See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/328353870

Characterization of Optical Thin Films by Spectrophotometry and Atomic

Force Microscopy

Article · April 2015

CITATIONS READS

6 59,001

4 authors, including:

Shuvendu Jena Raj Tokas

Bhabha Atomic Research Centre Bhabha Atomic Research Centre
67 PUBLICATIONS 803 CITATIONS 86 PUBLICATIONS 981 CITATIONS

SEE PROFILE SEE PROFILE

Sonika Thakur
Himachal Pradesh University
104 PUBLICATIONS 1,612 CITATIONS

SEE PROFILE

All content following this page was uploaded by Shuvendu Jena on 18 October 2018.

The user has requested enhancement of the downloaded file.

 SMC Bulletin Vol. 6 (No. 1) April 2015

Characterization of Optical Thin Films by Spectrophotometry and Atomic

Force Microscopy
S. Jena, R.B. Tokas, S. Thakur and N.K. Sahoo
Atomic & Molecular Physics Division, Bhabha Atomic Research Centre, Mumbai 400085, India
E-mail: [email protected]
Abstract:
The present article provides an overview of optical and morphological characterization of optical thin
films by using spectrophotometry and atomic force microscopy (AFM) respectively. The basic principle
of spectrophotometer for thin film transmission measurement is discussed briefly. The optical properties
such refractive index, extinction co-efficient and band gap along with film thickness is determined from
the transmission spectrum using two popular method (i) Envelope method and (ii) Inverse synthesis
method. The detail theoretical formulation of the two methods has been discussed and applied to a thin
film sample for example purpose. The working principle and different imaging modes of AFM is explained
briefly to gain basic understanding about this technique. The surface quality of any optical thin film is
qualified by probing its surface topography using AFM. The procedure for deriving important surface
parameters such as root mean square (RMS) roughness and grain size or particle size from the AFM image
is discussed with an example. The power spectral density (PSD) of AFM data provides a more thorough
description of the morphology than only rms roughness. The detail procedure of extracting different surface
characteristic parameters from the average PSD function derived from the topography has been described
and implemented for normally and glancing angle deposited HfO2 thin films as an example.

1. Introduction films is responsible for optical losses, because of scattering

Spectral characteristics and surface morphology are the mainly for UV and deep-UV optical devices. This optical
two most important characteristics, which mainly defines loss reduces the efficiency of the optical devices. Therefore
the quality of any thin film optical coatings. The spectral it is necessary to have minimum roughness for fabricating
characteristics (transmission, reflection and absorption) high quality optical devices. The optical scattering
generally measured by UV-VIS-NIR spectrophotometry, phenomena strongly depend on both the morphology as
while surface morphology is probed through atomic force well as the grain structure of the films. The AFM, since
microscopy (AFM). The spectral characteristics of optical its inception, has been used by various researchers in
thin films depends on the wavelength dependence of different fields to image and to obtain qualitative and
the optical constants, i.e. the refractive index n (λ) and quantitative information about surface morphology,
the extinction co-efficient k (λ) of thin films, and their grain structure, elasticity, viscoelasticity, frictional force,
physical thickness, d. The determination of the optical photoconductivity, chemical composition mapping etc. In
constants is of great importance in the design of any optical this article, we will only discuss the working principle and
interference filters. Few examples of optical interference potential application of AFM to characterize the surface
filters are band pass filters, beam splitters, neutral density roughness and grain structure of thin film optical coatings.
filters, antireflection coatings, high reflectivity mirrors Further, the parameters like micro-surface roughness,
etc. Successful fabrication of such filters is often directly lateral distributions of heights and surface microstructure
related to the accurate value of n (λ), k (λ) and d. There of thin films are determined using power spectral density
are numerous methods for the determination of optical (PSD) function4. The detail formulation of PSD to extract
constants of thin films. But we will focus on two routinely these parameters by fitting with suitable combination
used methods such as Envelope/Swanepoel method1,2 of theoretical models is discussed. The transparent
and Inverse synthesis method3, in which optical constants refractory oxide coating materials generally used for
and film thickness are derived from single transmission development of variety of optical interference filters is
measurement only. Surface roughness and grain structure HfO2, ZrO2, SiO2, TiO2, Ta2O5, Nb2O5, Sc2O3, Gd2O3 etc5.
has strong influence on many physical properties such They are used, because of their large transparent spectral
as optical, tribological, adhesion etc. Therefore accurate range from ultraviolet to infrared region, better thermal,
evaluation of such surface characteristics is necessary. chemical and mechanical stability and high laser damage
For optical coating applications, surface roughness of thin threshold5. Finally these two characterization techniques

