drift, image correlation is used to track predetermined

Fig. 11.4 The NL-means de-noising algorithm is applied on the SEM image. (a) Unprocessed
image. (b) Processed image. Adapted from [22]
11 A Review of Nanomanipulation in Scanning Electron Microscopes 353

Fig. 11.5 Line scanning

technique to speed up
position feedback acquisition

stationary points in an image [22]. If the selected features move, then the image is
shifted back using correlation-based template matching. Gong et al. [22] used the
NL-means method and image correlation to condition SEM visual feedback for an
automated nanoprobing task.

11.2.5 In-plane Imaging Speed

Although the frame rate of SEM imaging can be as high as 25 Hz, automation
and controls applications can benefit from a faster feedback rate. If the geometry
of the object is known, a line scan approach can be used instead of acquiring the
entire image to speed up automated tasks within the SEM, as shown in Fig. 11.5
[23]. Using this approach, Tiemerding et al. tracked moving rectangular features
and achieved a 1 kHz update rate [23].

11.2.6 Depth Sensing in SEM

Depth sensing, which is difficult to achieve in SEM due to its large depth of field,
is important in microscopy applications, especially when automation is required
in 3D because nanomanipulation tools are delicate and easy to break. Hence, it is
imperative that the vertical position of tools is precisely known. Opening the SEM
chamber to replace a broken tool also incurs lengthy pumping in order to restore
vacuum in the chamber. Because the method of depth sensing used in OMs in which
a tool tip is brought into focus on a reference plane of known depth is ineffective
in SEM due to its large depth of field, two main ways to detect depth have been
developed.
354 D.K. Luu et al.

Fig. 11.6 Vibrating bimorph

actuator for contact detection.
When contact is established,
the amplitude of vibration
drops

11.2.6.1 Contact Detection

Contact sensor-based depth sensing uses a probe or force sensor to touch a reference
plane of known depth and can be as accurate as 20 nm [24]. The AFM can be used to
detect contact because it inherently outputs force or height information, such as via
piezoresistive signals [25]. Eichhorn et al. [26] implemented a touch sensor using
a vibrating piezoelectric bimorph actuator with an integrated sensor as shown in
Fig. 11.6. The vibration amplitude distinctly drops upon contact and a sensor signal
drop is detected. However, because the tool tip is connected to the bimorph actuator,
there are potentially unwanted vibrations.
Ru and To [24] implemented a contact detection method by detecting tool tip
sliding. A probe is moved slowly towards a surface and upon contact, sliding
motion is detected from image processing. Electrical connections are also attached
to the probe, which can be used to electrically verify touch. Technologies such as
piezoresistive and capacitive sensors could provide high resolutions and reliability;
however, they are difficult to integrate with nanomanipulation tools [24].

11.2.6.2 Vision-Based Depth Sensing

Vision-based depth sensing utilizes various strategies to determine the depth of the
tool tip from images alone. Kasaya et al. [27] integrated an optical microscope to the
wall of the SEM chamber to detect the depth of the tool tip. Cvetanovic et al. used
a similar approach, but instead of using an optical microscope, an infrared camera
and a scale were fixed to the wall of the chamber [28] as shown in Fig. 11.7.
Stereo image information can be used to determine depth of the tool tip by tilting
the sample stage [29, 30], or by tilting the electron beam [31]. Stereo imaging
in SEM requires careful mechanical design. Tilting the electron beam requires
additional instrumentation near the beam emitter for beam control. Additionally,
correlating two or multiple images must be well handled due to SEM image noise.
Differently, Eichhorn et al. developed a shadow-based depth detection process [32,
33]. However, occlusions and poor shadow contrast from image background cause
difficulties in image segmentation.
11 A Review of Nanomanipulation in Scanning Electron Microscopes 355

Fig. 11.7 A scale is mounted

on the SEM chamber wall,
and an IR camera is used to
detect the tool tip position
relative to the scale [28]

11.2.7 Ultramicrotomy for 3D Image Reconstruction

Ultramicrotomy is a relatively new application in SEM nanomanipulation and can

be used to create 3D reconstructions of blocks of tissue. It is an important process
in histology, the study of the anatomy of cells and other microscopic structures
in plants and animals. In traditional ultramicrotomy, a highly skilled technician
painstakingly takes thin slices of tissue on the order of 70 nm in thickness that are
then collected on a TEM grid from a small container of water attached to the knife.
The technician risks either losing, wrinkling, or mixing the order of the slices.
To address these difficulties, Denk and Horstmann developed a technique called
Serial Block Face Scanning Electron Microscopy (SBFSEM) to automatically
slice and collect ultramicrotome images, and to reconstruct tissue in 3D from
images [34]. One major advantage of SBFSEM besides automation is that the
thin slices are discarded and the freshly exposed block face is imaged, eliminating
the need to collect the delicate sections, as shown in Fig. 11.8. After the cutting
process, computer vision algorithms are used to identify the outlines of structures
in each image, and then they are used to reconstruct the tissue in 3D, as shown
in Fig. 11.9. One exciting research direction following this development is the
mapping of neurons in brains. Researchers have reconstructed 3D blocks of tissue
from SBFSEM data [35] and are now striving for full brain maps [36].
356 D.K. Luu et al.

Fig. 11.8 SBFSEM process. (a) Starting position. (b) The knife moves in to take a slice from
the specimen. (c) The SEM images the exposed block face of the specimen. (d) The specimen
advances to prepare for another slice and the process repeats

Fig. 11.9 Reconstruction of tissue in 3D. (a) Cross sectional images are obtained and stacked
together. (b) Image segmentation is done by software. (c) From the results of image segmentation,
tissue is reconstructed in 3D [35]

11.3 Nanomanipulation Systems

Nanomanipulation systems are platforms integrating one or more nanopositioners

that are used in SEM to perform nanomanipulation tasks. Positioners are typically
driven by piezoceramic actuators, which are both vacuum compatible and antimag-
netic. Nanotools are attached to nanopositioners as “hands” to perform various
tasks. Nanomanipulation systems often incorporate a combination of coarse and
fine positioners to achieve greater range, speed, and accuracy. Coarse positioners
are often based on the slip-stick principle are used to achieve range and speed,
as shown in Fig. 11.10a–c. By expanding or contracting the piezoelectric stack
abruptly, the friction interface between the pusher and moving part slips. Then,
by actuating the piezoelectric stack slowly in the opposite direction, the moving
part “sticks” and pure displacements are produced. However, stick-slip actuation
has relatively low resolution (50 nm), and the driving principle causes vibrations.
When the tool tip is approximately in position, fine positioners, based on flexure
mechanisms, are used to accurately position the tool tip as shown in Fig. 11.10d.
A slip-stick mechanism can also be used instead of a flexure mechanism as a fine
positioner if the piezoelectric actuator is only slowly contracted and expanded, but
the motion range becomes much smaller.
11 A Review of Nanomanipulation in Scanning Electron Microscopes 357

Fig. 11.10 (a–c) The slip stick mechanism works by quickly expanding the piezo stack, then
slowly retracting it. The friction interface between the pusher and the moving part will slip during
the quick expansion, and stick during the slow retraction. (d) The flexure mechanism works as a
lever

Fig. 11.11 Examples of commercial nanomanipulation systems: (a) Zyvex S100. (b) Xidex
NanoBot. (c) Oxford Instruments OmniProbe 400. (d) Toronto Nanoinstrumentation LifeForce-
2000 (e) Imina miBot BT-11 (f) Kleindiek Nanotechnik MM3A

Using the piezoelectric actuator-based positioners, a number of commercial

nanomanipulation systems have been developed, each with different capabilities
such as joystick or game controller teleoperation, closed-loop controlled, in-situ
tool-tip changing, or specimen exchange port compatibility [37–42]. Some of the
commercial nanomanipulation systems are shown in Fig. 11.11 and Table 11.3.

11.3.1 Hybrid AFM/FIB/SEM Systems

A number of integrated systems have emerged to form powerful nanoscale laborato-

ries. SEM and AFM are complementary techniques for performing material surface
investigations [43]. The integration of AFMs inside SEMs combines advantages of
 358

Table 11.3 Summary of commercial nanomanipulation systems

Speed Motion range Axes Coarse resolution Fine resolution
Zyvex S100 3 mm/s 12 mm XYZ 12 100 nm open loop 5 nm open loop
Xidex NanoBot 1 mm/s 12 mm XYZ 6 50 nm open loop 1 nm open loop
NX-2000
Oxford Instruments N/A 4 mm XYZ 4 (1 rotation) 10 nm closed loop N/A
OmniProbe 400
TNI LF-2000 >10 mm/s 10 mm XY 12 <100 nm open loop 0.1 nm open loop
5 mm Z 1 nm closed loop
Imina miBot BT-11 2.5 mm/s Unlimited XY 2 translation 40 nm open loop 0.5 nm open loop
10 mm Z 2 rotation
Kleindiek MM3A 2 mm/s telescoping 12 mm telescoping 1 translation 5 nm open loop rotation N/A
10 mm/s rotation 240ı rotation 2 rotation 0.5 nm open loop
translation
D.K. Luu et al.
11 A Review of Nanomanipulation in Scanning Electron Microscopes 359

both instruments to perform high resolution analysis in real time, revealing details
that may be missed by AFM or SEM alone. FIB is a standard micromachining
tool using FIBs for cutting or depositing materials. Several hybrid AFM/SEM and
AFM/FIB/SEM systems have been developed [3, 4, 12], and commercial systems
are also available [44–48].
Conventional AFMs based on laser beam deflection are not typically integrated
inside a SEM due to the space and optical path constraints. A commercial hybrid
AFM–SEM combination system by DME–SPM was designed with a modified laser
path inside SEM [46], shown in Fig. 11.12a. However, laser alignment in such
a configuration is complicated, and the laser power must be kept low to limit
thermal generation in the vacuum chamber. Attocube Systems AG used a fiber-
optic configuration to construct an in-situ AFM for SEM (attoAFM/SEM) with a
laser interferometer [48], in which the laser source and detector are placed outside
the vacuum chamber to avoid the heating problem, shown in Fig. 11.12b. The optical
fiber, AFM cantilever, and sample are tilted by 45ı , enabling full SEM visual access
to both AFM tip and sample.

Fig. 11.12 Hybrid AFM/SEM and AFM/SEM/FIB systems. (a) AFM–SEM based on laser beam
deflection with modified laser path system by DME-SPM. (b) AFM–SEM with a fiber-optic
configuration by Attocube Systems AG. (c) AFM using tuning fork with a QPlus or Akiyama
probe by Trioptics. (d) 3TB4000 AFM/FIB/SEM from Nanonics Imaging Ltd
360 D.K. Luu et al.

In addition to optical-based methods, AFMs based on self-sensing cantilevers

and tuning forks have also been used for imaging in either contact mode or
tapping mode. Fatikow et al. developed an AFM subsystem using self-sensing
piezoresistive cantilevers integrated inside an SEM to perform both in-situ scanning
and manipulation [3], providing force feedback without using laser measurement.
Dynamic working mode AFMs using tuning fork with a QPlus or Akiyama probe
configuration were also integrated inside SEM to realize surface topography with
a large scanning area of 500 m  500 m [47], as shown in Fig. 11.12c. The
3TB4000 from Nanonics Imaging Ltd is an integration of AFM based on a tailor-
made Q-Plus tuning fork, SEM, and FIB, shown in Fig. 11.12d. This instrument
allows a large field of view imaging from SEM, followed by high resolution AFM
imaging and 3D fabrication with ion beam milling [44]. Fatikow et al. [12] designed
a hybrid AFM/FIB/SEM system to perform similar functions using piezoresistive
cantilevers in contact mode.

11.3.2 Nanotools as End-Effectors for Nanomanipulation

There are many nanotools available for nanomanipulation within SEM. The most
common tasks are nanoassembly, nanomaterial and cell mechanical and electrical
characterization, and cell nanosurgery.
AFM tips can be used as-is or modified to adapt to different applications in SEM
nanomanipulation. Due to commercial availability and existing hardware for precise
position control, hybrid AFM–SEM systems are used in nanomanipulation [3, 4,
12]. The AFM cantilever also inherently provides force feedback, which is useful for
detecting contact with specimens [3]. Combined with the fast visual feedback from
the SEM, AFM is suitable for both automation and teleoperation. A brief overview
of different AFM tips is given in Table 11.4. The modified AFM tips are typically
manufactured by FIB milling or deposition.
Non-contact tuning fork-based AFMs using an oscillating quartz tuning fork
from an ordinary wristwatch have also been used for imaging [55]. In non-contact
mode, the quartz tuning fork with a small probe attached to it is brought close to
the sample while it is resonating, as illustrated in Fig. 11.13a. When the tuning

Table 11.4 Different AFM nanotools and their applications

Tip Usage examples
Original tip • DNA contact pushing [49]
• Carbon nanotube contact pushing [50, 51]
• Nanoparticle contact pushing [2, 3]
Flat tip • Cell stiffness characterization [52]
Needle tip • Cell stiffness characterization [52]
Buffering beam • Cell stiffness characterization [53]
• Single-cell cutting [54]
11 A Review of Nanomanipulation in Scanning Electron Microscopes 361

Fig. 11.13 Non-contact AFM tools: (a) Tuning fork [55]. (b) Q-Plus probe [56]. (c) Tapping mode
AFM tool [57]

fork approaches close enough to the sample, it experiences intermolecular forces,

causing a decrease in vibration amplitude and increase in frequency that can be
detected as a change in a voltage signal. Another non-contact AFM called the Q-
Plus probe fixes one side of the tuning fork such that only one prong vibrates as
shown in Fig. 11.13b [56]. The advantage of the Q-Plus probe is that because there
is only one vibrating prong, the dynamics of the system can be easier to interpret.
Tapping mode AFMs operate with intermittent contacts. The Akiyama probe, a
type of tapping mode AFM tool, is a cantilever bonded to a tuning fork which moves
up and down in a tapping motion as the tuning fork vibrates, as shown in Fig. 11.13c
[57]. Noncontact and tapping mode AFM are more suitable for nondestructive
measurement to perform characterization on delicate samples because the probe
does not physically touch the sample as severely or often as in the contact mode. The
cantilever on the Akiyama probe can also be modified to adjust stiffness, allowing
delicate biological samples to be probed.
A tungsten probe is a piece of metal with a nanometer tip, originally used for
making electrical measurements [58, 59]. At nanometer scales, intermolecular and
electrostatic forces are more dominant [60], making probes a versatile tool for many
nanomanipulation tasks such as pick and place for nanoassembly, especially for
manipulating carbon nanotubes [61] and nanowires [10]. Additionally, microelec-
tromechanical systems (MEMS) micro/nano-grippers have also been developed and
used in SEM nanomanipulation for gripping nanoparticles [62].

