MEMS and Microsystems:

Design and Manufacture*

SOLUTION MANUAL
Tai-Ran Hsu, Professor**
Microsystems Design and Packaging Laboratory
Department of Mechanical and Aerospace Engineering
San Jose State University
San Jose, CA 95192-0087
USA

June 3, 2002

_____________________________________
* McGraw-Hill, Boston, 2002, ISBN 0-07-239391-2

** Telephone: (408)924-3905; Fax: (408)924-3995

E-mail: tairan@email.sjsu.edu

1
 Contents

Chapter 1 Overview of MEMS and Microsystems 3

Chapter 2 Working Principles of Microsystems 3

Chapter 3 Engineering Science for Microsystems Design

and Fabrication 8

Chapter 4 Engineering Mechanics for Microsystems Design 10

Chapter 5 Thermofluid Engineering and Microsystems Design 24

Chapter 6 Scaling Laws in Miniaturization 35

Chapter 7 Materials for MEMS and Microsystems 36

Chapter 8 Microsystems Fabrication Processes 40

Chapter 9 Overview of Micromanufacturing 49

Chapter 10 Microsystems Design 49

Chapter 11 Microsystem Packaging 49

2
 Chapter 1
Overview of MEMS and Microsystems
(P. 32)

1. (b); 2. (a); 3. (b); 4. (c); 5. (a); 6. (c); 7. (c); 8. (c); 9. (b); 10. (c)
11. (a); 12. (a); 13. (b); 14. (a); 15. (b); 16. (c); 17. (c); 18. (a); 19. (a); 20. (c)

Chapter 2
Working Principles of Microsystems
(P. 68)

Part 1. Multiple Choice

1.(1); 2. (3); 3. (2); 4. (2); 5. (1); 6. (1); 7. (1); 8. (2); 9. (1); 10. (3);
11. (3); 12. (2); 13. (2); 14. (1); 15. (3); 16. (1); 17. (2); 18. (3); 19. (2); 20. (1)

Part 2. Description Problems

Problem 2:

Transducers Advantages Disadvantages

Piezoresistors High sensitivity. Sensitive to temperature.
Small sizes. Produced by doping foreign
substances to silicon substrates.
Capacitors Simple in structure, hence less Exhibit nonlinear input/output
expensive to produce. relationship-require careful calibration
Not sensitive to temperature- prior to applications.
suitable for operations at elevated Much bulkier than piezoresistors-takes
temperatures. up precious space in micro devices.

Problem 3: The three principal signal transduction methods for micro pressure sensors are:

(a) Piezoresistors.
(b) Capacitors.
(c) Resonant vibrating beams.

Advantages of (a) and (b) have been presented in Problem 2. Advantage of (c) is high
resolution and sensitivity, especially for high temperature applications. Principal
disadvantages of this method are the high cost involved in manufacturing and the bulky size.

3
Problem 5:

The assembly of minute overlapped electrodes (known as “comb drives”) can produce
electrostatic forces. The scaling laws in Chapter 6 will prove that electrostatic force actuation
scale down two orders of magnitude better than electromagnetic force for actuation. A major
drawback of electrostatic forces is their low magnitudes, which make them impractical for
actuation in macro scale.

Problem 6:

The natural frequency of a device is related to its geometry, which governs the stiffness of the
device, and its mass. Varying the stress state in the device made of an elastic solid, such as the
sensing element of a micro pressure sensor will result in the change of its geometry, and thus the
shifting of its natural frequency.

Problem 7:

We may compute and tabulate the ratios of the output voltage, Vo to the input voltage, Vi vs. the
corresponding gaps between a pair of parallel electrodes and follow the procedure as outlined in
Example 2.2 on P. 47:

Gap, d 2 1.75 1.50 1.00 0.75 0.50

Vo/Vi 0 0.033 0.071 0.167 0.227 0.300

We may plot the relation of the gap, d versus Vo/Vi using the above data in the table. The curve
in Vo/Vi vs. the gap d is close to be a straight line. We realize that Vo/Vi   when d  0.

Problem 8:

The output voltage from a thermopile with 3 thermocouple pairs can be obtained from Eq. (2.4)
as:
V  NT

with N = 3, and T = (120 – 20) + 273 = 373 K, the Seebeck coefficient,  = 38.74x10-6 V/oC
for copper/Constantan from Table 2.3.

Thus, the output voltage is: V  3 x38.74 x10 6 x373  0.04335 volt or 43.35 mv

4
Problem 9:

Actuation techniques Advantages Disadvantages

Thermal force Simple in structure. Response may not be instant due to
thermal inertia of the material.
Shape-memory alloys Actuation is more precise. Same problem as in the thermal actuation
case. It is functional only with a thermal
source.
Piezoelectric Simple and it is less costly Cannot maintain the actuated movement
to produce. Usually for sustained period of time due to
provides precise actuation. overheating.
Electrostatic force Takes up the least amount Low in magnitudes.
of space. Actuation is
instant.

Problem 10:

We assume that there is no friction between the electrodes and the dielectric Pyrex glass. By
following the geometry and the dimensions given in Example 2.1 on P. 45 with:

L = W = 800x10-6 m; o = 8.85x10-9 F/m; r = 4.7 (Table 2.2); V = 70 v; and d = 2x10-6 m

From Eq. (2.10), we may compute the electrostatic force in the width-direction: Fw = 0.0815 N.
From Eq. (2.11), for the force in the length-direction: FL = 0.0815 N

Problem 11:

We will model the comb drive actuator from a simplified model as illustrated below:

Spring constant Spring constant

k k

Moving electrodes Moving electrodes

Fixed electrodes

5
 From Example 2.4, we have the spring constant, k = 0.05 N/m.

The required traveling distance of the moving electrodes is  = 10x10-6 m, which corresponds
to the spring force:

F = k = 0.05x10x10-6 = 0.5x10-6 N

There are five pairs of electrodes by each of the two moving electrodes. The force needs to
be generated by each pair of electrodes is thus equal to:

f = F/10 = 0.05x10-6 N
From Eq. (2.11),
1  r  oW 2
FL  V
2 d
with FL = f = 0.05x10-6 N; r = 1.0; o = 8.85x10-12 C/N-m2; W = 5x10-6 m; d = 2x10-6 m:

1 1x8.85 x10 12 x5 x10 6 2

0.05 x10 6  V
2 2 x10 6

We may solve for the required voltage to be V = 21.26 volts

Problem 12:

We need to make a few assumptions before embarking the solution of the problem. The very first
piece of information we need for the solution to this problem is the materials that make the
gripper. For simplicity, we assume that both the “drive arm” and the “closure arm” are made of
silicon with the surfaces of the “comb drive” “fingers” electrodes coated with thin electrically
conducting copper films. The dielectric medium in the electrode gaps is air. Properties of all the
materials are available in Table 7.3 on P. 247.

Next, we need to establish the dimensions of the gripper arms and the gaps between the
electrodes. We assume that the arrangements that are presented in Example 2.4 are applicable in
this Problem. We will consider the following geometry and dimensions:
“A”
Flexible “Drive Arm”
Req’d tip
movement:
“A” 5 m
Rigidly held
“Closure Arm” Width of electrodes, W = 5 m

Req’d tip
movement:
5 m
Gap, d = 2 m 10 m
150 m 38 m 5 m

300 m 10 m

View “A-A”

6
We will find the necessary voltage supply to the electrodes on both drive arms to provide a 5 m
movements at the free end of each of these two arms. We will treat the Drive arms as two elastic
cantilever beams and the generated electrostatic forces by the electrodes as concentrated forces
acting at the distance that equals to a distance b = 150 + 0.5x38 = 169 m away from the
support-end as illustrated below:
b = 169 m P

max = 5 m
L = 300 m

Since the expression for the maximum deflection at the free-end of the cantilever with a load, P
applied at a distant, b from the support (see the illustration above) is:
P b2
 max   3L  b 
6 EI
with the Young’s modulus, E = 1.9x1011 Pa from Table 7.3 for silicon, and the area moment of
inertia, I = 4.17x10-22 m4 (for the cross-section of the beam shown in View “A-A”in the sketch of
the gripper), we will have the following relationship for the equivalent force, P:

5 x10 6

 
2
P 169 x10 6 3 x300 x10 6  169 x10 6 
 
6 x 1.9 x1011 4.17 x10 22 

Solve for the equivalent applied force, P = 0.11385x10-3 N

We are now ready to estimate the voltage supply to the electrodes to generate the above actuation
force.

There are 5 pairs of electrodes for each arm. From Eq. (2.11), the electrostatic force is:

ro 21 W
FL  2 d V
with r = 1.0; o = 8.85x10-12 C/N-m2; W = 5x10-6 m; and d = 2x10-6 m

We thus have:
1 1x8.85 x10 12 x5 x10 6 2
0.11385 x10 3  5 V
2 2 x10 6

We may solve for the supply voltage to be V = 1434 volts, which is an unusually high voltage for
a microgripper. The postulated geometry and dimensions thus require substantial modifications

7
in order to bring down the required voltage. Reduction in the length, or the depth of the drive
arm would result in the reduction of the required voltage for actuation.

