hsu93912_ch07.

qxd 10/1/2001 12:45 PM Page 235

C h a p t e r 7
Materials for MEMS
and Microsystems

CHAPTER OUTLINE
7.1 Introduction
7.2 Substrates and Wafers
7.3 Active Substrate Materials
7.4 Silicon as a Substrate Material
7.4.1 The Ideal Substrate for MEMS
7.4.2 Single-Crystal Silicon and Wafers
7.4.3 Crystal Structure
7.4.4 The Miller Indices
7.4.5 Mechanical Properties of Silicon
7.5 Silicon Compounds
7.5.1 Silicon Dioxide
7.5.2 Silicon Carbide
7.5.3 Silicon Nitride
7.5.4 Polycrystalline Silicon
7.6 Silicon Piezoresistors
7.7 Gallium Arsenide
7.8 Quartz
7.9 Piezoelectric Crystals
7.10 Polymers
7.10.1 Polymers as Industrial Materials
7.10.2 Polymers for MEMS and Microsystems

235
hsu93912_ch07.qxd 10/1/2001 12:46 PM Page 236

236 MEMS and Microsystems: Design and Manufacture

7.10.3 Conductive Polymers

7.10.4 The Langmuir–Blodgett (LB) Film
7.11 Packaging Materials
Problems

7.1 | INTRODUCTION
In Chapter 1, we maintained that the current technologies used in producing MEMS
and microsystems are inseparable from those of microelectronics. This close rela-
tionship between microelectronics and microsystems fabrication often misleads en-
gineers to a common belief that the two are indeed interchangeable. It is true that
many of the current microsystem fabrication techniques are closely related to those
used in microelectronics. Design of microsystems and their packaging, however, is
significantly different from that for microelectronics. Many microsystems use micro-
electronics materials such as silicon, and gallium arsenide (GaAs) for the sensing or
actuating elements. These materials are chosen mainly because they are dimension-
ally stable and their microfabrication and packaging techniques are well established
in microelectronics. However, there are other materials used for MEMS and micro-
systems products—such as quartz and Pyrex, polymers and plastics, and ceramics—
that are not commonly used in microelectronics. Plastics and polymers are also used
extensively in the case of microsystems produced by the LIGA processes, as will be
described in Chapter 9.

7.2 | SUBSTRATES AND WAFERS

The frequently used term substrate in microelectronics means a flat macroscopic ob-
ject on which microfabrication processes take place [Ruska 1987]. In microsystems,
a substrate serves an additional purpose: it acts as signal transducer besides support-
ing other transducers that convert mechanical actions to electrical outputs or vice
versa. For example, in Chapter 2, we saw pressure sensors that convert the applied
pressure to the deflection of a thin diaphragm that is an integral part of a silicon die
cut from a silicon substrate. The same applies to microactuators, in which the actuat-
ing components, such as the microbeams made of silicon in microaccelerators, are
also called substrates.
In semiconductors, the substrate is a single crystal cut in slices from a larger
piece called a wafer. Wafers can be of silicon or other single crystalline material such
as quartz or gallium arsenide. Substrates in microsystems, however, are somewhat
different. There are two types of substrate materials used in microsystems: (1) active
substrate materials and (2) passive substrate materials, as will be described in detail
in the subsequent sections.
Table 7.1 presents a group of materials that are classified as electric insulators
(or dielectrics), semiconductors, and conductors [Sze 1985]. The same reference
classifies the insulators to have electrical resistivity ! in the range of ! " 108 #-cm;
semiconductors with 10$3 #-cm % ! % 108 #-cm; and conductors with ! % 10 –3
#-cm. We will find that common substrate materials used in MEMS such as silicon
hsu93912_ch07.qxd 10/1/2001 12:46 PM Page 237

CHAPTER 7 Materials for MEMS and Microsystems 237

(Si), germanium (Ge), and gallium arsenide (GaAs) all fall in the category of semi-
conductors. One major reason for using these materials as principal substrate materi-
als in both microelectronics and microsystems is that these materials are at the
borderline between conductors and insulators, so they can be made either a conduc-
tor or an insulator as needs arise. Indeed, the doping techniques that were described
in Chapter 3 can be used to convert the most commonly used semiconducting
material, silicon, to an electrically conducting material by doping it with a foreign
material to form either p- or n-type silicon for conducting electricity. All semi-
conductors are amenable to such doping. Another reason for using semiconductors is
that the fabrication processes, such as etching, and the equipment required for these
processes have already been developed for these materials.
A checklist of factors that help the designer in selecting substrate materials for
microsystems is available in Madou [1997].

