Lectures on MEMS and Microsystems Design and Manufacture

Chapter 7
Materials for MEMS and Microsystems

This chapter will cover the materials used in “silicon-based” MEMS and
microsystems. As such, silicon will be the principal material to be studied.

Other materials to be dealt with are silicon compounds such as: SiO2,
SiC, Si3N4 and polysilicon.

Also will be covered are electrically conducting of silicon piezoresistors

and piezoelectric crystals for electromechanical actuations and signal
transductions.

An overview of polymers, which are the “rising stars” to be used as MEMS

and microsystems substrate materials, will be studied too.
 Silicon – an ideal substrate material for MEMS

● Silicon (Si) is the most abundant material on earth. It almost always exists in
compounds with other elements.

● Single crystal silicon is the most widely used substrate material for MEMS
and microsystems.

● The popularity of silicon for such application is primarily for the following
reasons:

(1) It is mechanically stable and it is feasible to be integrated into electronics

on the same substrate (b/c it is a semiconducting material).

(2) Electronics for signal transduction such as the p or n-type piezoresistive

can be readily integrated with the Si substrate-ideal for transistors.

(3) Silicon is almost an ideal structure material. It has about the same
Young’s modulus as steel (∼ 2x105 MPa), but is as light as aluminum
with a density of about 2.3 g/cm3.
 Silicon – an ideal substrate material for MEMS-Cont’d

(4) It has a melting point at 1400oC, which is about twice higher than that of
aluminum. This high melting point makes silicon dimensionally stable
even at elevated temperature.

(5) Its thermal expansion coefficient is about 8 times smaller than that of
steel, and is more than 10 times smaller than that of aluminum.

(6) Silicon shows virtually no mechanical hysteresis. It is thus an ideal

candidate material for sensors and actuators.

(7) Silicon wafers are extremely flat for coatings and additional thin film
layers for either being integral structural parts, or performing precise
electromechanical functions.

(8) There is a greater flexibility in design and manufacture with silicon than
with other substrate materials. Treatments and fabrication processes
for silicon substrates are well established and documented.
 Single-Crystal Silicon
● For silicon to be used as a substrate material in integrated circuits and
MEMS, it has to be in a pure single-crystal form.
● The most commonly used method of producing single-crystal silicon is the
Czochralski (CZ) method.

The Czochralski method for producing single-crystal silicon

Puller Equipment: a crucible and a “puller”.

Quartz
crucible Graphite
Seed susceptor Procedure:
crystal (1) Raw Si (quartzite) + coal, coke, woodchips)
Silicon
boule are melted in the crucible.
Heating element

(2) A “seed” crystal is brought to be in contact

Heating element

with molten Si to form larger crystal.

Silicon melt (3) The “puller” slowly pulls the molten Si up
to form pure Si “boule” after the
solidification.
(4) The diameters of the “bologna-like” boules
vary from 100 mm (4”) to 300 mm (12”) in
diameters.
Chemical reaction for the process: SiC + SiO2 → Si + CO + SiO
Pure silicon wafers

Pure silicon boules of 300 mm

diameter and 30 ft long, can
weigh up to 400 Kg.
A pure silicon boule

These boules are sliced into

300 mm
thin disks (wafers) using 200 mm
wafer
diamond saws. wafer

Standard sizes of wafers are:

100 mm (4”) diameter x 500 µm thick.

150 mm (6”) diameter x 750 µm thick.
200 mm (8”) diameter x 1 mm thick
300 mm (12”) diameter x 750 µm thick (tentative).
Single Silicon Crystal Structure

● Single silicon crystals are basically of “face-cubic-center” (FCC)

structure.
● The crystal structure of a typical FCC crystal is shown below:
z
Atoms

Lattice

x
Note: Total number of atoms: 8 at corners and 6 at faces = 14 atoms
 Single Silicon Crystal Structure-Cont’d

● Single crystal silicon, however has 4 extra atoms in the interior.

● The situation is like to merge two FCC crystals together as shown below:

(a) Merger of two FCC (b) Merged crystal structure

● Total no. of atoms in a single silicon crystal = 18.

● The unsymmetrical distribution of atoms within the crystal make pure

silicon anisotropic in its mechanical properties.

● In general, however, we treat silicon as an isotropic material.

The Miller Indices
● Miller indices are commonly use to describe the faces of crystalline materials.
z
● A plane intersects x, y and z-coordinates at
P(x,y,z) a, b and c.
c y
b
● A point on the plane located at P(x,y,z)
a
● The equation defines the P(x,y,z) is:
x y z
x + + =1 (7.1)
a b c
Express Eq. (7.1) in a different form:
hx + ky + mz = 1 (7.2)
in which h = 1/a, k = 1/b and k = 1/c.

● Miller indices involve:

(hkm) = designation of a “face”, or a plane;

<hkm> = designation of a direction that is perpendicular to the (hkm) plane.

● NOTE: In a cubic crystal, such as silicon, a = b = c = 1

The 3 Distinct Planes of a Cubic Crystal
z
z z

y
y y

x
x x
Figure A Figure B Figure C

Top face: Plane (001) Diagonal face: Plane (110) Incline face:
Plane (111)
Right face: Plane (010)

Front face: Plane (100)

The 3 Principal Planes of a Silicon Crystal
(y) <001>
z
(001)

The (100) group (010)

(100)
y (x)<100>

x (z)<010>

The (110) group The (111) group

z
z

y
y

x
x
The 3 Principal Planes of a Silicon Crystal-Cont’d
0.543 nm 0.768 nm 0.768 nm

0.768 nm
(100) Plane (110) Plane (111) Plane
Diagonal Plane Inclined Plane
●Characteristics of silicon by principal planes:

(1) The (100) planes contain least number of atoms→ the weakest plane
→ easiest to work with.
(2) The (110) planes offers the cleanest surfaces in micro fabrications.
(3) The (111) contains shortest bonds between atoms → strongest plane
→ toughest to work with.
Miller Index for Orientation Young’s Modulus, E (GPa) Shear Modulus, G (GPa)

<100> 129.5 79.0

<110> 168.0 61.7

<111> 186.5 57.5

NOTE: The (100) plane makes an angle of 54.74o with the (111) plane.
 (Bulk) Mechanical and Thermophysical Properties of Silicon

Legend: σy = yield strength; E = Young’s modulus; ρ = mass density; C = specific heat;

k = thermal conductivity; α = coefficient of thermal expansion, TM = melting point.

σy E ρ C k α TM
(109 N/m2) (1011 N/m2) (g/cm3) (J/g-oC) (W/cm-oC) (10-6/oC) (oC)

Si 7.00 1.90 2.30 0.70 1.57 2.33 1400

SiC 21.00 7.00 3.20 0.67 3.50 3.30 2300
Si3N4 14.00 3.85 3.10 0.69 0.19 0.80 1930
SiO2 8.40 0.73 2.27 1.00 0.014 0.50 1700
Aluminum 0.17 0.70 2.70 0.942 2.36 25 660
Stainless Steel 2.10 2.00 7.90 0.47 0.329 17.30 1500
Copper 0.07 0.11 8.9 0.386 3.93 16.56 1080
GaAs 2.70 0.75 5.30 0.35 0.50 6.86 1238
Ge 1.03 5.32 0.31 0.60 5.80 937
Quartz 0.5-0.7 0.76-0.97 2.66 0.82-1.20 0.067-0.12 7.10 1710

* Principal source for semiconductor material properties: “Fundamentals of Microfabrication”, Marc

Madou, CRC Press, 1997
 Silicon Compounds
There are 3 principal silicon compounds used in MEMS and microsystems:
Silicon dioxide (SiO2), Silicon carbide (SiC) and silicon nitride (Si3N4) – each
Has distinct characteristic and unique applications.

Silicon dioxide (SiO2)

● It is least expensive material to offer good thermal and electrical insulation.

● Also used a low-cost material for “masks” in micro fabrication processes

such as etching, deposition and diffusion.

● Used as sacrificial material in “surface micromachining”.

● Above all, it is very easy to produce:

- by dry heating of silicon: Si + O2 → SiO2

- or by oxide silicon in wet steam: Si + 2H2O → SiO2 + 2H2

Silicon dioxide (SiO2) – cont’d

Properties Values
Density (g/cm3) 2.27
Resistivity (Ω-cm) ≥1016
Dielectric constant 3.9
Melting point (oC) ∼1700
Specific heat (J/g/oC) 1.0
Thermal conductivity (W/cm/oC) 0.014
Coefficient of thermal expansion (ppm/oC) 0.5

Silicon carbide (SiC)

Its very high melting point and resistance to chemical reactions make it ideal
candidate material for being masks in micro fabrication processes.

It has superior dimensional stability.