Chapter 3
Engineering Science for Microsystems Design and Fabrication
(P. 93)

Part 1: Multiple Choice:

1.(2); 2. (2); 3. (1); 4. (1); 5. (1); 6. (1); 7. (2); 8. (3); 9. (2); 10(2); 11. (3); 12. (1); 13. (1); 14.(3);
15. (2); 16. (1); 17. (1); 18. (3); 19. (3); 20. (1); 21. (2); 22. (1); 23. (2); 24. (2); 25. (2); 26. (1);
27. (3); 28. (3); 29. (3); 30. (2); 31. (3); 32. (1); 33. (2); 34. (2); 35. (2); 36. (1); 37. (2); 38. (2);
39. (3); 40. (3).

Part 2: Descriptive Problems:

Problem 1:

We have learned from this chapter that the mass of a proton in an atom is 1.67x10 -27 Kg, which is
1800 times greater than the mass of an electron. We may thus assume that the total mass of
protons in an atom to be the mass of the same atom. We are also aware of the fact that a neutron
in the nucleus of an atom has the same mass as that of a proton.

Since each hydrogen atom has one proton and one electron, and each silicon atom has 14 each
protons and neutrons, we may thus obtain the mass of a single hydrogen atom to be 1.67x10 -27
Kg, whereas (14+14)x1.67x10-27 = 46.76x10-27 Kg to be the mass of a silicon atom.

The radii of hydrogen and silicon atoms are available in Table 8.7, from which we may obtain
radii at 0.046 nm and 0.117 nm for hydrogen and silicon atoms respectively.

Problem 3:

A reasonable resistivity of a conductor is 10-5 -cm, the same as that of platinum.

Problem 4:

The negative sign in the equation means that the concentration of the diffused substance
decreases as the distance of diffusion increases.

Problem 5:

By following what is shown in Figure 3.11, the optimum temperatures for As, P, and B are the
temperatures at which the maximum solubility of diffusion take place. Thus, the corresponding
optimum diffusion temperatures are  1220oC, 1200oC and 1400oC for As, P and B.

8
Problem 6:

Equation (3.5) is used for the solution of this problem:

 x 
C  x, t   C s erfc 
 2 Dt 
The concentration in Example 3.1 at various instants into diffusion can be obtained by using the
following parameters:
x = 0.075; t = 1,1.5, 2, 2.5 h; Cs = 4.5x1020 atoms/cm3; and (D)1/2 = 0.085 m/h1/2, from which
we have the following relationship:

 0.075   0.4412 
C  0.075, t   4.5 x10 20 erfc   4.5 x10 20 erfc 
 2 x0.085 t   t 

Numerical solution of C(0.075, t) at various time, t can be tabulated as follows:

Time, t (h) 1.0 1.5 2.0 2.5
Concentration, C 2.64 3.01 3.22 3.36
(1020 atoms/cm3)

Problem 7:

Let the time required to dope boron into silicon substrate at a depth of 2 m to be tf. We obtained
the corresponding concentration of boron at resistivity of 10-3 -cm from Fig. 3.8 to be C = 1020
atoms/cm3. Thus at a diffusion temperature at 1000 oC as specified in Example 3.1, we will have
the concentration of C at 2 m from the silicon substrate surface at time tf to be:

 2 

C 2m, t f   C s erfc
2 D 

 tf 
from which, we obtained the following expression for the solution of tf:

 1 
10 20  4.5 x10 20 erfc 
 0.085
 t f 
with (D)1/2 = 0.085 m/h1/2 as in Example 3.1.

One may readily solve tf = 3.72 h using Table 3.4 for the values of error functions, erf(x).

9
 Chapter 4
Engineering Mechanics for Microsystem Design
(P. 159)

Part 1. Multiple Choice

1. (2); 2. (3); 3. (1); 4. (1); 5. (2); 6. (3); 7. (3); 8. (1); 9. (2); 10. (1)
11. (3); 12. (1); 13. (3); 14. (1); 15. (3); 16. (1); 17. (2); 18. (3); 19. (1); 20. (1)
21. (2); 22. (3); 23. (1); 24. (3); 25. (2); 26. (3); 27. (1); 28. (3); 29. (2); 30. (3)

Part 2. Computation Problems

Problem 1:

We have d = 600x10-6 m, a = d/2 = 300x10-6 m, and P = 20x106 N/m2.

The Young’s modulus, E = 0.7x1011 N/m2 for aluminum from Table 7.3 on P. 247. The Poisson’s
ratio,  = 0.3 for aluminum from metal handbook.

The maximum deflection of the circular diaphragm with a thickness, h = 13.887x10-6 m is

obtained from Eq. (4.7) as:

W max  

3W m 2  1 a2 
16E m2 h3
in which W = (a2)P = 3.14(300x10-6)2(20x106) = 5.652 N, m = 1/ = 1/0.3 = 3.33.

One may thus calculate the maximum deflection, Wmax = -1.4771x10-4 m, or 14.77 m.

Problem 2:

A few assumptions on the dimensions of the pressure sensor die need to be made before we can
embark on the computation. First, let us assume the geometry of the die as illustrated below:

3 mm
Applied Pressure, P

h
3 mm

H
a
L L
L = 3000 m

Plan View Cross-section of the Die

10
We will then assign the dimension of the thickness of the die and the size of the foot print as
shown below:
h

H = 500 m o
74
54.

L = 250 m a Edge Length of Diaphragm, a

L =3000 m

Detail Dimensions of Foot Print

The assigned die thickness, H = 500 m is the standard thickness of 100 mm diameter wafer as
indicated in Section 7.4.2 on P. 239, whereas the footprint L = 250 m is an assumed design
parameter.

In order to make use of Eq. (4.10) for the required thickness of the diaphragm, we need to
determine the edge length of the diaphragm, a, first. Referring to the diagram of the footprint as
illustrated above can do this.
( H  h)
It is clear from the diagram that  tan(54.74 o ) , in which H = 500 m. We thus have:
a
H h
a   353.6  0.707 h
tan(54.74 o )

Consequently, the edge length of the square diaphragm, a is:

a = L – 2L - 2a = 1792.786 + 1.414h

From Eq. (4.10) with max = 350 MPa, we will have:

0.308 p[(1792.78  1.414h) x10 6 ]2

2
 350 x10 6
h
350 x10 6 h2 1.1364 x10 21 h2
or p  12 2
 2
N / m2 or Pa
0.308x10 (1792.78  1.414h) (1792.78  1.414h)

We may tabulate the results of the diaphragm thickness vs. applied pressure as follows:

Diaphragm thickness (m) 500 300 200 100 50

Maximum pressure (MPa) 88.47 31.85 14.16 3.54 0.88

11
Any combination of maximum applied pressure and the diaphragm thickness will produce a
maximum stress of 350 MPa at the mid-span of the edges of the square diaphragm.

Problem 3:

The equivalent spring constant of elastic beams can be obtained from the following expression:
F
k eq 

where F = applied load to the beam
 = deflection of the beam under the load

Case 1 Simply-supported beams:

F

L
From the strength of materials theory, we have the deflection of the beam under the concentrated
force, F to be:
FL3
 
48 EI

from which we may obtain the equivalent spring constant, keq to be:
F 48EI
k eq   3
 L

where E = Young’s modulus of the beam material

I = Area moment of inertia of the beam cross-section

Case 2 Beams with fixed-ends:

The deflection of the beam under the concentrated force, F is:

F L3
 
192 EI
Hence the equivalent spring constant is:
F 192 EI
k eq  
 L
3

12
Case 3 Cantilever beams: F

The deflection of the beam at the free-end is:

F L3
 
3EI
which leads to the following expression for the equivalent spring constant:

F 3EI
k eq  
L L3

Problem 4:
The mass, m attached to the beam is 5 g, or 5x10-3 Kg; The equivalent beam spring constant keq
in the arrangement shown below, and from Case 2 of Problem 3 is 18240 N/m
20x10-6 m

(a) From Eq. (4.16), we have the equivalent natural frequency,

k eq 18240
n    1910 rad / s
m 5 x10 3
(b) The equivalent motion of the mass in the y-direction, according to Eq. (4.14) is:

13
 2
d y(t )
m 2  k eq y(t )  0
dt
with y(0) = 5x10-6 m, and y’(0) = 0. Substitute these values into the above equation:

2
d y(t )  3.6848x10 6 y(t )  0 (a)
d t2
The solution of the differential equation is:

y (t )  C1 cos1910t  C 2 sin 1910t (b)

From the condition y(0) = 5x10-6, we get C1 = 5x10-6

From y’(0) = 0, we have C2 = 0

Thus, the amplitude of vibration, y(t) is:

y (t )  5 x10 6 cos1910t (c)

The maximum amplitude of vibration is the coefficient of the cosine function in the solution in
Eq. (c), or ymax = 5x10-6 m, or 5 m.

Problem 5:

By referring to the forced vibration analysis in Section 4.3.2, we have the following differential
equation to solve for the amplitude of the vibrating mass:

d 2 y(t ) 2 F o cost
  n y (t ) 
d t2 m
with the specified conditions: y(0) = 5x10-6 m and y’(0) = 0.