11.3.3 Nanotool Changer

Nanomanipulation can require the use of different tool tips. Broken tools also
need to be replaced before manipulation can proceed. Thus, a toolbox array with
different tip morphologies was built, shown in Fig. 11.14a. The tools, called
Nanobits, are fabricated out of Si3 N4 or SiO2 by electron beam lithography and
362 D.K. Luu et al.

Fig. 11.14 Changeable toolbox inside SEM and ESEM: (a) A library of free-standing nitride
Nanobits with different tip morphologies. (b) Silicon dioxide Nanobit. (c) Silicon nitride nanobit
[64]

silicon processes and are 2–4 m long and 120–150 [63, 64]. Two typical tools
are shown in Fig. 11.14b, c. In operation, a microgripper detaches these tools from
the array and assembles them to an AFM probe or other end effectors. Another
nanotool exchange system was proposed in ESEM for biological applications [65].
Low-melting temperature metal was used to attach and hold different tools for
cell characterization and surgery. The Oxford OmniProbe 400 is a commercial
nanomanipulator which is also capable of in-situ tool exchange and probe tip
sharpening for repairing or replacing broken tools without opening the SEM
chamber [39]. Using an in-situ tool changer can reduce the amount of time required
to change a tool tip from hours to minutes.

11.4 Gas Injection for Nanomanipulation

Chemical reactions via gases injected into the SEM chamber can be used to perform
useful assembly tasks. Gases can be used to deposit structures using the EBID
process. Small amounts of gas can also be introduced near a non-conductive sample
to whisk away built-up charges.
11 A Review of Nanomanipulation in Scanning Electron Microscopes 363

11.4.1 Electron Beam Induced Deposition

In EBID, precursor particles adsorbed on a substrate are dissociated by a focused

electron beam, thereby depositing material on the substrate. The resulting deposited
structures are nanometers in scale. EBID’s ability to deposit metals, dielectrics, and
semiconductor materials makes this technology suitable for applications such as
nanoassembly to join materials [10], circuit, or photomask alterations or repairs
[66], and sensor applications. Visual inspection can be performed directly after the
EBID process.
An EBID system consists of two key parts: an electron beam source and a nozzle.
A substrate surface is placed inside a chamber with the nozzle and the electron beam.
The nozzle, a fine capillary connected to a gas source, injects precursor molecules
such as WO3 onto the substrate where they are adsorbed. The electron beam source
emits a directed beam at the locations where deposition is desired, dissociating
the precursor molecules. The deposit (e.g., tungsten) remains on the surface of the
substrate while the dissociated products (e.g., oxygen) are desorbed. A graphical
illustration of the EBID process is shown in Fig. 11.15.
The resolution of EBID in SEM can be as fine as 3 nm [67]. Although the electron
beam can be focused with a resolution finer than 1 nm, the action of dissociation is
actually done by secondary electrons diffracted from the surface of the sample. The
minimum feature size is therefore a function of the electron beam and the spread
of diffracted secondary electrons. The reason for this phenomenon is the so-called
cross-section for the precursor molecules. The cross-section is a measure of how
likely particles are to interact, and secondary electrons possess the optimal energy
to maximize cross-section. For example, the electron beam from an SEM has an
energy of thousands of eV, whereas the secondary electrons only have energies of
tens of eV.

Fig. 11.15 Gas source injects precursor molecules from a small nozzle (e.g., 0.5 mm diameter).
Molecules are adsorbed on substrate. Directed electron beam dissociates precursor molecule.
Particles are deposited on surface of substrate. Dissociated products are desorbed
364 D.K. Luu et al.

Table 11.5 Parameters that affect EBID

Parameter Effect
Beam energy (E) Use high energy for high resolution [69]
Beam current (I) Use high current for high deposition rate [70] according to rate
equation model
Dwell time (tD ) A long dwell time will cause undesired particles that have not
yet been desorbed to be deposited [69]
Replenishment time (tr ) Higher replenishment rate will increase the partial pressure of
the precursor, increasing resolution and deposition rate [71]

Table 11.6 Common EBID precursor materials

Material Remarks
Metal-halogen complexes (WF6 ) Highly toxic and corrosive
Tungsten hexacarbonyl (W(CO)6 ) For tungsten deposition
Metal carbonyls (Me(CO)x ) Metal deposition, carbon contamination
Trimethylcyclopentadienyl platinum (CpPtMe3 ) For platinum deposition

EBID can also deposit microstructures such as nanogranular structures, which

makes EBID suitable for developing strain sensors and magnetic sensors [68]. Other
possible deposit microstructures include amorphous and polycrystalline structures,
each of which exhibits different electric, magnetic, and mechanical properties [68].
The resulting structure depends on process control and the precursor molecule
used. Organometallic precursors with tendency to form carbides usually result in
amorphous structures [68]. Metallic structures unlikely to form carbides usually
result in nanogranular structures, composed of metallic particles in a dielectric
matrix. For certain precursors, a polycrystalline deposit microstructure may be
obtained by allowing only small carbon impurity depositions [68].
There are several parameters that affect the quality or speed of deposition. The
beam energy, current, dwell time, and replenishment time affect deposition rates and
resolution, as shown in Table 11.5.
However, the most important parameter in the EBID process is the selection of
the precursor. A good precursor material must produce a solid deposit after the
dissociation process, and the deposit must exhibit the required physical properties.
Common precursor gases and their applications are shown in Table 11.6.
There are commercial EBID systems for SEM, which consist of one or more
nozzles connected through the SEM feedthrough port to reservoirs of precursor
gases. A motion stage moves the EBID nozzle to points of interest. Examples of
commercial gas injection systems are the Xidex PGIS [72], Oxford Instruments
OmniGIS II [73], and the Orsay Physics GIS-5 [74], which can all supply at
least three precursor gases. Some commercial gas injection systems are shown in
Fig. 11.16.
Nanowire growth via field emission is a nanoscale fabrication technique based
on the decomposition of precursor gas molecules absorbed on a substrate by field
11 A Review of Nanomanipulation in Scanning Electron Microscopes 365

Fig. 11.16 Examples of EBID gas injection systems: (a) Oxford instruments OmniGIS. (b) Orsay
physics MonoGIS. (c) Orsay Physics GIS-5

Fig. 11.17 High-purity platinum nanowires are grown on a cathode from gas injection. Adapted
from [76]

emission current. The precursor molecules can decompose into positive ions and
negative ions, which are then attracted and deposited to cathodes and anodes
by electrostatic force. This technique produces high purity and hybrid metallic
nanowire growth [75, 76], and various types of nanowires can be grown with the
introduction of different precursors.
A high-purity platinum nanowire was fabricated via exposure inside a field
emission electron microscope using this technique [75]. A multi-walled CNT
(MWCNT) worked as a cathode, providing an emitter for a low working field
emission voltage, and a tungsten probe was employed as a field emission anode.
After introducing trimethylcyclopentadienyl platinum (CpPtMe3 ), the platinum
was deposited on the emitter to form a platinum nanowire, shown in Fig. 11.17.
In a similar procedure, a metallic sensor consisting of a platinum and tungsten
dual nanowire electrode for intracellular pH detection was fabricated with the
introduction of CpPtMe3 and tungsten hexacarbonyl (W(CO)6 ) respectively, on two
different MWCNT emitters [77, 78]. The tungsten nanowire is used as a work
electrode and the platinum nanowire is used as a counter electrode. The performance
of the pH sensor was tested inside acid and alkali buffer solutions [78].
366 D.K. Luu et al.

11.5 Nanomanipulation Application Examples

The nanometer positioning resolution of SEM nanomanipulation systems has

proven useful for a number of applications such as mechanical and electrical
properties characterization, nanoassembly, and cell nanosurgery.

11.5.1 Nanomaterials Characterization

Characterization of nanoscale materials such as silicon nanowires, carbon nan-

otubes (CNT), and graphene sheets are important for investigating their properties
and exploring applications of these materials. Various nanoscale mechanical and
electrical characterization experiments have been performed with nanorobotic
manipulation inside SEM.
Fukuda et al. performed an in-situ mechanical property characterization exper-
iment to determine the Young’s modulus of an individual MWNT, as shown
in Fig. 11.18a. A bundle of MWNTs was soldered on an AFM cantilever by
EBID and then buckled on a substrate by pushing the cantilever down [61]. In a
similar procedure, Zhu et al. demonstrated in-situ tensile testing for mechanical
characterization of silver nanowires. After two ends of the nanowire were fixed to

Fig. 11.18 Manipulation of nanoscale materials for property characterization. (a) A bundle of
MWNTs was buckled for Young’s modulus determination. Adapted from [61] (b) A nanospring
was stretched between a probe and an AFM cantilever. [80] (c) Tuning fork AFM to characterize
InP membrane [81] (d) Nanoindentation test for mechanical characterization of a graphene flake.
Adapted from [84] (e) Schematic illustration for piezoresistivity characterization of a suspended Si
nanowire with tensile testing. Adapted from [86] (f) Electrical characterization of graphene flakes.
Adapted from [87]
11 A Review of Nanomanipulation in Scanning Electron Microscopes 367

an AFM cantilever and a tungsten probe by carbonaceous material deposition by

EBID, tensile testing was performed by moving the cantilever and tungsten probe
apart. A range of mechanical properties was measured, including Young’s modulus,
yield strength, and ultimate tensile strength [79]. An InGaAs/GaAs nanospring was
stretched for tensile tests using a similar method for mechanical characterization
[80], shown in Fig. 11.18b.
Figure 11.18c shows an example application for the characterization of 2D
nanomaterials. Abrahamians et al. [81] performed a nondestructive measurement
method using a tuning-fork-based dynamic force sensor inside a SEM to determine
the stiffness of a batch of suspended InP micromembranes. The tuning fork
with a QPlus configuration was controlled with frequency modulation mode, and
the frequency shift was used to calculate the stiffness of the sample [82]. The
characteristic force-displacement curves were measured to determine the Young’s
modulus of the membrane. Mikczinski et al. [83] fixed a 2D nanopaper made
of microfibrillated cellulose at both ends and drove it against a capacitive force
sensor to perform nanoindentation experiments to determine its Young’s modulus.
In Fig. 11.18d, Fatikow et al used a self-sensing piezoresistive AFM cantilever to
conduct nanoindentation experiments on graphene flakes which were transferred
and soldered to be suspended on a standard aluminum TEM grid for mechanical
characterization [84, 85].
The schematic in Fig. 11.18e illustrates tensile experiments for piezoresistivity
characterization of a suspended Si nanowire performed by Lugstein et al. The two
ends of the Si nanowire under test were anchored with a suspended cantilever and
an insulated Si pad respectively. The probe in green is used to apply stress to the
freestanding cantilever, generating strain on the nanowire. The two probes in red
were applied to form electrical connections to measure the resistance changes [86].
Zimmermann et al. [87] used an L-shaped four-point probe fabricated using FIB for
electrical characterization of a graphene flake fabricated by mechanical exfoliation,
illustrated in Fig. 11.18f).

11.5.2 Construction of Nanoscale Devices

Not only has nanorobotic manipulation become an enabling technology for charac-
terizing the properties of nanomaterials, but it has also been further applied for the
construction of nanoscale building blocks and devices.
Figure 11.19a shows individual gold nanowires that were picked and placed by
two probes, and then welded and assembled to form a nanoscale pattern “NANO” by
Peng et al. [63]. A 3D letter was N formed on a MWCNT by oxygen assisted cutting,
bending, and manipulation for assembly, as shown in Fig. 11.19b [88]. Oxygen gas
can be introduced into the vicinity of a CNT in SEM through a glass nozzle while
irradiating it with a low-energy electron beam to cut and bend the CNT in order
to from 3D structures, as shown in [89, 90]. The cutting mechanism is opposite to
the deposition method. By exposing a CNT to an electron beam, carbon heats up
368 D.K. Luu et al.

Fig. 11.19 Assembly and construction for nanoscale blocks and devices. (a) Nanowires were
picked and placed to spell “NANO” [63]. (b) A MWCNT was used to form a 3D letter “N” using
oxygen assisted fabrication [88]. (c) Holes are pierced in polymer microcapsules to release medical
agents [92]. (d) Microspheres are stacked using two cooperative manipulators [25]. (e) A tool is
mounted on an AFM cantilever [64]. (f1–f2) Photonic plates are removed from a substrate, picked
and placed on a structure to produce a photonic crystal [93]

and ionizes, reacting with oxygen to generate carbon monoxide or carbon dioxide
to cut the CNT. Yuzvinsky et al. have also demonstrated CNT cutting with water
vapor [91] and have also achieved partial cuts in CNT, making hinge-like structures.
By increasing the acceleration voltage of the electron beam, placing the gas nozzle
further from the sample, or decreasing irradiation time, CNTs can be bent instead
of cut [88]. Nanoscale holes were produced by piercing of polymer microcapsules
by precise control of a microtool of microelectrode made from glass capillary
[92], as indicated by the arrow in Fig. 11.19c, to release encapsulated medical
active agents for targeted drug delivery. Tiemerding et al. created a 3D pyramidal
structure consisting of stacked silica spheres via cooperative manipulation of two
probes, as shown in Fig. 11.19d. An electrochemically etched tungsten probe
with a high-aspect ratio tip and a modified piezoresistive AFM cantilever with a
spherical adhesion pocket tip were fabricated to pick and place the spheres [25]. In
Fig. 11.19e, a nanoscale structure was manipulated by a microgripper to approach
an AFM cantilever, and then soldered and assembled onto an AFM tip using EBID
to construct new tools [64]. For a photonics application, Aoki et al. separated a
photonic plate from a substrate after breaking the connection bridge with a probe,
11 A Review of Nanomanipulation in Scanning Electron Microscopes 369

then picked and transferred the plate onto a substructure, as shown in Fig. 11.19f1.
More plates were stacked and aligned using the same procedure to form a 3D
photonic crystal, shown in Fig. 11.19f2, showing potential for the production of
optical wavelength photonic crystal devices [93].