 Silicon nitride (Si3N4)
● Produced by chemical reaction:
3SiCl2H2 + 4NH3 → Si3N4 + 6HCL + 6H2
● Used as excellent barrier to diffusion to water and ions.
● Its ultra strong resistance to oxidation and many etchants make it a
superioro material for masks in deep etching.
A
● Also used as high strength electric insulators.
● Selected properties Si3N4 films are as follows:

Properties LPCVD* PECVD**

Deposition temperature (oC) 700-800 250-350
Density (g/cm3) 2.9-3.2 2.4-2.8
Film quality Excellent Poor
Dielectric constant 6-7 6-9
Resistivity (Ω-cm) 1016 10 -1015
6

Refractive index 2.01 1.8-2.5

Atom % H 4-8 20-25
Etch rate in concentrated HF 200 A/min
Etch rate in boillng HF 5-10A/min
Poisson’s ratio 0.27
Young’s modulus (GPa) 385
Coefficient of thermal expansion, ppm/oC 1.6
* Low pressure chemical vapor deposition; ** Plasma enhanced chemical vapor deposition
 Polycrystalline silicon

● It is usually called “Polysilicon”.

● It is an aggregation of pure silicon crystals with randomly orientations

deposited on the top of silicon substrates:
Random small
polysilicon grains

A few microns
Oxide layer

Silicon substrate ∼µm

● These polysilicon usually are highly doped silicon.

● They are deposited to the substrate surfaces to produce localized

“resistors” and “gates for transistors”

● Being randomly oriented, polysilicon is even stronger than single silicon

crystals.
Polycrystalline silicon – cont’d

Comparison of Mechanical Properties of Polysilicon with Other Materials

Materials Young’s modulus Poisson’s ratio Coefficient of

(GPa) thermal expansion
(ppm/oC)
As substrates:
Silicon 190 0.23 2.6
Alumina 415 8.7
Silica 73 0.17 0.4

As thin films:
Polysilicon 160 0.23 2.8
Thermal SiO2 70 0.2 0.35
LPCVD SiO2 270 0.27 1.6
PACVD SiO2 2.3
Aluminum 70 0.35 25
Tungsten 410 0.28 4.3
Polymide 3.2 0.42 20-70
 Silicon Piezoresistors

Piezoresistance = a change in electrical resistance of solids when

subjected to stress fields. Doped silicon are
piezoresistors (p-type or n-type).
Relationship between change of resistance {∆R} and stresses {σ}:

{∆R} = [π] {σ} (7-6)

where {∆R} = { ∆Rxx ∆Ryy ∆Rzz ∆Rxy ∆Rxz ∆Ryz}T represents the
change of resistances in an infinitesimally small cubic piezoresistive
crystal element with corresponding stress components:
{σ} = {σxx σyy σzz σxy σxz σyz}T and [π] = piezoresistive coefficient matrix.
Mechanical
load y
σyy
A silicon piezoresistance
σyx
subjected to a stress field: p- or n-type σyz σxy σ
silicon σzy xx
y σzz σzx σxz x
x
z
z
 Silicon Piezoresistors – Cont’d
● Due to equilibrium condition, there are six independent stress components:
3 normal stress components and 3 shearing stress components.
● Consequently, the piezoresistive coefficient matrix has the components:
⎡π 11 π 12 π 12 0 0 0 ⎤
⎢ ⎥
⎢ π 12 π 11 π 12 0 0 0 ⎥
⎢ 0 ⎥
[π ] = ⎢π 12 π 12 π 11
0 0 (7.7)
⎥
⎢0 0 0 π 44 0 0 ⎥
⎢0 0 0 0 π 44 0 ⎥
⎢ ⎥
⎢⎣0 0 0 0 0 π 44⎥⎦
● Expanding Eq. (7.6) result in the following:
∆ R xx = π 11σ xx + π 12 (σ yy + σ zz ) ∆ R xy = π 44 σ xy

∆ R yy = π 11σ yy + π 12 (σ xx + σ zz ) ∆ R xz = π 44 σ xz

∆ R zz = π 11σ 11 + π 12 (σ xx + σ yy ) ∆ R yz = π 44 σ yz

● Note: Only 3 piezoresistive coefficients are required; π11 and π12 associated
with normal stresses and π44 with shearing stresses.
 Silicon Piezoresistors – Cont’d

Numerical values of piezoresistive coefficients

Silicon piezoresistors at room temperature
Materials Resistivity
(Ω-cm)
π11* π12* π44*

p-silicon 7.8 +6.6 -1.1 +138.1

n-silicon 11.7 -102.2 +53.4 -13.6

Silicon piezoresistors
Leads x
y T-direction
∆R
irec t ion = π Lσ L + π T σ T
L-d R
p- or n-type Si
 Silicon Piezoresistors – Cont’d

Numerical values of piezoresistive coefficients

Leads x
∆R
y T-direction
= π Lσ L + π T σ T
ec tion R
ir
L-d
p- or n-type Si

Piezoresistive coefficients of p-type silicon piezorestors in various directions

Crystal Planes Orientation Orientation πL πT

<x> <y>
(100) <111> <111> +0.66π44 -0.33π44
(100) <110> <100> +0.5π44 ∼0
(100) <110> <110> +0.5π44 -0.5π44
(100) <100> <100> +0.02π44 +0.02π44
Example 7.3 Estimate the change of resistance in silicon piezoresistors
attached to the diaphragm of a pressure sensor in Example 4.4.
From Example 4.4, the corresponding maximum stress
at the mid-span of each of the 4 edges is:
Piezoresistor, A σmax = 186.81 MPa with an applied pressure at 70 MPa.

(100 Plane) We may thus let: σL = σT = σmax = 186.8 MPa, or

B
C
σ =σ
= 186.8x106 Pa (N/m2)
L max
σT = σmax
Since the diaphragm is on (100) face, the
two piezoresistive coefficients are:
D
266 µm
π L = πT = 0.02 π44
480 µm
783 µm
1085 µm But π44 = 138.1 x 10-11 Pa-1 from the Table,
2500 µm we thus have:
∆R
= π L σ L + π T σ T = 2π 44σ max = 2 × 0.02(138.1× 10 −11 )(186.8 × 10 6 ) = 0.01032 Ω/Ω
R
 Temperature sensitivity of silicon piezoresistors

A major deficiency of silicon piezoresistors is its sensitivity of temperature

as indicated in the table:

Doping p-Type TCR p-Type TCP n-Type TCR n-Type TCP

concentration (% per oC) (% per oC) (% per oC) (% per oC)
(1018/cm3)

5 0.0 -0.27 0.01 -0.28

10 0.01 -0.27 0.05 -0.27
30 0.06 -0.18 0.09 -0.18
100 0.17 -0.16 0.19 -0.12

TCR = temperature coefficient of resistance;

TCP = temperature coefficient of piezoresistivity.

Example: a p-type silicon piezoresistor with a doping of 1019 atoms/cm3,

The loss of piezoresistivity is 0.27%/oC. In an operating temperature of 120oC,
It would lose (120-20)x0.27% = 27% of the value of the piezoresistivity coefficient.
 Gallium Arsenide (GaAs)

● GaAs is a compound simiconductor with equal number of Ga and As atoms.

● Because it is a compound, it is more difficult to process.

● It is excellent material for monolithic integration of electronic and

photonic devices on a single substrate.

● The reason for being excellent material for photoelectronics is its high
electron mobility (7 times more mobile than silicon):

Materials Electron Mobility, m2/V-sec

Aluminum 0.00435
Copper 0.00136
Silicon 0.145
Gallium Arsenide, GaAs 0.850
Silicon oxide ≅0
Silicon nitride ≅0
Gallium Arsenide (GaAs)-Cont’d

● GaAs is also a good thermal insulator.

● Low yield strength (only 1/3 of that of silicon) – “bad”.

● A comparison of GaAs and silicon as substrate materials in

micromachining:

Properties GaAs Silicon

Opto-electronics Very good Not good
Piezoelectric effect Yes No
Piezoelectric coefficient (pN/oC) 2.6 Nil
Thermal conductivity Relatively low Relatively high
Cost High Low
Bonding to other substrates Difficult Relatively easy
Fracture Brittle, fragile Brittle, strong
Operating temperature High Low
Optimum operating temp. (oC) 460 300
Physical stability Fair Very good
Hardness (GPa) 7 10
Fracture strength (GPa) 2.7 6
Quartz Z-axis
Si
● Quartz is a compound of SiO2.
● The single-unit cell is in shape of tetrahedron:
O2
● Quartz crystal is made of up to 6 rings with
O2
6 silicon atoms.
O2

Properties Value || Z Value ⊥ Z Temperature

Dependency
Thermal conductivity 29x10-3 16x10-3 ↓with T
(Cal/cm/sec/oC)
Relative permittivity 4.6 4.5 ↓with T
Density (Kg/m3) 2.66x103 2.66x103
Coefficient of thermal 7.1 13.2 ↑with T
expansion (ppm/oC)
Electrical resistivity (Ω/cm) 0.1x1015 20x1015 ↓with T
Fracture strength (GPa) 1.7 1.7 ↓with T
Hardness (GPa) 12 12
 Quartz-Cont’d

● Quartz is ideal material for sensors because of its extreme dimensional

stability.

● It is used as piezoelectric material in many devices.