The proper differential equation becomes:

2
d y(t )  3.648x10 6 y(t )  5 cos1910t
d t2 5 x10 3

14
 6
in which the natural frequency of the beam spring system,  n  3.648x10  1910 rad / s
The solution of the above differential equation is:

t
y(t )  C1 cos n t  C 2 sin  n t  F o sin  n t
2m  n
with  = n = 1910 Rad/s at the resonant vibration situation and Fo = 5 N

Use the first condition, i.e. y(0) = 5x10-6 m will result in C1 = 5x10-6. The other condition y’(0)
will result in C2 = 0. Thus the solution for the amplitude of the vibration mass being:

y (t )  5 x10 6 cos1910t  0.2618 t sin 1910t

Now, if we let tf = the time at which the beam spring breaks at y(tf) = 1 mm = 10-3 m, we will
have:

10 3  5 x10 6 cos1910t f  0.2618t f sin 1910 t f

We may solve for tf from the above equation, or by an approximate relationship of 10-3 0.2618tf
from the above expression. This approximation is justified by letting sin1910tf = 1.0 and
cos1910tf = 0. This approximation leads to tf = 3.82 ms.

Problem 6:

The beam is loaded as illustrated in Example 4.8

1 m
Dynamic force

h m
L=600m

Beam Cross-section

The area moment of inertia of the beam cross-section is:

10 24 3 4
I 
1
12
10 6  h3 
4

12
h m with h in micrometers.

The equivalent spring constant, keq is as computed in Case 2 of Example 4.8 and 4.9 for fixed-
ends as:

15
k eq 
192EI

  
192 1.9 x1011 10 24 h3
 0 .014074 h
3
N /m
L
3

12 600 x10 6 3

The corresponding circular frequency of the balanced force accelerometer is:

2 k eq2 x0.014074 h3
  6
 96.864 h3 Rad / s
m 3x10
From Example 4.9, the amplitude of vibration of the beam is:

X (t )  C1 cost  C 2 sin t
in which the arbitrary constants C1 and C2 can be determined by the initial conditions:

dX (t )
X (t ) t  0  0 and  50 km / h  13.8888 m / s
dt t 0

We thus have:

C1 = 0 and C2 = 13.8888/ = 0.1434 h-3/2

Thus, we have the amplitude of vibration to be:

X (t )  0.1434 h1.5 sin 96.864 h3 t

At X(tf) = 5 mm = 5x10-3 m for the failure of the beam:

5 x10 3  0.1434 h1.5 sin 96.864 h3 t f

0.1434
or h1.5  3
sin (96.864 h1.5 t f )
5 x10
The approximate value of tf is when sin 96.864 h1.5 t f  1.0 for a maximum value of h, which
leads to h = 9.37 m.

Problem 8: Beam Springs

We may illustrate the balanced force accelerometer system below:

5 m m
0
40 Beam Mass
16

10 m
From Example 4.11 on P. 125, we get the damping coefficients for the balanced force
accelerometer to be:

cair = 1.3125x10-8 N-s/m for air as the damping fluid, and

csi = 51.8x10-8 N-s/m for silicone oil as the damping fluid.

We further have the mass of the silicon beam to be: m = v, in which the mass density,  = 2.3
g/cm3 or 2.3x103 Kg/m3 from Table 7.3 on P. 247, and the volume of the beam = v.

By referring to the geometry and dimensions of the beam in Example 4.11, we have

v = bHL = (5x10-6) (10-5) (7x10-4) = 35x10-15 m3

Consequently, the beam has a mass, m = (2.3x103)(35x10-15)= 80.5x10-12 Kg

We will use the model illustrated in Fig. 4.7(b) on P. 107 to assess the motion of the beam mass,
and Eq. (4.19) with the spring constant k = 2keq in Example 4.8 is used to compute the
displacement of the beam mass, X(t) in the equation.

The solution of Eq. (4.19) depends on the cases with the values of (2 - 2) as described in Eq.
(4.20a), (4.20b) or (4.20c). We will thus need to compute both 2 and 2 first in order to select
which of the above solutions for the case under consideration.

Let us assume that both beam springs have fixed ends, and the equivalent spring constants can be
computed from the following expressions as presented in Case 2 on P. 118:

192 EI
k eq  3
L

with E = 1.9x1011 N/m2 (Table 7.3) and I 

1
12
 
5 x10 6 10 x10 6  3
 4.167 x10 22 m4

Thus, k eq 
 
192 1.9 x1011 4.167 x10 22   237.52 N /m
400 x10  6 3

17
 2 k eq 2 x 237.52 6
and    2. 43 x10 Rad / s
m 80.5 x10 12

which leads to 2 = 5.9x1012 Rad2/s2

The damping parameters:

c air 1.3125x10 8
 air  
2m 12
 81.52 for air
2 x80.5 x10

c 51.8 x10 8
 si  
si
12
 3217.39 for silicone oil
2m 2 x80.5 x10
from which we have:

 air     81.52  5.9 x10  0

2 2 2 12

 si     3217.39  5.9 x10  0

2 2 2 12

The above values of (2 - 2) for the two distinct damping media of air and silicone oil will lead
to the use of Eq.(4.20c) for the movement of the beam mass. The movement will be of
oscillatory nature.

Problem 9:

The balanced-force accelerometer is illustrated in Fig. 4.25, and also as below:

m
L = 700

h = 10
m
Beam Mass, m

b = 5 m
Dimensions of the Beam Mass

The dimensions of the two beam springs are not given in the problem. We may either assume the
unspecified dimensions are identical to those given in Example 4.8 and 4.9, or by using an open
Dimensions
size of the beam springs that will withstand of the Beam
the specified Springs as described in Example
conditions
4.12. We will assume the dimensions of the beam springs as shown below:
600 m

18
10 m
5 m
The equivalent spring constant for beam springs with fixed-ends is keq = 1.76 N/m as in Case 2 of
Example 4.8.

Since the maximum deceleration of the car in the present case is X   22.22 m / s 2 from
Example 4.12. By neglecting the mass of the beam springs, we may express the dynamic force
associated with the moving beam mass as:

  t 
F (t )  mX

The mass of the beam mass, m = v, in which  = 2.3 g/cm3 is the mass density of silicon from
Table 7.3, and the volume of the beam mass, V is:

V = bhL = (5x10-6)(10x10-6)(700x10-6) = 35x10-15 m3

The mass of the beam mass is thus computed to be:

m = (2.3x103 Kg/m3)(35x10-15 m3) = 80.5x10-12 Kg

The force acting on both beam springs at the time of deceleration of –22.22 m/s2 is:

  
  80.5 x10 12 Kg 22.22 m / s 2  1788.71x10 12 Kg  m / s 2  1788.71x10 12
F  mX N

The induced deflection of the beam springs by the above dynamic force of the magnitude is:

P = F/2 = 1788.71x10-12/2 = 894.4x10-12 N is  = P/keq.

with L = 600x10-6 m, I = 10.42x10-24 m4 (from Example 4.8), E = 1.9x1011 N/m2 (Table 7.3), and
keq = 1.76 N/m, we will find that the maximum movement of the beam mass to be:

max = 508 m

Problem 10:

This bi-layer strip is subjected to a uniform temperature rise, T as illustrated below:

19
 SiO2 strip

1000 m h = 10 m

Silicon strip h

The radius of curvature,  from Eq. (4.50) is:

2h

3 2   1T

where 1 and 2 are coefficients of thermal expansion of SiO2 and silicon strips respectively
(available in Table 7.3), and h is the thickness of the individual strips.

Let us express the radius of curvature of the bi-layer strip in a different form from the above
expression:
C
 
T
2h 2 x10 x10 6
in which C    3.643
3 2   1 3 2.33  0.5 x10 6

From Example 4.14, we have the movement of the free-end,  to be:

   1  cos  
360 L
where   with L = 1000x10-6 m
2
360 x1000 x10 6 5.7325 x10 2
Hence   
6.28  

We may tabulate the values of the temperatures vs. the movement of the free-end of the beam
actuator as follows:

T (oC)  = C/T (m) (o)  = (1 - cos) (m)

10 0.3643 0.1574 1.373
20 0.1822 0.3147 2.747
30 0.1214 0.4720 4.120
40 0.0911 0.6294 5.496
50 0.0729 0.7880 6.890

Problem 11:

20
The beam has the following geometry and dimensions:
y
b = 5 m

Silicon x H = 10 m
2h = 10 m 0
Beam

L = 1000 m

The temperature variation in the beam is:

T(z) = 2x106z + 30 o
C

At the top face, i.e. z = 5x10-6 m, we have T(5x10-6) = 40 oC, and

at the bottom face at z = -5x10-6 m, the temperature is T(-5x10-6) = 20 oC.

Material properties of the silicon beam are given in Example 4.15 on P. 142:

Mass density,  = 2.3 g/cm3; Specific heats, c = 0.7 J/g-oC; Thermal conductivity, k = 1.57 J/cm-
o
C-s; Coefficient of thermal expansion,  = 2.33x10-6/oC; Young’s modulus, E = 1.9x1011 N/m2;
Poisson’s ratio,  = 0.25.

