JOURNAL OF APPLIED PHYSICS VOLUME 36.

NUMBER 12 DECEMBER 1965
General Relationship for the Thermal Oxidation of Silicon
B. E. DEAL AND A. S. GROVE
Fairchild Semiconductor, A Division of Fairchild Camera and Instrument Corporation,
Palo Alto, California
(Received 10 May 1965; in final form 9 September 1965)
The thermal-oxidation kinetics of silicon are examined in detail. Based on a simple model of oxidati?n
which takes into account the reactions occurring at the two boundaries of the oxide layer as well as dIf-
fusion process, the general relationship x02+:4xo=B(t+r) .is derived. This relationship is be m
cellent agreement with oxidation data obtamed over a ;VIde range of temperature -1300 C), partial
pressure (0.1-1.0 atm) and oxide thickness (300-20000 A) for both oxygen and water.
eters A, B, and r are shown to be related to the physico-chemical.constant.s of the reaction m the
predicted manner. Such detailed analysis also leads t? regardmg the nature of the
transported species as well as space-charge effects on the Imtlal phase of OXidatIOn.
1. INTRODUCTION
O
WING to its great importance in planar silicon-
device technology, the formation of silicon dioxide
layers by thermal oxidation of single-crystal silicon
has been studied very extensively in the past several
years,l-15 Now, with the availability of large of
experimental data, it appears that there IS much
contradiction and many peculiarities in the store of
knowledge of silicon oxidation. For instance, reported
activation energies of rate constants vary between 27
and 100 kcal/mole for oxidation in dry oxygen; pressure
dependence of rate constants has been reported as
linear as well as logarithmic. While most of the data on
silicon oxidation have been evaluated using the parabolic
rate law, certain authors have taken recourse to using
empirical power-law dependence,14 xon= kt, where both n
and k were complex functions of temperature, pressure,
and oxide thickness.
The problems associated with the latter approach can
be illustrated by considering Figs 1 and 2. These figures
1 J. T. Law, J. Phys. Chern. 61, 1200 (1957).
2 M. M. Atalla, Properties of and ComRound Semi-
conductors, edited by H. Gatos (Intersclence PublIshers, Inc.,
New York, 1960), Vol. 5, pp. .
3 J. R. Ligenza and W. G. Spitzer, J. Phys. Chern. Solids 14,
131 (1960).
4 J. R. Ligenza, J. Phys. Che!D' 65, 2011 (1961). .
6 W. G. Spitzer and J. R. Llgenza, J. Phys. Chern. Solids 17,
196 (1961).
6 M. O. Thurston, J. C. C. Tsai, and K. D. Kang, "Diffusion of
Impurities into Silicon Through an Oxide Report 896-
Final, Ohio State University, Research FoundatIOn, U. S. Army
Signal Supply Agency Contract DA-36-039-SC-83874, March 1961.
7 P S Flint "The Rates of Oxidation of Silicon," Paper pre-
sented the'Spring Meeting of The Electrochemical Society,
Abstract No. 94, Los Angeles, 6-10 May 1962.
8 P. J. Jorgensen, J. Chern. Phys. 37, 874 (1962).
9 J. R. Ligenza, J. Electrochem. Soc. 109, 73 (1962).
10 B. E. Deal, J. Electrochem. Soc. 110, 527 (1963).
11 H. Edagawa, Y. Morita, S. Maekawa, and Y. lnuishi, J.
Appl. Phys. (Japan) 2, 765 (1963).
12 N. Karube, K. Yamamoto, and M. Kamiyama, J. App!. Phys.
(Japan) 2, 11 (1963).
13 H. C. Evitts, H. W. Cooper, and S. S. Flaschen, J. Electro-
chern. Soc. 111, 688 (1964).
14 C. R. Fuller and F. J. Strieter, "Silicon Oxidation," Paper
presented at the Spring Meeting of The Electrochemical Society
Abstract No. 74, Toronto, 3-7 May 1964.
16 B. E. Deal and M. Sklar, J. Electrochem. Soc. 112, 430
(1965).
contain a summary of data obtained in these laboratories
which are in good general agreement with the cor-
responding data of Fuller and Strieter
l4
and of Evitts,
Cooper, and FlaschenY (The experimental are
dealt with in detail later.) The plots are logarIthm of
oxide thickness vs the logarithm of oxidation time for
dry and wet oxygen (9S0C H
2
0) at various tempera-
tures. The slope of the lines corresponds to the exponent
n in the above power law. These values are indicated
at the limiting position of some of the curves. In the
case of wet oxygen (Fig. 1), n ranges from 2 for thicker
oxides at 1200°C to 1 for the thinner oxide region of the
920°C data. However, for dry oxygen (Fig. 2), the value
of n at 1200° approaches 2 as the oxide thickness in-
creases above 1.0 ).I.; but at lower temperatures and
oxide thicknesses the value of n decreases only to about
1.S and then appears to increase again. Obviously the
data cannot be represented by a simple power law.
Most of the previous theoretical treatments of the
kinetics of the oxidation of metals emphasize only two
limiting types of oxidation mechanisms.
16
In one, the
(I)
(I)
'"
z
" !:!
:t:
...
'" eO.1