1
SMC Bulletin Vol. 6 (No. 1) April 2015

are implemented on few thin film optical coatings to show over time. The advantages to a single beam configuration
their potential applications to characterize their optical and are that there are often fewer moving parts. This makes the
morphological properties. instrument simpler and less likely to have parts wear out
or get out of alignment. This type of spectrophotometer is
2. Spectrophotometry: usually less expensive.
Spectrophotometry is a technique to measure the
Double beam spectrophotometer: It overcomes certain
transmission, reflection and absorption characteristics
limitations of the single beam spectrophotometers
of a material as a function of wavelength of light.
and is therefore preferred over them. A double beam
Spectrophotometry deals with near UV, visible and near IR
spectrophotometer has two light beams, one of which
light. The application of spectrophotometers is enormous
passes through the sample while other passes through
in different field of science and technology. It includes
a reference cell as shown in Fig. 1. This allows more
semiconductors and laser technology, optics and photonics,
reproducible measurements as any fluctuation in the light
printing and forensic examination, as well as in laboratories
source or instrument electronics appears in both reference
for the study of materials, chemicals and biological
and the sample and therefore can easily be removed from
samples. There are two types of spectrophotometer (i)
the sample spectrum by subtracting the reference spectrum.
single beam spectrophotometer6 and (ii) double beam
Modern instruments can perform this subtraction
spectrophotometer7 , which can be used for measuring the
automatically. Simultaneous recording at multiple
transmission in low absorbing materials.
wavelengths allows recording of the entire spectrum at
Single beam spectrophotometer: The schematic diagram once. Double beam spectrophotometer is generally used
of a single-beam spectrophotometer is shown in Fig. 1. to investigate the spectral characteristics of thin films or
Light from the source lamp is collimated and then passed bulk materials. For transmission measurements in thin film
through a monochromator, which diffracts the light. The samples, the spectrophotometer quantitatively compares
narrow bandwidths of this diffracted spectrum are then the fraction of light that passes through a reference (air in
passed through a mechanical slit on the output side of the our case) and the thin film, then electronically compares the
monochromator. Then the intensity of the light transmitted intensities of the two signals and computes the percentage
through the sample is measured by detectors and converted of transmission of the sample compared to the reference.
to absorbance or transmittance. Generally Halogen lamp is
used for VIS-NIR light source while deuterium lamp is used 2.1 Optical Properties:
for UV-light source. The most commonly used detectors The optical properties can be extracted from
in the UV-VIS and NIR region are the photomultiplier transmission as well as from reflection spectrum depending
tubes (PMT) and PbS detector respectively. For better on the materials. The methods of extracting optical
performance of a single beam spectrophotometer, the light constants from only transmission spectrum can be classified
source, detector, and electronics must be reasonably stable into two categories. The first one is an Envelope method
and the second one is Inverse synthesis
method. The envelope method, one of the
algebraic methods, has been developed and
refined by several workers after the work of
Manifacier1 et al. Inverse synthesis method
is based on the optical structural modeling
followed by the fitting of the experimental
data with the theoretical one. In this
following section, we briefly describe the
general procedure of the two methods with
their advantage and disadvantage. These
two methods are applied to determine the
refractive index, extinction co-efficient and
thickness of ZrO2-MgO thin film deposited
using reactive electron beam evaporation
for example purpose. The transmission of
the film is measured in the range of 190-
Figure 1: Schematic of Single and double beam spectrophotometer 1200 nm with a wavelength resolution of

2
 Figure 3: Transmission spectrum of ZrO2-MgO
SMC Bulletin Vol.
thin film. Curves Ts, TM,
6 (No. 1) April 2015

The procedure of the Envelope method can be briefly describ

1 nm using SHIMADZU made UV-VIS-NIR (dual beam) The procedure of the Envelope method can be briefly
spectrophotometer. described
should
where misasis
wherembe follows.
referred
an an toThe
thedetails
integer
integer for for should
previous
maxima
maxima and be referred
reports.
and
half half
integer to minima
The integer
extremes
for of m
for
the previous reports. The extremes of the measured
2.1.1 Envelope Method: transmission spectrum
spectrum occur
wavelength. T and T occur
according to according
thethe
define to the condition
condition
envelopes passing tangentiall
wavelength. MTM andm Tm define the envelopes passing tan
Fig. 2 shows schematically the optical transmission
measurement on a film/substrate system, in which the 2nd masO shown in Fig. 3. The refractive index of the
minima (1)transparent
substrate is transparent and the film is semitransparent or minima as shown in Fig. 3. The refractive index of the tra
transparent. The n, k and d are refractive index, extinction
fromwhere m is an integer for maxima and half integer
the transmission spectrum Ts via the relation
co-efficient and thickness of the film, while ns, ks=0 and for minima
from and λ is the corresponding
the transmission spectrum Ts via wavelength. TM
the relation
ds are that of the substrate. The refractive index of air is and Tm define the envelopes passing tangentially through
the 1 1 minima as shown in Fig. 3. The refractive
assumed to be n0=1. If the film thickness is uniform, then n s maxima and 1
indexTof 1 transparent
s the T
21
the interference effects give rise to a spectrum as shown ns  s  1 substrate can be calculated from
in Fig. 3 (ZrO2-MgO thin film). These interference fringes T
the transmission
s s
2
T spectrum Ts via the relation
can be used to calculate the optical constants as well as Above the strong absorption region, the expression for refracti
thickness of the film using Envelope method. 1 1
ns = + −1 (2)
Above the Tstrong
Ts extreme
different
2 absorption region, the expression for
s points is given by