We will first compute the thermal force, NT and the thermal moment, MT from the respective Eq.
(4.54a) and (4.54b) as:
N T   2.33x10 1.9 x10  5 x10 6  2 x10 z  30dz  132.81 N
6
6 11 5 x10 6

M T   2.33x10 1.9 x10  5 x10 6  2 x10 z  30zdz  73.78 x10

6
6 11 5 x10 6 6
N m

From Example 4.15, we have A = 5x10-11 m2 and I = 4.167x10-22 m4.

From Eq. (4.55), we have the thermal stress along the x-direction to be:
bN z (b M )
 xx  x, z    ET  z   A T  I T
with xx,max occurs at z = 5x10-6 m.

Thus, xx,max = (x,5x10-6) = -2600 Pa

We will compute the associate thermal strains from Eq. (4.56a) and (4.56b) with maximum
values occurring at z = 5x10-6 m:
1 b N z 
 xx  x, z   E  A T  I  b M T  
 

21
which leads to xx,max = xx(x,5x10-6) = 0.00932%
 b N z  1  
 zz  x, z    E  A T  I  b M T     E T ( z )
   

results in zz,max = zz(x,5x10-6) = -0.0023%

The deflection of the beam in the x-direction = u (x,z) can be computed from Eq. (4.57a) as:
x b NT z 
u  x, z     b 
MT 
E  A I 
with umax at x = 500x10-6 m and z = 5x10-6 m:

umax = u(500x10-6, 5,10-6) = 0.0466 m

The deflection of the beam in the z-direction, w(x,z) is obtained from Eq. (4.57b):
b M T 2  b NT z
2  1   z
w x, z    x   z  b M T     0 T ( z )dz
2 EI E A 2I   E 
with wmax occurs at x = 500x10-6 m and z = 5x10-6 m, we have:

wmax = w(500x10-6, 5x10-6) = -0.582 m

Problem 12:

The width of the beam has been increased to 100x10-6 m. The “wide” beam now is effectively a
“plate”. As such, the thermal stress formulation for thin plates will be used to solve this problem,
with the temperature variation across the plate thickness, i.e.

T(z) = 2.1x106z + 28.8 in degree C

We realize that the thermal force, NT = 127.5 N and the thermal moment, MT = 77.4725x10-6 N-
m remain unchanged as in Example 4.15 on P. 145.

The thermal stresses in both x- and y-directions can be computed from Eq. (4.51) as:

 xx   yy   5.9027x10  2.1x10 z  28.8  17 x10  1.2396x10 z

5 6 6 12

The associated thermal strains are obtained from Eq. (4.52a):

22
 5
 xx   yy  6.711x10  4.893z
 zz   0.3509x10 12.75x10  0.9297x10 z   3.8833x10  2.1x10 z  28.8
11 6 12 6 6

with xy = yz = zx = 0

The induced displacements of the plate in the x-direction, u(z) and that in the y-direction, v(z),
and w(x,y,z) in the z-direction can be computed from Eq. (4.53a,b and c):

u z  
x
1.9 x10

11
12.75 x10 6  0.9297 x1012 z 
v z  
y
1.9 x10

11
12.75 x10 6  0.9297 x1012 z 
and
  
w x, y, z    2.4465 x2  y  0.7018x10 11 96.84  6.375x106 z  0.2324x1012 z 2
2

The maximum values of stress, strains and displacements occur at: x = 500x10-6 m, y = 50x10-6
m and z = 5x10-6 m. Thus, we will have the following maximum stress, strains and
displacements:
xx,max = yy,max = 4000 Pa
xx,max = yy,max = zz,max = 0.00915%
umax = 0.046 m; vmax = 0.0046 m; wmax = -0.6173 m

We have realized that by extend the beam into a plate with a width of 100 m has not produced
significant difference in the results from those obtained from a beam with a width of 5 m.

Problem 14:

We have the dimensions of the specimen as shown in the diagram below, in which s = 1 cm =
10-2 m; b = 5 mm = 5x10-3 m; and the width, B = 24x10-4 m. We need to assume the crack length,
c = 100 m = 10-4 m.

Pcr = 40 MN
B = 2400 m

b = 5000 m

s = 10000 m

The critical load, Pcr that breaks the specimen is 40x106 N/m2.

23
We will use Eq. (4.64a) for the function F(c/b) as s/b = 2 < 4, as indicated in Section 4.5.2 on P.
149:
2 3 4
c c c c c
F    1.09 1.735   8.2   14.18   14.57 
b b b b b

with c/b = 0.02

Hence F(c/b) = 1.0586

Eq. (4.63) is used to compute fracture toughness:

c
K c   c c F  
b
The c in the above expression is obtained from the bending stress in a “solid” beam subjected
to three-point bending as follows:

Mc
c  I
Bb 3
where c  b / 2  2.5 x10 3 m and I  250 x10 13 m 4
12
s 5
The bending moment, M  Pcr    2 x10 N  m
2
Thus, we have the critical stress corresponding to Pcr to be:

c 
 2 x10  2.5x10   2000
5 3
Pa
250 x10 13

which leads to the fracture toughness, Kc to be:

K c  2000 3.14 x10 1.0586  37.52 Pa m

4

Problem 15:

For the width of the specimen, B to be increased to 100x240 m, we will have I = 2.5x10-9 m4.
This new value will change the critical stress according to the following expression:

c 
2 x10 2.5x10   20
5 3
Pa
2.5 x10 9

Kc = 0.375 Pa m , which is 100 times smaller than the case in Problem 14. This result, of
course, is computed on the basis that the enlarged specimen breaks at the same critical load, P cr,
which is not quite a realistic hypothesis. We would expect a much greater value of P cr for larger
specimens. It nevertheless underlines the importance of the size effect on the measurement of the

24
fracture toughness of specific materials. A credible Kc for design purpose must be independent
of the specimen geometry and size. A great deal of research effort is needed in the measurements
of Kc for microsystems materials in micro scale.

Chapter 5
Thermofluid Engineering and Microsystem Design
(P. 163)

Part 1. Multiple Choice

1.(3); 2. (1); 3. (2); 4. (2); 5. (3); 6. (1); 7. (2); 8. (1); 9. (1); 10. (3); 11. (2); 12. (3); 13. (2); 14.
(1); 15. (3); 16. (1); 17. (2); 18. (3); 19. (3); 20. (2); 21. (1); 22. (1); 23. (3); 24. (2); 25. (1); 26.
(3); 27. (1); 28. (1); 29. (2); 30. (1); 31. (3); 32. (1); 33. (1); 34. (3); 35. (2); 36. (2); 37. (1); 38.
(2); 39. (1); 40. (2); 41. (3); 42. (3); 43. (1); 44. (2); 45. (3); 46. (1); 47. (3); 48. (1); 49. (1); 50.
(2); 51. (3); 52. (3); 53. (3); 54. (1); 55. (1)

Part 2. Computational Problems

Problem 2:

We have d1 = 500x10-6 m and d2 = 50x10-6 m

The flow rate is Q = 1x10-6 cm3/min = 1.67x10-14 m3/s

A1 = (500x10-6)2/4 = 19.64x10-8 m2

A2 = (50x10-6)2/4 = 19.64x10-10 m2

Q 1.67 x10 14

V1   8
 0.085 m / s
A1 19.64 x10

Q1.67 x10 14

V2   10
 8.503 m / s
A2 19.64 x10
Problem 3:

The opening of the valve may be illustrated as follows:

m
0
30
=
,W
th

Valve Plate Thickness: , Ve

id

ty
eloci
W

4 m
00 
m E x it v
L=4
25
15o

Valve Opening, H
Fluid Flow
(a) The opening of the valve is H = (L sin15o)cos15o = 400x10-6x sin15ocos15o = 100x10-6 m

(b) We will next estimate the velocity of the gas flow at the exit of the valve, i.e.Ve. Base on the
law of continuity, we have:

M x1
Ve 
 Ae
in which  = mass density of the H2 gas = 0.0826 Kg/m3 (from Example 5.2)
Mx1 = the mass flow rate in the direction of Ve = 15.3x10-6 Kg/s (from Example 5.2)

The exit cross-sectional area, Ae = HW, in which W is the width of the plate valve = 300 m, or
300x10-6 m.

Hence Ae = (100x10-6)(300x10-6) = 3x10-8 m2.

Thus, the exit velocity is:

15.3 x10 6
Ve   6174.33 m / s
0.0826 x3 x10 8

NOTE: This exit velocity is unrealistically high for a micro valve. This high value on the
velocity is a result of extremely small opening at the exit (A e = 3x10-8 m2), and large mass flow
rate (Mx1 = 15.3x10-6 Kg/s)

(c) The volumetric flow at the exit can be computed as follows:

6
Q  M x1 
15.3 x10
 185.23 x10 6 m3 / s
 0.0826

or Q = 11,113.8 cm3/min, which is significantly smaller than the intended design capacity of
30,000 cm3/min.

Problem 4:

26
The uniformly distributed load that is required to bend a cantilever beam (plate) such as the
closure plate with a free-end displacement of H = 100x10 -6 m in Problem 3 (see illustration
below) can be obtained by the following expression:

L = 400 m
W L4
 max 
8EI
W N/m

in which max is the maximum deflection of a cantilever beam at the free-end due to uniformly
distributed load, W per unit length; E is the Young’s modulus of the beam material; I is the area
moment of inertia of the beam cross-section.