● It is also excellent material for microfluics systems used in biomedical

applications.

● It offers excellent electric insulation in microsystems.

● A major disadvantage is its hard in machining. It is usually etched in

HF/NH4F into desired shapes.

● Quartz wafers up to 75 mm diameter by 100 µm thick are available

commercially.
 Piezoelectric Crystals
● Piezoelectric crystals are solid ceramic compounds that produce
piezoelectric effects:
Induced Mechanical
Mechanical Deformation
Forces

Applied Voltage, V
V

Mechanical force induced Electric voltage induced

electric voltage mechanical deformation

● Natural piezoelectric crystals are: quartz, tourmaline and sodium

potassium tartrate.

● Synthesized crystals are: Rochelle salt, barium titanate and lead

zirconate.
 Piezoelectric Crystals – Cont’d
Induced Mechanical
Mechanical Deformation
Forces

Applied Voltage, V
V

Mechanical force induced Electric voltage induced

electric voltage mechanical deformation

Mechanical strain by electric field: Electric field by stress:

ε=dV V=fσ
where ε = induced strain where V = generated voltage in
d = piezoelectric coefficient volts/m
V = applied voltage, V/m σ = applied stress in Pa

1
=E
fd
 Piezoelectric Crystals – Cont’d
Piezoelectric coefficients:

Piezoelectric Crystals Coefficient, d Electromechanical

(10-12 m/volt) conversion factor, K**
Quartz (crystal SiO2) 2.3 0.1
Barium titnate (BaTiO3) 100-190 0.49
Lead zirconate titanate, PZT 480 0.72
(PbTi1-xZrxO3)
PbZrTiO6 250

PbNb2O6 80

Rochelle salt 350 0.78

(NaKC4H4O6-4H2O)
Polyvinylidene fluorid, PVDF 18

**K 2 = Output of mechanical energy

or K2 =
Output of electrical energy
Input of electrical energy Input of mechanical energy
Example 7.4

A thin piezoelectric crystal film, PZT is used to transduce the signal in a micro
accelerometer involving a cantilever beam made of silicon. The accelerometer
is design for maximum acceleration/deceleration of 10 g.

The PZT transducer is located at the support of the cantilever beam where the
maximum strain exists (near the support) during the bending of the beam
as illustrated below.

Determine the electrical voltage output from the PZT film at a maximum
acceleration/deceleration of 10 g. g th
len
e am
PZT crystal 2 µm B
Mass: 10 µ
(see detail A) m σmax
m = 10 mg
A 10 µm
µm
4
A 50 µm
1000 µm σmax
View A-A Detail A
Example 7.4 – Cont’d
Solution:
Use Newton’s 2nd law to find the equivalent dynamic force with an acceleration
of 10 g: Peq = ma = (10x10-6)x(10x9.81) =981x10-6 N
Peq

L = 1000 µm

The maximum bending moment is:

Mmax = PeqL = (981x10-6)(1000x10-6) = 0.981x10-6 N-m and it occurs at the built-in end

The corresponding maximum stress is:

M max C(0.981x10 −6 )(25 x10 −6 ) 6 Pa
σ max = = = 235 . 36 x10
I (0.1042 x10 −18 )
=
σ =
235.36 x10 6
= −5
ε m/m
max
and the maximum strain is: max 11
123. 87 x10
E 1.9 x10
The voltage generated in the PZT piezoelectric crystal is:
ε ε max 123.87 x10 −5
V= = = −12
= 0.258 x10 7
Volts/m
d d 480 x10
or v = Vl = (0.258x107)(4x10-6) = 10.32 volts
 Example 7.5

Piezoelectric
Determine the required electric Teflon Actuator
voltage for ejecting a droplet of ink Coating
2000 µm φ
from an inkjet printer head using PZT
piezoelectric crystal as a pumping V
mechanism.
10 µm

The ejected ink will have a resolution of Inkwell

300 dpi (dots per inch). The ink droplet
is assumed to produce a dot with a film
thickness of 1 µm on the paper.
Ejection nozzle
The geometry and dimension of the diameter, d
printer head is illustrated below. Ink droplet:
Sphere with diameter, d
Assume that the ink droplet takes a
shape of a sphere and the inkwell is 1 µm Dot on paper
Paper
always re-filled after ejection. D
 Example 7.5 – Cont’d
Piezoelectric
Teflon Actuator
Solution: Coating
2000 µm φ
● Determine the ejection nozzle diameter, d:
V

The diameter of the dot film on the paper is: 10 µm

Inkwell
D = 1/300 inch = 0.084666 mm = 84.67 µm

By equating the volumes of the dot sphere

the flat dot on the paper, we have: Ejection nozzle

4 3 ⎛π 2⎞
πr = ⎜ D ⎟(t )
diameter, d
Ink droplet:
Sphere with diameter, d
3 ⎝4 ⎠ 1 µm Dot on paper
Paper
D
from which, we get the radius of the
dot, r = 11.04x10-6 m, with D = 84.7 µm and t = 1 µm
 Example 7.5 – Cont’d

● We assume that:
Piezoelectric
Teflon Actuator
Volume of an ink droplet Coating
leaving the ink well 2000 µm φ

V
= Volume created by vertical 10 µm
expansion of the PZT cover
Inkwell

Let W = vertical expansion of the PZT cover

induced by the applied voltage, V Ejection nozzle
∆ = diameter of the PZT cover = 2000 µm diameter, d
Ink droplet:
Sphere with diameter, d

1 µm Dot on paper
We will have: Paper
D

4 V dot4 x5629.21x10 −18 −12

W = = = 1791 . 83 x10 m
π ∆ 2 3.1416(2000 x10 −6 ) 2
 Example 7.5 – Cont’d

Piezoelectric
● The corresponding strain in the PZT Teflon Actuator
Coating
piezoelectric cover is: 2000 µm φ

V
−12
W 1791.83x10
ε= = = 179.183x10 −6 m/m 10 µm
L 10 x10 −6
Inkwell

The piezoelectric coefficient of the PZT

crystal is d = 480x10-12 m/v, leading to
the required voltage to be:
Ejection nozzle

179.183x10 −6
diameter, d
ε 6 volts/m
Ink droplet:
V= = = 0.3733 x10 Sphere with diameter, d
d 480 x10 −12 1 µm Dot on paper
Paper
or D

( )( )
v = LV = 10 x10 −6 0.3733 x10 6 = 3.733 volts
 Polymers

What is polymer?
Polymers include: Plastics, adhesives, Plexiglass and Lucite.

Principal applications of polymers in MEMS:

● Currently in biomedical applications and adhesive bonding.
● New applications involve using polymers as substrates with
electric conductivity made possible by doping.

Molecular structure of polymers:

● It is made up of long chains of organic (hydrocarbon) molecules.
● The molecules can be as long as a few hundred nm.

Characteristics of polymers:
● Low melting point; Poor electric conductivity
● Thermoplastics and thermosets are common industrial products
● Thermoplastics are easier to form into shapes.
● Thermosets have higher mechanical strength even at temperature
up to 350oC.
Polymers as industrial materials

Polymers are popular materials used for many industrial products for the
following advantages:

● Light weight

● Ease in processing

● Low cost of raw materials and processes for producing polymers

● High corrosion resistance

● High electrical resistance

● High flexibility in structures

● High dimensional stability

 Polymers for MEMS and microsystems

(1) Photo-resist polymers are used to produce masks for creating desired
patterns on substrates by photolithography technique.
(2) The same photoresist polymers are used to produce the prime mold with
desirable geometry of the MEMS components in a LIGA process in micro
manufacturing.
(3) Conductive polymers are used as “organic” substrates for MEMS and
microsystems.
(4) The ferroelectric polymers that behave like piezoelectric crystals can be used
as the source of actuation in micro devices such as in micro pumping.
(5) The thin Langmuir-Blodgett (LB) films can be used to produce multilayer
microstructures.
(6) Polymers with unique characteristics are used as coating substance to
capillary tubes to facilitate effective electro-osmotic flow in microfluidics.
(7) Thin polymer films are used as electric insulators in micro devices,
and as dielectric substance in micro capacitors.
(8) They are widely used for electromagnetic interference (EMI) and radio
frequency interference (RFI) shielding in microsystems.
(9) Polymers are ideal materials for encapsulation of micro sensors and
the packaging of other microsystems.
Conductive Polymers

● Polymers are poor electric conducting materials by nature.