different
Above extreme points
the strong
2
is given
absorption by the expression
region,
n N  N n s
for refractive index at wavelength of different extreme
points is given by 2
n N  N T n s T n s2  1
M m
where N 2n s 
n = N + N − nTs2M Tm 2 (3)
TM  Tm 2 n s2  1
where N 2(1),
Using n T −T n +1
where equation
N = 2n s s Mwe can
m estimate the film thickness(4)by the relat
+ s
TMTTMmTm 2 2
O 1O 2
d Using equation (1), we can estimate the film thickness
Using
2[O nequation
(O1 )  O 2 n(O(1),
2 )] we can estimate the film thickness by
by the 1relation

where Ȝ1 andO 1Ȝ1

O 2are the wavelength of two adjacent maxima (or m
d (5)
2[O1n(O1 )  O 2 n(O 2 )]
thickness of the thin film can be obtained by following the proc
where λ1 and λ1 are the wavelength of two adjacent
Figure 2: Schematic of light transmission through a thin film/substrate where
The
maxima
Ȝ1 and
accurate Ȝ1 are
thickness
(or minima). ofthe
The thewavelength
moreZrO of two
2-MgOthickness
accurate film adjacent maxi
is 413±4
of thenm. Finally
system (T=IT/I0).
thin film can be obtained by following the procedure
thickness
and ofthickness
the final
described the thin
elsewhere. offilm
The canthickness
the film,
accurate be accurate
the obtained by
ZrOfollowing
- index n
refractive
of the 2
MgO film is 413±4 nm. Finally, from the exact value of m
each
and maximum
Thethe
accurate and minimum
thickness
final thickness using
offilm,
of the the ZrO Eq. (1) as
2-MgO
the accurate shown
film in Fig.nm
is 413±4
refractive 3
index n can again be calculated for each maximum and
function
and the offinal
minimum wavelength
using n (Ȝ)
Eq. (1) as
thickness of can
shown be
thein then3.the
Fig.
film, obtained
The by fitting
refractive
accurate the es
refractive
index as a function of wavelength n (λ) can be then obtained
8
Fig. 5 (a) the
by fitting using Cauchyvalues
estimated dispersion
as shown in Fig.. 5 (a) using
equation
each maximum and minimum using Eq. (1) as shown in
Cauchy dispersion equation8.
The extinction co-efficient (k) is given by
The extinction co-efficient (k) is given by
function of wavelength n (Ȝ) can be then obtained by fittin
D O ª (n  1)(n  n s )( TM  Tm ) º
k O  ln « » 8 (6)
Fig.4S5 (a) using
4Sd « (Cauchy
n  1)(n  ndispersion equation
s )( TM  Tm ) »
.
¬ ¼
The calculated extinction co-efficient at different
Thecalculated
The extinction
extreme
co-efficient
extinction
of the transmission
(k) is given
co-efficient by extreme of the tr
at different
spectrum along with its fitting
Figure 3: Transmission spectrum of ZrO2-MgO thin film. Curves Ts, using Cauchy dispersion model to get k (λ) is shown in
TM, and Tm, are as per the text. with
Fig. 5its
D(b).fitting using
O ªCauchy
(n  1)(n dispersion  Tmto) ºget k (Ȝ) is shown
 n s )( TM model
k O  ln « »
4S 4Sd « (n  1)(n  n s )( TM  Tm ) »
¬ ¼ 3

The calculated extinction co-efficient at different extreme

 SMC Bulletin Vol. 6 (No. 1) April 2015

This method works well for thick film having

interference fringes extremes. It fails at strong absorption
edge. This method also does not suitable if there are
local absorption bands in between interference fringes
extremes. It also fails if the film has high absorption
such that interference fringes are not visible unlike at
absorption edge. The accuracy of the method decreases
with decreasing film thickness. Because with decreasing
thickness, the spacing between interferences extremes
increases, this makes interpolation between extremes
more difficult9. Instead of all this pitfalls, the Envelope
method is widely used as routine analysis technique
for determination of optical constants of transparent or
semiconducting optical materials.
Figure 4: Experimental and fitted transmission spectrum of the
2.1.2 Inverse synthesis Method: ZrO2–MgO thin film.
The optical constants (n & k) of the films have been
For an example point of view, the experimental
obtained by fitting the measured transmission spectrum
transmission spectrum along with its fitting for ZrO2-MgO
with a theoretically generated spectrum. The reliability
thin film using single oscillator and Urbach dispersion
of inverse synthesis mainly depends on the validity of
model for refractive index and extinction co-efficient
optical model and the fitting statistics. The procedure of
respectivelyhad
oscillator is shown in Fig.contribution
the largest 4. Single oscillator
to n andmodela single ter
this method is as follows.
assumes that the material is composed of individual dipole
(i) Measure the transmission of thin film deposited on a oscillators that are set to forced vibration by the incident
dispersion of refractive index equation given by:
weakly absorbing substrate. light. Wemple and DiDomenico Jr. proposed that the lowest
oscillator
energy had the
oscillator largest
had contribution
the largest to n and
contribution to na and
single
a term Sel
(ii) The optical parameters and thickness of thin film as single term Sellmeier E0 E
relation describes the dispersion of
d
well as substrate should be assumed properly. The (O )
ndispersion
refractive
1index
of refractive
equation
2 index
2 equation
given by: given by:
transmission of the film/substrate structure as shown E 0  (hc / O )
in Fig. 1can then be readily calculated from the complex
E0 E d (8)
refractive index of the film (n+ik) and thickness (d) n(O ) h 1is Planck’s
Where (8
constant, c is the velocity of the light
using the expression given elsewhere10. E 2  (hc / O ) 2
0