11.5.3 Cell Nanosurgery in ESEM

Environmental scanning electron microscopes (ESEMs) enable high resolution

(e.g., 3.5 nm) and real-time imaging and observation of water-containing samples
and electrically insulating materials using a specially built secondary electron detec-
tor [94]. Both TEMs and conventional SEMs work under high-vacuum conditions;
therefore, they cannot directly observe water-containing samples (e.g., biological
cells in liquids). ESEM overcomes these limitations to allow manipulation and
characterization of biological, water-containing samples [9, 95].
A nanomanipulator system was constructed inside an ESEM with three units,
two units for manipulation with 7 DOFs and one cooling stage to hold samples
and control their temperature. This system can simultaneously perform real-time
observation and manipulation of biological samples. Commercial AFM cantilevers
were modified by FIB to fabricate various tools for manipulation and cell char-
acterization. A soft buckling nano-needle was made based on the modification of
a commercial AFM cantilever to perform indentation experiments, as shown in
Fig. 11.20a. Cell stiffness can be calculated after imaging processing [94, 95]. A
nano-fork and nano/micro-putter were fabricated by FIB to perform cell lifting
and side pushing experiments for investigating adhesion between the cell and
substrate [96, 97], as shown in Fig. 11.20b, c. A nano-picker was also constructed to
perform cell–cell adhesion force measurements [98]. An in-situ single cell cutting
experiment was performed with a nano-knife, and the cutting force and sample
slice angle were evaluated by the buffering beam deformation [54], shown in
Fig. 11.20d. The stiffness of fabricated tools can be calibrated experimentally using
nanomanipulation techniques, by comparing the deflection of the tools to a reference
cantilever with a known stiffness [96]. A dual nano-probe was also fabricated
from a standard piezoresistive cantilever to conduct cell electrical measurements to
determine viability [99], shown in Fig. 11.20e. An AFM cantilever was modified by
FIB into a sharp tip with a neck to deliver fluorescent materials into Caenorhabditis
elegans, and can be broken at the neck point as an embedded nanoscale injector after
insertion into the biological sample [100], as shown in Fig. 11.20f).

11.5.4 Hybrid Nanorobotic Manipulation in TEM and SEM

TEMs provide real-time, sub-nanometer imaging. A hybrid nanorobotic manipula-

tion system integrated with an 8-DOF SEM manipulator subsystem and a 6-DOF
370 D.K. Luu et al.

Fig. 11.20 Various tools for single cell measurement under ESEM. (a) Cell stiffness testing.
Adapted from [94]. (b–c) Nano fork and nano putter for testing cell adhesion. Adapted from [96,
97]. (d) Nano knife for cutting single cells. Adapted from [54]. (e) Testing cell viability using a
dual nanoprobe. Adapted from [99]. (f) Nanoinjection of C. Elegans. Adapted from [100]

TEM manipulator inside a SEM and TEM, respectively, was constructed to take
advantages of these two types of electron microscopes. This hybrid system is
capable of performing complicated sample preparation by taking advantage of
SEM’s large workspace and high imaging resolution of TEM [5, 6]. A schematic
of the SEM and TEM hybrid system is shown in Fig. 11.21. The TEM manipulator
is prepared on the SEM manipulator. An example application for CNT mechanical
characterization was performed as a demonstration. A CNT was pre-manipulated
for positioning the sample onto the TEM manipulator inside the SEM, followed by
stimulating the CNT for measurement with the TEM manipulator inside the TEM.
The experiment showed improved accuracy in measuring the Young’s modulus of
the CNT [6].

11.5.5 Automated Tasks

Manual nanomanipulation by joystick is time-consuming and skill dependent.

Success rate can vary significantly due to poor reproducibility and inconsistency
among operators. There is a steep learning curve, and intense focus is required
for operators to avoid damaging samples or the fragile end effectors. Automated
manipulation and assembly tasks using visual tracking and servoing techniques can
be more reliable, efficient, and reproducible than manual operation.
11 A Review of Nanomanipulation in Scanning Electron Microscopes 371

Fig. 11.21 Nanorobotic manipulators inside TEM and SEM. Adapted from [5].

11.5.5.1 Pick/Place and Assembly

Ye et al. [10] used a nanomanipulation system to automatically pick and place an

individual silicon nanowire and then transfer it to a MEMS device for characteriza-
tion, shown in Fig. 11.22.
The system selects a nanowire from the growing substrate using image process-
ing, as shown in Fig. 11.22a. To pick up the nanowire, the probe moves to contact
the nanowire, and then EBID is used to solder the nanowire onto the probe at the
contact point, as shown in Fig. 11.22b. To place the nanowire, the probe moves
the nanowire to the desired location. The nanowire is EBID soldered in place as
shown in Fig. 11.22c, d. Both mechanical and electrical characterization experi-
ments were performed on the MEMS device following this pick-place procedure
[10, 101]. Zhang et al. [102] demonstrated a method for nanodevice construction by
controlling a nanomanipulation system to perform nanowire selection.

11.5.5.2 Electrical Probing

Nanoprobing inside the SEM allows nanoprobe tips to be precisely positioned on

nanostructures for electrical characterization, which is important for failure analysis,
372 D.K. Luu et al.

Fig. 11.22 Pick-place of nanowire. (a) Image processing software automatically detects and
selects a nanowire. (b) Probe removes nanowire from substrate. (c) Nanowire is EBID soldered to
one side of the gap. (d) Nanowire is EBID soldered to other side of the gap. Electrical breakdown
between probe 1 and nanowire releases the nanowire from the probe. Adapted from [10]

quality control, and process development in the semiconductor industry. Automated

nanoprobing was performed as shown in Fig. 11.23 to probe nanostructures on
a SEM metrology chip. Algorithms were developed to de-noise the SEM image
and compensate for drift in real time, enabling robust visual tracking and servo
control for probing nanostructures. Using the SEM as a vision sensor for automation
and contact detection [24], Ru et al. landed four probes on a silicon nanowire to
perform four-point probing of single nanowires with a 100 % success rate over
50 trials [58, 103]. A feedforward controller was applied to improve the response
time for controlling multiple nanomanipulators [58, 103]. Besides nanowires, this
process can also be used to probe electrical properties of nanostructures [59] or
other nanomaterials. To enable faster and more accurate positioning without relying
on SEM visual feedback, Zhou et al. developed a strain gauge-based position
feedback system, which demonstrated a closed-loop positioning accuracy better
than 3 nm [8].
11 A Review of Nanomanipulation in Scanning Electron Microscopes 373

Fig. 11.23 Automated probing of nanostructure and silicon nanowire. (a) User identifies points to
probe from SEM image. (b) Nanomanipulator probes the specified points. Adapted from [58]

Fig. 11.24 Chromatin extraction process. (a) DNA extraction setup in SEM. (b) and (c) SEM
and fluorescence imaging correlation. (d) Land probe on selected cell. (e) Press nano-spatula into
surface and scrape off cell fragment. (f) Lift nano-spatula from sample and EBID “glue” cell
fragment to nano-spatula [104]

11.5.5.3 DNA Extraction

Nanomanipulation was demonstrated to perform cell nanosurgery inside SEM. Chen

et al. [104] utilized cell nanosurgery to infer functions of cell sub-nuclear structures
by examining nearby gene loci by using a “nano-spatula” tool to scoop up a single
chromatin as shown in Fig. 11.24. To facilitate the process, SEM images were
correlated with fluorescence microscope images to determine target locations to
scoop [105, 106]. Because gene locations within the nucleus are not random, this
technique can be used to discover functions and gene regulation mechanisms of
sub-nuclear structures.
374 D.K. Luu et al.

11.6 Conclusion

Nanoscale laboratories inside SEM have been constructed and integrated with
AFMs, FIBs, optical microscopes, multi-tool changers, and other new nanoscale
fabrication techniques. These technologies provide a foundation for bottom-up
fabrication and a powerful platform for characterizing nanoscale materials, forming
nanoscale building blocks, and assembling nanodevice prototypes. Technologies
such as EBID and FIB permit micro/nanodevice prototypes to be directly machined
or deposited, shortening the development time. These tools are expected to help
make new discoveries in biology, material science, and semiconductor industries.
Technically, new automation techniques will continue to be developed and will
emerge to achieve enhanced efficiency and reproducibility.

References

1. D.M. Eigler, E.K. Schweizer, Positioning single atoms with a scanning tunnelling microscope.
Nature 344(6266), 524–526 (1990)
2. S. Kim, F. Shafiei, D. Ratchford, X. Li, Controlled AFM manipulation of small nanoparticles
and assembly of hybrid nanostructures. Nanotechnology 22 (11), 115301 (2011)
3. U. Mick, M. Weigel-Jech, S. Fatikow, Robotic workstation for AFM-based nanomanipulation
inside an SEM, in 2010 IEEE/ASME International Conference on Advanced Intelligent
Mechatronics (AIM) (2010), pp. 854–859
4. U. Mick, V. Eichhorn, T. Wortmann, C. Diederichs, S. Fatikow, Combined nanorobotic
AFM/SEM system as novel toolbox for automated hybrid analysis and manipulation of
nanoscale objects, in 2010 IEEE International Conference on Robotics and Automation
(ICRA) (2010), pp. 4088–4093
5. T. Fukuda, M. Nakajima, P. Liu, M.R. Ahmad, Bringing the nanolaboratory inside electron
microscopes. Nanotechnol. Mag. IEEE 2(2), 18–31 (2008)
6. M. Nakajima, F. Arai, T. Fukuda, In situ measurement of Young’s modulus of carbon
nanotubes inside a TEM through a hybrid nanorobotic manipulation system. IEEE Trans.
Nanotechnol. 5(3), 243–248 (2006)
7. T. Fukuda, M. Nakajima, M. Ahmad, Y. Shen, M. Kojima, Micro- and nanomechatronics.
IEEE Ind. Electron. Mag. 4(4), 13–22 (2010)
8. C. Zhou, Z. Gong, B.K. Chen, M. Tan, Y. Sun, Closed-loop controlled nanoprobing inside
SEM, in 2014 IEEE 14th International Conference on Nanotechnology (IEEE-NANO) (2014),
pp. 45–48
9. T. Fukuda, M. Nakajima, H. Tajima, Y. Shen, T. Yue, Micro-nanomanipulation system toward
biological cell analysis and assembly, in 2012 First International Conference on Innovative
Engineering Systems (ICIES) (2012), pp. 31–36
10. X. Ye, Y. Zhang, C. Ru, J. Luo, S. Xie, Y. Sun, Automated pick-place of silicon nanowires.
IEEE Trans. Autom. Sci. Eng. 10(3), 554–561 (2013)
11. S. Fahlbusch, S. Mazerolle, J.-M. Breguet, A. Steinecker, J. Agnus, R. Pérez, J. Michler,
Nanomanipulation in a scanning electron microscope. J. Mater. Process. Technol. 167(2–3),
371–382 (2005)
12. S. Fatikow, V. Eichhorn, M. Bartenwerfer, F. Krohs, Nanorobotic AFM/SEM/FIB system
for processing, manipulation and characterization of nanomaterials, in 2013 IEEE 18th
Conference on Emerging Technologies Factory Automation (ETFA) (2013), pp. 1–4
11 A Review of Nanomanipulation in Scanning Electron Microscopes 375

13. J.I. Goldstein, D.E. Newbury, P. Echlin, D.C. Joy, A.D.R. Jr, C.E. Lyman, C. Fiori, E. Lifshin,
Introduction, in Scanning Electron Microscopy and X-Ray Microanalysis, Springer US, 1992,
pp. 1–19
14. “University of Glasgow:: Schools:: School of Geographical and Earth Sciences,”
University of Glasgow:: Glasgow, Scotland, UK. [Online]. Available: http://www.gla.
ac.uk/schools/ges/research/researchfacilities/isaac/services/scanningelectronmicroscopy/
imaging/secondaryelectronseandbackscatteredelectronbse/
15. J. Portilla, V. Strela, M.J. Wainwright, E.P. Simoncelli, Image denoising using scale mixtures
of Gaussians in the wavelet domain. IEEE Trans. Image Process. 12, 1338–1351 (2003)
16. S.M. Smith, J.M. Brady, SUSAN—a new approach to low level image processing. Int. J.
Comput. Vis. 23(1), 45–78 (1997)
17. F. Catté, P.-L. Lions, J.-M. Morel, T. Coll, Image selective smoothing and edge detection by
nonlinear diffusion. SIAM J. Numer. Anal. 29(1), 182–193 (1992)
18. C. Vogel, M. Oman, Iterative methods for total variation denoising. SIAM J. Sci. Comput.
17(1), 227–238 (1996)
19. N. Ouarti, B. Sauvet, S. Régnier, High quality real-time video with scanning electron micro-
scope using total variation algorithm on a graphics processing unit. Int. J. Optomechatronics
6(2), 163–178 (2012)
20. A. Buades, B. Coll, A non-local algorithm for image denoising, in In CVPR (2005), pp. 60–65
21. F. Goudail, P. Réfrégier, Statistical Image Processing Techniques for Noisy Images (Springer
US, Boston, 2004)
22. Z. Gong, B.K. Chen, J. Liu, Y. Sun, Automated nanoprobing under scanning electron
microscopy, in 2013 IEEE International Conference on Robotics and Automation (ICRA)
(2013), pp. 1433–1438
23. T. Tiemerding, C. Diederichs, S. Zimmermann, S. Fatikow, Closing the loop: high-speed
visual servoing and control of a commercial nanostage inside the SEM, in 2013 International
Conference on Manipulation, Manufacturing and Measurement on the Nanoscale (3M-
NANO) (2013), pp. 35–39
24. C. Ru, S. To, Contact detection for nanomanipulation in a scanning electron microscope.
Ultramicroscopy 118, 61–66 (2012)
25. T. Tiemerding, S. Zimmermann, S. Fatikow, Robotic dual probe setup for reliable pick and
place processing on the nanoscale using haptic devices, in 2014 IEEE/RSJ International
Conference on Intelligent Robots and Systems (IROS 2014) (2014), pp. 892–897
26. V. Eichhorn, S. Fatikow, T. Wich, C. Dahmen, T. Sievers, K.N. Andersen, K. Carlson, P.
Bøggild, Depth-detection methods for microgripper based CNT manipulation in a scanning
electron microscope. J. Micro-Nano Mechatron. 4(1–2), 27–36 (2008)
27. T. Kasaya, H.T. Miyazaki, S. Saito, K. Koyano, T. Yamaura, T. Sato, Image-based autonomous
micromanipulation system for arrangement of spheres in a scanning electron microscope. Rev.
Sci. Instrum. 75(6), 2033–2042 (2004)
28. A. Cvetanovic, A. Cvetanovic, A. Deutschinger, I. Giouroudi, W. Brenner, Design of a novel
visual and control system for the prevention of the collision during the micro handling in a
SEM chamber. Microelectron. Eng. 87(2), 139–143 (2010)
29. E. Ribeiro, M. Shah, Computer vision for nanoscale imaging. Mach. Vis. Appl. 17(3),
147–162 (2006)
30. S. Fatikow, T. Wich, T. Sievers, M. Jähnisch, V. Eichhorn, J. Mircea, H. Hülsen, C. Stolle,
Automatic nanohandling station inside a scanning electron microscope. Proc. Inst. Mech.
Eng. B J. Eng. Manuf. 222(1), 117–128 (2008)
31. M. Jähnisch, S. Fatikow, 3D vision feedback for nanohandling monitoring in a scanning
electron microscope. Int. J. Optomechatronics 1, 4–26 (2007)
32. V. Eichhorn, S. Fatikow, T. Wortmann, C. Stolle, C. Edeler, D. Jasper, O. Sardan, P. Boggild,
G. Boetsch, C. Canales, R. Clavel, NanoLab: a nanorobotic system for automated pick-and-
place handling and characterization of CNTs, in IEEE International Conference on Robotics
and Automation, 2009. ICRA ’09 (2009), pp. 1826–1831
376 D.K. Luu et al.