● A comparison of electric conductivity of selected materials are:

Materials Electric Conductivity, S/m*

Conductors:
Copper, Cu 106-108
Carbon 104

Semiconductors:
Germanium, Ge 100
Silicon 10-4-10-2

Insulators:
Glass 10-10-10-8
Nylon 10-14-10-12
SiO2 10-16-10-14
Polyethlene 10-16-10-14

* S/m = siemens per meter = Ω-1 = A2-s3/Kg-m2

Conductive Polymers – Cont’d
Some polymers can be made electrically conductive by the following
3 methods:

(1) Pyrolysis:
at 600oC
Pyropolymer-base Conductive polymer
Phthalonitrile resin Amine =
as high as 2.7x104 S/m

(2) Doping:
Introducing metal atoms into molecular matrices of polymers
→ Conductive polymers
Polymers groups Dopants
Polyacetylenes (PA) Br2, I2, AsF5, HClO4 and H2SO4 for p-type
Sodium naphthalide in tetrahydrofuran for n-type
Polyparaphenylenes (PPP) AsF5 for p-type; alkali metals for n-type
Polyphenylene sulfide (PPS) AsF5

(3) Insertion of conductive fibers:

Fibers made of Au, Ag, stainless steel, aluminum fibers and flakes.
 Langmuir-Blodgett (LB) films

● The process was first introduced by Langmuir in 1917 and was later
refined by Blodgett. That was why it is called Langmuir-Blodgett
process, or LB films.

● The process involves the spreading volatile solvent over the surface-active
substrate materials.

● The LB process can produce more than one single monolayer by depositing
films of various compositions onto a substrate to produce a multilayer
structure.

● LB films are good candidate materials for exhibiting ferro (iron)- , pyro (heat)-
and piezoelectric properties. LB films may also be produced with controlled
optical properties such as refractive index and anti reflections.

They are thus ideal materials for micro sensors and optoelectronic devices.
 Langmuir-Blodgett (LB) films – Cont’d

● Following are a few examples of LB film applications in microsystems:

(1) Ferroelectric (magnetic) polymer thin films:

● The one in particular is the Poly-vinylidene fluoride (PVDF).

● Applications of this type of films include:
- Sound transducers in air and water,
- Tactile sensors,
- Biomedical applications such as tissue compatibility,
cardio-pulmonary sensors and implantable transducers
and sensors for prosthetics and rehabilitation devices.
● As a piezoelectric source. The piezoelectric coefficient of PVDF is given
in Table 7-14.

(2) Coating materials with controllable optical properties:

● Broadband optical fibers that transmit light at various wavelengths.
Langmuir-Blodgett (LB) films – Cont’d

(3) Microsensors:

Many electrically conducting polymeric materials are sensitive to the

exposed gas and other environmental conditions. So they are suitable
materials for micro sensors.

Its ability of detecting specific substances relies on the reversible and

specific absorption of species of interest on the surface of the polymer
layer and the subsequent measurable change of conductivity of the polymer.
Gas Sample
Conductive
Dielectric Polymer
Plastic A gas sensor:
Encapsulant Electrical conductivity changes
V with absorption of the exposed
p-Silicon Substrate
gas.
 SU-8 Photoresists

• It is a negative epoxy-based polymer sensitive to UV light (λ = 350-400 nm)

• It is used for thin-film production with thickness from 1 µm to 2 mm
• Reasons for it being popular in MEMS:
• Can be built to thick films for 3-D MEMS structures (aspect ratio to 50)
• Much lower production costs than thick films by silicon
• It is commercially available in liquid form
• SU-8 films can be produced by a spin-process:
Dispenser
Resist Edge
Resist Photoresist
spray bead
puddle Wafer Catch cup Wafer
Vacuum
Vacuum chuck chuck
Spinner
SU8-100 SU8-50
motor
To drain & 250
Film thickness
(micrometer)

To vacuum exhaust 200

pump 150
100
50
0
1000 1500 2000 2500 3000
Spin speed (rpm)
 Mechanical Properties of SU-8 Polymer
Young’s modulus 4400 MPa

Poisson’s ratio 0.22

Viscosity 0.06 Pa-s (40% SU-8 – 60% solvent)

1.50 Pa-s (60% SU-8 – 40% solvent)
15.0 Pa-s (70% SU-8 – 30% solvent)

Coefficient of thermal expansion* 0.183 ppm /oC

Thermal conductivity 0.073 W/cm-oC

Glass transition temperature 200oC

Reflective index 1.8 at 100 GHz

1.7 at 1.6 THz
Absorption coefficient 2/cm at 100 GHZ
40/cm a 1.6 THz
Relative dielectric constant 3 at 10 MHz

Source: Guerin 2005. * in comparison to 2.33 ppm/oC for silicon

 Typical Process Flow for Constructing SU-8 Films

(1) (2) (3)

Substrate Spin-coating Soft
treatment SU-8 photoresist bake
(e.g. silicon wafer, washed inH2SO4
+H2O2, baked at 200oC for 5 min) (spin @ desired speed for 30 s)

(4) (5) (6)

Exposure to Post expose Develop
UV light bake SU-8 photoresist
(required time by supplier) (following instruction by supplier)
(in chemical
solvent)

(7) (8) (9)

Rinse Hard SU-8 film
& dry bake removal
(in de-ionized water) (@150-200oC) (in special liquid bath, or laser
ablation, and pyrolosis, or as
instructed by the supplier)
 Packaging Materials

Unlike IC packaging in which plastic or ceramic are extensively used as

encapsulate materials for the delicate IC circuits, MEMS packaging
involve a great variety of materials-varying from plastic and polymers
to stainless steel, as can be seen in a specially packaged micro
pressure sensor:
Cylindrical metal casing

Metal wires Square silicon die

& pad
Adhesive die bond
Metal interconnect pins
SiO2 insulator

Pyrex glass
Constraint base
 Lectures on
MEMS and MICROSYSTEMS DESIGN AND MANUFACTURE

Chapter 8
Microsystems Fabrication Processes
To fabricate any solid device component, one must first select materials
and adequate fabrication method.

For MEMS and microsystems components, the sizes are so small that
no machine tools, e.g. lathe, milling machine, drilling press, etc. can do
the job. There is simply no way one can even grip the work piece.

Consequently, radically different techniques, non-machine-tool techniques

need to be used for such purpose.

Most physical-chemical processes developed for “shaping” and fabricating

ICs are adopted for microsystems fabrications. This is the principal
reason for using silicon and silicon compounds for most MEMS and
microsystems – because these are the materials used to produce ICs.
 Microfabrication Traditional Manufacturing
by physical-chemical processes by machine tools
 Microfabrication Processes

● Photolithography

● Ion implantation

● Diffusion

● Oxidation

● Chemical vapor deposition

● Physical vapor deposition (Sputtering)

● Deposition by expitaxy

● Etching
 Photolithography
Photolithography process involves the use of an optical image and a
photosensitive film to produce desired patterns on a substrate.

The “optical image” is originally in macro scale, but is photographically

reduced to the micro-scale to be printed on the silicon substrates.

The desired patterns are first printed on light-transparent mask, usually

made of quartz.

The mask is then placed above the top-face of a silicon substrate coated
with thin film of photoresistive materials.

The mask can be in contact with the photoresistave material, or placed

with a gap, or inclined to the substrate surface:
Light Light Light
Mask

Si substrate
Si substrate
Si substrate
Photoresistive coating
 Photolithography-Cont’d

Positive resist:
Substrate (a)

Photoresist
(b)
Substrate

UV light or (c)
other sources
Mask
Negative resist:
Substrate (a)

Processes: (b)
(a) Development
(b) Etching
(c) Photoresist removal (c)
 Photolithography- Cont’d

The two kinds of photoresists:

● Positive resists:

There are two kinds of positive resists:

(1) the PMMA (polymethymethacrylate) resists,
(2) the two-component DQN resist involving diazoquinone ester (DQ) and
phenolic novolak resin (N).

In the latter kind, the first component takes about 20-50% by weight in the
compound.

Positive resists are sensitive to UV lights with the maximum sensitivity at a

wavelength of 220 nm.

The PMMA resists are also used in photolithography involving electron beam,
ion beam and x-ray. Most positive resists can be developed in alkaline solvents
such as KOH (potassium peroxide), TMAH (tetramethylammonium hydroxide),
ketones or acetates.
 Photolithography- Cont’d

● Negative resists:

(1) Two-component bis (aryl) azide rubber resists, and

(2) Kodak KTFR (azide-sensitized polyisotroprene rubber).

Negative resists are less sensitive to optical and x-ray exposures but more
sensitive to electron beams.

Xylene is the most commonly used solvent for developing negative resists.

Line definitions of photoresists:

In general, positive resists provide more clear edge definitions than the
negative resists. So, it is a better option for high resolution patterns for
micro devices.

(a) by negative resists (b) by positive resists

 Photolithography- Cont’d
Application of photoresists
● The process begins with securing the substrate wafer onto the top of a vacuum chuck.
● A resist puddle is first applied to the center portion of the wafer from a dispenser.
● The wafer is then subjected to high speed spinning at a rotational speed from 1500 to
8000 rpm for 10 to 60 seconds. The speed is set depending on the type of the resist,
the desired thickness and uniformity of the resist coating.
● The centrifugal forces applied to the resist puddle cause a uniform spread of the fluid
over the entire surface of the wafer.
● Typically the thickness is between 0.5 – 2 µm with ±5 nm variation.
For some microsystems applications, the thickness had been increased to 1 cm.