(iii) The wavelength dependency of n (λ) and k (λ) are the Where h is Planck’s
dispersion energy constant,
and can cbe is the velocity
closely of the
associated with t
Where h is Planck’s constant, c is the velocity
generally expressed by different dispersion models light, E0 is the oscillator energy, Ed is the dispersion energy of the light, E0 i
and can be closely associated with the structural order
such as (1) Cauchy-Urbach model11 (2) single oscillator imaginary partenergy
the dispersion of theandrefractive
canofbe index,associated
theclosely
i.e., the extinction
with the stru
model (3) Adachi model (4) Forouhi and Bloomer parameter.
12 13 The imaginary part refractive index, i.e.,
model (5) Tauc-Lorentz model (6) Cody-Lorentz
14 15 the extinction coefficient (k) of the composite is assumed
(iv) Using
model16the above procedure, the fitting has been carried assumed
etc. imaginary
to
outfollow topart
the follow
Urbach
by minimizing thesquared
of the
the
modelUrbach
refractive
as givenmodel
index,
below.as given
i.e., below. coeffi
the extinction

(iv) Using the above procedure, the fitting has been carried
difference (Ȥ2) between the experimentally kassumed
measured (Oand to follow
1240the
) Dcalculated
exp( (1Urbach
/ O of1 /model
Evalues J )) as given below.(9)
out by minimizing the squared difference (χ2) between
the experimentally measured and calculated values of k (O ) D exp(1240 E (1 / O  1 / J )) (9
transmission given by:
transmission given by:
(a) (b
F 2 1
(2 N  P )
¦
N
i 1
>T i
Exp
 T i Cal
2
@ (7) (7) (a) (b)

Exp Cal
Where Ti and Cal T are the experimental and
Where Ti Exp and Ti i are
theoretical transmittances
the experimental and theoretical transmittances respectively.
respectively. N is the number
of data points, P is the number of model parameters and
the Nminimization
is the number has of data
been done usingPthe
points, is Levenberg-
the number of model
Figure parameters
5: (a) Refractive indexand theextinction co-efficient spectra of
and (b)
Marquardt algorithm. After the fine adjustment of model the ZrO2–MgO thin film.
minimization
parameters, a goodhas beencan
fitting done
be using the Levenberg-Marquardt algorithm. After the fine
achieved.

adjustment of model parameters, a good fitting can be achieved.

4
 SMC Bulletin Vol. 6 (No. 1) April 2015

Fig. 5(a) shows the refractive index of ZrO2-MgO thin where K is constant depends on the transition
film determined using both Envelope method and Inverse probability, Eg is the optical band gap and m determines
synthesis method. Fig. 5(b) shows the extinction co-efficient the type of transition (m=1/2 for direct allowed transition
spectra determined by both the methods. The refractive and m=2 for indirect allowed transition).
index and thickness value obtained by these two methods
The best fit of the experimental result is found using
are almost close, while the extinction co-efficient differs a
Eq. (11) with m=2 i.e. optical band gap is determined by
lot. The Inverse synthesis method is very powerful and
plotting (αhν)½ as a function of the photon energy hν and
can yield very accurate values of optical constants (n &
extrapolating the linear curve progression to zero. Fig.
k) for thin films or even multilayer thin films with some
6 shows the Tauc plot for HfO2 thin film prepared by
know parameter like film thickness or material parameters
electron beam evaporation to determine its optical band
as guess value. The only limitation of this method is that
procedure
one
9
. The
has to make advantageabout
assumptions of inverse synthesis
the type methodgap
of dispersion
(Eg=5.56
is that it caneV)
befor example purpose.
applicable to any type
relation before start the fitting procedure . The advantage 3. Atomic force microscopy (AFM):
9

