EXAMPLE 1.
7-1: Bent-beam Actuator
One design of a micro-scale, lithographically fabricated (i.e., MEMS) device that can produce inplane motions is called a bent-beam actuator (Que (2000)). A V-shaped structure (the bent-beam
in Figure 1) is suspended between two anchors. The anchors are thermally staked to the
underlying substrate and therefore keep the ends of the bent-beam at room temperature (Ta =
20). An elevated voltage is applied to one pillar and the other is grounded. The voltage
difference causes current (I) to flow through the bent-beam structure. The temperature of the
bent-beam rises as a result of ohmic heating and the thermally induced expansion causes the
apex of the bent-beam to move outwards. The result is a voltage-controlled actuator capable of
producing in-plane motion.
Figure 1: Bent-beam actuator.
The anchors of the bent-beam actuator are placed La = 1 mm apart and the beam structure has a
cross-section of W = 10 m by th = 5 m. The slope of the beams (with respect to a line
connecting the two pillars) is = 0.5 rad, as shown in Figure 2. The bent-beam material has
conductivity k = 80 W/m-K, electrical resistivity e = 1x10-5 ohm-m and coefficient of thermal
expansion CTE = 3.5x10-5 K-1. You may neglect radiation from the beam and assume all of the
heat that is generated is convected to the surrounding air at temperature T = 20C with average
heat transfer coefficient h = 100 W/m2-K or transferred conductively to the pillars (which
remain at Ta = 20 C). The actuator is activated with I = 10 mA of current.
Figure 2: Dimensions and conditions associated with bent-beam actuator.
a.) Is it appropriate to treat the bent-beam as an extended surface?
The input parameters for the problem are entered into EES:
"EXAMPLE 1.7-1: Bent-beam Actuator"
$UnitSystem SI MASS RAD PA C J
$Tabstops 0.2 0.4 0.6 0.8 3.5
"Inputs"
L_a=1 [mm]*convert(mm,m)
w=10 [micron]*convert(micron,m)
th=5 [micron]*convert(micron,m)
I=0.010 [Amp]
theta=0.5 [rad]
T_a=converttemp(C,K,20 [C])
T_infinity=converttemp(C,K,20 [C])
h_bar=100 [W/m^2-K]
k=80 [W/m-K]
rho_e=1e-5 [ohm-m]
CTE=3.5e-6 [1/K]
"distance between anchors"
"width of beam"
"thickness of beam"
"current"
"slope of beam"
"temperature of pillars"
"temperature of air"
"heat transfer coefficient"
"conductivity"
"electrical resistivity"
"coefficient of thermal expansion"
The extended surface approximation requires that the 3-D temperature distribution within the
bent-beam be approximated as 1-D; that is, temperature gradients within the beam that are
perpendicular to the surface will be ignored so that the temperature may be approximated as a
function only of s, the coordinate that follows the beam (see Figure 2). The resistance that must
be neglected in order to use the extended surface approximation is conduction in the lateral
direction (Rcond,lat). The extended surface approximation is justified provided that the lateral
conduction resistance is small relative to the resistance that is being considered, convection from
the outer surface (Rconv). The Biot number is therefore:
Bi =
Rcond ,lat
Rconv
The heat transfer will take the shortest path to the surface and therefore it is appropriate to use
the smallest lateral dimension (th/2) to compute the lateral conduction resistance.
th h W L th h
Bi =
=
2kW L 1 2k
where L is the length of the beam from pillar to apex (see Figure 2).
Bi=th*h_bar/(2*k)
"Biot number"
The Biot number is small (3x10-6) and therefore the extended surface approximation is justified.
b.) Develop an analytical solution that can predict the temperature of one leg of the bent-beam as
a function of position along the beam, s.