Dispenser Reduce size by controlled

spinning speed
Resist Edge
Resist Photoresist
spray bead
puddle Wafer Catch cup Wafer
Vacuum
Vacuum chuck chuck
Spinner
motor
To drain &
To vacuum exhaust
pump
 Photolithography- Cont’d
Light sources

Photoresist materials used in micro fabrication are sensitive to light

with wavelength ranging from 300 to 500 nm.

Most popular light source for photolithography is the mercury vapor lamps.
This light source provides a wavelength spectrum from 310 to 440 nm.

Deep UV (ultra violet) light has a wavelength of 150-300 nm and the

UV light source has wavelengths between 350-500 nm.

In special applications for extremely high resolutions, x-ray is used.

The wavelength of x-ray is in the range from 4 to 50 Angstrom.
(an Angstrom, = 0.1 nm or 10-4 µm).
 Photolithography- ends

Photoresist development
Dispenser
● The same spinner may be used for Resist Edge
development after exposure with Resist spray bead
Photoresist
dispensing development solvent. puddle Wafer Catch cup Wafer
● A rising distilled water follow Vacuum chuck
Vacuum
chuck
the development. Spinner
● Developers agent for +ve resistor motor
To drain &
are KOH or TMAH. Xylene is the To vacuum exhaust
agent for –ve resistors, pump

Photoresist removal and postbaking

● After development and the desired pattern in created in the substrate,

a descumming process takes place.
● The process uses O2 plasma to remove the bulk of photoresist.
● Postbaking to remove the residue of solvent at 120oC for 20 minutes.
● Etching will remove all residue photoresist.
 Ion Implantation

● It is physical process used to dope silicon substrates.

● It involves “forcing” free charge-carrying ionized atoms of B, P of As
into silicon crystals.
● These ions associated with sufficiently high kinetic energy will be
penetrated into the silicon substrate.
● Physical process is illustrated as follows:

High
energy beam Beam
Beam Accelerator
controller controller

Ion beam Shield High energy

(e.g. SiO2) Distribution
of dopant ion beam
Ion Source
(ionized dopant
atoms) Si Substrate
Silicon substrate
(See Figure 3.4)
Required energy for ion implantation:

Dopants p- or n-type Ionization Energy (eV)

Phosphorus (P) n 0.044

Arsenic (As) n 0.049
Antimony (Sb) n 0.039
Boron (B) p 0.045
Aluminum (A ) p 0.057
Gallium (Ga) p 0.065
Indium (In) p 0.160
 Ion Implantation – Cont’d

Density distribution in depth

Energized ion beam

Shield Rp
N(x)
Doped
zone
∆Rp
Silicon substrate
x

Q ⎡ − ( x − R p) 2 ⎤
N ( x) = exp ⎢ ⎥ (8.1)
2π ∆ R p ⎣ 2 ∆ R p
2
⎦
where Rp = projected range in µm,
∆Rp = scatter or “straggle” in µm, and
Q = dose of ion beam (atoms/cm2)
 Ion Implantation – Cont’d

Density distribution in depth – Cont’d

(a) At 30 KeV energy level:

Ion Range, Rp (nm) Straggle, ∆Rp (nm)

Boron (B) 106.5 39.0
Phosphorous (P) 42.0 19.5
Arsenic (As) 23.3 9.0

(b) At 100 KeV energy level:

Ion Range, Rp (nm) Straggle, ∆Rp (nm)

Boron (B) 307.0 69.0

Phosphorous (P) 135.0 53.5
Arsenic (As) 67.8 26.1
 Example 8.1

A silicon substrate is doped with boron ions at 100 KeV. Assume the
maximum concentration after the doping is 30 x 1018/cm3. Find:
(a) the dose, Q in Eq. (8-1),
(b) the dopant concentration at the depth 0.15 µm,
(c) the depth at which the dopant concentration is at 0.1% of the
maximum value.
Solution:

(1) To find the dose:

From Table 8.2, we find Rp = 307 nm = 307x10-7 cm and ∆Rp = 69x10-7 cm at 100 KeV
energy level.

Since we have the maximum concentration, Nmax = 30 x 1018/cm3 at x = Rp in Eq. (8-1),

so from the same equation, we will have:
Q
N max =
2π ∆ R p
from which, we have the dose:

Q = (2π)0.5(∆Rp)Nmax = (6.28)0.5(69 x 10-7 cm)(30 x 1018 cm-3) = 5.2 x 1014/cm2

 Example 8.1 – Cont’d

(2) To find the concentration at x = 0.15 µm:

We may use the following relationship derived from Eq. (8-1):

⎡ (0.150 − 0.307) 2 ⎤ ⎛ 0.0246 ⎞

N (0.15µm) = N max exp ⎢− ⎥ = ( 30 x10 18
) exp⎜ − ⎟ = 2.27 x10 18
/ cm 3

⎣ 2(0.069) 2 ⎦ ⎝ 0.009522 ⎠

(3) To find x = xo at which the concentration N(xo) = (0.1%)Nmax = 3 x 1016/cm3:

We may solve xo from the following equation:

5.2 x 1014 ⎡ ( xo − 307 x10 −7 ) 2 ⎤

N ( xo) = exp ⎢ − ⎥ = 3 x10 16

2 x 3.14 x 69 x10 − 7 ⎣ 2(69 x10 − 7 ) 2 ⎦

from which, we solve for xo = 563.5x10-7 cm or 0.5635 µm.

 Diffusion

● Diffusion is another common technique for doping silicon substrates.

● Unlike ion implantation, diffusion takes place at high temperature.

● Diffusion is a chemical process.

● The profile of the spread of dopant in silicon by diffusion is different from

that by ion implantation:

High temp. Ion beam at RT

dopant gas

Mask
Silicon substrate

Dopant profile by Diffusion Dopant profile by ion implantation

 Diffusion – Cont’d

Design analysis of diffusion

Fick’s law governs diffusion (Chapter 3):

∂N ( x )
F = −D (8.2)
∂x
where F = Dopant flux, which is the number of dopant atoms passing through
a unit area of the substrate in a unit time (atoms/cm2-sec)
D = Diffusion coefficient or diffusivity of the substrate to the dopant (cm2/sec)
N = Dopant concentration in the substrate per unit volume. (atoms/cm3)

The distribution of dopant, N(x,t) in the “depth” direction of the substrate

at time t is obtained from the “Diffusion equation”:

∂N ( x, t ) ∂ 2 N ( x, t )
=D (8.3)
∂t ∂x 2
 Diffusion – Cont’d
Solution of diffusion equation

∂N ( x, t ) ∂ 2 N ( x, t ) (8.3)
=D
∂t ∂x 2
Initial condition (t = 0):

N(x,0) = 0, meaning there is no impurity in the substrate when the

diffusion process begins.

Boundary conditions (at x = 0 and ∞ ):

N(0,t) = Ns, which is the concentration at the surface exposed to the gaseous dopant.
N(∞, t) = 0, meaning the diffusion of foreign substance is highly localized, and that
the concentration far away from the contacting surface is negligible.

The solution of the diffusion equation in (8.3) with these conditions is:
⎡ x ⎤
N ( x, t ) = N s erfc ⎢ ⎥ (8.4)
⎣ 2 Dt ⎦
where erfc(x) is the complementary error function that has the form:
x 2
2 −
(8.5)
∫e
erfc( x ) = 1 − erf ( x ) = 1 − y dy
π 0
 Diffusion – Cont’d
The diffusivity, D:

The diffusivity of silicon with common dopants can be obtained by

an empirical formula:
ln ( D ) = aT '+ b (8.6)

in which the temperature, T’ = 1000/T with T = diffusion temperature in K.

The constants a and b for common dopants can be obtained from:

Dopants Constant, a Constant, b

Boron -19.9820 13.1109

Arsenic -26.8404 17.2250
Phosphorus (Ns = 1021/cm3) -15.8456 11.1168
Phosphorus (Ns = 1019/cm3) -20.4278 13.6430
 Example 8.2
A silicon substrate is subjected to diffusion of boron dopant at 1000oC with a dose
1011/cm2. Find:
(a) the expression for estimating the concentration of the dopant in the substrate,
(b) the concentration at 0.1 µm beneath the surface after one hour into the diffusion
process. The substrate is initially free of impurity.
Solution:

First, to find the diffusivity, D with temperature T’ = 1000/(1000+273) = 0.7855.