0.1 1.0 10
OXIDATION TIME (hours)
FIG. 1. Oxidation of silicon in wet oxygen (95°C H
2
0).
16 N. Cabrera and N. F. Mott, Rept. Progr. Phys.12, 163 (1948).
3770
Downloaded 26 Jul 2010 to 203.110.243.22. Redistribution subject to AIP license or copyright; see http://jap.aip.org/jap/copyright.jsp
RELATIONSHIP FOR OXIDATION OF Si 3771
oxide thickness is small in comparison to the extent of
possible space-charge regions within the oxide, and the
oxidation kinetics is strongly influenced by the space
charge or by a voltage drop across the oxide film due to
contact potential differences. In the other, the rate of
diffusion of either the oxidizing species or the metal
across the oxide film determines the oxidation kinetics.
This latter condition leads to a parabolic relationship,
Xo a: tt. These two approaches have been emphasized
to such an extent that deviations from parabolic oxida-
tion have been attributed incorrectly to space-charge
effects.
In this paper the oxidation kinetics are examined in
greater detail. It is shown that when the reactions
occurring at the two boundaries of the oxide layer are
taken into account, a general relationship can be ob-
tained with parameters which are related to
the physico-chemical constants of the oxidation system.
This general relationship is shown to be in excellent
agreement with data obtained by various groups of
investigators over a wide range of conditions. It is
shown that the parameters of this relationship follow
the predictions of the model. In addition, further in-
formation is obtained regarding the mechanism of
oxidation: the nature of the transported species, and the
role of space-charge effects.
2. GENERAL OXIDATION EQUATION
Consider silicon covered by an oxide layer of thick-
ness Xo, as indicated in Fig. 3. In accordance with the
experimental evidence for silicon, it is assumed that
oxidation proceeds by the inward movement of a species
of oxidant rather than by the outward movement of
silicon.
2
,3,8,17 The transported species must go through
the following stages:
(1) It is transported from the bulk of the oxidizing
gas to the outer surface where it reacts or is adsorbed.
(2) It is transported across the oxide film towards the
silicon.
(3) It reacts at the silicon surface to form a new
layer of Si0
2

It is assumed in this paper that the oxidation process
is beyond an initial transient period with the conse-
quence that the fluxes of oxidant in each of the above
three steps are, at all times, identical. (The flux of
oxidant is the number of oxidant molecules crossing a
unit surface area in a unit time.) The validity of this
steady-state assumption is shown in the Appendix.
The steady-state fluxes are approximated as follows.
The flux of the oxidant from the gas to the vicinity
of the outer surface is taken to be
F1=h(C*-Co),
(1)
11 W. A. Pliskin and R. P. Gnall, J. Electrochem. Soc. Ill,
872 (1964).



'"
'"
....
z
" !:!
:I:

....