Table 7.1 | Typical electrical resistivity of insulators, semiconductors, and conductors

Approximate electrical
Materials resistivity !, "-cm Classification

Silver (Ag) 10!6 Conductors

Copper (Cu) 10!5.8
Aluminum (Al) 10!5.5
Platinum (Pt) 10!5
Germanium (GE) 10!3–101.5 Semiconductors
Silicon (Si) 10!3–104.5
Gallium arsenide (GaAs) 10!3–108
Gallium phosphide (GaP) 10!2–106.5

Oxide 109 Insulators

Glass 1010.5
Nickel (pure) 1013
Diamond 1014
Quartz (fused) 1018

7.3 | ACTIVE SUBSTRATE MATERIALS

Active substrate materials are primarily used for sensors and actuators in a micro-
system (Fig. 1.5) or other MEMS components (Fig. 1.7). Typical active substrate ma-
terials for microsystems include silicon, gallium arsenide, germanium, and quartz.
We realize that all these materials except quartz are classified as semiconductors in
Table 7.1. These substrate materials have basically a cubic crystal lattice with a tetra-
hedral atomic bond. [Sze 1985]. These materials are selected as active substrates pri-
marily for their dimensional stability, which is relatively insensitive to environmental
conditions. Dimensional stability is a critical requirement for sensors and actuators
with high precision.
As indicated in the periodic table (Fig. 3.3), each atom of these semiconductor
materials carries four electrons in the outer orbit. Each atom also shares these four
electrons with its four neighbors. The force of attraction for electrons by both nuclei
hsu93912_ch07.qxd 10/1/2001 12:46 PM Page 238

238 MEMS and Microsystems: Design and Manufacture

holds each pair of shared atoms together. They can be doped with foreign materials
to alter their electric conductivity as described in Section 3.5.

7.4 | SILICON AS A SUBSTRATE MATERIAL

7.4.1 The Ideal Substrate for MEMS
Silicon is the most abundant material on earth. However, it almost always exists in
compounds with other elements. Single-crystal silicon is the most widely used sub-
strate material for MEMS and microsystems. The popularity of silicon for such ap-
plication is primarily for the following reasons:
1. It is mechanically stable and it can be integrated into electronics on the same
substrate. Electronics for signal transduction, such as a p- or n-type
piezoresistor, can be readily integrated with the Si substrate.
2. Silicon is almost an ideal structural material. It has about the same Young’s
modulus as steel (about 2 ! 105 MPa), but is as light as aluminum, with a mass
density of about 2.3 g/cm3. Materials with a high Young’s modulus can better
maintain a linear relationship between applied load and the induced
deformations.
3. It has a melting point at 1400°C, which is about twice as high as that of
aluminum. This high melting point makes silicon dimensionally stable even at
elevated temperature.
4. Its thermal expansion coefficient is about 8 times smaller than that of steel, and
is more than 10 times smaller than that of aluminum.
5. Above all, silicon shows virtually no mechanical hysteresis. It is thus an ideal
candidate material for sensors and actuators. Moreover, silicon wafers are
extremely flat and accept coatings and additional thin-film layers for building
microstructural geometry or conducting electricity.
6. 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.

7.4.2 Single-Crystal Silicon and Wafers

To use silicon as a substrate material, it has to be pure silicon in a single-crystal form.
The Czochralski (CZ) method appears to be the most popular method among several
methods that have been developed for producing pure silicon crystal. The raw sili-
con, in the form of quartzite are melted in a quartz crucible, with carbon (coal, coke,
wood chips, etc). The crucible is placed in a high-temperature furnace as shown in
Figure 7.1 [Ruska 1987]. A “seed” crystal, which is attached at the tip of a puller, is
brought into contact with the molten silicon to form a larger crystal. The puller is
slowly pulled up along with a continuous deposition of silicon melt onto the seed
crystal. As the puller is pulled up, the deposited silicon melt condenses and a large
bologna-shaped boule of single-crystal silicon several feet long is formed. Figure 7.2
shows one of such boules produced by this method. The diameter of the boules
ranges from 100 mm to 300 mm.
hsu93912_ch07.qxd 10/1/2001 12:46 PM Page 239

CHAPTER 7 Materials for MEMS and Microsystems 239

Figure 7.1 | The Czochralski method for growing

single crystals. (Ruska [1987].)