ofinverse
of thin film, whilemethod
synthesis Envelope method
is that it canisbesuitable fortotransparent
applicable or semitransparent thick films.
AFM19 is most widely used technique to investigate
any type of thin film, while Envelope method is suitable surface morphology of any kind of materials. It uses near-
2.1.3
for Optical or
transparent band gap:
semitransparent thick films. field forces between atoms of the probe tip apex ad the
2.1.3 surface to generate signals of surface topography. This
The Optical bandgap
optical band gap:(E ) of the films is determined from
g the obtained
near field values
interactions of Waals
(van der absorption
force for topography)
The optical band gap (Eg) of the films is determined
eliminates the resolution limit associated with optical
from the obtained
co-efficient values of
(Į). Around absorption
strong co-efficient
absorption region,(α).
the optical absorption is computed from the by far-field
microscopy, because its resolution is limited
Around strong absorption region, the optical absorption
interactions between light and specimens. AFM is a high
is computed from the measured transmission (T) and 17
measured transmission (T) and reflection (R) 17spectra using the following
resolution technique,. which is less-destructive
imagingformula
reflection (R) spectra using the following formula .
than either transmission electron microscope or scanning
§ · electron microscope. AFM is capable of measuring the
1 ¨ 1 R ¸ vertical dimension (Z-direction) of the sample i.e. the
D ln¨ ¸ (10) (10)
d ¨T ¸ height or depth of the surface features, while optical and
© Tsub ¹
electron microscope does not have this scope.
Where d is the film thickness and Tsub is the transmission
ofWorking
the barePrinciple
substrate.: The operation of AFM for is
20
Where
of d issubstrate.
the bare the film thickness
The absorption and Tco-efficient
sub is the transmission
of thin The absorption
governed by three basic components (i) Cantilever tip (ii)
films (behave like amorphous semiconductor) in the high
piezoelectric scanner and (iii) feedback loop control system.
co-efficient
absorption of (α
region thin
≥ 10films
4
cm-1) (behave like amorphous
is given according to Tauc’s semiconductor) in the high absorption
A schematic diagram of a typical AFM system is shown in
relation18.
4 Fig.187. A sharp tip is mounted at the end of a cantilever. The
region
αhν (Į • 10
= K (hν-E ) m cm-1) is given according to Tauc’s relation
(11) .
cantilever beam is attached at one end to a piezoelectric
g
scanner. The force detection is based on a beam deflection
ĮhȞ = K (hȞ-Eg)m method. A (11)
diode laser is bounced off from the back side
of the cantilever (coated with metallic thin film for better
where K is constant depends on the transition probability, Eg isonto
reflection) the aoptical band gap and
position-sensitive m photodiode
quadrant
detector. This detector measures the bending of cantilever
determines the type of transition (m=1/2 for direct allowed during the transition and over
tip scanning m=2the forsample
indirect
surface. As the tip
is scanned across the sample in the x-y direction, the force of
allowed transition). the interaction between the sample and the tip is measured
by the cantilever. A feedback loop control system is used
to maintain a desired force between the tip and the sample.
When the measured force is larger or smaller than this set
point, the piezoelectric scanner in the z-direction moves
the probe away from or towards the surface, bringing
the force back to the set point. The displacement of piezo
scanner directly corresponds to the height of the sample.
A topographic image of the surface can be generated by
Figure 6: Representative Tauc plot for optical band gap determination scanning the tip over the sample surface and recording the
of HfO2 thin film displacement of the piezo-scanner as a function of position.

5
SMC Bulletin Vol. 6 (No. 1) April 2015

Figure 7: Schematic diagram of atomic force microscopy with beam Figure 8: Schematic of force-distance curve for interaction between tip
deflection method. and sample surface.

The cantilever is typically silicon or silicon nitride with a 3.1 Surface morphology:
tip radius of curvature on the order of nanometers. The surface morphology of optical thin films is
For surface topography, the dominant interactions at measured using AFM. AFM images have a lateral (x,
short tip-sample distances in the AFM are van der Waals y) resolution and a vertical (z) resolution. The radius of
interactions. The typical force-distance curve21 for van der curvature of the end of the tip will determine the highest
Walls interactions is shown in Fig. 8. Depending on the lateral resolution obtainable with a specific tip. The vertical
distance between tip-sample, there are 3 types of imaging resolution is necessary for determining surface roughness.
modes22 in AFM. They are (i) Contact mode (ii) Intermittent Root mean square (rms) surface roughness describes the
contact mode and (iii) Non-contact mode. finish or smoothness of optical surfaces. It represents
the standard deviation of surface profile heights23. The
Contact mode: The tip physically touches the surface
lateral resolution is necessary for determination of grain
and the force (10-6 N < F < 10-9) is repulsive as shown in Fig.
or particle size24. However, another factor that needs to
8. The force between the tip and sample remains constant
be considered during image analysis is the number of
by maintaining a constant cantilever deflection using the
data points, or pixels, present in an image in the x and y
feedback loops and an image of the surface is obtained. It scan-direction. For example, in acquiring a 5 μm × 5 μm
is a fast scanning mode and good for rough samples. It is area images with 512 pixels, the pixel size is ≈ 10 nm (5
not much suitable for soft samples as the tip may damage μm/512 pixels). In this case, it is not possible to resolve
the sample. features smaller than 10 nm at a 5 μm scan size. Thus, it is
Intermittent contact mode (Tapping): In this mode, the important to consider the grain size when choosing the scan
tip lightly “taps” on the sample surface at its resonant size. After AFM imaging, the image can be processed for
frequency during scanning. The tip contacts the surface determination of surface roughness as well as grain size.
at bottom of its oscillation. The force between the tip The surface roughness is derived from the vertical height,
and sample remains constant by maintaining constant while the grain size is derived from the lateral size of the
oscillation amplitude and an image of the surface is grains or particles in the image. A cross-sectional line can
obtained. It gives high resolution imaging for soft samples. be drawn across any part of the image to see lateral size
It is not suitable for imaging in liquids and the scan speed of grains, and the vertical profile along that line is also
is slower. displayed. By making several cross-sectional line profiles
through different grains, it is possible to calculate the grain
Non-contact mode: The tip oscillates above the sample or particle size. This procedure is applied to determine
at a distance of 10-100 nm during scanning. The surface the surface roughness and grain size of HfO2 thin film
topography can be measured by monitoring changes in the deposited using reactive electron beam evaporation. NT-
amplitude of oscillation due to attractive van der Waals MDT, P-47H AFM system has been used for measuring
forces (10-9 N < F < 10-12) as shown in Fig. 8 using a feedback the topography of the HfO 2 thin film morphological
loop. This mode extends the tip life time due to its very low measurements of obliquely deposited HfO2 thin films. A
force. It generally gives low resolution imaging. super sharp diamond like carbon (DLC) coated Si probe