33. V. Eichhorn, S. Fatikow, Ch. Dahmen, Ch. Edeler, C. Stolle, D. Jasper (Presenter), Automated
microfactory inside a scanning electron microscope, in Proc. of the 6th International
Workshop on Microfactories (IWMF), Evanston, 5–7 October 2008
34. W. Denk, H. Horstmann, Serial block-face scanning electron microscopy to reconstruct three-
dimensional tissue nanostructure. PLoS Biol.2(11), e329 (2004)
35. B. Andres, U. Koethe, T. Kroeger, M. Helmstaedter, K.L. Briggman, W. Denk, F.A.
Hamprecht, 3D segmentation of SBFSEM images of neuropil by a graphical model over
supervoxel boundaries. Med. Image Anal. 16(4), 796–805 (2012)
36. V. Marx, Neurobiology: brain mapping in high resolution. Nature 503(7474), 147–152 (2013)
37. “Nanotechnology at Zyvex,” Nanotechnology at Zyvex, 2010. [Online]. Available: http://
www.zyvex.com/Products/S100_Features.html
38. J. Zuverza, “NanoBot® Nanomanipulator System from Xidex,” Xidex Nanobot® Nanomanip-
ulator and Nanotechnology Products, 2011. [Online]. Available: http://xidex.com/products/
nanobot.html
39. “Nanomanipulators - OmniProbe 400 - Oxford Instruments,” Home Page - Oxford
Instruments, 2015. [Online]. Available: http://www.oxford-instruments.com/products/
nanomanipulation-nanofabrication/nanomanipulator/omniprobe-400
40. B. K. Chen, LifeForce Nanomanipulation System: LF-2000
41. “Micromanipulator miBotTM BT-11 j Imina Technologies SA,” Imina Technologies SA j High
Precision Robots for Microscopes, 2014. [Online]. Available: http://imina.ch/technology
42. “Kleindiek Nanotechnik: Electron Microscopy,” Kleindiek Nanotechnik, 2015. [Online].
Available: http://www.nanotechnik.com/mm3a-em.html
43. P. Russel, D. Batchelor, SEM and AFM: complementary techniques for surface investigations.
Microsc. Anal. 2001, 9–14 (2001)
44. “NSOM/SNOM, TERS, Low Temperature AFM & SPM Solutions from Nanonics.” [Online].
Available: http://www.nanonics.co.il/
45. “Klocke Nanotechnik, Aachen.” [Online]. Available: http://nanomotor.de/index.html
46. “Hybrid AFM SEM Combination, the DME BRR Microscope,” Manufacturer of Atomic
Force Microscopes (AFM), Scanning Tunneling Microscopes (STM), 2013. [Online]. Avail-
able: http://www.dme-spm.com/remafm.html
47. “Test of camera modules, autocollimator, MTF system - TRIOPTICS,” 2015. [Online].
Available: http://www.trioptics.com/
48. “Atomic force microscope / electronic - attoAFM/SEM - attocube systems AG,” DirectIn-
dustry - The Virtual Industrial Exhibition: sensor - automation - motor - pump - handling -
packaging, 2015. [Online]. Available: http://www.directindustry.com/prod/attocube-systems-
ag/atomic-force-microscope-electronic-50096-1321683.html
49. F. Long, C. Wang, M. Lü, F. Zhang, J. Sun, J. Hu, Optimizing single DNA molecules
manipulation by AFM. J. Microsc. 243(2), 118–123 (2011)
50. T. Hertel, R. Martel, P. Avouris, Manipulation of individual carbon nanotubes and their
interaction with surfaces. J. Phys. Chem. B 102(6), 910–915 (1998)
51. G. Li, N. Xi, H. Chen, C. Pomeroy, M. Prokos, ‘Videolized’ atomic force microscopy for
interactive nanomanipulation and nanoassembly. IEEE Trans. Nanotechnol. 4(5), 605–615
(2005)
52. M.R. Ahmad, M. Nakajima, S. Kojima, M. Homma, T. Fukuda, In-situ single cell mechanical
characterization of W303 Yeast cells inside Environmental-SEM, in 7th IEEE Conference on
Nanotechnology, 2007. IEEE-NANO 2007 (2007), pp. 1022–1027
53. Y. Shen, M. Nakajima, M. Homma, T. Fukuda, Auto nanomanipulation system for single
cell mechanical property characterization inside an environmental SEM, in 2012 IEEE/RSJ
International Conference on Intelligent Robots and Systems (IROS) (2012), pp. 646–651
54. Y. Shen, M. Nakajima, Z. Yang, S. Kojima, M. Homma, T. Fukuda, Design and characteriza-
tion of nanoknife with buffering beam for in situ single-cell cutting. Nanotechnology 22(30),
305701 (2011)
11 A Review of Nanomanipulation in Scanning Electron Microscopes 377

55. V. Thanh, S.A. Chizhik, T. Xuan, N. Trong, V.V. Chikunov, Tuning fork scanning probe
microscopes – applications for the nano-analysis of the material surface and local physico-
mechanical properties, in Scanning Probe Microscopy-Physical Property Characterization at
Nanoscale, ed. by V. Nalladega (InTech, 2012)
56. F.J. Giessibl, Principles and Applications of the qPlus Sensor, in Noncontact Atomic Force
Microscopy, ed. by S. Morita, F.J. Giessibl, R. Wiesendanger (Springer, Berlin, Heidelberg,
2009), pp. 121–142
57. T. Akiyama, U. Staufer, N. de Rooij, Self-sensing and self-actuating probe based on quartz
tuning fork combined with microfabricated cantilever for dynamic mode atomic force
microscopy. Appl. Surf. Sci. 210(1–2), 18–21 (2003)
58. C. Ru, Y. Zhang, Y. Sun, Y. Zhong, X. Sun, D. Hoyle, I. Cotton, Automated four-point probe
measurement of nanowires inside a scanning electron microscope. IEEE Trans. Nanotechnol.
10(4), 674–681 (2011)
59. Z. Gong, B.K. Chen, J. Liu, Y. Sun, Robotic probing of nanostructures inside scanning
electron microscopy. IEEE Trans. Robot. 30(3), 758–765 (2014)
60. L. Dong, B.J. Nelson, Tutorial – robotics in the small part II: nanorobotics. IEEE Robot.
Autom. Mag. 14(3), 111–121 (2007)
61. T. Fukuda, F. Arai, L. Dong, Assembly of nanodevices with carbon nanotubes through
nanorobotic manipulations. Proc. IEEE 91(11), 1803–1818 (2003)
62. B.K. Chen, Y. Zhang, D.D. Perovic, Y. Sun, MEMS microgrippers with thin gripping tips. J.
Micromech. Microeng. 21(10), 105004 (2011)
63. Y. Peng, T. Cullis, B. Inkson, Bottom-up nanoconstruction by the welding of individual
metallic nanoobjects using nanoscale solder. Nano Lett. 9(1), 91–96 (2009)
64. R.T.R. Kumar, S.U. Hassan, O.S. Sukas, V. Eichhorn, F. Krohs, S. Fatikow, P. Boggild,
Nanobits: customizable scanning probe tips. Nanotechnology 20(39), 395703 (2009)
65. M. Nakajima, T. Kawamoto, M. Kojima, T. Fukuda, Nanotool exchanger system using
low-melting metal under environmental SEM, in 2011 International Symposium on Micro-
NanoMechatronics and Human Science (MHS), 2011, pp. 69–74
66. A. Garetto, C. Baur, J. Oster, M. Waiblinger, K. Edinger, Advanced process capabilities for
electron beam based photomask repair in a production environment, in Proc. SPIE 7122
(2008), p. 7122k
67. L. van Kouwen, A. Botman, C.W. Hagen, Focused electron-beam-induced deposition of 3 nm
dots in a scanning electron microscope. Nano Lett. 9(5), 2149–2152 (2009)
68. M. Huth, F. Porrati, C. Schwalb, M. Winhold, R. Sachser, M. Dukic, J. Adams, G. Fantner,
Focused electron beam induced deposition: a perspective. Beilstein J. Nanotechnol. 3, 597–
619 (2012)
69. I. Utke, S. Moshkalev, P. Russell, Nanofabrication Using Focused Ion and Electron Beams:
Principles and Applications (Oxford University Press, New York, 2012)
70. V. Scheuer, H. Koops, T. Tschudi, Electron beam decomposition of carbonyls on silicon.
Microelectron. Eng. 5(1–4), 423–430 (1986)
71. S.J. Randolph, J.D. Fowlkes, P.D. Rack, Focused, nanoscale electron-beam-induced deposi-
tion and etching. Crit. Rev. Solid State Mater. Sci. 31(3), 55–89 (2006)
72. “Nanobot Nanomanipulator - Parallel Gas Injection System,” Xidex Nanobot Nanomanip-
ulator and Nanotechnology Products, 2011. [Online]. Available: http://xidex.com/products/
parallel-gas-injection-system.html
73. “Gas Injection Systems: OmniGIS II – Oxford Instruments,” 2015. [Online]. Avail-
able: http://www.oxford-instruments.com/products/nanomanipulation-nanofabrication/gas-
injection-system/omnigis-ii
74. “GIS-5: Multi-lines Gas Injection System,” Orsay Physics, 2015. [Online]. Available: http://
www.orsayphysics.com/product-gis-5.html
75. Z. Yang, M. Nakajima, Y. Saito, Y. Ode, T. Fukuda, Isolated high-purity platinum nanowire
growth via field emission from a multi-walled carbon nanotube. Appl. Phys. Express 4(3),
035001 (2011)
378 D.K. Luu et al.

76. Z. Yang, M. Nakajima, Y. Ode, T. Fukuda, Tungsten/platinum hybrid nanowire growth via
field emission using nanorobotic manipulation. J. Nanotechnol. 2011, e386582 (2011)
77. Z. Yang, M. Nakajima, Y. Ode, T. Fukuda, Nanoassembly of pH sensor nanoprobe by
multiple-metallic nanowires, in 2010 10th IEEE Conference on Nanotechnology (IEEE-
NANO) (2010), pp. 352–355
78. Z. Yang, M. Nakajima, Y. Ode, Z. Zhang, T. Fukuda, Fabrication and evaluation of nano
probe pH sensor based on nanorobotic manipulation, in 2010 International Symposium on
Micro-NanoMechatronics and Human Science (MHS), 2010, pp. 284–289
79. Y. Zhu, Q. Qin, F. Xu, F. Fan, Y. Ding, T. Zhang, B.J. Wiley, Z.L. Wang, Size effects on
elasticity, yielding, and fracture of silver nanowires: in situ experiments. Phys. Rev. B 85(4),
045443 (2012)
80. D.J. Bell, L. Dong, B.J. Nelson, M. Golling, L. Zhang, D. Grützmacher, Fabrication and
characterization of three-dimensional InGaAs/GaAs nanosprings. Nano Lett. 6(4), 725–729
(2006)
81. J.-O. Abrahamians, B. Sauvet, J. Polesel-Maris, R. Braive, S. Regnier, A nanorobotic system
for in situ stiffness measurements on membranes. IEEE Trans. Robot. 30(1), 119–124 (2014)
82. A. Castellanos-Gomez, N. Agraït, G. Rubio-Bollinger, Dynamics of quartz tuning fork force
sensors used in scanning probe microscopy. Nanotechnology 20(21), 215502 (2009)
83. M.R. Mikczinski, G. Josefsson, G. Chinga-Carrasco, E.K. Gamstedt, S. Fatikow, Nanorobotic
testing to assess the stiffness properties of nanopaper. IEEE Trans. Robot. 30(1), 115–119
(2014)
84. S. Zimmermann, V. Eichhorn, S. Fatikow, Nanorobotic transfer and characterization of
graphene flakes, in 2012 IEEE/RSJ International Conference on Intelligent Robots and
Systems (IROS) (2012), pp. 640–645
85. S. Zimmermann, T. Tiemerding, T. Li, W. Wang, Y. Wang, S. Fatikow, Automated mechanical
characterization of 2-D materials using SEM based visual servoing. Int. J. Optomechatronics
7(4), 283–295 (2013)
86. A. Lugstein, M. Steinmair, A. Steiger, H. Kosina, E. Bertagnolli, Anomalous piezoresistance
effect in ultrastrained silicon nanowires. Nano Lett. 10(8), 3204–3208 (2010)
87. S. Zimmermann, S.A.G. Barragan, S. Fatikow, Nanorobotic processing of graphene: a
platform tailored for rapid prototyping of graphene-based devices. IEEE Nanotechnol. Mag.
8(3), 14–19 (2014)
88. P. Liu, K. Kantola, T. Fukuda, F. Arai, Nanoassembly of nanostructures by cutting, bending
and soldering of carbon nanotubes with electron beam. J. Nanosci. Nanotechnol. 9(5), 3040–
3045 (2009)
89. P. Liu, F. Arai, T. Fukuda, Cutting of carbon nanotubes assisted with oxygen gas inside a
scanning electron microscope. Appl. Phys. Lett. 89(11), 113104 (2006)
90. P. Liu, F. Arai, T. Fukuda, Nanofabrication of carbon nanotubes assisted with oxygen gas,
in Sixth IEEE Conference on Nanotechnology, 2006. IEEE-NANO 2006, vol. 2 (2006),
pp. 540–543
91. T.D. Yuzvinsky, A.M. Fennimore, W. Mickelson, C. Esquivias, A. Zettl, Precision cutting of
nanotubes with a low-energy electron beam. Appl. Phys. Lett., 86(5), 053109 (2005)
92. E. Meyer, H.-G. Braun, Micro- and nanomanipulation inside the SEM. J. Phys. Conf. Ser.
126, 012074 (2008)
93. K. Aoki, H.T. Miyazaki, H. Hirayama, K. Inoshita, T. Baba, K. Sakoda, N. Shinya, Y. Aoyagi,
Microassembly of semiconductor three-dimensional photonic crystals. Nat. Mater. 2(2),
117–121 (2003)
94. M.R. Ahmad, M. Nakajima, S. Kojima, M. Homma, T. Fukuda, The effects of cell sizes,
environmental conditions, and growth phases on the strength of individual W303 yeast cells
inside ESEM. IEEE Trans. Nanobioscience 7(3), 185–193 (2008)
95. M.R. Ahmad, M. Nakajima, S. Kojima, M. Homma, T. Fukuda, In situ single cell mechanics
characterization of yeast cells using nanoneedles inside environmental SEM. IEEE Trans.
Nanotechnol. 7(5), 607–616 (2008)
11 A Review of Nanomanipulation in Scanning Electron Microscopes 379