Puller

Quartz
crucible Seed Graphite
crystal susceptor
Silicon
boule
Heating element

Heating element
Silicon melt

The silicon crystal boule produced by the CZ method is then ground to a perfect
circle on its outside surface, then sliced to form thin disks of the desired thickness by
fine diamond saws. These thin disks are then chemically-lap polished to form the fin-
ished wafers.
Principal materials in the silicon melt are silicon oxide and silicon carbide.
These materials react at high temperature to produce pure silicon, along with other
gaseous by-products as shown in the following chemical reaction:
SiC ! SiO2 → Si ! CO !SiO
The gases produced by the above reaction escape to the atmosphere and the liquid Si
is left and solidifies to pure silicon. Circular pure-crystal silicon boules are produced
by this technique in three standard sizes: 100 mm (4 in), 150 mm (6 in) and 200 mm
(8 in) in diameters. A larger size of boule at 300 mm (12 in) in diameter is the latest
addition to the standard wafer sizes. Current industry standard on wafer sizes and
thicknesses are as follows:
100 mm (4 in) diameter " 500 #m thick
150 mm (6 in) diameter " 750 #m thick
200 mm (8 in) diameter " 1 mm thick
300 mm (12 in) diameter " 750 #m thick (tentative)
hsu93912_ch07.qxd 10/1/2001 12:46 PM Page 240

240 MEMS and Microsystems: Design and Manufacture

Figure 7.2 | A 300-mm single-crystal

silicon boule cooling on a
material-handling device.

(Courtesy of MEMC Electronic Materials Inc., St. Peters, Missouri.)

The size difference between a 200-mm and a 300-mm wafer is shown in Figure 7.3.
The latter size wafer has 2.25 times more surface area than the 200 mm wafer and
thus provides significant economic advantage for accommodating many more sub-
strates on a single wafer.
Silicon substrates often are expected to carry electric charges, either in certain
designated parts or in the entire area, as in the resonant frequency pressure sensors
described in Section 4.3.6. Substrates thus often require p or n doping of the wafers.
The doping of p- and n-type impurities, as described in Section 3.5, can be done
either by ion implantation or by diffusion, as will be described in detail in Chapter 8.
Common n-type dopants of silicon are phosphorus, arsenic, and antimony, whereas
boron is the most common p-type dopant for silicon.

7.4.3 Crystal Structure

Silicon has an uneven lattice geometry for its atoms, but it has basically a face-
centered cubic (FCC) unit cell as illustrated in Figure 7.4. A unit cell consists of
atoms situated at fixed locations defined by imaginary lines called a lattice. The di-
mension b of the lattice is called the lattice constant in the figure. In a typical FCC
crystal, atoms are situated at the eight corners of the cubic lattice structure, as well as
at the center of each of the six faces. We have shown those “visible” atoms in black
and the “invisible” or “hidden” ones in gray in Figure 7.4. For silicon crystals, the
hsu93912_ch07.qxd 10/1/2001 12:46 PM Page 241

CHAPTER 7 Materials for MEMS and Microsystems 241

lattice constant b ! 0.543 nm. In an FCC lattice, each atom is bonded to 12 nearest-
neighbor atoms.

Figure 7.3 | Size difference between a 200-mm wafer and a 300-mm wafer.

(Courtesy of MEMC Electronic Materials Inc., St. Peters, Missouri.)

Figure 7.4 | A typical face-center-cubic unit cell.

z
Atoms

Lattice

x
hsu93912_ch07.qxd 10/1/2001 12:46 PM Page 242

242 MEMS and Microsystems: Design and Manufacture

The crystal structure of silicon, however, is more complex than that of regular
FCC structure illustrated in Figure 7.4. It can be considered the result of two inter-
penetrating face-centered cubic crystals, FCC A and FCC B, illustrated in Figure 7.5a
[Angell et al. 1983]. Consequently, the silicon crystal contains an additional four
atoms as shown in Figure 7.5 b.