The cross-section of the plate is 300 m wide x 4 m thick, which leads to an area moment of
inertia, I to be:
I 
1
12
  3

300 x10 6 4 x10 6  16 x10  22 m4

with a Young’s modulus, E = 1.9x1011 N/m2 from Table 7.3 for silicon, and max = H = 100x10-6 m
as shown in the figure in Problem 3, we may determine the required load, W from the following
relation:

100 x10   8 x1W.9 x 400 x10  6 4

6
10 16 x10
11
  22

which leads to W = 9.5 N/m

The force required to lift the plate of 400 m long at the free end is WL, or 9.5x(400x10 -6) =
3800x10-6 N. However, there is a fluid-induced force acting on the plate too. This force is Fy =
40x10-8 N as computed from Example 5.2. The net required electrostatic force is thus equal to the
difference of the above two forces, or Fd = 3800x10-6 – 40x10-8  3.8 mN.

The corresponding required voltage to generate such force can be obtained by using Eq. (2.8) to
give:

2
V 
2d2 Fd


2 x 6 x10 6 3.8 x10 3  2
 29.443x10 4

 12 6
 r  o WL 1x 8.85x10 300x10 350x10  6
 
which leads to V = 542 v – a rather high voltage.

27
Problem 5:

The problem is illustrated below:

m
0.1
L=
V2

/s
0 m m
V1 = 60
5 0
d 2=
m
100
d1= d ave

Assume that the average velocity of the fluid is computed at the cross-sectional area of the
conduit at its mid-section is used. The 30o inclination is neglected.

From Example 5.3 on P. 175,

V2 = 2.4x10-3 m/s.

We may calculate the Vave in the mid-cross section to be:

Vave = 0.5 (V1 + V2) = 1.5x10-3 m/s

Let dave = 0.5 (d1 + d2) = 75x10-6 m, which leads to the radius at the mid-section, aave = 37.5x10-6
m. The pressure drop, P in the conduit using the Hagen-Poiseuille equation in Eq. (5.17) is:

8LQ
P 
 a4

where  = dynamic viscosity of the fluid = 1199.87x10-6 N-s/m2 (Table 4.3 on P. 124 for alcohol)

L = length of the conduit = 0.1 m, and

    
Q  AaveV ave from Eq.  5.6   a2ave V ave  3.14 37.5 x10 6 1.5 x10 3  66.2343x10 13 m3 / s
2

We will have the approximate pressure drop, P to be:

8 x1199.87 x10 6 x 0.1x 66.2343x10 13

P   1024 Pa

3.14 37.5 x10 6  4

28
The pressure drop, P obtained from the Hagen-Poiseulle’s equation is about 2.5 larger than that
by the Bernoulli’s equation in Example 5.3. This indicates that the scaled down effect of the
conduits on the pressure drop of a fluid flowing in a small conduit is significant.

Problem 6:

The purpose of this problem is to compare the estimated pressure drop in a fluid (water in this
case) flowing through a capillary tube computed by using the Hagen-Poiseuille’s equation and
that induced by the surface tension in the minute water cylinder. The capillary tube section is
illustrated below:
1000 m
The radius of the capillary tube is
a = d/2 = (20x10-6)/2 = 10-5 m

20 m dia.

Since the volumetric flow rate of the water is not given in the problem, we will first work on an
hypothesis that water flows in the capillary tube in a laminar flow pattern. This pattern of flow of
water requires the Reynolds number, Re, be kept below 1000, which leads to the following
relationship:

DV
Re   1000

with the density of water,  = 1000 Kg/m3, and the inside diameter of the tube, D = 20x10-6 m,
and the dynamic viscosity,  = 1001.65x10-6 N-s/m2 from Table 4.3, we will have, from the above
expression, the velocities of flow to be:

V = 50 m/s for Re = 1000, and

V = 5 m/s for Re = 100.

If we use the lower bond velocity, V = 5 m/s with Re = 100, the volumetric flow, Q would be:

Q  AV 

4
 2

20 x10 6 x5  15.708 x10 10 m3 / s

The corresponding pressure drop, P by the Hagen-Poiseuilli’s equation is:

8 x 1001.65 x10 6 10 3 15.708 x10 10 

P  4
 400,000 Pa
 20 x10 6 
3.14 
 2 
To maintain a water flow with this enormous pressure drop is beyond the capability of most
volumetric pumping devices. Consequently, let us assume a typical velocity flow at V = 10 m/s
in the capillary tube.

29
This water flow velocity of V = 10 m/s, which leads to a volumetric flow rate, Q = AV =
31.4x10-16 m3/s, in which A is the cross-sectional area of the tube.

Substituting the above values into the Hagen-Poiseuille’s equation, will lead to a pressure drop of:

P 
  
8 x 1001.65 x10 6 10 3 31.4 x10 16   0.8013 Pa
4
 20 x10 6 
3.14 x 

 2 

Now, we will consider the pressure difference between the inside and outside of the water
cylinder in that section of the capillary tube, namely the surface tension of the water. We will
recognize that the pressure required to overcome the surface tension is the sum of that for the
length of the two ends of the “water cylinder” in the tube, and the same surface tension between
the circumferential surface and the tube wall. The surface tension of water in these areas can be
found in Eqs. (5.24a) and (5.24b), or the required pressure is:

3
Pst 
a

in which  = surface tension coefficient of water = 0.073 N/m as in Eq. (5.23) and a = the radius
of the tube.

We will thus have the required pressure:

3 x 0.073
 P st   21,900 Pa
10 5

We realize that the pressure required to overcome the surface tension of the water in the capillary
tube section is much greater than that for driving that tiny volumetric flow of water through the
tube as predicted by the Hagen-Poiseulle’s equation. The above computations have underlined
the dominance of the effect of surface tension in water (liquid) flowing in capillary tubes.
Problem 7:

The situation of a capillary tube inserted in the water is illustrated in Fig. 5.16 on P. 183. The
tube has a diameter of 20 m, which gives a radius, a = 10-5m. By following the same procedure
for the solution in Example 5.6, we have the rise of water level in the capillary tube, h to be:

2 cos  2 x0.073x cos(0)

h   1.488 m, which is unrealistically high, as the total length of
wa 9810 x10 5
the capillary tube in only 1 mm.

Problem 8:

30
The rarefied gas flows in a nano scaled tube section is illustrated below:

50x10-9 m
30x10-9 m dia.

N2 gas

P = 0.5 Pa

The use of the Hagen-Poiseuille’s equation in the form of:

8LQ
P  will lead to the volumetric flow rate as:
 a4

 a 4 P
Q with the dynamic viscosity of the N2 gas,  = 17.48x10-6 N-s/m2 from Table 4.3
8 L
on P. 124, the length of the tube section, L = 50x10-9 m, and the radius of the tube, a = 15x10-9 m.
Thus with a pressure drop of P = 0.5 Pa, we will have the volumetric flow rate, Q to be:

Q 
 
3.14 x 15 x10 9  0.5
4
 1.1367 x10  20 3
 
8 x 17.48 x10 6 50 x10 9  m /s

and the mass flow rate m  = Q = 1.2506x1.1367x10-20 Kg/s = 1.422x10-20 Kg/s with the value
 = 1.2506 Kg/m for N2 gas. We notice that this mass flow rate is about two orders of
3

magnitude less than m  = 3.116x10-18 Kg/s as in Example 5.7 for rarefied gas flow. The Hagen-
Poiseuilli’s equation is thus inadequate for capillary and rarefied gas flow.

Problem 9:

We have the problem as illustrated below:

1 cm = 10-2 m
10x10-6 m dia.
Air flow

P = 5 Pa

Following the same procedure as in Problem 8, we have the volumetric flow rate of air in the
tube by the following expression:

31
  a 4 P
Q with a = 5x10-6 m, P = 5 Pa,  = 18.75x10-6 N-s/m2 from Table 4.3, and L = 10-2
8 L
m. We will thus have the numerical value of Q to be:

Q 
 
3.14 x 5 x10 6 x5
4
 65.42 x10 16 3
 
6
8 x 18.75 x10 x10 2 m /s

The density of the air at 20oC is  = 1.2929x10-3 g/cm3 = 1.2929 Kg/m3, which leads to the mass
flow rate of the air to be:

 air  Q  1.2929 x65.42 x10 16  0.8458x10 14

m Kg / s

We find that the mass flow rate of the air as shown above is about two orders of magnitude
smaller than the value of 0.1352x10-12 Kg/s from Example 5.8. We thus conclude that the
Hagen-Poiseuille’s equation is not suitable for assessing the rarefied gas flow.

Problem 10:

Thermal diffusivity defined in Eq. (5.39) should be used as effective measure of materials’
response in thermal actuation. The listing should thus be constructed on this basis.

Problem 11:

We assume that the copper film is so thin and ductile that it only generates heat to the SiO2/Si
bilayer strip, but does not impose any mechanical constraint on the overall structure. One end of
the strip is maintained at 20oC whereas the other end and the top and bottom surfaces are
surrounded by stagnant air as illustrated in the figure below.