96. M.R. Ahmad, M. Nakajima, M. Kojima, S. Kojima, M. Homma, T. Fukuda, Nanofork

for single cells adhesion measurement via ESEM-nanomanipulator system. IEEE Trans.
Nanobioscience 11(1), 70–78 (2012)
97. Y. Shen, M. Nakajima, S. Kojima, M. Homma, M. Kojima, T. Fukuda, Single cell adhesion
force measurement for cell viability identification using an AFM cantilever-based micro
putter. Meas. Sci. Technol. 22(11), 115802 (2011)
98. Y. Shen, M. Nakajima, S. Kojima, M. Homma, T. Fukuda, Study of the time effect on the
strength of cell-cell adhesion force by a novel nano-picker. Biochem. Biophys. Res. Commun.
409(2), 160–165 (2011)
99. M.R. Ahmad, M. Nakajima, M. Kojima, S. Kojima, M. Homma, T. Fukuda, Instantaneous
and quantitative single cells viability determination using dual nanoprobe inside ESEM. IEEE
Trans. Nanotechnol. 11(2), 298–306 (2012)
100. T. Hirano, M. Nakajima, M. Kojima, N. Hisamoto, M. Homma, T. Fukuda, Selective nano-
injection using nano-probe based on nanomanipulation under hybrid microscope, in 2011
International Symposium on Micro-NanoMechatronics and Human Science (MHS) (2011),
pp. 216–221
101. Y. Zhang, X. Liu, C. Ru, Y.L. Zhang, L. Dong, Y. Sun, Piezoresistivity characterization
of synthetic silicon nanowires using a MEMS device. J. Microelectromech. Syst. 20(4),
959–967 (2011)
102. Y.L. Zhang, J. Li, S. To, Y. Zhang, X. Ye, L. You, Y. Sun, Automated nanomanipulation for
nanodevice construction. Nanotechnology 23(6), 065304 (2012)
103. C. Ru, Y. Zhang, Y. Sun, Y. Zhong, X. Sun, D. Hoyle, I. Cotton, Automated four-point probe
measurement of nanowires inside a scanning electron microscope, in 2010 IEEE Conference
on Automation Science and Engineering (CASE) (2010), pp. 533–538
104. B.K. Chen, D. Anchel, Z. Gong, R. Cotton, R. Li, Y. Sun, D.P. Bazett-Jones, Gene
organization: nano-dissection and sequencing of DNA at single sub-nuclear structures (Small
16/2014). Small 10(16), 3266–3266 (2014)
105. Z. Gong, B.K. Chen, J. Liu, C. Zhou, D. Anchel, X. Li, D.P. Bazett-Jones, Y. Sun, Correlative
microscopy for nanomanipulation of sub-cellular structures, in 2014 IEEE International
Conference on Robotics and Automation (ICRA) (2014), pp. 5209–5214
106. Z. Gong, B.K. Chen, J. Liu, C. Zhou, D. Anchel, X. Li, J. Ge, D.P. Bazett-Jones, Y.
Sun, Fluorescence and SEM correlative microscopy for nanomanipulation of subcellular
structures. Light Sci. Appl. 3(11), e224 (2014)
Chapter 12
Nanopositioning for Lithography
and Data Storage

Wenbo Zhang, Ming Pang, and Changhai Ru

The history of the science shows that there is usually one strong social demand
behind a new technology. The new technology, which can solve the key problem,
must be got people’s praise and attraction. Nanotechnology is also facing the
similar historical opportunity. Nanotechnology, which is understanding matter and
the control of matter at dimensions of 100 nm or less, appeared in recent years
[1–4]. The inventions of the scanning tunneling microscope (STM) [5] and the
atomic force microscope (AFM) [6], which are widely used in biology [7, 8],
chemistry [9], materials science [10, 11], and physics [12], greatly promoted the
research of the nanotechnology areas. The wide applications all depend on the
high-precision nanopositioning stages. Based on this, nanopositioning is one of
the key requirements of nanotechnology [13]. Nanopositioning involves precision
control and manipulation of devices and materials at a nanoscale. Nanopositioners
are precision mechatronic systems applied to move objects over a small range with
a resolution down to a fraction of an atomic diameter. Hence, nanopositioners
need to possess high resolution, accuracy, stability, and fast response. To realize
manipulation at the nanometer scale with SPM, nanopositioning systems are
necessary [14, 15]. In SPMs, nanopositioning system is used to control the probe to
scan over a sample surface. Accurate position sensor and feedback control system
are key to successful nanopositioning.
Nanopositioning systems are vital in optical alignment systems [16, 17] and
next-generation space telescopes [18, 19]. Such nanopositioning is also crucial

W. Zhang • M. Pang ()

College of Automation, Harbin Engineering University, Harbin 150001, China
e-mail: [email protected]
C. Ru
College of Automation, Harbin Engineering University, Harbin 150001, China
Research Center of Robotics and Micro System & Collaborative Innovation Center of Suzhou
NanoScience and Technology, Soochow University, Suzhou 215021, China

© Springer International Publishing Switzerland 2016 381

C. Ru et al. (eds.), Nanopositioning Technologies,
DOI 10.1007/978-3-319-23853-1_12
382 W. Zhang et al.

for biology areas, such as cell tracking and nanomaterials testing [20, 21], DNA
analysis, the manufacturing of small objects [22], and nanoassembly [23]. Nanopo-
sitioning systems are the core of lithography tools, which might replace traditional
optical lithography systems [24, 25]. The new lithography technology can con-
struct semiconductor integrated circuits (ICs) having less than 35 nm nodes [26].
Similarly, the servo system of hard-disk drives (HDD) also requires nanometer-
scale precision [26, 27]. The nanopositioning system can be used to realize the
areal densities higher than 1 Tb/in2 in probe-storage device [28, 29]. Novel
nanopositioning technology is also required for semiconductor inspection [30, 31].
The vast range of applications necessitates nanopositioning systems with high
resolution, high bandwidth, and robust control designs [32–38].
This chapter introduces some typical applications of nanopositioning, including
lithography, data storage, and fault analysis.

12.1 Nanopositioning for Lithography

Lithography technology is the key of the semiconductor industry. Improvement of

lithography has been a key driver for improving integrated circuit (IC) performance.
Therefore, new generation lithography systems based on high precision nanoposi-
tioning and nanometer precise placement have attracted much attention [39]. At
the same time, low-cost technologies for mass producing nanometric structures
are in dire demand, especially for future manufacturing of integrated optics [40].
Diffraction grating is an important dispersion element in modern physics [41], and
the diffraction efficiency of grating has strict requirements on the notch groove depth
and shape [42, 43]. Mechanical characterization of chips is the best way to make
diffraction grating by ruling engine. The requirements of ruling engine are large
useful area, high precision and multifunction. Large useful area means large ruling
width and ruling depth. High precision means the positional accuracy of optical
grating. Multifunction means multiple inputs get different ruling results.
Electron beam, which is called electron projection lithography (EPL) [44, 45], is
used for patterning the substrate to replace traditional optical lithography. Scanning
electron beam can realize 10 nm lithography [46, 47]. Fischer and Chou used a
modified scanning electron microscope to realize 10 nm electron beam lithography
and sub-50 nm overlay [47]. They repeatedly achieved gratings of 10 nm wide on
thick GaAs substrates, as shown in Fig. 12.1.
The disadvantage of scanning electron microscope is uneconomical and has
low throughput due to its inherent drawback. X-ray lithography can realize high
throughput with 50 nm resolution [48]. However, X-ray lithography tools are rather
expensive. Chou et al. developed nanoimprint lithography (NIL), which is a low-
cost mass manufacturing technology, for fabricating sub-25 nm features [40]. The
fabrication of the mold used for NIL process is one of the most important research
hotspots of NIL.
12 Nanopositioning for Lithography and Data Storage 383

Fig. 12.1 Scanning electron

micrograph of a 40 nm period
Ni/Au grating with 10 nm
linewidth on bulk GaAs [47].
Copyright 1993. Reproduced
with permission from
Applied physics letters

Vorbringer-Doroshovets et al. demonstrated nano-scale lithography by using a

scanning probe combined with a Nanopositioning and Nanomeasuring Machine
[49], as shown in Fig. 12.2.
For the scanning probe lithography experiments, the researchers used a modified
hybrid Atom Force Microscope (AFM) set-up to operate at ambient conditions and
room-temperature. The results are shown in Fig. 12.3.
For the application of lithography in grating ruling machine, Harrison success-
fully applied interference to the notch groove positional servo system and developed
MIT-A [50] and MIT-B [51]. For MIT-A, the notch groove position is confirmed
based on the optical wavelength. The area and precision of the optical grating were
improved. Thereafter, grating interference measuring technique, laser interference
measuring technique, piezoceramic micro-positioning technology, and computer
control technology were developed and applied to optical grating. In 1966, Harrison
et al. developed MIT-C [52] based on the experience of MIT-A and MIT-B. This
grating ruling machine was constructed out of the forth all-purpose measuring
machine produced by Moore Company, as shown in Fig. 12.4.
Compared with traditional grating ruling machine, the measuring elements for
feedback were implemented by the single frequency laser interferometer and the
feedback actuator was implemented by the differential gear train. MIT-C has double
work stage structure, which is connected by flexure hinge, the inner stage is used
to correct the straightness error of the guide rail. Now MIT-C has ruled the optical
grating, of which the max size is 340  440 mm.
Since Harrison and Stroke successfully used the interference servo system in
the grating ruling machine, many researchers developed various kinds of ruling
384 W. Zhang et al.

Fig. 12.2 (a) Basic concept and main components of the Nanopositioning and Nanomeasuring
Machine (NPM Machine). (b) Additional angular sensors for measurement [49]. Copyright 2013.
Reproduced with permission from SPIE Advanced Lithography

machines. Stiff et al. developed interference servocontrol grating ruling engine [53].
In 1992, Kita and Harada in the Central Lab of Hitachi developed digital control
grating ruling engine [54], as shown in Fig. 12.5.
It has both macro- and micro-positioning systems for the first time. The macro-
positioning system is driven by the screw nut structure and the micro-positioning
system with upper and lower layers of platform structure is driven by piezoelectric
ceramic. Its pivot swinging angle of workbench in 300 mm stroke is 0.2. The
maximum grating groove spacing is 10,000 g/mm and the dividing control precision
is 5 nm [54].
12 Nanopositioning for Lithography and Data Storage 385

Fig. 12.3 AC-AFM mode images of the SPL-structures generated in calixarene resist. All
lithographic steps and AFM-images were done step by step with the same cantilever (and without
any development steps). The two bottom AFM topography images show line pattern written
with 45 nm half-pitch into 20 nm thick C-MC4R resist [49]. Copyright 2013. Reproduced with
permission from SPIE Advanced Lithography

Fig. 12.4 MIT-C grating ruling machine

The control strategy is not only decided by the drive method but also the
controlled member. There are mainly two control strategies of nanopositioning for
lithography, including feedback control strategy and feedforward control strategy.
PID Control is a classical feedback control method. The integral controller of PID
386 W. Zhang et al.