6
 SMC Bulletin Vol. 6 (No. 1) April 2015

numbers of publications which describe the

PSD based surface statistics thoroughly. We
have adopted the formulation described in
references4,25 for the computation of PSDF,

1 N N
S 2 ( f x f y ) = 2 ∑∑ Z mn e
− 2πi∆L ( f x m + f y n )
(
L  m=1 n=1
2
1 N N 
S 2 ( f x f y ) = 2 ∑∑ Z mn e
− 2πi∆L ( f x m + f y n )
(∆L) 2  (12)
L  m=1 n=1 
Here S2 is the 2-dimensional PSDF, L2
is the scan surface area, N is the number
of data points in both X and Y direction
of scanned area, Zmn is the surface profile
height at position (m, n), fx, and fy are the
spatial frequencies in X and Y directions
respectively and ΔL (L/N) is the sampling
interval. Computation of PSDF is further
Figure 9: Representative (a) 2D, (b) 3D AFM morphology for determination of (c) grain
followed by transition to polar coordinates
size and (d) surface roughness of HfO2 thin film.
in frequency space and angular averaging
(φ)
having tip curvature 1-3 nm, resonance frequency 198 kHz
2π
and force constant 8.8 N/m has been used. DLC coated 1
AFM probe being very hard and anti-abrasive, was chosen S2 ( f ) =
π ∫S 2 ( f , ϕ )dϕ (13)
to get the consistency in measurements. 0

Measurements with different scan sizes such as 2.5x2.5 As the PSDF depends only on one parameter, it will
μm2, 5x5μm2 and 10x10 μm2 with spatial resolution of be plotted in all our figures as a ‘slice’ of the 2-D PSD
512x512 points have been taken for the film. The scan size function with unit ‘(length)4’. Several mathematical models
2.5x2.5 μm2 with 512x512 points can give better spatial alone and in combinations has been proposed and used
resolution of ≈ 5 nm, which would be useful for accurate by researchers to interpret experimental PSD. The most
determination of grain size or particle size. 9(a) & (b) used extended model for PSD of thin films is the sum
show the 2D and 3D AFM topography of electron beam of Henkel transforms of the Gaussian and exponential
evaporated HfO2 thin film. The grains are spherical in size autocorrelation functions. But such model fails when
with uniform size distribution. The estimated average grain wide spatial frequency range is considered. To describe
size is found to be 58.7 ± 0.6 nm as shown in Fig. 9(c). The roughness over large spectral frequencies, PSD model
RMS surface roughness determined from the height counts should comprises contribution from substrate, pure thin
is 4.24 nm as shown in Fig. 9(d). film and aggregates or superstructures. PSD of substrates
generally follows inverse power law 4,25 with spatial
3.2 Power Spectral Density: frequency (assuming
characterized by abc orfractal like surfaces).
k-correlation PSD
model4,25 function
, while PSD of aggre
The most important surface morphology parameter of pure thin film is conventionally characterized by abc
is the rms roughness. However this parameter does not or k-correlation model4,25, while PSD of aggregates or
characterized by Gaussian function with its maximum shiftin
account for characterized
the spatial distribution model4,25,superstructures
of peaks or valleys
by abc or k-correlation while PSD of is aggregates or superstructures
characterized is
by Gaussian function with
present in the surface. A more complete description is its maximum 4,25 shifting to a nonzero spatial frequency4,25. The
frequency . The combination of these PSD models is given by:
provided bycharacterized
power spectralby density (PSD) function
Gaussian of the surface
with its maximumof shifting
combination these PSDtomodels
a nonzero
is givenspatial
by:
topography. It describes both the aspects of surface
K a
topography such
frequency 4,25
as the spread. Theofcombination
heights from aofmean
and the lateral distance over which the height variation or
theseplane
PSD models
PSDTotal is given by:
f J 1 2 2 (c1) / 2
(1 b f )
¦ SVag,mWag,m exp[S 2Wag 2
,m( f  fag,m)
m
mutual correlation occurs.K For thin filmaoptical coatings the
PSD function isPSD
especially
Total
f 1