Figure 7.5 | Structure of a silicon crystal.

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

Figure 7.6 shows a three-dimensional model of the crystal structure of silicon. A

closer look at this structure will reveal that these four additional atoms in the interior
of the FCC (the white balls in Fig. 7.6) form a subcubic cell of the diamond lattice
type as illustrated in Figure 7.7 [Ruska 1987]. A silicon unit cell thus has 18 atoms
with 8 atoms at the corners plus 6 atoms on the faces and another 4 interior atoms.
Many perceive the crystal structure of silicon to be a diamond lattice at a cubic lat-
tice spacing of 0.543 nm [Kwok 1997].

Figure 7.6 | Photograph of a

silicon crystal
structure

The spacing between adjacent atoms in the diamond subcell is 0.235 nm [Brysek
et al. 1991, Sze 1985, Ruska 1987]. Four equally spaced nearest-neighbor atoms that
lie at the corners of a tetrahedron make the diamond lattice, as shown in the inset in
hsu93912_ch07.qxd 10/1/2001 12:46 PM Page 243

CHAPTER 7 Materials for MEMS and Microsystems 243

Figure 7.7 | A subcubic cell of the diamond lattice type in a

silicon crystal.

Figure 7.7. One may also perceive the silicon crystal as being stacked layers of
repeating cubes. Each cube has an atom at each corner and at the center of each face
(FCC structure). These cubes are interlocked with four neighboring cubes in bulk
single-crystal silicon boules, and the wafers are sliced from the boules.
Because of the asymmetrical and nonuniform lattice distance between atoms,
single-crystal silicon exhibits anisotropic thermophysical and mechanical character-
istics that need to be understood for the benefits of handling and manufacturing.
These orientation-dependent material characteristics can be better expressed by using
the Miller indices [Ruska 1987, Sze 1985].

EXAMPLE 7.1
Estimate the number of atoms per cubic centimeter of pure silicon.

Solution
Since the lattice constant b ! 0.543 nm ! 0.543 " 10#9 m, and there are 18
atoms in each cubic cell, the number of atoms in a cubic centimeter, with 1 cm !
0.01m, is
V
! " ! 0.01
"
3
N! v n! " 18 ! 1.12 " 1023 atoms/cm3
0.543 " 10#9

In the above computation, V and v represent respectively the bulk volume of sili-
con in the question and the volume of a single crystal and n is the number of
atoms in a single unit crystal of silicon.

7.4.4 The Miller Indices

Because of the skew distribution of atoms in a silicon crystal, material properties are
by no means uniform in the crystal. It is important to be able to designate the princi-
pal orientations as well as planes in the crystal on which the properties are specified.
A popular method of designating crystal planes and orientations is the Miller indices.
These indices are effectively used to designate planes of materials in cubic crystal
families. We will briefly outline the principle of these indices as follows.
hsu93912_ch07.qxd 10/1/2001 12:46 PM Page 244

244 MEMS and Microsystems: Design and Manufacture

Let us consider a point P(x, y, z) in an arbitrary plane in a space defined by the

cartesian coordinate system x-y-z. The equation that defines the point P is
x y z
a!b!c"1 (7.1)

where a, b, and c are the intercepts formed by the plane with the respective x, y, and
z axes.
Equation (7.1) can be re-written as:
hx ! ky ! mz " 1 (7.2)
It is apparent that h " 1/a, k " 1/b, and m " 1/c in Equation (7.2).
Now, if we let (hkm) designate the plane and #hkm$ designate the direction
that is normal to the plane (hmk), then we may designate the three planes that apply
to a cubic crystal. These three groups of planes are illustrated in Figure 7.8. We will
assume that the cubic structure has unit length with the intercepts a " b " c " 1.0.