Fig. 12.5 Hitachi Grating Ruling Engine. (a) Optical layout of the laser interferometer. (b) Control
system block diagram: LPF low-path filter, PZT piezoelectric device, AMP amplifier [54].
Copyright 1992. Reproduced with permission from OSA Publishing

can provide high gain feedback in low-frequency stage, which can be used to
overcome creep and hysteresis. It is suitable to be used in nanopositioning systems.
Kouno et al. and Cuttino et al. used traditional PID feedback controller and dynamic
compensation module to reduce the hysteresis-caused error of the piezoelectric
actuators in the single point diamond lathes less than 150 nm [55, 56] (Fig. 12.6).
Tan et al. designed a learning controller, which can auto-control the parameters of
PID on line, which improved the robustness of the integral controller [57]. Heertjes
et al. proposed a nonlinear PID controller, which can modify the parameters of PID
on line [58]. Compared with traditional PID controller, this nonlinear PID controller
can suppress interference and improve positioning accuracy, which was proved by
positioning on the wafer table. Maeda et al. designed a controller suitable for super
precision machine tool with traditional PID, feedforward and band-pass filter, which
reached a high positioning accuracy to track a circle at 500 Hz [59] (Fig. 12.7).
The disadvantage of PID is that it needs higher feedback gain to realize
positioning accuracy on nanometer in high frequency band while piezoelectric
actuator can only offer low gain at high frequencies. Numasato and Tomizuka used
Notch filter between the two grades drivers of HDD, which are voice coil motor
and piezoelectric ceramics, to solve the problem. The Notch filter can alter the first
harmonic peak of the control system and improve control precision [60] (Fig. 12.8).
Leang and Devasia added Notch filter to the control system of Atomic Force
Microscopy, which improved the gain margin of piezoelectric actuators and sup-
pressed the disadvantages of piezoelectric actuation, such as hysteresis, creep, and
vibration [61], as shown in Fig. 12.9.
12 Nanopositioning for Lithography and Data Storage 387

Fig. 12.6 Experimental results [56]

It can be seen that closed-loop control is widely used in nanopositioning

for lithography, but it is limited by parameter uncertainty and high-frequency
interference. Hence, other control methods such as adaptive control androbust
control were considered. Cruz-Hernández and Haywar used lead-lag control to
compensate the lag in the energy transfer process of smart materials [62]. Shakir
and Kim used lead lag controller and LQG controller to realize a resolution of 5 nm
[63]. The result is shown in Fig. 12.10.
Schitter et al. proposed a control strategy based on H1 theory, which is used in
AFM (Atom Force Microscopy) control system and making the scanning speed five
times higher than traditional PID control while the error was also reduced [35]. For
a given scan rate, the controller error of the H1 controller was always lower than
that of the PID controller. The average error (measured at the edges of patch) of the
H1 controller was from one-fifth to one-third of the PID controller at 20 and 4 Hz,
respectively. Furthermore, H1 -controlled AFM had better damping of the stick slip
effect. The experimental results are shown in Fig. 12.11.
Helfrich et al. combined H1 control with iterative learning control based
on feedforward and obtained relatively good positioning effect [64]. Bashash
and Jalili used robust adaptive control method on coupled parallel piezoelectric
bending nanopositioning stage to suppress the lag of piezoelectric actuator and
coupled nonlinearities. The positioning accuracy of the system was significantly
improved [65]. Sebastian and Salapaka designed an H1 robust control method
388 W. Zhang et al.

a 150
b 150

100 100

50 100 Hz 50 100 Hz
200 Hz
Y nm

200 Hz
250 Hz

Y nm
0 250 Hz 0
300 Hz 300 Hz
400 Hz 400 Hz
−50 −50 500 Hz
500 Hz

−100 −100

−150
−150 −150 −100 −50 0 50 100 150
−150 −100 −50 0 50 100 150
X nm X nm
K2 = 1. K2 = 0.87.
Standard feedforward gain Experimentally adjusted feedforward gain

100 nm radius circular motion

a b 11
6
X displacement

3 Sharp trajectory 10 Sharp trajectory

Sinusoidal trajectory
Y displacement μm

0 9 Sinusoidal trajectory
μm

−3 8
−6 7
6

Y displ. μm
5.2
30
5
15 5.1
4
X error

0
nm

−15 Sharp trajectory 3 5.0

−30 Sinusoidal trajectory 2 4.8 4.9 5.0
0.0 0.1 0.2 0.3 1 X displ. μm
Time s 0
X axis square motion −1
−6 −5 −4 −3 −2 −1 0 1 2 3 4 5 6
X displacement μm
square motion trajectory

Square motion in the micrometer range

Fig. 12.7 100 nm radius circular motion and square motion in the micrometer range [59]

for AFM nanopositioning. The control system not only eliminated the influence
of nonlinearity but also much improved the positional accuracy and speed [66].
In addition, to use the model information better, many nanopositioning systems
used feedforward control or combined feedforward control and feedback control
in the controlling process. Kuhnen and Janocha used feedforward control method to
reduce nonlinear errors from 30 to 3 % [67]. Figure 12.12 shows the compensated
characteristics of the overall system given by the serial combination of the inverse
feedforward controller and the magnetostrictive actuator.
The one-dimensional ultra-precision stage developed by Tokyo Institute of
Technology, which is controlled by the combination of conventional PID and
feedforward compensation and driven by PMLSML (Permanent Magnet Linear
Synchronous Motor), realized 2 nm step positioning [68]. Yong et al. used feed-
forward control on the X-Y nanopositioning stage of a scanning probe microscopy,
and achieved maximum error of 2 % [69].
12 Nanopositioning for Lithography and Data Storage 389

Fig. 12.8 Block diagrams of control systems and settling response of dual-actuator system with a
step [60]

a b
30 35
Measured Response Along x- axis (μm)

30
Displacement Along x- axis (μm)

20

25
10 Max. Output
Desired Displacement = 25.00 μm
Hysteresis = 1.04 μm
20 Uncompensated Displacement = 33.41 μm
Max. Output Compensated Displacement = 25.37 μm
0 Hysteresis = 6.29 μm
15

−10
10

−20 Without Feedback 5 Without Feedback

With Feedback With Feedback
Desired Reponse Desired Reponse
−30 0
−30 −20 −10 0 10 20 30 0 2 4 6 8 10 12 14 16
Desired Response xd (μm) Time (minutes)

Fig. 12.9 Experimental results showing (a) hysteresis and (b) creep compensation using high-gain
feedback. Dashed line is without compensation; solid line is with compensation; and dotted line is
the desired response [61]. Copyright 2002. Reproduced with permission from the POL publication

12.2 Nanopositioning for Data Storage

There is a great need to develop ultra-high density, small form factor, and low cost
memory devices since these data-storage technologies can bring enormous impact
to the areas of computer science and mobile handheld devices. Due to the limits of
390 W. Zhang et al.

Fig. 12.10 Experimental results [63]

physics, the scale and storage capacity of conventional storage technologies, such
as magnetic recording and solid-state flash memory, are reaching their limits. New
technologies are proposed to address future data-storage requirements.
High capacity, small-form-factor, and low-power are the primary advantages of
probe storage. With a sharp tip scanned in close proximity to a surface, the STM
and AFM have the ability to image or modify surfaces with spatial resolution
ranging from angstroms to tens of nanometers [70, 71]. STM and AFM have
potential for data-storage device and can realize higher areal data storage densities
than conventional methods [72]. In AFM, a sharp tip is fixed on the end of
a soft cantilever spring to sense and apply atomic force [6]. AFM uses high-
frequency, piezoresistive silicon cantilevers to mechanically read back topographic
data (Fig. 12.13).
The example is shown in Fig. 12.14, where individual xenon atoms were
positioned to spell out “IBM.”
12 Nanopositioning for Lithography and Data Storage 391

Fig. 12.11 Purple membrane imaged with an H1 -controlled AFM [(a, c, e, and g)] and a PID-
controlled AFM [(b, d, f, and h)]. All images were recorded from right to left and are 1.5 by
1.5 m2 . The line scan rate is 20 Hz for (a, b, e, and f) and 4 Hz for (c, d, g, and h). The sample
topography is shown in (a–d) [scaling from black to white is 23 nm for (a, c) and 25 nm for (b,
d)]. The cantilever’s deflection signal is shown in (e, h) [scaling is 4 nm from black to white] [35].
Copyright 2001. Reproduced with permission from AIP publisher

Fig. 12.12 Compensated

hysteretic displacement-given
displacement relation [67]
392 W. Zhang et al.

Fig. 12.13 Concept of topographic data storage using an AFM tip. The tip rides over the surface
of the substrate, causing deflection of the cantilever as the tip moves over a topographic feature.
The deflection is detected via a piezoresistive sensor [72]

Fig. 12.14 Atomic-scale message demonstrating the ability to position single atoms with angstom
precision. Each bump is an individual xenon atom positioned with the tip of a scanning tunneling
microscope operating at 4.2 K. This represents perhaps the ultimate in data-storage density, with
a single atomic “bit” (courtesy of D. Eigler, IBM Research Division, Almaden Research Center)
[72]

STM scans the tip while monitoring the tunnel and maintaining tip position servo
current constant to map out the surface topography [73]. The actuator can move in
the vertical and cross-track directions. A closed-loop servo system, of which the
bandwidth is nearly 1000 Hz, used the signal from the piezoresistive cantilever as
the error signal to control the vertical load on the cantilever [74].
However, the limitation of the feedback speed and low tunneling currents makes
the atomic-level storage based on AFM difficult to realize. The micro-electro-
mechanical systems (MEMS)-based arrays of cantilevers operating in parallel was
employed to as a solution, with each cantilever performing WRITE/READ/ERASE
operations in an individual storage field. In MEMS-actuated magnetic probe-based
storage system, a magnetic storage medium is positioned in the x/y plane [75].
Writing is achieved by an array of probe tips, each tip was positioned by a z-axis
positioning system. The position servo can realize displacements up to 55 % of the
initial actuation gap.
12 Nanopositioning for Lithography and Data Storage 393

Fig. 12.15 Illustration of the Millipede concept [77]

Vettiger and Eleftheriou presented the scanning-probe-based data-storage con-

cept, which is called the “millipede” [76, 77]. The technology uses the thermo-
mechanical concept to record and recover data in thin polymer films, as shown in
Fig. 12.15.
It is encoded as nanometer-scale indentations, for which the presence or absence
of indentations corresponds to logical ones or zeros, respectively. Early experimen-
tal results demonstrated that the overall chip tip-apex height control was within
500 nm. It is realized by using feedback control [78]. The main drawback of
thermomechanical probe-based data is the low data rate per microcantilever. To
solve this, WRITE/READ operations depend on a mechanical parallel x/y scanning
of either the entire cantilever array chip or the storage medium [77]. A row/column
time-multiplexed addressing scheme is proposed to ensure the parallel operations
efficient. It addresses the array column by column with full parallel WRITE/READ
operation within one column [76]. The cantilever moves from one bit position to
the next and is driven by the pulse. During reading process, each tip is always
positioned over the center of a desired track. Similarly during writing, the tips should
be positioned on the written marks, which are aligned in a predefined way. The servo
loop for track following is driven by an appropriate error signal, which is called PES.
The stability and robustness of the associated tracking servo loop is directly affected
by the quality of the PES [79]. Magnetic recording provides the concept of mutually
vertically displaced to provide two signals in quadrature, which can be combined to
provide a robust PES [79]. As shown in Fig. 12.16a, servo marks are placed in bursts
labeled A and B for the in-phase signal and C and D for the quadrature signal. The
tip moves from the edge of the top mark toward its center, then toward its bottom.
The combination of the two signals (I and Q), shown as solid lines in Fig. 12.16b,
has zeros crossing as all track center locations and constant slope, which qualifies it
as a valid PES. At even-numbered tracks, the zero of the in-phase signal indicates the
track center. In turn, the zeros of the quadrature signal can be uniquely mapped into
a position estimate by examining the polarity of the in-phase signal at corresponding
positions.
394 W. Zhang et al.

Fig. 12.16 Only two signal marks per burst are shown here to simplify the presentation while in
fact, each burst typically consists of many marks in order to enable averaging of the corresponding
readout signals. The solid horizontal lines depict the track centerlines and circles represent written
marks [77]

The experimental results of the principle of PES generation based on servo and
the resulting in-phase and quadrature signals are shown in Fig. 12.17.
Experiments with single cantilevers demonstrated that data can be recorded at
a density of 641 Gb/in2 and read back with raw error rates better than 104 when
the x/y actuation is performed by an off-the-shelf piezoscanner with closed-loop
position control via internal capacity sensors [80]. The error-rate performance as a
function of off-track position with 641 Gbit/in2 is showed in Fig. 12.18.
The displacement control problem of microscanner can be divided into X and Y
directions, with low interacting. So the microscanner design can be used for parallel-
probe-based mobile storage devices or other nanopositioning system [81]. Because
the displacement of the scanner makes a change in the heater temperature, two
pairs of thermal position sensors are fixed on the cantilever-array chip to sense a
displacement of the object by temperature, as shown in Fig. 12.19.
12 Nanopositioning for Lithography and Data Storage 395

Fig. 12.17 Experimental results [77]

Fig. 12.18 Raw bit-error rate

as a function of distance from
the track centerline [80]

A current-to-voltage amplifier is used to measure the changes in the resulting

current. The sensors are used in pairs using a differential configuration to minimize
drift effects. The experiments result that the linearity, drift, and noise limit perform
good, even the device is simple. These sensors’ bandwidth can reach 4 kHz and the
displacement resolution is less than 1 nm [83].
Because the fabrication of the electrostatic microactuators is easy, the elec-
trostatic microactuators are widely used in the MEMS field and data storage.
A complete electrostatic microactuators has two parts. The stator is fixed while
the shuttle is movable. When voltage is applied, the electrical force is generated
between the two combs. The overlap area of the fingers and the distance between
the electrodes increase, as shown in Fig. 12.20.
Engelen et al. design a prototype x/y scanner using electrostatic comb drive
actuation, as shown in Fig. 12.21.
In the z direction, etching the device through the full thickness of the wafer
(400 m) leads to a large out-of-plane stiffness for passive shock. The large etch
396 W. Zhang et al.