The general solution for an extended surface with a constant cross-sectional area and spatially
uniform generation was derived in Section 1.7.2:
T = C1 exp ( m s ) + C2 exp ( m s ) + T +
g Ac
h per
(1)
For the bent-beam actuator, the perimeter (per), cross-sectional area (Ac), and fin parameter (m)
are
per = 2 (W + th )
Ac = W th
h per
k Ac
m=
per=2*(W+th)
A_c=W*th
m=sqrt(h_bar*per/(k*A_c))
"perimeter"
"area"
"fin parameter"
The volumetric generation, g , is related to ohmic heating. The electrical resistance of the bentbeam structure (Re) is:
Re =
e 2 L
Ac
where
L=
La
2 cos ( )
The volumetric rate of electrical dissipation is the ratio of ohmic dissipation to the volume of the
structure:
g =
L=L_a/(2*cos(theta))
R_e=rho_e*L*2/A_c
g```_dot=I^2*R_e/(2*L*A_c)
I 2 Re
2 L Ac
"length of half-beam"
"resistance of beam structure"
"volumetric generation"
The constants C1 and C2 in Eq. (1) are determined using the boundary conditions.
temperature of the beam where it meets the pillar is specified:
The
Ts =0 = Ta
(2)
Substituting Eq. (1) into Eq. (2) leads to:
C1 + C2 + T +
g Ac
= Ta
h per
(3)
A half-symmetry model of the bent-beam actuator will be developed; because both legs of the
bent-beam see identical conditions there is nothing to drive heat from one leg to the other and
therefore there will be no conduction through the end of the half-beam (at s = L):
qs = L = k
dT
ds
=0
s=L
or
dT
ds
=0
(4)
s=L
Substituting Eq. (1) into Eq. (4) leads to:
C1 m exp ( m L ) C2 m exp ( m L ) = 0
(5)
Equations (3) and (5) can be entered in EES and used to determine C1 and C2.
T_infinity+C_1+C_2+g```_dot*A_c/(h_bar*per)=T_a
C_1*m*exp(m*L)-C_2*m*exp(-m*L)=0
"from boundary condition at s=0"
"from boundary condition at s=L"
A variable s_bar is defined as s/L so that s_bar = 0 corresponds to the pillar and s_bar = 1 to the
apex; the variable s_bar is defined only for convenience so that it is easy to generate a parametric
tables in which s is varied from 0 to L even if parameters such as and La change.
s_bar=s/L
"non-dimensional position"
The temperature is evaluated using Eq. (1).
T=T_infinity+C_1*exp(m*s)+C_2*exp(-m*s)+g```_dot*A_c/(h_bar*per)
T_C=converttemp(K,C,T)
"temperature"
"in C"
A parametric table is generated that includes the variables s_bar and T_C. The temperature
distribution through a half-leg of the beam is shown in Figure 3.
Figure 3: Temperature as a function of dimensionless position along half-length of beam.
c.) The thermally induced elongation of a differential segment of the beam (of length ds) is
given by:
dL = CTE (T Ta ) ds
Estimate the displacement of the apex of the beam. Plot the displacement as a function of
voltage.
The total displacement of the beam (L) is obtained by integrating the differential elongation
along the beam:
L
L = CTE (T Ta ) ds
(6)
Substituting the solution for the temperature distribution, Eq. (1) into Eq. (6) leads to:
L
g Ac
L = CTE C1 exp ( m s ) + C2 exp ( m s ) + T +
Ta ds
h per
Evaluating the integral:
L
g Ac
C1
C2
L = CTE T Ta +
x + exp ( m s ) exp ( m s )
h per
m
m
Substituting the integration limits:
g Ac
C1
C2
L = CTE T Ta +
exp ( m L ) 1
L + exp ( m L ) 1
h per
m
m
DELTAL=CTE*((T_infinity-T_a+g```_dot*A_c/(h_bar*per))*L+C_1*(exp(m*L)-1)/m-C_2*(exp(-m*L)-1)/m)
"displacement of beam"
Assuming that the joint associated with the apex does not provide a torque on either leg of the
beam, the displacement of the apex can be estimated using trigonometry (Figure 4).
Figure 4: Trigonometry associated with apex motion
The original position of the apex (y) is given by:
L
y = L a
2
therefore, the motion of the apex (y) is:
2
L
L
y = ( L + L ) a L2 a
2
2
DELTAy=sqrt((L+DELTAL)^2-(L_a/2)^2)-sqrt(L^2-(L_a/2)^2)
DELTAy_micron=DELTAy*convert(m,micron)
"displacement of apex"
"in m"
The voltage across the beam (V) is:
V = I Re
V=I*R_e
"voltage"
Figure 5 illustrates the actuator displacement as a function of voltage. This plot was generated
using a parametric table including the variables DELTAy_micron and V; the variable I was
commented out in order to make the table.
Figure 5: Actuator displacement as a function of the applied voltage.