From the Table, we have constants, a = -19.982 and b = 13.1109.
Thus, D is: 0.005676 µm2/h = 1.5766x10-6 µm2/s, which is in a ”ball park”
agreement with that from Figure 3.12 on P. 95

(a) Since initially N(x,0) = 0, and N(0,t) = Ns = 1011 atoms/cm2, and N(∞,t) = 0,
from Eq. (8.4), we have:
⎡ x ⎤ ⎡ x ⎤ ⎛ 398.21x ⎞
N ( x, t ) = N s erfc ⎢ ⎥ = 1011
erfc ⎢ ⎥ = 1011
erfc ⎜⎜ ⎟⎟
−6
⎣ 2 Dt ⎦ ⎢⎣ 2 1.5766 x10 t ⎥⎦ ⎝ t ⎠
⎡ ⎛ 398.21x ⎞⎤
N ( x, t ) = 1011 ⎢1 − erf ⎜⎜ ⎟⎟⎥
⎣ ⎝ t ⎠⎦
Example 8.2 – cont’d

(b) For x= 0.1 µm at t = 1 h, or 3600 seconds into the diffusion,

the concentration is:

⎡ ⎛ 398.21x0.1 ⎞⎤
N (0.1µm,1hour ) = 1011 ⎢1 − erf ⎜⎜ ⎟⎟⎥ = 1011 [1 − erf (0.6637 )]
⎣ ⎝ 3600 ⎠⎦
= 1011 (1 − 0.6518) = 3.482 x 1010 / cm 3

The numerical value of the error function, erf (0.6637) is obtained from
Figure 3.14 on P. 96
 Oxidation
SiO2 is an important element in MEMS and microsystems. Major application
of SiO2 layers or films are:

(1) To be used as thermal insulation media

(2) To be used as dielectric layers for electrical insulation

● SiO2 can be produced over the surface of silicon substrates either by:
(1) Chemical vapor deposition (CVD), or
(2) Growing SiO2 with dry O2 in the air, or wet steam by the following
two chemical reactions at high temperature:
Si (solid) + O2 (gas) → SiO2 (solid)
Si (solid) + 2H2O (steam) → SiO2 (solid) + 2H2 (gas)

Resistance heater Controlled

air chamber
O2 or H2 O +
carrier gas Fused quartz cassette to vent
Wafers

Fused quartz
Resistance heater
tubular furnace
 Oxidation – Cont’d
Principle of thermal oxidation:
It is a combined continuous physical diffusion and chemical reactions
Oxidizing species: O2 or steam
Oxidizing species: O2 or steam

SiO2 layer

Silicon substrate
(1) At the inception of oxidation (2) Formation of oxide layer by chemical reaction

Oxidizing species: O2 or steam

Growing
SiO2 layer
Instantaneous
SiO2/Si boundary
(3) Growth of oxide layer with diffusion
and chemical reactions
 Oxidation – Cont’d
Rate of thermal oxidations:

Because of the extreme complexity of thermal oxidation, analytical methods

for predicting the rate of this process is impossible. The following quasi-
empirical formulas are used for the thickness of SiO2 layer (x):
B (8.9)
For small time, t: x = (t + τ )
A
For larger time, t: x = B (t + τ ) (8.10)
⎛ d 2 + 2D d ⎞
in which ⎜ o o⎟ (8.11)
⎜ ⎟ N1
τ=⎝ ⎠
ks
2D N o

D = diffusivity of oxide in silicon, e.g. D = 4.4x10-16 cm2/sec at 900oC.

do = initial oxide layer (∼ 200 in dry oxidation, = 0 for wet oxidation)
ks = surface reaction rate constant.
No = concentration of oxygen molecules in the carrier gas.
= 5.2x1016 molecules/cm3 in dry O2 at 1000oC and 1 atm.
= 3000x1016 molecules/cm3 in water vapor at the same temperature and pressure.
N1 = number of oxidizing species in the oxide.
= 2.2x1022 SiO2 molecules/cm3 with dry O2,
= 4.4x1022 SiO2 molecules/cm3 in water vapor.
 Oxidation – Cont’d
Rate of thermal oxidations – Cont’d
Determination of constant A and B in Eqs. (8.9) and (8.10):
For small time, t in Eq. (8.9):

⎛B⎞ (8.12)
log ⎜ ⎟ = aT '+ b
⎝ A⎠
For larger time, t in Eq. (8.10):
(8.13)
ln ( B ) = aT '+ b
where T’ = 1000/T, with temperature T in Kelvin, K. Constant a and b are from the Table:

Constants Coefficient (a) Coefficient (b) Conditions

Linear rate -10.4422 6.96426 Dry O2, Ea = 2 eV, (100) silicon
constant -10.1257 6.93576 Dry O2, Ea = 2 eV, (111) silicon
Eq. (8-12) -9.905525 7.82039 H2O vapor, Ea = 2.05 eV, (110) silicon
-9.92655 7.948585 H2O vapor, Ea = 2.05 eV, (111) silicon

Parabolic rate -14.40273 6.74356 Dry O2, Ea = 1.24 eV, 760 Torr vacuum
constant -10.615 7.1040 H2O vapor, Ea = 0.71 eV, 760 Torr
Eq. (8-13) vacuum
 Example 8.3
Estimate the thickness of the SiO2 layer over the (111) plane of a “clean”
silicon wafer, resulting from both “dry” and “wet” oxidation at 950oC for
one and half hours.
Solution:
Since oxidation take place on a new “clean” silicon wafer, we can use the
condition that ‫ = ح‬0. Thus, we may use:
B
x= t for small time, and x = Bt for larger time.
A
The constants, A and B in the above expressions are selected from Table 8.4
as follows:
a b Conditions

Eq. (8-12) -10.1257 6.9357 Dry O2

Eq. (8-12) -9.9266 7.9486 Wet steam
Eq. (8-13) -14.4027 6.7436 Dry O2
Eq. (8-13) -10.6150 7.1040 Wet steam
Example 8.3 – Cont’d

Since we have T’ = 1000/(950+273) = 0.8177, we obtained the constants

B/A and B from Eqs. (8.12) and (8.13) as:

Dry oxidation Wet oxidation

B/A (µm/hr) 0.04532 0.6786
B (µm/hr) 0.006516 0.2068

The rates of oxidation in two conditions after 1.5 hours are thus obtained
from the two equations available as:

Dry oxidation (µm) Wet oxidation (µm)

B for small time 0.068 1.018
x= t
A
x = Bt for larger time 0.0989 0.5572

We observe that wet oxidation results in much fast rates in oxidation

than dry oxidation.
SiO2 thickness by observations - A quick way to tell!

● Both SiO2 and Si3N4 layers have distinct color from the silicon
substrates in which these layers grow.
● In the case of SiO2 layers, they are essentially transparent but with
distinct light refraction index from that of the silicon substrate.
● Consequently, when the surface is shone by the rays of white light
one can view the surface exhibiting different colors corresponding
to the layer’s thickness.
● The color of the surface of a SiO2 layer is the result of the interference
of the reflected light rays.
● However, the same color may repeat with different layer thickness,
as shown below:

SiO2 layer 0.050 0.075 0.275 0.310 0.50 0.375 0.390

thickness 0.465 0.493
(µm)

Color Tan Brown Red Blue Green to Green Yellow

violet Yellow yellow
green
 Chemical Vapor Deposition

● Chemical vapor deposition (CVD) is the most important process in

microfabrication.

● It is used extensively for producing thin films by depositing many

different kind of foreign materials over the surface of silicon
substrates, or over other thin films that have already been deposited
to the silicon substrate.

● Materials for CVD may include:

(a) Metals: Al, Ag, Au, W, Cu, Pt and Sn.
(b) Organic materials: Al2O3, polysilicon, SiO2, Si3N4, piezoelectric ZnO,
SMA TiNi, etc.

● There are three (3) available CVD processes in microfabrication:

(a) APCVD: (Atmospheric-pressure CVD);
(b) LPCVD (Low-pressure CVD), and
(c) PECVD (Plasma-enhanced CVD).

● CVD usually takes place at elevated temperatures and in vacuum in

high class clean rooms.
 Chemical Vapor Deposition – Cont’d

Working principle of CVD:

● CVD involves the flow of a gas containing diffused reactants (normally

in vapor form) over the hot substrate surface

● The gas that carries the reactants is called “carrier gas”

● The “diffused” reactants are foreign material that needed to be deposited

on the substrate surface

● The carrier gas and the reactant flow over the hot substrate surface,
the energy supplied by the surface temperature provokes chemical
reactions of the reactants that form films during and after the reactions

● The by-products of the chemical reactions are then let to the vent

● Various types of CVD reactors are built to perform the CVD processes
 Chemical Vapor Deposition – Cont’d

Reactors for CVD:

By-products
and gas out
Resistance heater

Substrate Horizontal Reactor

Reactant
and gas in
Susceptor

Substrates

Vertical Reactor
Resistance
heater

Reactant To exhaust
and gas in
 Chemical Vapor Deposition – Cont’d

Chemical reactions in CVD:

● CVD of SiO2 on silicon substrates:
SiH4 + O2 → SiO2 + 2H2 at 400 – 500oC
Carrier gases are: O2 (such as in the above reaction), NO, NO2, CO2 and H2.
The diffused reactant in the reaction is Silane (SiH4) -a common reactant in CVD.