o
O'O'O"".2,-l-L-J-LL.L\-';I.O.---'--'-J.-J....1..LU,IO!.---L-J.--'-'-'-'.J.I#,1OO
OXIDATION TIME (hours)
FIG. 2. Oxidation of silicon in dry oxygen (760 Torr).
where h is a gas-phase transport coefficient, Co is the
concentration of the oxidant at the outer surface of the
oxide at any given time, and C* is the equilibrium con-
centration of the oxidant in the oxide. (This linearized
approximation to the transport rate is analogous to
Newton's law of cooling.) Note that when Co=C*,
F1=O. The equilibrium concentration of the oxidant is
assumed to be related to the partial pressure of the
oxidant in the gas by Henry's law,
C*=Kp. (2)
Henry's law only holds in the absence of dissociation or
association of the oxidant at the outer surface. Thus,
the oxidant is assumed to be molecular O
2
, and H
2
0
in the cases of dry- and wet-oxygen oxidation, re-
spectively. The experimental evidence will be seen to
justify this assumption.
GAS OXIDE SILICON
...---.. ------1
C
C· ---
x
IN STEADY STATE, F,·Fz·F.
FIG. 3. Model for the oxidation of silicon.
Downloaded 26 Jul 2010 to 203.110.243.22. Redistribution subject to AIP license or copyright; see http://jap.aip.org/jap/copyright.jsp
3772 B. E. DEAL AND A. S. GROVE
The flux of the oxidant across the oxide layer is
assumed to be given by Fick's law,
(3)
at any point x within the oxide layer, where Deff is the
effective diffusion coefficient and dC/ dx is the concen-
tration gradient of the oxidizing species in the oxide.
It follows from the assumption of steady-state
oxidation that F2 must be the same at any point x
within the oxide film, or that dFddX =0. Consequently,
the concentration of the oxidant within the oxide layer
is linear as indicated in Fig. 3, and the flux F 2 is given by
F 2= Deff(CO-Ci )/ XO, (4)
where Ci is the concentration of the· oxidant near the
oxide-silicon interface.
If the transported species are ionic, they set up
a space charge within the oxide film. The effective
diffusion coefficient Deft incorporates the effect of such
space charges in enhancing the rate of transport. It was
shown .by Cabrera and Mott,16 as well as by Bardeen,
Brattam, and Shockley,I8 that if the oxide thickness is
large enough in comparison to the thickness of the
space-charge regions, Deff is approximately twice the
actual diffusion coefficient. The extent of the space-
charge region in Si0
2
films is shown later to be of
the order of 200 A, or less, so that this condition is
met in most practical cases.
Finally, the flux corresponding to the oxidation re-
action is expressed by the first-order relation
(5)
Noting t ? ~ t Fl=F2 and F2=Fa due to the steady-
state conditlon, and solving for Ci and Co between
these two equations, one obtains
1
C* l+k/h+kxo/Deff'
and
(6)
Co l+kxo/Deff
C* l+k/h+kxo/Deff'
(7)
Note that in the limit where the diffusivity becomes
very small relative to the rate constants associated
with surfaces, k and h (i.e., Deu/kxo ~ 0), C
i
~ 0
and C o ~ C*. This condition is commonly referred to
as "diffusion controlled."
With C and Co eliminated, the flux is given by
kC*
F=Fl=F2=Fa=------
l+k/h+kxo/Deff
(8)
. If Nl i ~ the number of oxidant molecules incorporated
mto a umt volume of the oxide layer, the rate of growth
18 J. Bardeen, W. H. Brattain, and W. Shockley, J. Chern.
Phys. 14, 714 (1946).
of the oxide layer is described by the differential
equation
dxo F kC*/Nl
dt Nl l+k/h+kxo/Deff
(9)
To arrive at a general initial condition to this equation,
the total oxide thickness Xo is taken to consist of two
parts: an initial layer of oxide Xi that might have been
present on the silicon prior to the oxidation step under
consideration, and the additional thickness grown during
the step under consideration. Thus the initial condition
for this step is:
XO=Xi at t=O.
(10)
Such division of the oxide layer is not only important
because it permits consideration of multiple oxidations,
but also because Xi can be regarded as the thickness of
the layer grown before the approximations of the above
treatment become valid. Thus, an initial oxide layer
may be formed by mechanisms involving fields and
space charges within the oxide layer.
16
The solution of the differential equation, Eq. (9),
subject to the initial condition, Eq. (10), obtained by a
straightforward integration, is
X02+AxO=Bt+x?+Axi,
(11)
which can be rewritten in the form
xo
2
+Axo=B(t+T),
(12)
where
A = 2Deff (l/k+ l/h),
(12a)
B=2DeffC*/Nl, (12b)
and
T= (X?+AXi)/B.
(12c)
The quantity T corresponds to a shift in the time co-
ordinate which corrects for the presence of the initial
oxide layer Xi. The form of this mixed parabolic
relationship was first proposed by Evans.
19
,20
Solving the quadratic equation (12) yields the form
Xo [ t+r J'
A/2 = 1+ A2/4B -1,
(13)
which gives the oxide thickness as a function of time.
It is interesting to examine two limiting forms of
Eq. (13). At relatively large times, i.e., t»A2/4B, and
also t»T,
Xo ( t )!
A/2'.::::. A2/4B
or
(14)
Thus, the general relation, Eq. (13), reduces to the
well-known parabolic oxidation law for relatively long
19 U. R. Evans, Trans. Electrochem. Soc. 46, 247 (1924).
20 U. R. Evans, The Corrosion and Oxidation of M etaJs (Edward
Arnold and Company, London, 1960), pp. 819-859.
Downloaded 26 Jul 2010 to 203.110.243.22. Redistribution subject to AIP license or copyright; see http://jap.aip.org/jap/copyright.jsp
RELATIONSHIP FOR OXIDATION OF Si
3773
times. The coefficient B, as given by Eq. (12b), is
identical with that derived by Cabrera and MotU
6
It is referred to as the parabolic rate constant.
At the other extreme, for relatively small oxidation
times, i.e., t<<.A
2
/4B,
or