Figure 7.8 | Designation of the planes of a cubic crystal.

z
z
z

y
y y

x x x
(a) Normal planes (b) Diagonal plane (c) Inclined plane

We can designate various planes in Figure 7.8 by using Equations (7.1) and (7.2)
as follows:
Top face in Figure 7.8a: (001)
Right face in Figure 7.8a: (010)
Front face in Figure 7.8a: (100)
Diagonal face in Figure 7.8b: (110)
Inclined face in Figure 7.8c: (111)
The orientations of these planes can be represented by #100$, which is perpendic-
ular to plane (100), #111$, which is normal to the plane (111), etc. Figure 7.9 shows
these three planes and orientations for a unit cell in single-crystal silicon.
The silicon atoms in the three principal planes, (100), (110) and (111), are il-
lustrated in Figure 7.10. The atoms shown in open circles are those at the corners of
the cubic, solid circles in gray are the ones at the center of the faces, and the gray cir-
cles in dotted lines are the atoms at the interior of the unit cell.
As shown in Figures 7.6 and 7.10, the lattice distances between adjacent atoms
are shortest for those atoms on the (111) plane. These short lattice distances between
hsu93912_ch07.qxd 10/1/2001 12:46 PM Page 245

CHAPTER 7 Materials for MEMS and Microsystems 245

Figure 7.9 | Silicon crystal structure and planes and

orientations.
(y) <010>

(010)

(100)
(001)
(x) <100>

(z) <001>

Figure 7.10 | Silicon atoms on three designated planes.

0.543 nm 0.768 nm 0.768 nm

0.768 nm
(100) Plane (110) Plane (111) Plane

atoms make the attractive forces between atoms stronger on this plane than those on
the other two planes. Also, this plane contains three of the four atoms that are situ-
ated at the center of the faces of the unit cell (in gray). Thus, the growth of crystal in
this plane is the slowest and the fabrication processes, e.g., etching, as we will learn
in Chapters 8 and 9, will proceed slowest.
Because of the importance of the orientation-dependent machinability of silicon
substrates, wafers that are shipped by suppliers normally indicate in which directions,
i.e., !110", !100", or !111" the cuts have been made, by “flats” as illustrated in
Figure 7.11. The edge of silicon crystal boules can be ground to produce single pri-
mary flats, and in some cases, with additional single secondary flats. The wafers that
are sliced from these crystal boules may thus contain one or two flats as shown in
Figure 7.11. The primary flats are used to indicate the crystal orientation of the wafer
structure, whereas the secondary flats are used to indicate the dopant type of the
wafer. For example, Figure 7.11a indicates a flat that is normal to the !111" orien-
tation with p-type of silicon crystal. The wafer with additional second flat at 45° from
the primary flat in Figure 7.11b indicates an n-type doping in the wafer. Other
arrangements of flats in wafers for various designations can be found in [van Zant
1997, Madou 1997].
hsu93912_ch07.qxd 10/1/2001 12:46 PM Page 246

246 MEMS and Microsystems: Design and Manufacture

Figure 7.11 | Primary and secondary flats in silicon wafers.

Primary Primary
flat flat

45#

Secondary
flat
(a) p-type (b) n-type

7.4.5 Mechanical Properties of Silicon

Silicon is mainly used as an integrated circuit carrier in microelectronics. For micro-
systems, it is the prime candidate material for sensors and actuators, as well as com-
mon substrates for microfluidics. The technology that is used in implementing the IC
on silicon dies is used in a similar way for microsystems. However, silicon, as the
material of components of generally three-dimensional geometry, needs to withstand
often-severe mechanical and thermal loads, in addition to accommodating electrical
instruments such as piezoresistive integrated into it. It is thus important to have
a good understanding of the mechanical characteristics of silicon as a structural
material.
Basically, silicon is an elastic material with no plasticity or creep below 800°C.
It shows virtually no fatigue failure under all conceivable circumstances. These
unique characteristics make it an ideal material for sensing and actuating in mi-
crosystems. However, it is a brittle material. Therefore, undesirable brittle fracture
behavior with weak resistance to impact loads needs to be considered in the design
of such microsystems. Another disadvantage of silicon substrates is that they are
anisotropic. This makes accurate stress analysis of silicon structures tedious, since di-
rectional mechanical property must be included. For example, Table 7.2 indicates the
different Young’s modulus and shear modulus of elasticity of silicon crystals in dif-
ferent orientations [Madou 1997]. For most cases in microsystem design, the bulk
material properties of silicon, silicon compounds, and other active substrate materi-
als presented in Table 7.3 are used.