● CVD of Si3N4 on silicon substrates:

3SiH4 + 4NH3 → Si3N4 + 12 H2 at 650 – 750oC

3SiCl4 + 4NH3 → Si3N4 + 12 HCl
3SiH2Cl2 + 4NH3 → Si3N4 + 6HCl + 6H2

● CVD of polysilicon on silicon substrates:

It is essentially a pyrolysis process of Silane at 600 – 650oC.
SiH4 → Si + 2H2
 Chemical Vapor Deposition – Cont’d
Rate of CVD Build-up:
● CVD is the principal technique for building the desired 3-D geometry of
many MEMS and microsystems by means of thin film deposition.
● The rate of the build-up of these thin films obviously is a concern to
process design engineers.
● Quantification of the rate of CVD is extremely complicated. A quasi-
quantitative assessment of such rate of build-up may begin with the
understanding of the physical-chemical principles on which CVD operates.
● Two major factors affect the rate of CVD:
(a) The velocity of carrier gas and the diffused reactant,
as measured by the Reynold’s number (Re) and the
associated boundary layer (δ) thickness at the substrate-gas interface.
(b) The Diffusion flux of the reactant (N).
y
Reactant and gas flow Reactant and gas flow
V(x)
x G Boundary layer

s δ(x)
x
x x
Hot silicon substrate Hot silicon surface
 Chemical Vapor Deposition – Cont’d
Rate of CVD Build-up-Cont’d:
Velocity, V
The Reynolds number:
Reactant and gas flow ρLV (x)
Re = (8.18)
x µ

x where L = length of the substrate

Hot silicon substrate

Dynamic Viscosity (µ) of Carrier Gases Used in CVD Processes

Gas Viscosity (micro poises, µP)
@0oC @490oC @600oC @825oC
Hydrogen, H2 83 167 183 214
Nitrogen, N2 153 337 - 419
Oxygen, O2 189 400 - 501
Argon, Ar 210 448 - 563

1 Poise (P) = 1 dyne-s/cm2 = 0.1 N-s/m2 = 0.1 Kg/m-s

 Chemical Vapor Deposition – Cont’d
Rate of CVD Build-up-Cont’d:
● A boundary layer is created at the interface between the flowing carrier gas and
the hot substrate surface
● The carrier gas (together with the reactant) diffuses through the boundary layer
to reach the hot substrate surface for chemical reactions
● The thickness of the boundary layer thus has significant effect on the rate of CVD
y
Reactant and gas flow
V(x)

G Boundary layer
The thickness of boundary layer:
δ(x)
s x
x
x δ ( x) = (8.17)
Hot silicon surface Re( x )
r
The diffusion flux of reactant (N ) across the boundary layer is:
r D
N = ( N G − N S ) atoms or molecules/m2-s (8.19)
δ
where D = diffusivity of reactant in the carrier gas (cm2/s)
NG = Concentration of reactant at the top of boundary layer (molecules/m3)
Ns = concentration of reactant at the substrate surface (molecules/m3)
 Chemical Vapor Deposition – Cont’d
Rate of CVD Build-up-Cont’d:
Determination of NG and Ns:

Step 1: Use Avogadro’s theory:

(a) The volume occupied by one mole of any gas is 22.4 x10-3 m3 at standard
conditions (20oC or 293 K at 1 atm, or 101.33 KPa), for ALL gases.
(b) A molar density of 44.643 moles/m3, and
(c) Avogadro number = 6.022 x 1023 molecules in one mole of gas.

Step2: Find molar mass of commonly used gases:

Gases Hydrogen, H2 Nitrogen, N2 Oxygen, O2 Argon, Ar

Molar mass (g) 2 28 32 40

Step 3: Use ideal gas law to determine the molar density of the gas at the
operating temperature and pressure.

Step 4: The concentrations, NG and Ns can be determined by multiplication of

the Avogadro’s number in Step 1 and molar density in Step 3.
 Chemical Vapor Deposition – Cont’d
Rate of CVD Build-up-Cont’d:
Example 8.4
A CVD process involves the reactant being diluted at 2% in the carrier oxygen gas
at 490oC. Find the number of molecules in a cubic meter volume of the carrier gas.
Pressure variation in the process is negligible.
Solution:
We may follow the steps outlined above begun with Step 3 for finding the volume
of the carrier gas at 490oC using the ideal gas law:
P1V 1 = T 1
P2 V 2 T 2
Let V1 = 22.4 x 10-3 m3/mol = the volume of the gas at the standard pressure (1 atm)
and temperature (20oC or 293 K).

Since P1 ≈ P2 and T2 = 490 + 273 = 763 K, we have V2 = V1T2/T1= 58.33 x 10-3 m3/mol

The molar density is d2 = 1/V2 = 1/58.33 x 10-3 = 17.1433 mol/m3

The concentration NG defined as the molecules of gas per m3 is:

NG = (Avogadro’s number) x d2 = 6.022 x 1023 x 17.1433 = 103.24 x 1023 molecules/m3

 Chemical Vapor Deposition – Cont’d
Rate of CVD Build-up-Cont’d:

Since CVD is a process with coupled diffusion and chemical reactions,

the amount of carrier gas and reactant mixture that can diffuse through
The boundary layer thus has significant effect on the rate of deposition.
r
Let N = the flux of carrier gas and reactant mixture,
v (8.20)
N = Ks Ns
in which Ns = concentration of carrier gas and reactant at the substrate surface.
Ks = surface reaction rate constant may be determined by:
⎛ Ea ⎞
ks = k ' exp ⎜− ⎟ (8.21)
⎝ kT ⎠
where k’ = constant relating to the reaction and reactant concentration,
Ea = activation energy,
k = Boltzmann constant,
T = absolute temperature of the chemical reaction.
 Chemical Vapor Deposition – Cont’d
Rate of CVD Build-up-Cont’d:

The flux of the carrier gas and the reactant, in Eq. (8-19):
r D
N = (N G − N S)
δ
may be expressed
v in terms of the surface reaction rate, ks with the substitution
of Eq. (8-20) N = K s N s as:
r D NG ks
N= (8.22)
D + δ ks
in which δ is the mean thickness of the boundary layer as expressed in Eq. (8-17).
The rate of the growth of the thin film over the substrate surface, r, in m/sec,
can be estimated by the following expressions:
For δks ≥ D: D NG (8.23a)
r=
γδ
For δks << D r=
NG kS (8.23b)
γ
where γ is the number of atoms or molecules per unit volume of the thin film.
The value of γ may be estimated by a postulation that the thin film is “densely”
packed by atoms or molecules in spherical shapes with the radius according
to the selected materials as listed in the following Table:
Rate of CVD Build-up-Cont’d:
Estimation of the number of atoms or molecules per unit volume of the thin film:
1 1
γ = =
v 4 3 in which v = volume of the thin film with radius a from the Table:
πa
3
Reactant materials Atomic radius (nm) Ionic radius (nm)
Hydrogen 0.046 0.154
Helium 0.046 0.154
Boron 0.097 0.02
Nitrogen 0.071 0.02
Oxygen 0.060 0.132
Aluminum 0.143 0.057
Silicon 0.117 0.198
Phosphorus 0.109 0.039
Argon 0.192

Iron 0.124 0.067

Nickel 0.125 0.078
Copper 0.128 0.072
Gallium 0.135 0.062
Germanium 0.122 0.044
Arsenic 0.125 0.04
 Chemical Vapor Deposition – Cont’d
Example 8.5

A CVD process is used to deposit SiO2 film over a silicon substrate. Oxygen is
used as the carrier gas following the chemical reaction:
SiH4 + O2 → SiO2 + 2H2

Other conditions are identical to that given in Example 8-4 (i.e. the reactant, SiH4
is diluted to 2% in the O2 gas with silicon substrate at 490oC).

The CVD process is carried out in a horizontal reactor as illustrated below:

Resistance heaters

T = 490oC
a ct a nt, SiH 4
Tubular re
g as , O 2 + 0 m m/s ec
r 5
reactor Ca rrie city, V =
diameter: Velo
D = 200 mm
m
150 m
 Chemical Vapor Deposition – Cont’d

Example 8.5 – Cont’d

The supply carrier gas has a flux of N = 1024 molecules/m2-s.
2 -s ec
o l ec ule s/m Substrates
24 m
x: N=1 0
Flu

Determine the following:

(1) The density of the carrier gas (ρ),

(2) The Reynolds number of the gas flow (Re),
(3) The thickness of the boundary layer over the substrate surface (δ),
(4) The diffusivity of the carrier gas and reactant to the silicon substrates (D),
(5) The surface reaction rate (Ks),
(6) The deposition rate (r).
 Chemical Vapor Deposition – Cont’d

Example 8.5 – Cont’d

Solution:
(1) The density of the carrier gas (ρ):
We obtained the molar density of the carrier gas in Example 8.4 to be:
d2 = 17.1433 mol/m3, and the molecular mass of O2 is 32 g, which leads
to the density, ρ to be:
ρ = (32 g/mole)(17.1433 mole/m3) = 548.586 g/m3

(2) The Reynolds number, (Re):