Fig. 12.19 Microscanner and thermal position sensors [82]

Stator

Motion

Motion
interdigitated Shuttle parallel-plates actuator
comb actuator

Fig. 12.20 Electrostatic micro-actuators [13]

depth is necessary for the use of a relatively large gap between fingers (25 m)
but it reduced the actuation force. To increase the force without decreasing the
minimum etch trench width, a taped shape [85, 86] and a new ‘stepped’ shape,
shown in Fig. 12.22.
The simulation result of the three finger shapes is shown in Fig. 12.23.
From the simulation result it can be got that stepped shape is better than the
other two. To provide the margin for control operations, the range of achievable
displacement needs to be increased by increasing the force available at large positive
displacements. The drawback of the interdigitated comb drives is that it generates
weak force while it needs high voltage.
12 Nanopositioning for Lithography and Data Storage 397

Fig. 12.21 Photograph of a fabricated device (2  2 cm). The wires are electrical connections to
the isolated stators and to ground the moving parts of the device (bottom right) [84]
50 mm

50 mm
50 mm

50 mm

50 mm
50 mm

straight tapered stepped

Fig. 12.22 The three investigated finger shapes, showing the new stepped finger shape on the right
[84]
398 W. Zhang et al.

a 5 b 60

4 50 stepped
Available force (mN)

Displacement (μm)
3
stepped 40
2
only springs 30
1 Fcomb tapered
tapered
20
0

−1 10
straight
−2 0
−60 −40 −20 0 20 40 60 0 20 40 60 80 100 120 140
Position (μm) Voltage (V)
The simulation result of the three finger shapes Displacement measurements on a device with
tapered and a device with stepped fingers

Fig. 12.23 Simulation results [84]

Fig. 12.24 Schematic depiction of the X-Y scanner with the MR-based position sensors [87].
Copyright 2010. Reproduced with permission from the POL publication

Kartik et al. develop a high-speed x/y-positioning scanner design with high-

bandwidth piezoelectric actuator [87], as shown in Fig. 12.24. A suitable high-speed
scanner for nanopositioning system should have high first resonances and a low
coupling between the three orthogonal directions.
The high first resonances are realized by the high stiffness of the flexures and
low carrier. The low cross-coupling is achieved by low mass and uncomplicated
mechanical connections. The scanner comprises one stage each in the X- and
Y-scan directions, which is actuated by piezoelectric stacks. Piezo-electric stack
actuators have high stiffness and ability to generate large actuation forces, which is
12 Nanopositioning for Lithography and Data Storage 399

X position (μm)

4
Y position (μm)

0
0 5 10 15 20
Time (ms)

Fig. 12.25 The response of the scanner along the X- and the Y-scan directions to a 250 Hz trian-
gular drive signal [87]. Copyright 2010. Reproduced with permission from the POL publication

suitable for such application. Due to the requirement for high scan speed and high
actuation bandwidth, the allowable capacitances of the selected piezostacks have an
upperbound. The maximum achievable elongation and blocking force of the piezo-
actuator is also constrained. The effective mass carried by the X-Y scan table has to
be reduced in view of the limit imposed by the requirement for speed on the flexural
rigidity of the scanner. So in the Z-direction, the scan table is not actuated. Piezo-
actuators have hysteresis characteristic, which has bad effect on the scanner. When
a triangular drive signal at a frequency of 250 Hz applied at either of the X- and
Y-directions, the displacement of the scan table is shown in Fig. 12.25.
From the trace of Fig. 12.25, it is apparent to see the presence of hysteresis
in the displacement response relative to the reference. It is a behavior typical
of piezo-actuators when driven with a voltage amplifier. To solve this, Fleming
et al. place charge amplifiers as the driving source instead of voltage amplifiers
[88]. The positional signal is provided by Magneto-Resistance (MR) sensors. MR
sensors are traditionally used as magnetic sensing [89]. The external magnetic field
changes the electrical resistance of spintronics materials. The magneto-resistance
effect is caused by the scattering of electrons which is measured by their mean
free path length. Giant magneto-resistance (GMR) sensors, which is a frequently
used advanced MR sensors, consist of ultra-thin layers and pinned magnetic layers.
Daughton et al. have demonstrated that the MR sensors have high sensitivity, low
noise, and high bandwidth [90], which achieve high resolution position sensing.
The key problem is how to translate the motion of the scanner into a change in the
400 W. Zhang et al.

Displacement

GMR N
Sensing
Elements

GMR Sensing Elements

(Shielded)

Fig. 12.26 Schematic depiction of the magneto-resistance based position sensing concept [87].
Copyright 2010. Reproduced with permission from the POL publication

magnetic field, which can be seen by an MR sensor. Sahoo et al. proposed a similar
idea for topography sensing using micro-cantilevers [91]. The schematic depiction
of the MR-based position sensor is shown in Fig. 12.26.
Four GMR sensors are configured into a Wheatstone bridge. Two of them are
active GMR sensing elements while the other two are shielded GMR sensing
elements. The sensor is sensitive in one direction in its plane with a cosine scale off
in sensitivity as the sensor is rotated away from its sensing direction. The open-loop
dynamics of the high-speed scanner were characterized by both the laser Doppler
vibrometer and the MR-based position sensor. A laser Doppler vibrometer was used
to calibrate the MR-based position sensors. When a white noise excitation is applied
to the actuating piezo-stack in the corresponding direction, the open-loop frequency
response of the stage in the X- and Y-directions is measured by the MR-based
position sensor. The result is shown in Fig. 12.27.
The measured frequency responses for both axes can be fitted as sixth-order
transfer functions, which can be used for controller design. A block diagram
describing the closed loop is shown in Fig. 12.28.
The X- and Y-axes were controlled independently using H1 control framework,
as shown in Fig. 12.29.
Wp and Wu first-order weighting functions to specify the closed loop transfer
functions. Wp captures the requirement on tracking, and Wu ensures that the control
effort is within limits and that the higher unmodelled resonance modes are not overly
excited. The signal w1 is known as the “exogenous input” and z1 and z2 are known as
“exogenous outputs.” The “tracking error” and “control signal” are denoted by v and
u, respectively. The H1 control design problem is to find a stabilizing controller K1
such that kTzw k1 < 1 , where 1 is a constant approximately equal to 1 [92]. Here
12 Nanopositioning for Lithography and Data Storage 401

Fig. 12.27 Left: Open-loop frequency response of the scanner along the X-scan direction,
measured with the MR sensor (solid line-type). Right: Open-loop frequency response of the scanner
along the Y-scan direction, measured with the MR sensor (solid line-type); the corresponding sixth-
order transfer function fit (dashed line-type) accurately captures the first two resonant modes [87].
Copyright 2010. Reproduced with permission from the POL publication

Fig. 12.28 Block diagram

depiction of the closed loop
architecture for each axis
[87]. Copyright 2010.
Reproduced with permission
from the POL publication

Fig. 12.29 Block diagram

depiction of the H1 control
formulation [87]. Copyright
2010. Reproduced with
permission from the POL
publication

 
z1
Tzw is the closed-loop transfer function matrix relating z with w, where z D
z2
and w D Œw1 . The H1 synthesis was used as the eighth order controllers, realized
by a digital signal processor with a sampling period of 10 s [87]. The experimental
frequency response of the closed loop system for both the X- and Y-scan axes is
shown in Fig. 12.30.
The closed loop system performance is shown in Fig. 12.31 when a triangular
signal at 250 Hz, which is a typical reference for imaging applications, is applied
along the X-scan direction, while stepping in the Y-scan direction by 100 nm. Since
both the scan directions are controlled with very high bandwidths, minimal cross-
coupling between the axes is seen.
Positioning for data storage system has a strict requirement for accuracy. Single
position sensor is easily effected by drift and low-frequency noise. Pantazi et al.
developed a medium-derived position-sensing concept which provides positional
402 W. Zhang et al.

Magnitude (dB) 0

−20

−40

0
Phase (°)

−100
−200
−300
102 103 104
Frequency (Hz)

Fig. 12.30 Experimental closed loop frequency response for the X- (solid line-type) and Y-scan
directions (dashed line-type) [87]. Copyright 2010. Reproduced with permission from the POL
publication

3
Y position (μm)

0
0 1 2 3 4
X position (μm)

Fig. 12.31 A closed loop scan operation along the X-scan direction performed while stepping in
the Y-scan direction by 100 nm [87]. Copyright 2010. Reproduced with permission from the 5th
IFAC Symposium on Mechatronic Systems

information in the cross-track direction along the y-scan direction [93]. The “self-
servo-write” process relies entirely on the thermal position sensors to move the
scanner in closed-loop mode to the right position for inscribing the servo pattern
[77]. Therefore, position accuracies are below the resolution of these sensors. Using
12 Nanopositioning for Lithography and Data Storage 403

Fig. 12.32 Left: Closed-loop transfer functions corresponding to the x-axis and the controller.
Right: Closed-loop transfer functions corresponding to the x-axis and the resonant controller [95]

Fig. 12.33 2DOF control architectures: The schemes (1) and (2) are equivalent as control designs
in that one can be retrieved exactly in terms of the other. Practical implementable designs for the
controllers in (1) and (2) can easily be derived from the control design in (3), however the opposite
situation may require certain factorization procedures [96]

this concept of directed shaping of the noise sensitivity transfer function, positioning
accuracies can realize far below the noise levels of the position sensors without
incurring a signification compromise on the tracking bandwidth [94]. When the
scanner was only controlled in the frequency area where control was essential, the
measurement noise doesn’t impact the positioning accuracy in other frequencies
[95]. So the noise was shaped. H1 control design approach and resonant controllers
were implemented for the self-servo write process. The response of different signals
is shown in Fig. 12.32.
Lee and Salapaka propose two-degree-of-freedom controllers, as shown in
Fig. 12.33, to decouple the requirements for tracking performance and tolerance
to measurement noise [96].
The drawback is that complexity of system increased in linear feedback control.
Tuma et al. proposed a novel nonlinear control based on impulsive control to
improve the tracking performance without compromise on bandwidth [97]. The
principle of impulsive control is to change the controller states impulsively at
discrete time instants, as shown in Fig. 12.34.
Using triangular waveform to evaluate this method, the simulation results of a
range of controller gains and for a triangular reference signal of varying frequency
were shown in Fig. 12.35.
404 W. Zhang et al.

Fig. 12.34 Left: Linear feedback loop without impulsive state multiplication (ISM). Right: The
same feedback loop with ISM [97]. Copyright 2010. Reproduced with permission from the POL
publication

Fig. 12.35 Left: Reference signal. Right: Tracking error in steady state [97]. Copyright 2010.
Reproduced with permission from the POL publication

For different controller gains ki , the comparison of the tracking error for a
triangular reference signal using the feedback loops with and without impulsive
state multiplication (ISM) is shown in Fig. 12.36.
In the experiment presented, it can be shown that impulsive control offers better
transient performance and easier implements. For piecewise affine reference signals,
the stability conditions and performance bounds of impulsive control were derived
[98]. Experiments indicate that in a high-speed AFM system, an impulse controller
improved the tracking performance without increasing the bandwidth of the linear
feedback loop [99].
The development of the nanotechnology has opened a new application area,
especially in nanopositioning. This chapter has introduced some classical actuators,
sensors, and control strategies applied in nanopositioning for lithography and data
storage. It can be seen that nanopositioning must be applied in more area in the
future.
12 Nanopositioning for Lithography and Data Storage 405

Fig. 12.36 The comparison of tracking error [97]. Copyright 2010. Reproduced with permission
from the POL publication

Acknowledgement Thanks to the organizations and individuals for granting the pictures.

References

1. B. Shapiro, Workshop on control of micro-and nano-scale systems (2005)

2. T. Sheetz, J. Vidal, T.D. Pearson et al., Nanotechnology: awareness and societal concerns.
Technol. Soc. 27(3), 329–345 (2005)
3. Bhushan B. in Springer Handbook of Nanotechnology, ed. by B. Bhushan. Nanotechnology,
vol 99 (Springer Heidelberg, 2004), p. 145
4. B. Bhushan, Nanotechnology: Molecular Speculation on Global Abundance (MIT Press,
Cambridge, 1996)
5. G. Binnig, H. Rohrer, Scanning tunneling microscopy. IBM J. Res. Dev. 44(1-2), 279–293
(2000)
6. G. Binnig, C.F. Quate, C. Gerber, Atomic force microscope. Phys. Rev. Lett. 56(9), 930 (1986)
7. Q. Zou, K.K. Leang, E. Sadoun et al., Control issues in high-speed AFM for biological
applications: collagen imaging example. Asian J. Control 6(2), 164–178 (2004)
8. R. Kassies, K.O. Van der Werf, A. Lenferink et al., Combined AFM and confocal fluorescence
microscope for applications in bio-nanotechnology. J. Microsc. 217(1), 109–116 (2005)
9. K.D. Jandt, M. Finke, P. Cacciafesta, Aspects of the physical chemistry of polymers, biomate-
rials and mineralised tissues investigated with atomic force microscopy (AFM). Colloids Surf.
B Biointerfaces 19(4), 301–314 (2000)
10. I. Schmitz, M. Schreiner, G. Friedbacher et al., Phase imaging as an extension to tapping mode
AFM for the identification of material properties on humidity-sensitive surfaces. Appl. Surf.
Sci. 115(2), 190–198 (1997)
406 W. Zhang et al.

11. K. Yamanaka, A. Noguchi, T. Tsuji et al., Quantitative material characterization by ultrasonic

AFM. Surf. Interface Anal. 27(5-6), 600–606 (1999)
12. M. Tsukada, N. Kobayashi, M. Brandbyge et al., Physics of artificial nano-structures on
surfaces. Prog. Surf. Sci. 64(3), 139–155 (2000)
13. S. Devasia, E. Eleftheriou, S.O.R. Moheimani, A survey of control issues in nanopositioning.
IEEE Trans. Control Syst. Technol. 15(5), 802–823 (2007)
14. M. Yves, Scanning Probe Microscopes (SPIE, Bellingham, 1995)
15. R. Wiesendanger, T. Mulvey, Scanning probe microscopy and spectroscopy. Meas. Sci.
Technol. 6(5), 600 (1995)
16. Z. Rihong, X. Daocai, Y. Zhixing, C. Jinbang, Research on systems for measurements of CCD
parameters, in Proceedings of SPIE Detectors, Focal Plane Arrays, Imaging Devices II (1998),
pp. 297–301
17. D. Krogmann, H.D. Tholl, P. Schreiber, et al., Image multiplexing system on the base of
piezoelectrically driven silicon microlens arrays, in 3rd International Conference on Micro
Opto Electro Mechanical Systems, MOEMS, vol 99 (1999), pp. 178–185
18. J.C. Fasik. An inchworm actuator for the next generation space telescope (Burleigh Instru-
ments, Inc., Fishers, 1998)
19. B. Zhang, Z. Zhu, Developing a linear piezomotor with nanometer resolution and high stiffness.
Mechatron. IEEE/ASME Trans. 2(1), 22–29 (1997)
20. Handbook of micro/nano tribology[M] (CRC, 1998)
21. S.V. Kalinin, D.A. Bonnell, Effect of phase transition on the surface potential of the BaTiO3
(100) surface by variable temperature scanning surface potential microscopy. J. Appl. Phys.
87(8), 3950–3957 (2000)
22. D.R. Meldrum, W.H. Pence, S.E. Moody, et al., Automated, integrated modules for fluid
handling, thermal cycling and purification of DNA samples for high throughput sequencing
and analysis, in Advanced Intelligent Mechatronics, 2001. Proceedings. 2001 IEEE/ASME
International Conference on. IEEE (2001), vol 2, pp. 1211–1219
23. G.M. Whitesides, L.J. Christopher, The art of building small. Sci. Am. 285(3), 38–47 (2001)
24. Nanolithography: a borderland between STM, EB, IB, and X-ray lithographies[M] (Springer
Science & Business Media, 2013)
25. P. Vettiger, U. Staufer, D.P. Kern, Preface. Microelectron. Eng. 32(1), 1–2 (1996)
26. L.R. Harriott, Limits of lithography. Proc. IEEE 89(3), 366–374 (2001)
27. R. Horowitz, T.L. Chen, K. Oldham, et al., in Springer Handbook of Nanotechnology,
Microactuators for dual-stage servo systems in magnetic disk files (Springer Berlin Heidelberg,
2004), pp. 951–981
28. S. Hosaka, A. Kikukawa, H. Koyanagi et al., SPM-based data storage for ultrahigh density
recording. Nanotechnology 8(3A), A58 (1997)
29. A. Sebastian, A. Pantazi, G. Cherubini, et al., Nanopositioning for probe storage, in American
Control Conference, 2005. Proceedings of the 2005. IEEE (2005), pp. 4181–4186
30. D.L. White, O.R. Wood II, Novel alignment system for imprint lithography. J. Vac. Sci.
Technol. B 18(6), 3552–3556 (2000)
31. S. Verma, W. Kim, H. Shakir, Multi-axis maglev nanopositioner for precision manufacturing
and manipulation applications. Ind. Appl. IEEE Trans. 41(5), 1159–1167 (2005)
32. A.J. Fleming, S.O.R. Moheimani, Precision current and charge amplifiers for driving highly
capacitive piezoelectric loads. Electron. Lett. 39(3), 282–284 (2003)
33. D. Croft, S. Devasia, Vibration compensation for high speed scanning tunneling microscopy.
Rev. Sci. Instrum. 70(12), 4600–4605 (1999)
34. D. Croft, G. Shed, S. Devasia, Creep, hysteresis, and vibration compensation for piezoactua-
tors: atomic force microscopy application. J. Dyn. Syst. Meas. Control 123(1), 35–43 (2001)
35. G. Schitter, P. Menold, H.F. Knapp et al., High performance feedback for fast scanning atomic
force microscopes. Rev. Sci. Instrum. 72(8), 3320–3327 (2001)
36. S. Salapaka, A. Sebastian, J.P. Cleveland et al., High bandwidth nano-positioner: a robust
control approach. Rev. Sci. Instrum. 73(9), 3232–3241 (2002)
12 Nanopositioning for Lithography and Data Storage 407