We further postulate that heat flows in the strip predominantly in the y-direction with some
dissipation through the left end at x = 0. This postulation on the heat flow is justifiable as the
dimension of the strip in the x-direction far exceeds that in the y-direction. Further, the
assumption of thermal insulation of the right end at x = 1000 m and the bottom surface (y =60
m) is also reasonable. These surfaces are in contact with the surrounding stagnant air at 20oC.
The surrounding air temperature is not expected to change significantly enough during the brief
period of actuation to induce a natural convection that will dissipate heat from the strip through
these surfaces.
Heat supply
2 m thick copper film
x
Si O2 4 m
20oC

silicon
20 m

1000 m
32
y Stagnant air at 20oC
We will assign T1(x,y,t) and T2(x,y,t) to be the temperature distributions in the SiO2 and silicon
strips respectively.

Equation (5.48a) and (5.48b) are used to determine the respective temperature distributions,
T1(x,y,t) and T2(x,y,t) with the following initial and boundary conditions:

The initial conditions:

T 1  x, y, t  t 0  T 2  x, y, t  t 0  20 C
o

The boundary conditions for the temperature distribution in the SiO2 layer are:

T 1  x, y, t  x 0  20 C
o
at the left end, and

 T 1  x, y , t 
 0 thermally insulated at the right end, and
x x 1000 m

T 1  x , y , t  y  4 m  T 2  x , y , t  y  4 m for compatibility at SiO2 and silicon interface, and

 T 1  x, y , t  i
2
R
 for heat input at the top surface of SiO2 layer, in which R = the
y y 0 k1
electric resistance of copper film, (), and i = the passing electric current (amp), and k1 =
thermal conductivity of SiO2 (W/m-oC)

The boundary conditions for temperature distribution in the silicon strip are:
T 2  x, y, t  x 0  20 C
o
at the left end, and

 T 2  x, y , t 
 0 thermally insulated at the right end, and
x x 1000 m

33
 T 2  x, y , t   T 1  x, y , t 
 for compatibility at the silicon and SiO2 interface,
y y  4 m
y y  4 m
and

 T 2  x, y , t 
0 for thermally insulated at the bottom surface.
y y  24 m

Problem 12:

Equation (5.45) provides:


T  r , t  h  h
  T r,t  
r  rs  Tf
n  
r  rs k k

which leads to the following special cases:

(A) When h = 0:

Equation (5.45) becomes:


T  r , t 
  0
n  
r  rs

which is equivalent to have qin and qout = 0 in Eq. (5.44a) and (5.44b) respectively. This means
 
that no heat is allowed to flow across the boundary at r  rs .

(B) When h  :

By dividing each term in Eq. (5.45) by h and then letting h  , we will have:

T  r , t  r r  T f
s

which is the prescribed surface temperature boundary condition as shown in Eq. (5.43).

Problem 13:

The respective heat conduction equations for SiO2 and silicon stripes are:

34
 T 1  x, y , t    T 1  x, y , t   1  T 1  x, y , t 
2 2

 x2 y
2
 1 t
for 0  x  1000 m and 0 y  4 m; t > 0

and

 2 T 2  x, y , t   2 T 2  x, y , t  1  T 2  x, y , t 
 
 x2 y
2
 2 t
for 0  x  1000 m, 4 m  y  24 m, and t > 0.

The constant 1 and 2 in the above differential equations are the thermal diffusivities of SiO2
and silicon respectively. The appropriate initial and boundary conditions are presented in
Problem 11.

Problem 14:

The differential equations are as shown in Problem 13, and the initial conditions are given in
Problem 11. The following boundary conditions apply:

(A) In SiO2 strip:

T 1  x, y, t  x 0  20 C
o

 T 1  x, y , t  h h
 T 1  x, y, t  x 1000 m   20
x x 1000 m k1 k1
 T 1  x, y , t  i
2
R

y y 0 k1

T 1  x, y , t  y  4 m  T 2  x, y , t  y  4 m

35
 T 1  x, y , t   T 2  x, y , t 

y y  4 m
y y  4 m

(B) In silicon strip:

T 2  x, y, t  x 0  20 C
o

 T 2  x, y , t  h h
x
 T 2  x, y , t    20
x 1000 m k2 x 1000 m k2

 T 2  x, y , t  h h
y
 T 2  x, y , t    20
y  24 m k2 y  24 m k2
where k1 and k2 are thermal conductivity of SiO2 and silicon respectively, and h is the specified
heat transfer coefficient.

Problem 16:

The relaxation time,  is defined in Eq. (5.54) as  = /V, in which  = average mean free path of
phonons or electrons and V = average velocity of the heat carrier.

For metals such as gold, silver and copper, electrons are the principal energy (heat) carriers.
Thus, the value of  = 10-8 m are used for electrons and the velocity of the heat carrier, V = 106
m/s listed in Table 5.2 is used. These values lead to a “relaxation time”,  to be:

 = 10-8/106 = 10-14 seconds

Chapter 6
Scaling Laws in Miniaturization
(P.234)

1.(1); 2. (2); 3. (1); 4. (3); 5. (2); 6 (1); 7. (1); 8. (1); 9. (1); 10. (3); 11. (1); 12. (3); 13. (1); 14.
(1); 15. (2)

36
 Chapter 7
Materials for MEMS and Microsystems
(P.268)

Part 1. Multiple Choice

1.(2); 2. (3); 3. (2); 4. (1); 5. (1); 6. (1); 7. (2); 8. (1); 9. (2); 10. (1); 11. (2); 12. (3); 13. (3); 14.
(2); 15. (1); 16. (1); 17. (2); 18. (1); 19. (3); 20. (1); 21. (3); 22. (2); 23. (3); 24. (3); 25. (2); 26.
(3); 27. (1); 28. (3); 29. (2); 30. (3); 31. (1); 32. (2); 33. (1); 34. (3); 35. (1); 36. (3); 37. (3); 38.
(1); 39. (1); 40. (2)

Part 2. Computational Problems

Problem 2:

The planar area of a circular wafer, A, can be computed by:

 2
A d in which d = the diameter of the wafer.
4

The ratio of plane areas of wafers with 300 mm and 200 mm diameters is:

2
A2  Area of 300 mm wafer   300   2.25
A1 Area of 200 mm wafer  200 
Hence a wafer with 300 mm diameter has 2.25 times greater area than that of a 200 mm
wafer.

Problem 3:

By following the same expression used in Example 7.1, the number of atoms per cubic mm of
silicon is:
3
V   0.001 
N  n  9 
x18  1.12 x10 20 atoms / mm 3
v  0.543 x10 
in which v = the volume of a single silicon crystal.

37
Likewise, the number of atoms per cubic micrometer of silicon is:
3
 10 6 
N   9
 x18  1.12 x1011 atoms / m 3

 0.543 x10 

Problem 4:

A piezoresistor has the following geometry and dimension:

max m
10 max = 235.36x106 Pa

2 m 4 m

The area on which the maximum normal stress exists is: A = 2 x 10 = 20 m2 = 20x10-12 m2.

From Eq. (7.8), we have:

R
  L L  TT
R

Since the piezoresistor is attached to the cantilever beam as illustrated in Fig. 7.17, we will
have: L = max = 235.36x106 Pa, and T = 0 as in Example 7.4.

Piezoresistive coefficients for several orientations of p-type silicon crystals is available in

Table 7.9.

Let us assume that the piezoresistor of (100) plane in the <100> orientation is used in this
case. We will have the coefficient L = 0.0244, with 44 = 138.1x10-11 Pa-1 from Table 7.8.
We will thus have the piezoresistive coefficient L = 2.762x10-11 Pa-1.

The corresponding rate of the change of electric resistance by the piezoresistor is:

R 6
  L  L  (2.762 x10 11 ) x(235.36 x10 )  6.5 x10 3
R

But since the resistance of a material is defined as:

L
R in which  is the resistivity of the material, which is a p-type piezoresistor. We find
A
the values of  vary from 10-3 to 104.5 -cm from Table 7.1. We will adopt a value of  = 7.8
-cm = 7.8x102 -m from Table 7.8.

38
 Thus, with L = 4x10-6 m and A = 20x10-12 m2, the resistance R in the piezoresistor is:

R 
7.8x10  4 x10   1.56 x10
2 6
8
 or 156 M
12
20 x10

The net change of resistance in the piezoresistor at 235.36 MPa stress is:

  
R  6.5 x10 3 1.56 x10 8  10.14 x10 5  or 1.014 M

Problem 5:

The piezoelectric coefficient, d, for PVDF polymer films can be found to be 18x10-12 m/v from
Table 7.14. Consequently, the induced voltage by the induced strain of 123.87x10-5 m/m from
Example 7.4 is:

  max 123.87 x10 5

V    12
 6.88 x10 7 v / m
d d 18 x10
with the piezoelectric film being 4 m long as shown in Fig. 7.17, the output voltage is:

v = V = (6.88x107)(4x10-6) = 275.3 v

Problem 6:

If the length of the imaginary lattice is (a) in the (100) plane, then 2a  1.414a is the lattice
for both diagonal (110) and inclined (111) planes in Fig. 7.8.