Jimportant
(1 b2 and
f
¦
 SVaag
2 (crepresents
) 1) / 2 ,mWag,m
very exp[
The
2 2
1stSterm
Wag,mis ( f power 2
 fag,m)law,
] 2nd term is abc-model
(14) 3rd tem is shif
and (14)
m
popular way to describe the surface quality. There are large
K is spectral strength of fractal and Ȗ is fractal spectral indices. The
The 1st term is power law, 2nd term is abc-model and 3rd tem is shifted Gaussian model. Here
7
related to intrinsic rms roughness (ıabc) and correlation length (IJab
K is spectral strength of fractal and Ȗ is fractal spectral indices. The parameters a, b and c are
SMC Bulletin Vol. 6 (No. 1) April 2015

resolution of 512x512 points have been

taken for the film. Power spectral density
function (PSDF) has been computed for
each scan size and extended PSDF has
been obtained by combining PSDF of
different scans together by a suitable
methodology as described elsewhere25.
A representative 2D and 3D AFM image
having scan size 1x1 μm2 for both normal
and glancing angle deposited HfO2 film
is shown in Fig. 10.
The experimental PSD functions
are fitted with combined PSD model as
described by Eq. (14) for the thin films
as shown in Fig. 11. It shows that PSD
function of GLAD thin film depicts
higher roughness for full range of spectral
frequencies as compared to normal
deposited film. The intrinsic roughness
which is the only contribution from pure
Figure 10: 2D and 3D AFM morphology of (a) normally deposited and (b) glancing angle
thin film i.e. σabc (0.79 nm) is higher for
deposited (GLAD) HfO2 thin film.
GLAD film as compared to that (0.17 nm)
The 1st term is power law, 2nd term is abc-model and 3rd of normal deposited film. This can be explained in the light
tem is shifted Gaussian model. Here K is spectral strength of atomic shadowing and re-emission & diffusion of ad-
of fractal and γ is fractal spectral indices. The parameters atoms. For glancing angle depositions, atomic shadowing
a, b and c are related to intrinsic rms roughness (σabc) and effect is very high and it leads to very high roughening
correlation length (τabc). The σag corresponds to the height effect to the surface of film. Consequently, film surface
and τag corresponds to the size of the aggregates or super roughening due to atomic shadowing dominates the
structures. For an example purpose, normal and glancing smoothing effects due to re-emission of ad-atoms diffusion
angle deposited HfO2 thin films have been studied using processes and results in the higher surface roughness as
PSD function. compared to the normal deposited film. It is known that
correlation length (τabc) represents the size of the grains on
Three different measurements having scan sizes i.e.
the surface The value τabc estimated from the PSD fitting
1x1 μm2, 2.5x2.5 μm2, 5x5μm2 and 10x10 μm2 with spatial
is 26.7 nm and 74.6 nm for normal and GLAD HfO2 thin
film respectively, which can be clearly verified by the
AFM image in Fig. 10. The highest grain size of the GLAD
film is the consequence of dominant atomic shadowing at
glancing angles. In fact, diameter of slanted columns is very
high at glancing angle which ultimately leads to bigger
grain size as reported in our earlier work26. This shows the
capability of PSD to extract different surface parameters
of thin films having different microstructures, which is
helpful to quantify the quality of optical surface.

4. Conclusion:
Characterization of optical properties using
spectrophotometry and surface morphology using AFM
are discussed in this article. The most commonly used
methods such as Envelope method and Inverse synthesis
Figure 11: Experimental and fitted extended PSDF of normal and GLAD method for determination of optical properties and film
deposited HfO2 thin films. thickness are described with an example. The fundamental