We realize the facts: the diameter of the reactor for the flow, D = 200 mm
= 0.2 m; the velocity of flow, V = 50 mm/s = 0.05 m/s; and the dynamic viscosity
µ = 400 x 10-6 P (Table 8.6) = 0.04 g/m-s. These values lead to the Reynolds
number to be:
ρDV (548.586)(0.2)(0.05)
Re = = = 137.147
µ 0.04
 Chemical Vapor Deposition – Cont’d
Example 8.5 – Cont’d
(3) The thickness of the boundary layer over the substrate surface (δ):
From Eq. (8.17) using the length of the sled, L = 150 mm = 0.15 m, we have:
L 0.15
δ = = = 0.01281 m
Re 137.147
(4) The diffusivity of the carrier gas and reactant to the silicon substrates (D):
This value can be obtained from Eq. (8.19) with a dilution factor,
η = % of dilution (= 2% in the present case):
r
δN
D=
η ( N G − N s)
r
In which N= given carrier gas flux = 1024 molecules/m2-sec
NG = Equivalent density of gas molecules = 103.24 x 1023 molecules/m3
from Example 8-4.
Ns = 0 (an assumed value for complete diffusion within the film)
Thus, we may calculate the diffusivity, D of the carrier gas-reactant mixture as:
(0.01281)(10 24 )
D= 23
= 0.062 m2/s
(0.02)(103.24 x10 )
 Chemical Vapor Deposition – Cont’d
Example 8.5 – Cont’d
(5) The surface reaction rate (Ks):
We may use Eq. r (8.22) to calculate Ks.
DN (0.062)(10 24 )
ks = r = = 0.09884 m/sec
D N G − δN (0.062)(103.24 x10 ) − (0.01281)(10 )
23 24

(6) The deposition rate (r):

We may use either Eq. (8.23a) or (8.23b) to predict the rate of CVD, depending on
the value of δks. In this case, we have: δks = (0.01281)(0.09884) = 0.0013 << D = 0.062.
So, Eq. (8.23b) will be used for the rate prediction, with a dilution factor, η:
NG ks
r =η
γ
The density of SiO2 in the thin film, γ is:
1 1
γ = = = 4.3074 x10 28 molecules/m3
4 4
π (a SiO2 ) 3 (3.14)(0.177 x10 −9 ) 3
3 3
with the radius of O2 atoms, aO2 = 0.06 nm+ radius of Si = 0.117 nm to be
The radius of SiO2 molecules, aSiO2 = 0.177 x10-9 m from Table 8.7.

Hence, we have the rate of CVD to be:

NG ks (103.24 x10 23 )(0.09884) −6
r= = (0.02) = 0. 4738 x10 m/s or 0.47 µm/s
γ 4.3074 x10 28
 Chemical Vapor Deposition – Cont’d

Enhanced CVD
The working principles of CVD process leads to the observations that
the rates of CVD is proportional to the following physical parameters:
● The temperature, T3/2.
● The pressure of the carrier gas, P-1.
● The velocity of gas flow, V-1.
● The distance in the direction of gas flow, x1/2, in which x is shown in
Fig. 8-10 (b):
y
Reactant and gas flow
V(x)

G Boundary layer

s δ(x)
x
x
Hot silicon surface
 Chemical Vapor Deposition – Cont’d
Enhanced CVD – Cont’d
Low-Pressure CVD (LPCVD)
Low-Pressure CVD (LPCVD):
From the previous observation, we may formulate the parameters that
affect the rate of CVD:
(T 1.5 )( x 0.5 )( D )
r∝
( P)(V )(δ )
A few possibilities exist to enhance the rate of CVD:

(1) To raise the process temperature, T would normally increase diffusivity, D.

However, it will harm the substrate.
(2) To decrease the velocity, V may enhance the rate r, but also results in
lower Reynolds number that will increase the boundary layer thickness, δ.
These two effects may cancel out each other. So, it is not a positive option.
(3) This leaves with the last option to decrease the pressure of the gas, P
with expectation to enhance the rate, r of the CVD.

Thus, LPCVD operates in vacuum at about 1 torr (1 mm Hg) has become

a popular CVD process in microfabrication.
The CVD operates at atmospheric pressure is called APCVD.
 Chemical Vapor Deposition – Cont’d
Enhanced CVD – Cont’d
Plasma Enhanced CVD (PECVD)
● Both APCVD and LPCVD operate at elevated temperatures, which often
damage the silicon substrates..

● High substrate surface temperature is required to provide sufficient

energy for diffusion and chemical reactions. RF source

● The operating temperatures may be

avoided if alternative form of energy
supply can be found.
Electrode Substrates
● CVD using plasmas generated from high
energy RF (radio-frequency) sources
Rotating susceptor
is one of such alternative methods.
Resistance
heater Heating elements
● This popular deposition method is called
“Plasma Enhanced CVD” or PECVD. Out to
Reactant vacuum pump
● A typical PECVD reactor is shown: and gas in
 Chemical Vapor Deposition – Cont’d
Summary and Comparison of 3 CVD Processes

CVD Pressure/ Normal Advantages Disadvantages Applications

Process Temperature Deposition
Rates,
(10-10
m/min)
APCVD 100-10 KPa/ 700 for SiO2 Simple, high Poor step Doped and
350-400oC rate, low coverage, undoped
temperature particle oxides
contamination
LPCVD 1-8 Torr/ 50-180 for Excellent High temperature Doped and
550-900oC SiO2 purity and and low undoped
30-80 for uniformity, deposition rates oxides, silicon
Si3N4 large wafer nitride,
100-200 for capacity polysilicon,
polysilicon and tungsten.
PECVD 0.2-5 Torr/ 300-350 for Lower Vulnerable to Low-
300-400oC Si3N4 substrate chemical temperature
temperature; contamination insulators over
fast, good metals, and
adhesion. passivation.
 Sputtering
● Sputtering is a form of Physical Vapor Deposition.

● It is used to deposit thin metal films in the order of 100 A (1A = 10-10 m)
onto the substrate surface.

● Metal films are used as electrical circuit terminals as illustrated below:

Metal wire bond Metal layer Metal layers
Insulation layer Wire bond Insulation
Piezoresistors
Silicon diaphragm surface

Die bond: Pressurized medium

Adhesive
Constraint
Base
Metal Diffused piezoresistor
leads

● Sputtering process is carried out with plasmas under very low pressure
in high vacuum up to 5x10-7 torr and at room temperature.

● No chemical reaction is involved in the deposition process.

 Sputtering – Cont’d
● Metal vapor is created by the plasma generated by the high energy
RF sources, such as the one illustrated below.

Target plates

e- + M →M+ + 2e-
Argon gas +
Metal vapor,M +
+
M-vapor

High velocity
Electrodes
Metal vapor

Substrate

● Inert Argon gas is used as the carrier gas for metal vapor.

● The metal vapor forms the metal films after condensation of the substrate
surface.
 Deposition by Epitaxy

● Both CVD and PVD processes are used to deposit dissimilar materials
on the silicon substrate surfaces.

● Epitaxy deposition process is used to deposit polysilicon films on silicon

substrate surfaces.

● Most polisilicons are doped pure silicon crystals randomly oriented. They are
used to conduct electricity at desired locations on silicon substrates.

● This process is similar to CVD with carrier gas with reactants that release
the same material as the substrates.

● One may deposit GaAs to GaAs substrates using this technique.

● There are four epitaxy deposition methods available.

● We will focus our attention on the popular “Vapor-phase epitaxy” (VPE) process.
 Deposition by Epitaxy – Cont’d

The reactant vapors

Reactant vapors Normal Normal Required Remarks

process deposition energy
temperature rate supply
(oC) (µm/min) (eV)
Silane (SiH4) 1000 0.1 – 0.5 1.6 – 1.7 No pattern shift
Dichlorosilane (SiH2Cl2) 1100 0.1 – 0.8 0.3 – 0.6 Some pattern shift

Trichlorosilane (SiHCl3) 1175 0.2 – 0.8 0.8 – 1.0 Large pattern shift

Silicon tetrachloride 1225 0.2 – 1.0 1.6 – 1.7 Very large pattern shift
(SiCl4)

Typical chemical reaction:

SiH4 → Si (solid) + 2H2 (gas)

 Deposition by Epitaxy – Cont’d
Reactors for epitaxy deposition
● Very similar to those used in CVD, except that many of the carrier gas used is H2.
● For safety reason, N2 gas is used to drive out any O2 gas in the system
before the process begins.
● The two types of reactors are illustrated below:
Resistant heating elements
Vertical reactor
Vent
Gas flow Wafers Substrates

Susceptor Carrier gas +

reactant
Dopant (if any) + H2
Reactant + H2

H2

N2
Horizontal reactor
Susceptor

Exhaust
 Etching

● MEMS and microsystems consist of components of 3-dimensional geometry.

● There are two ways to create 3-dimensional geometry:

● by adding materials at the desired locations of the substrates

using vapor deposition techniques, or
● by removing substrate material at desired locations
using the etiching methods.

● There are two types of etching techniques:

● Wet etching involving the use of strong chemical solvents (etchants), or
● Dry etching using high energy plasmas.

● In either etching processes, masks made of strong-resisting materials are

used to protect the parts of substrate from etching.

● Both etching methods will be presented in detail in the subsequent chapter on

Micromanufacturing.