Problem 7:

The lattices for the three planes in a single silicon crystal are:

a 0.707a 0.707a
7a
0
0.7
7a

a L a
0
0.7

La L La
a La
0.707a 0.707a 0.707a 0.707a

(a) The (100) Plane: (b) The (110) Plane: (c) The (111) Plane

39
 1 3
L 2a  0.707a La  4 a  0.433a
2

Problem 8:

The angle is 54.74 degree.

Problem 9:

We have been using  = 0.25 as the Poisson’s ratio for silicon in our problems solving. By using
this value for the Poisson’s ratio and the shear modulus of elasticity, G in Table 7.2, we will have
the following values for the Young’s moduli, Eth, of silicon in the three orientations by using the
relationship: Eth = 2(1 + )/G:

Orientations G , GPa  Eth, GPa Etable , GPa

<100> 79.0 0.25 197.50 129.5
<110> 61.7 0.25 154.25 168.0
<111> 57.5 0.25 143.75 186.5

We may make the following observations:

(1) The Young’s moduli Etable in the above Table are the measured values as given in Table 7.2.
These values are lower than those calculated from linear theory of elasticity in the <100>
orientation, but are higher in the other two orientations.
(2) The computed Young’s moduli, Eth in the <111> has the lowest value of the three. This is
contrary to the measured values.

We thus conclude that the three elastic properties, E, G and  of silicon do not follow the
relationship established for isotropic elastic materials.

Problem 10:

We will use the geometry and the dimensions of the inkjet printer head as presented in Fig. 7.19
in Example 7.5.

For a printing resolution of 600 dots per inch (DPI), we should have the diameter of the dots
to be D = 1 inch/600 = 25.4 mm/600 = 42.333 m.

The corresponding radius of the spherical ink dot (r) that is ejected by the printer head is:

40
 4 3  2 
r   D  t in which t is the thickness of ink dot on the paper.
3 4 

Again we will use t = 1 m as in Example 7.5. This assumption will lead to r = 6.954x10-6 m

The volume of the ink dot is computed by using the right-hand-side of the above expression
to be Vdot = 1408x10-18 m3.

The corresponding expansion of the piezoelectric cover for the ejection of ink volume, Vdot
is:

4Vdot 4 x1408 x10 18

W    448 x10 12 m
 2

3.14 2000 x10 
6 2

The corresponding strain in the piezoelectric cover is

W 448 x10 12

    44.8 x10 6 m / m
L 10 x10 6

The required voltage for 1 m thick cover is:

 44.8 x10 6
V   12
 9.3418 x10 4 v/m
d 480 x10
The required voltage for the present case for a 10 m thick cover is thus:

v = LV = (10x10-6)(9.3418x104) = 0.9342 v

Chapter 8
Microsystem Fabrication Processes
(P. 305)

Part 1. Multiple Choice

1.(3); 2. (3); 3. (2); 4. (2); 5. (2); 6. (3); 7. (2); 8. (1); 9. (1); 10. (2); 11. (2); 12. (1); 13. (1); 14.
(2); 15. (2); 16. (1); 17. (2); 18. (1); 19. (1); 20. (2); 21. (3); 22. (2); 23. (2); 24. (1); 25. (3); 26.
(3); 27. (3); 28. (1); 29. (3); 30. (2); 31. (3); 32. (2); 33. (1); 34. (2); 35. (3); 36. (1); 37. (2); 38.
(2); 39. (2); 40. (3); 41. (3); 42. (2); 43. (1); 44. (3); 45. (1); 46. (3); 47. (3); 48. (3); 49. (3); 50.
(1)

Part 2. Computational Problems

41
Problem 1:

We have phosphorous as the dopant and the doping is carried out with 30 KeV energy.

From Table 8.2, we will have:

Rp = 42x10-9 m, and Rp = 19.5x10-9 m = 19.5x10-7 cm

We further have the maximum concentration, Nmax = 30x1018 atoms/cm3 as given in Example 8.1.
It is at x = Rp = 42x10-9 m = 0.042 m.

(a) The supplied dose is:

Q  2 Rp N max  6.28 x(19.5 x10 7 )(30 x1018 )  1.466 x1014 atoms / cm 2

(b) We will use the following relationship to find the concentration at x = 0.15 m:

  0.150.042  2 
N (0.15 m)  30 x1018 exp  2 
 6.57 x1012 atoms / cm 3
 2 x(0.0195) 

(c) Let xo be the depth at which the dopant concentration is 0.1% of the maximum value. This
depth may be obtained by solving the following equation:

18 1.466 x1014  ( xo  0.042) 2 

N ( xo)  0.0001x(30 x10 )  7
exp 2 
2 x3.14 x19.5 x10  2 x(0.0195) 

Solve for xo = 0.1257 m

Problem 2:

The estimation of time required to dope a silicon substrate using ion implantation technique
requires the understanding of the physics of this process. The movements of dopants in a
substrate is extremely complicated as described in S.M. Sze’s book on “Semiconductor
Devices”, John-Wiley & Sons, New York, 1985, pp.405-415. The method that we will propose
Ion mass
here is a very “raw” approximation that is based on several somewhat
Implantation not-too-realistic 2
N atoms/cm
Energy, E, Joules
assumptions as presented below.
Impact Velocity, Vo, m/s
(i) As illustrated in the figure, we assume the
mass of ions to be implanted is the
concentration at a depth, i.e. N(x) atoms/cm2
at a velocity V(x). V(t)
x
(ii) The implantation energy is E in Joules.
Rp
(iii)The maximum penetration is 2Rp, which is

42 Silicon Substrate
 the projected distance of maximum concentration
f almost all dopants.
(iv) The dopant ions penetrate into silicon substrate
without colliding with silicon atoms, so there is no
scattering with changes of directions in their motion.
The silicon substrate however does offer resistance to ion movements. This resistance is
responsible for the stopping of the moving ions after having traveled a distance 2Rp.

The deceleration of the dopant ions is estimated by the following derivation with above
hypotheses.
1
We have the input energy for the implantation, E  M V o2 in Joule or N-m, in which M is the
2
mass of the ions in Kg and Vo is the initial impact velocity in m/s.

From the above expression, we will have the impact velocity to be:

2E
Vo  (a)
M

N  x  dx
2R p

in which M 
0
with N(x) given in Eq. (8.1)
2Rp
We assume that all ions stop at x = 2Rp and the following relationship holds:

V f  V o  2 a av (2 R p )
2 2

in which aav is the average deceleration of the ions.

Since the final ion velocity Vf in the above expression is zero, so we have:

2
aav  Vo (b)
4R p

For the problem on hand, we have Nmax = 20x1020 atoms/cm3 at a depth of 0.2 m and Rp =
307x10-9 m and Rp = 69x10-9 m with E = 100 KeV as presented in Table 8.2.

By using Eq. (8.1), we establish the dose, Q:

Q   0.2 x10 6  307 x10 9  2 

20 x10 20
 exp  
2 69 x10 9 
2
6.28 x69 x10 9  

Solve for Q = 1.151x1015 atoms/cm2

43
Now if we use the relations in Eqs. (a) and (b) for aav and Vo and apply them to the current case
with:

2
1
x  V o t  aav t 2  x
2E V
t  o t2 (c)
2 M 8Rp

in which x = 2 m = 2x10-6 m; E = 100 KeV = 1.6097x10-14 N-m.

The total mass of the ions, M with a concentration of 20x1020 atoms/cm3 is:

M = (20x1020)(10.81x1.66057x10-27) = 3.59x10-5 Kg

We need to work out the following:

2E 2 x1.6097 x10 14

 5
 2.995 x10 5
M 3.59 x10

and

V o   2.995x10 5   3.6515x10 4
2 2

9
8Rp 8 x307 x10

By using Eq. (c), we may solve for the time required to have a concentration 20x1020 atoms/cm3
as:

2x10-6 = 2.995x10-5 t – 3.6515x10-4 t2

Solve the above equation for t = 0.27 s

Now, we are ready to find the depth of doping at which the concentration is asymptotically zero.

Let us first find the maximum concentration in the silicon substrate with a dose of Q =
1.151x1015 atoms/cm2. We may determine this value by following the expression presented in
Step 1 in Example 8.1 as:
Q 1.151x1015
N max   7
 6.6565x1019 atoms / cm 2
2 Rp 6.28 x69 x10

We assume that the concentration of the dopant is asymptotically zero at the depth xo at which
the concentration N(xo) is 10-6 of Nmax, i.e. N(x0) = 6.6565x1013 atoms/cm3. We will then have
the corresponding depth, xo from solving the following equation as in Eq. (8.1):

44
6.6565 x10 13

1.151x1015 
exp 

x o  307 x10 7  2


6.28 x69 x10 7 
 2 x 69 x10 7 2
  


We may solve xo from the above equation to be xo = 669.7x10-7 cm or 0.67 m

Problem 3:

From Table 8.3, we have the constants required to evaluate the diffusion coefficient, D in Eq.
(8.6) to be: a = -19.982 and b = 13.1109.

(a) From Eq. (8.6), n( D )  aT '  b with T’ = 1000/T = 0.8525, in which T = 900 + 273.