8
 SMC Bulletin Vol. 6 (No. 1) April 2015

principle and different imaging modes of AFM are briefly 11. D. V. Likhachev, Thin Solid Films, 2014, 562, 90.
presented. The surface morphological properties such as 12. S. H. Wemple, M. DiDomenico, Phy. Rev. B, 1971, 3, 1338.
RMS roughness and grain size are probed by analysing 13. S. Adachi, Phy. Rev. B, 1987, 35, 7454.
AFM image. The power spectral density of AFM image are 14. A. R. Forouhi, I. Bloomer, Phy. Rev. B, 1986, 34, 7018.
15. G. E. Jellison, F. A. Modine, Appl. Phys. Lett., 1996, 69, 371.
discussed and analysed for micro roughness parameters
16. J. Price, P. Y. Hung, T. Rhoad, B. Foran, A. C. Diebold, Appl.
of thin films. Phys. Lett., 2004, 85, 1701.
17. S. Jena, R. B. Tokas, N. M. Kamble, S. Thakur, N. K. Sahoo,
References: Appl. Opt., 2014, 53, 850.
1. J. C. Manifacier, J. Gasiot, J. P. Fillard, J. Phy. E, 1976, 9, 18. R. Naik, S. Jena, R. Ganesan, N. K. Sahoo, Phys. Status Solidi
1002. B, 2014, 251, 661.
2. R. Swanepoel, J. Phy. E, 1983, 16, 1214. 19. G. Binnig, C. F. Quate, C. Gerber, Phys. Rev. Lett., 1986, 56,
3. J. A. Dobrowolski, F. C. Ho, A. Waldorf, Appl. Opt., 1983, 22, 930.
3191. 20. E. Meyer, Prog. Surf. Sci., 1992, 41, 3.
4. J. Ferré-Borrull, A. Duparré, E. Quesnel, Appl. Opt., 2001, 40, 21. B. Cappella, G. Dietler, Surf. Sci. Rep., 1999, 34, 1.
2190. 22. S. N. Magonov, M.-H. Whangbo, Surface analysis with STM
5. M. Friz, F. Waibel, in Optical Interference Coatings, edited by and AFM: experimental and theoretical aspects of image analysis,
N. Kaiser, H. Pulker, Springer: Berlin Heidelberg, 2003, Vol. VCH Publishers: New York, 2008.
88, pp. 105. 23. N. K. Sahoo, S. Thakur, M. Senthilkumar, N. C. Das, Appl.
6. K. C. Kao, T. W. Davies, J. Phy. E, 1968, 1, 1063. Surf. Sci., 2003, 206, 271.
7. M. W. Jones, K. C. Kao, J. Phy. E, 1969, 2, 331. 24. J. Grobelny, F. DelRio, N. Pradeep, D.-I. Kim, V. Hackley, R.
8. D. Y. Smith, I. Mitio, K. William, J. Phys.: Condens. Matter, Cook, in Characterization of Nanoparticles Intended for Drug
2001, 13, 3883. Delivery, edited by Scott E. McNeil, Humana Press, 2011, Vol.
9. P. Dirk, S. Philippe Frederic, J. Phys. D: Appl. Phys., 2003, 36, 697, pp. 71.
1850. 25. M. Senthilkumar, N. K. Sahoo, S. Thakur, R. B. Tokas, Appl.
10. S. Jena, R. B. Tokas, N. Kamble, S. Thakur, D. Bhattacharyya, Surf. Sci., 2005, 252, 1608.
N. K. Sahoo, Thin Solid Films, 2013, 537, 163. 26. R. B. Tokas, S. Jena, P. Sarkar, S. R. Polaki, S. Thakur, S. Basu,
N. K. Sahoo, Mater. Res. Express, 2015, 2, 035010.

Shri. Shuvendu Jena joined Atomic & Molecular Physics division (A&MPD), BARC in the year 2010
after graduating from 53rd batch of BARC training school. He is working on Optical coatings and
devices for laser and spectroscopic applications for various DAE projects.

Shri. R. B. Tokas joined Atomic & Molecular Physics Division (A&MPD), BARC in the year 2003 after
graduating from 46th batch of BARC training school. He has been involved in activities like design
and development of interference multi-layer devices for various laser applications, investigation of
various properties of thin films of HfO2, SiO2, TiO2, Al2O3MgO, ZrO2 etc., used for the fabrication
of multilayer devices. Glancing angle deposition of oxide thin films, Super-hydrophilic anti-reflection
broadband coatings for solar applications
Dr. Sudhakar Thakur is working in Atomic & Molecular Physics Division (A&MPD), BARC. He
has been involved in activities like, development of beam splitters for selective dye laser pumping,
anti reflection coatings for CVL and other dye laser experiments, narrow band Fabry Perrot type
transmitting filters etc.

Dr. N.K Sahoo is an Outstanding Scientists and Head, Atomic & Molecular Physics Division of
Bhabha Atomic Research Centre. Dr. Sahoo, after graduating from Ravenshaw College completed his
post-graduation from Utkal University with the specialization in Solid-state Physics. Subsequently
he joined the 25th batch physics discipline OCES course of Training School of BARC. Subsequently,
he joined “Multi-Disciplinary Research Section” of BARC and started his career in the field of
thin film physics and technology primarily required then for high power pulse laser spectroscopy
activities. He completed his PhD in Physics titled “Thin Film Optical Coatings for High Power Laser
Applications”, from University of Mumbai. Subsequently he was selected by National Academy
of Sciences (NAS, USA) as a National Research Council (NRC) research associate to carry out post-
doctoral research at NASA/Marshall Space Flight Center, Huntsville, USA in the field of “new
generation multilayer thin film laser & space coatings and devices

View publication stats