Table 7.2 | The diverse Young’s moduli and shear moduli of elasticity of silicon crystals

Miller index Young’s modulus Shear modulus

for orientation E, GPa G, GPa
!100" 129.5 79.0
!110" 168.0 61.7
!111" 186.5 57.5
hsu93912_ch07.qxd 10/1/2001 12:46 PM Page 247

CHAPTER 7 Materials for MEMS and Microsystems 247

Table 7.3 | Mechanical and thermophysical properties of MEMS materials*

!y, E, ", c, k, #, TM,

Material 109 N/m2 1011 N/m2 g/cm3 J/g-°C W/cm-°C 10$6/°C °C
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
Legend: !y " yield strength, E " Young’s modulus, # " mass density, c " specific heat, k " thermal conductivity,
$ " coefficient of thermal expansion, TM " melting point.

EXAMPLE 7.2
As indicated in Chapter 5, the thermal diffusivity of a material is a measure of how
fast heat can flow in the material. List the thermal diffusivities of silicon, silicon di-
oxide, aluminum, and copper, and make an observation on the results.

Solution
The thermal diffusivity $ is a function of several properties of the material as
shown in Equation (5.39):
k
$ " #c

where the properties k, #, and c for the four materials are given in Table 7.3. They
are listed in slightly different units in Table 7.4.

Table 7.4 | Thermal diffusivity of selected materials for microsystems

k, ", c, Thermal diffusivity,

Material J/sce-m-°C g/m3 J/g-°C #, m2/s
Si 157 2.3 % 106 0.7 97.52 % 10&6
SiO2 1.4 2.27 % 106 1.0 0.62 % 10&6
Aluminum 236 2.7 % 106 0.94 93 % 10&6
Copper 393 8.9 % 106 0.386 114.4 % 10&6

By substituting the material properties tabulated in the three left columns in

Table 7.4 into Equation (5.39), we can compute the thermal diffusivities of the ma-
terials as indicated in the right column in the same table. It is not surprising to
observe that copper has the highest thermal diffusivity, whereas silicon and alu-
minum have about the same value. Useful information from this exercise is that sil-
icon oxide conducts heat more than 150 times slower than silicon and aluminum.
hsu93912_ch07.qxd 10/1/2001 12:46 PM Page 248

248 MEMS and Microsystems: Design and Manufacture

We may thus conclude that copper films are the best material for fast heat trans-
mission in microsystems, whereas silicon dioxide can be used as an effective
thermal barrier.

7.5 | SILICON COMPOUNDS

Three silicon compounds are often used in microsystems: silicon dioxide, SiO2; sili-
con carbide, SiC; and silicon nitride, Si3N4. We will take a brief look at each of these
compounds as to the roles they play in microsystems.

7.5.1 Silicon Dioxide

There are three principal uses of silicon oxide in microsystems: (1) as a thermal and
electric insulator (see Table 7.1 for the low electric resistivity of oxides), (2) as a
mask in the etching of silicon substrates, and (3) as a sacrificial layer in surface
micromachining, as will be described in Chapter 9. Silicon oxide has much stronger
resistance to most etchants than silicon. Important properties of silicon oxide are
listed in Table 7.5.

Table 7.5 | Properties of silicon dioxide

Properties Values

Density, g/cm3
2.27
Resistivity, !-cm "1016
Relative permittivity 3.9
Melting point, °C !1700
Specific heat, J/g-°C 1.0
Thermal conductivity, W/cm-°C 0.014
Coefficient of thermal expansion, ppm/ ° C 0.5

Source: Ruska [1987].

Silicon dioxide can be produced by heating silicon in an oxidant such as oxygen

with or without steam. Chemical reactions for such processes are:
Si # O2 → SiO2 (7.3)
for “dry” oxidation, and
Si # 2H2O → SiO2 # 2H2 (7.4)
for “wet” oxidation in steam.
Oxidation is effectively a diffusion process, as described in Chapter 3. There-
fore, the rate of oxidation can be controlled by similar techniques used for most other
diffusion processes. Typical diffusivity of silicon dioxide at 900°C in dry oxidation
is 4 $ 10%19 cm2/s for arsenic-doped (n-type) silicon and 3 $ 10%19 cm2/s for boron-
doped (p-type) silicon [Sze 1985]. The process can be accelerated to much faster