Hence n D    19.982x 0.8525  13.1109   3.924  D  0.01976 or D = 0.0003904

m2/h, or D = 0.10844x10-6 m2/s

(b) Eq. (8.4) can be used to obtain the concentration function, N(x,t) follows:

 x   x   1518x 
N  x, t   1011 erfc  11
  10 erfc 
11
  10 erfc 
 2 Dt   2 0.10844x10  6 t   t 
or

  1518 x 
N  x, t   1011 1  erf  
  t 

For x = 0.1 m and t = 1 h = 3600 s:

  151.8 
N  0.1m,1h   1011 1  erf    10 1  erf  2.53   10 1  0.99999  0
11 11

  60 
1000
(c) For diffusion to take place at 800oC, we have T '  800  273  0.932 . This will lead to the
computation of the diffusivity, D as follows:

n D   19.982 x0.932  13.1109   5.51165  D  0.000163 m 2 / h

The diffusivity, D is so low in this case that it leads to a negligible concentration at x = 0.1 m
after 1 hour into the diffusion process.

(d) Let us raise the diffusion temperature to 1100oC:

We will have T’ = 1000/(1100 + 273) = 0.72833, which leads to D = 0.05584 m2/h,

45
or 15.5x10-6 m2/s. This diffusivity will result in a concentration at x = 0.1 m and t = 1 h to be:
  0.1 
N  0.1 m,1 h   1011 1  erf    1011 1  erf  0.21167   
6 
  2 15.5 x10 x3600 
1011 1  0.2352   7.648 x1010 atoms / cm 3

Tabulation of the results on the concentrations at x = 0.1 m after 1 h into the diffusion process
at various temperatures is given below:

Temperature, oC 800 900 1000 1100

Diffusivity, D, m2/h 0.000163 0.0003904 0.005676 0.05584
Concentration, N(0.1 m, 1h), 0 0 3.482x1010 7.648x1010
atoms/cm3

It is clear from the above tabulation of results that the higher the diffusion temperature, the
higher the diffusivity. Consequently, one can expect much higher concentration of the dopants
beneath the surface of the substrate at higher temperatures.

Problem 4:

The time required reaching the same concentration of dopant of 3.482x1010 atoms/cm3 as in
Example 8.2 at 0.2 m beneath the surface at 1000oC can be obtained by solving the following
equation:
  398.21x 0.2 
3.482 x1010  1011 1 erf  
  t 

By using Table 3.4 on P. 83 and solve for the time, t = 13661.73 s, or 3.795 h

Problem 5:

Estimate the required time to achieve 1 m thick SiO2 on a silicon substrate.

The constants used in estimating the rate of oxidation in Eqs. (8.9) and (8.10) are available in the
Table established in Example 8.3:

Dry oxidation Wet oxidation

B/A, m/h 0.04532 0.6786
B, m/h 0.006516 0.2068

Now if we let x = 1 m in Eq. (a) and (b) in Example 8.3, we will have the time required to
oxidize the silicon substrate with 1 m thick SiO2 obtained from Eqs. (8.9) and (8.10) to be:

46
 Time for dry oxidation, h Time for wet oxidation, h
Eq. (8.9) for small time 22.065 1.474
Eq. (8.10) for larger time 153.47 4.836

Problem 6:

The dilution of the hydrogen gas is  = 1%, and the deposition takes place at 800oC. We assume
that the process is used to deposit thin SiO2 film, and that other conditions for this CVD process
are identical to those specified in Example 8.5.

(a) The number of molecules in one cubic meter volume of the gas mixture (NG):

We may follow the procedure in Example 8.4 and find the molar density of the gas mixture at
800oC to be:

 T 1   20  273 
d 2   d1  
3
  x44.643  12.1905 mol / m
 T 2   800  273 
The concentration of H2 per cubic meter is:

NG = (6.022x1023)x12.1905 = 73.4111x1023 molecules/m3

(b) The density () of the carrier gas, H2 is:

 = (2 g/mol)(12.1905 mol/m3) = 24.38 g/m3

(c) The Reynolds number (Re):

DV
Re 


with the gas density,  = 24.38 g/m3, the diameter of the reactor, D = 20 cm = 0.2 m, the gas
velocity, V = 50 mm/s (as given in Example 8.5), and the viscosity,  = 214 P = 0.0214
g/m-s from Table 8.6 for H2 gas at 825oC.

We may compute the Reynolds number to be Re = 11.39

(d) The thickness of the boundary layer ():

L 0.2
    0.0593 m
Re 11 .39

47
 (e) The diffusivity of the carrier gas (D):

We may use the same equation presented in Example 8.5 as shown below:

N
D
 N G  N s 

with N = 1024 molecules/m2-s (given)
NG = 73.4111x1023 molecules/m3 (in Part (a))
Ns = 0 and  = 1% = 0.01

We thus have the diffusivity, D to be:

0.0593x10 24
D   0.8078 m 2 / s

0.01 73.4111 x10 23  0 
(f) The surface reaction rate (ks):

Following the expression used in Example 8.5, this rate can be computed from the
expression:

DN 0.8078x10 24
ks     0.1376 m / s
D N G  N 0.8078x73.4111 x10 23  0.0593x10 24

(g) The deposition rate (r):

We first compute ks = 0.0593x0.1376 = 0.00816 << D = 0.8078

This will justify us using the expression for estimating the rate of deposition, r 
NG ks

From Table 8.7 and Example 8.5, we have the number of SiO2 molecules per unit film
volume to be:  = 4.3074x1028, which leads to the rate of deposition, r to be:

0.01x 73.4111 x10 23 x 0.1376

r   0.2345 m / s
4.3074 x10 28

Problem 7:

Since the rate of the deposition is 0.2345 m/s as calculated in Problem 6, a deposition of 0.5 m
thick film will take 0.5/0.2345 = 2.132 s.

48
Problem 8:

The following changes in computing the rate of deposition will take place with a 490oC process
temperature:

(a) The number of molecules in one cubic meter volume of gas mixture (NG):

 T 1   293 
d 2    d1    x44.643  17.1433 mol / m
3

 T 2   763 
The corresponding concentration is

N G  6.022x10 17.1433  103.24 x10 molecules / m

23 23 3

(b) The density:

  2 x17.1433  34.2866 g / m 3

(c) The Reynolds number:

DV 34.2866 x 0.2 x0.05

Re    20.53
 0.0167

(d) The thickness of the boundary layer:

L 0.15
    0.033 m
Re 20.53

(e) The diffusivity of the carrier gas:


N 0.033x10 24
D   0.3196 m 2 / s
  N G  N s  0.01103.24 x10  0
23

(f) The surface reaction rate:


DN 0.3196 x10 24
k s
  
D N G  N 0.3196 x103.24 x10 23  0.033x10 24
 0.0978 m / s

(g) The deposition rate:

49
 N G ks 0.01x103.24x10 23 x0.0978
r    0.2344 m / s
 28
4.3074x10

Chapter 9
Overview of Micromanufacturing
(P.330)

1. (1); 2. (2); 3. (2); 4. (1); 5. (3); 6. (2); 7. (3); 8. (1); 9. (2); 10. (2); 11. (2); 12. (2); 13. (1); 14.
(1); 15. (2); 16. (1); 17. (1); 18. (2); 19. (3); 20. (1); 21. (3); 22. (3); 23. (2); 24. (3); 25. (2); 26.
(1); 27. (2); 28. (3); 29. (2); 30. (3); 31. (1); 32. (3); 33. (3); 34. (2); 35. (1); 36. (1); 37. (2); 38.
(2); 39. (2); 40. (3); 41. (2); 42. (2); 43. (1); 44. (2); 45. (3)

Chapter 10
Microsystem Design
(P.385)

1. (1); 2. (2); 3. (3); 4. (1); 5. (3); 6. (2); 7. (2); 8. (2); 9. (1); 10. (2); 11. (3); 12. (1); 13. (3); 14.
(2); 15. (1); 16. (3); 17. (2); 18. (3); 19. (2); 20. (3); 21. (1); 22. (2); 23. (1); 24. (2); 25. (3); 26.
(1); 27. (3); 28. (2); 29. (1); 30. (2); 31. (3); 32. (3); 33. (1); 34. (2); 35. (1); 36. (2); 37. (2); 38.
(1); 39. (2); 40. (3); 41. (2); 42. (1); 43. (1); 44. (3); 45. (2); 46. (1); 47. (1); 48. (2); 49. (3); 50.
(2); 51. (1); 52. (3); 53. (1); 54. (2); 55. (3); 56. (3); 57. (3); 58. (3); 59. (1); 60. (3)

Chapter 11
Microsystem Packaging
(424)

1. (3); 2. (2); 3. (3); 4. (1); 5. (3); 6. (3); 7. (1); 8. (1); 9. (2); 10. (2); 11. (1); 12. (2); 13. (3); 14.
(1); 15. (2); 16. (3); 17. (2); 18. (2); 19. (2); 20. (1); 21. (1); 22. (2); 23. (3); 24. (2); 25. (1); 26.
(1); 27. (2); 28. (3); 29. (1); 30. (1); 31. (2); 32. (2); 33. (3); 34. (3); 35. (3); 36. (2); 37. (2); 38.
(2); 39. (1); 40. (3)

50