37. O. M. El Rifai, K. Youcef-Toumi. Design and control of atomic force microscopes, in American
Control Conference, 2003. Proceedings of the 2003. IEEE (2003) vol. 5, pp. 3714–3719
38. A. Stemmer, G. Schitter, J.M. Rieber et al., Control strategies towards faster quantitative
imaging in atomic force microscopy. Eur. J. Control 11(4), 384–395 (2005)
39. T. Osada, M. Godwin, International technology roadmap for semiconductors (1999)
40. S.Y. Chou, P.R. Krauss, P.J. Renstrom, Imprint of sub-25 nm vias and trenches in polymers.
Appl. Phys. Lett. 67(21), 3114–3116 (1995)
41. J. Strong, The Johns Hopkins university and diffraction gratings. JOSA 50(12), 1148–1151
(1960)
42. D.J. Schroeder, An echelle spectrometer-spectrograph for astronomical use. Appl. Optics
6(11), 1976–1980 (1967)
43. R.G. Tuli, Instrumentation in astronomy VIII (SPIE, Washington, USA, 1994), pp. 674–668
44. S.D. Berger, J.M. Gibson, New approach to projection-electron lithography with demonstrated
0.1 m linewidth. Appl. Phys. Lett. 57(2), 153–155 (1990)
45. H.C. Pfeiffer, W. Stickel, PREVAIL-an e-beam stepper with variable axis immersion lenses.
Microelectronic Eng. 27(1), 143–146 (1995)
46. A.N. Broers, J.M.E. Harper, W.W. Molzen, 250-Å linewidths with PMMA electron resist.
Appl. Phys. Lett. 33(5), 392–394 (1978)
47. P.B. Fischer, S.Y. Chou, 10 nm electron beam lithography and sub-50 nm overlay using a
modified scanning electron microscope. Appl. Phys. Lett. 62(23), 2989–2991 (1993)
48. K. Early, M.L. Schattenburg, H.I. Smith, Absence of resolution degradation in X-ray lithogra-
phy for œ from 4.5 nm to 0.83 nm. Microelectronic Eng. 11(1), 317–321 (1990)
49. N. Vorbringer-Doroshovets, F. Balzer, R. Fuessl, et al., 0.1-nanometer resolution positioning
stage for sub-10 nm scanning probe lithography, in SPIE Advanced Lithography. International
Society for Optics and Photonics (2013), pp. 868018-868018-10
50. G.R. Harrison, N. Sturgis, S.C. Baker, G.W. Stroke, Ruling of large diffraction gratings with
interferometric control. J. Opt. Soc. Am. 47(1), 15–22 (1957)
51. G.R. Harrison, S.W. Thompson, Large diffraction gratings ruled on a commercial measuring
machine controlled interferometrically. JOSA 60(5), 591–595 (1970)
52. G.R. Harrison, S.W. Thompson, H. Kazukonis, et al., 750-mm ruling engine producing large
gratings and echelles. JOSA 62(6), 751–756 (1972)
53. C. Mitchell, Diffraction grating fabrication in Australia. Spectrochimia Acta Part B, 54,
2041–2204 (1999)
54. T. Kita, T. Harada, Ruling engine using a piezoelectric device for large and high-groove density
gratings. Appl. Optics 31(10), 1399–1406 (1992)
55. E. Kouno, P.A. McKeown, A fast response piezoelectric actuator for servo correction of
systematic errors in precision machining. CIRP Ann. Manuf. Technol. 33(1), 369–372 (1984)
56. J.F. Cuttino, A.C. Miller Jr., D.E. Schinstock, Performance optimization of a fast tool servo for
single-point diamond turning machines. Mechatron. IEEE/ASME Trans. 4(2), 169–179 (1999)
57. K.K. Tan, T.H. Lee, H.X. Zhou, Micro-positioning of linear-piezoelectric motors based on a
learning nonlinear PID controller. Mechatron. IEEE/ASME Trans. 6(4), 428–436 (2001)
58. M.F. Heertjes, X.G.P. Schuurbiers, H. Nijmeijer, Performance-improved design of N-PID
controlled motion systems with applications to wafer stages. Ind. Electron. IEEE Trans. 56(5),
1347–1355 (2009)
59. G.J. Maeda, K. Sato, H. Hashizume et al., Control of an XY nano-positioning table for a
compact nano-machine tool. JSME Int. J., Ser. C 49(1), 21–27 (2006)
60. H. Numasato, M. Tomizuka, Settling control and performance of a dual-actuator system for
hard disk drives. Mechatron IEEE/ASME Trans. 8(4), 431–438 (2003)
61. Leang K K, Devasia S. Hysteresis, creep, and vibration compensation for piezoactuators:
Feedback and feedforward control, in Proceedings of the Second IFAC Conference on
Mechatronic Systems (Berkeley, CA, 2002), pp. 9–11
62. J.M. Cruz-Hernández, V. Hayward, Phase control approach to hysteresis reduction. Control
Syst. Technol., IEEE Trans. 9(1), 17–26 (2001)
408 W. Zhang et al.

63. H. Shakir, W.J. Kim, Nanoscale path planning and motion control with maglev positioners.
Mechatron. IEEE/ASME Trans. 11(5), 625–633 (2006)
64. B.E. Helfrich, C. Lee, D.A. Bristow et al., Combined-feedback control and iterative learning
control design with application to nanopositioning systems. Control Syst. Technol. IEEE Trans.
18(2), 336–351 (2010)
65. S. Bashash, N. Jalili, Robust adaptive control of coupled parallel piezo-flexural nanoposition-
ing stages. Mechatron. IEEE/ASME Trans. 14(1), 11–20 (2009)
66. A. Sebastian, S.M. Salapaka, Design methodologies for robust nano-positioning. Control Syst.
Technol. IEEE Trans. 13(6), 868–876 (2005)
67. K. Kuhnen, H. Janocha, Inverse feedforward controller for complex hysteretic nonlinearities in
smart-material systems. Control Intell. Syst. 29(3), 74–83 (2001)
68. H. Shinno, H. Hashizume, H. Sato, Nanometer positioning of a linear motor-driven ultrapreci-
sion aerostatic table system with electrorheological fluid dampers. CIRP Ann. Manuf. Technol.
48(1), 289–292 (1999)
69. Y.K. Yong, S.S. Aphale, S.O.R. Moheimani, Design, identification, and control of a flexure-
based xy stage for fast nanoscale positioning. Nanotechnol. IEEE Trans. 8(1), 46–54 (2009)
70. U. Staufer, in Scanning Tunneling Microscopy II. Surface modification with a scanning
proximity probe microscope (Springer Berlin Heidelberg, 1992), pp. 273–302
71. C.F. Quate, Manipulation and modification of nanometer scale objects with the STM,
in Highlights in Condensed Matter Physics and Future Prospects (Springer US, 1991),
pp. 573–630
72. H.J. Mamin, R.P. Ried, B.D. Terris et al., High-density data storage based on the atomic force
microscope. Proc. IEEE 87(6), 1014–1027 (1999)
73. G. Binnig, H. Rohrer, C. Gerber et al., 77 reconstruction on Si (111) resolved in real space.
Phys. Rev. Lett. 50(2), 120 (1983)
74. B.D. Terris, S.A. Rishton, H.J. Mamin et al., Atomic force microscope-based data storage:
track servo and wear study. Appl. Phys. A Mater. Sci. Process. 66, S809–S813 (1998)
75. A. Davidson, MEMS-actuated magnetic probe-based storage, in Magnetic Recording Confer-
ence, 2002. Digest of the Asia-Pacific. IEEE (2002), pp. CE3–CE3
76. P. Vettiger, G. Cross, M. Despont et al., The “millipede”-nanotechnology entering data storage.
Nanotechnol. IEEE Trans. 1(1), 39–55 (2002)
77. E. Eleftheriou, T. Antonakopoulos, G.K. Binnig et al., Millipede-a MEMS-based scanning-
probe data-storage system. Magn. IEEE Trans. 39(2), 938–945 (2003)
78. M. Lutwyche, C. Andreoli, G. Binnig et al., 5 5 2D AFM cantilever arrays a first step towards
a Terabit storage device. Sensors Actuators A 73(1), 89–94 (1999)
79. P. Vettiger, G. Cross, M. Despont et al., The “millipede”-nanotechnology entering data
storage[J]. IEEE Trans. Nanotechnol. 1(1), 39–55 (2002)
80. H. Pozidis, W. Haberle, D. Wiesmann et al., Demonstration of thermomechanical recording at
641 Gbit/in 2. Magn. IEEE Trans. 40(4), 2531–2536 (2004)
81. M.A. Lantz, H.E. Rothuizen, U. Drechsler et al., A vibration resistant nanopositioner for
mobile parallel-probe storage applications. Microelectromech. Syst. J. 16(1), 130–139 (2007)
82. M. Lantz, H.E. Rothuizen, U. Drechsler et al., A vibration resistant nanopositioner for mobile
parallel-probe storage applications[J]. J. Microelectromech. Syst. 16(1), 130–139 (2007)
83. M.A. Lantz, G.K. Binnig, M. Despont et al., A micromechanical thermal displacement sensor
with nanometre resolution. Nanotechnology 16(8), 1089 (2005)
84. M. Lantz, H.E. Rothuizen, U. Drechsler et al., A vibration resistant nanopositioner for mobile
parallel-probe storage applications [J]. Microelectromech. Syst. J. 16(1), 130–139 (2007)
85. M.A. Rosa, S. Dimitrijev, H.B. Harrison, Enhanced electrostatic force generation capability of
angled comb finger design used in electrostatic comb-drive actuators. Electron. Lett. 34(18),
1787–1788 (1998)
86. J.B.C. Engelen, H.E. Rothuizen, U. Drechsler et al., A mass-balanced through-wafer electro-
static x/y-scanner for probe data storage. Microelectron. Eng. 86(4), 1230–1233 (2009)
87. S. Devasia, E. Eleftheriou, S.O.R. Moheimani, A survey of control issues in nanopositioning[J].
IEEE Trans. Control Syst. Technol. 15(5), 802–823 (2007)
12 Nanopositioning for Lithography and Data Storage 409

88. A.J. Fleming, S.O.R. Moheimani, A grounded-load charge amplifier for reducing hysteresis in
piezoelectric tube scanners. Rev. Sci. Instrum. 76(7), 073707 (2005)
89. S. Parkin, X. Jiang, C. Kaiser et al., Magnetically engineered spintronic sensors and memory.
Proc. IEEE 91(5), 661–680 (2003)
90. J. Daughton, Spin-dependent sensors. Proc. IEEE 91(5), 681–686 (2003)
91. Sahoo D R, Sebastian A, Haberle W, et al. Magnetoresistive sensor based scanning probe
microscopy, in. Nanotechnology, 2009. IEEE-NANO 2009. 9th IEEE Conference on. IEEE
(2009), pp. 862–865
92. S. Skogestad, I. Postlethwaite, Multivariable feedback control: analysis and design (Wiley,
New York, 2007)
93. A. Pantazi, A. Sebastian, G. Cherubini et al., Control of MEMS-based scanning-probe data-
storage devices. Control Syst. Technol. IEEE Trans. 15(5), 824–841 (2007)
94. E. Eleftheriou, Nanopositioning for storage applications. Annu. Rev. Control 36(2), 244–254
(2012)
95. A. Sebastian, A. Pantazi, S.O.R. Moheimani et al., Achieving subnanometer precision in a
MEMS-based storage device during self-servo write process. Nanotechnol. IEEE Trans. 7(5),
586–595 (2008)
96. C. Lee, S.M. Salapaka, Robust broadband nanopositioning: fundamental trade-offs, analysis,
and design in a two-degree-of-freedom control framework. Nanotechnology 20(3), 035501
(2009)
97. T. Tuma, A. Pantazi, J. Lygeros, et al., Tracking of high frequency piecewise affine signals
using impulsive control, in Proceedings of 5th IFAC symposium on mechatronic systems
(2010), pp. 90–95
98. T. Tuma, A. Pantazi, J. Lygeros, et al., Impulsive control for nanopositioning: stability
and performance, in Proceedings of the 14th international conference on Hybrid systems:
computation and control. ACM, 2011, pp. 173–180
99. T. Tuma, A. Sebastian, W. Häberle et al., Impulsive control for fast nanopositioning. Nanotech-
nology 22(13), 135501 (2011)