A/2-2 A2/4B
B
xO::=::-(t+T).
A
(15)
Thus, the general relationship reduces to a linear law.
The coefficient
B kh (C*)
A = k+h Nt
is referred to as the linear rate constant. The general
relationship and its two limiting forms are shown in
Fig. 12, together with experimental points which are
discussed later. It should be noted at this point, how-
ever, that for most of the range the parabolic limit does
not provide a good approximation. For this reason,
simply dividing xo
2
by t does not necessarily yield the
true parabolic rate constant B, but rather it gives
some quantity which mayor may not approximate B
depending on the oxidation conditions and the time.
Thus, it appears that many of the contradictions re-
ferred to in the Introduction may well have been due to
forcing the parabolic law in regions where it could not
provide a meaningful fit.
3. EXPERIMENTAL METHODS
The oxidation apparatus used in this program was
similar to that used in previous investigations.I°,15
Either purified dry oxygen (watercontentlessthanSppm)
or wet oxygen (oxygen bubbled through water at
95°C) could be supplied to the quartz oxidation tube.
The partial pressure of water in the wet-oxygen ambient
was estimated to be 640 Torr. Silicon slices were placed
flat on a quartz boat during oxidation. Oxidation tem-
perature was controlled to ± 1°C over the range 700°
to 1200°C.
Silicon used was in the form of circular slices 22 mm
in diameter and mechanically polished to 200 Il thick-
ness. It was (111) surface oriented, Czochralski pulled,
with an initial dislocation count of less than 100 cm-
2

Silicon resistivity was 1.3 Q·cm, p-type boron doped
(C
B
= 1.4SXlO
16
cm-
3
). Cleaning procedures prior to
oxidation included both organic and inorganic rinses
using ultrasonic agitation to insure noncontaminated
surfaces.
Oxide thicknesses were determined using multiple-
beam interferometric techniques. Accuracy of the
measurement for oxides thicker than 0.20 Il was
±40 A. For thinner oxides, the accuracy was ±2S A.
-0.60 1.0 2.0 3.0 4.0 5.0 6.0
VXo (hr IILI
FIG. 4. Evaluation of rate constants for oxidation of silicon in
wet oxygen (95°C H
2
0). Slopes of lines correspond to B in Eq.
(12), intercepts at t/xo=O correspond to -A in Eq. (12).
4. EXPERIMENTAL RESULTS
A. Wet Oxygen
To check how well the general oxidation equation,
Eq. (12), fits the data, the following procedure was
employed.
20
First, a detailed plot of ox.ide thickness vs
oxidation time was made to determine what initial
oxide thickness Xi the data extrapolated to at t=O.
In the case of wet-oxygen oxidation Xi was found to be
zero at all temperatures. Then the oxide thickness Xo
was plotted vs the quantity t/xo. As can be seen from
Eq. (12), if that relationship holds and T=O, such a
plot should yield straight lines with intercept equal to
- A, and slope equal to B.
Wet oxygen data from Fig. 1 are plotted in this
manner in Fig. 4. Straight lines are indeed obtained for
all four temperatures with the absolute values of A
increasing with decreasing temperatures. At the same
time, values of the slope B decrease with decreasing
temperature. Values of the constants A and B deter-
mined from this plot for the four temperatures are
tabulated in Table I. Analysis of the data has indicated
that the precision of the constant B is ±2% over the
entire range of measurements. For larger values of A
(low temperatures) the precision is ±2%. In the case
of high temperatures, the precision of A is ± 12%. It
should be noted that this larger deviation at higher
temperatures is not especially critical, since A itself
TABLE 1. Rate constants for oxidation of silicon in
wet oxygen (95°C H
2
0). xo2+Axo=B(t+r).
Oxidation
temperature (OC) A (I')
1200° 0.05 0.720 14.40
1100° 0.11 0.510 4.64
1000° 0.226 0.287 1.27
920° 0.50 0.203 0.406
r(h)
0
0
0
0
Downloaded 26 Jul 2010 to 203.110.243.22. Redistribution subject to AIP license or copyright; see http://jap.aip.org/jap/copyright.jsp
3774
B. E. DEAL AND A. S. GROVE
WET-OXIDATltlN TIME (hours)
__
.18
o
:; .IS
! .14
.12
.10
.08
.OS

§ .02

DRY-OXIDATION TIME (hours)
FIG. 5. Comparison of initial oxide growth at 920°C for oxidation
in wet-(95°C H
2
0) and dry-oxygen ambients.
is very small and hence its influence on the oxidation
rate becomes less important.
Special experiments demonstrated that the values of
A and B are essentially the same when the carrier-gas
oxygen is replaced by argon. Thus, the effect of the
oxygen in the wet-oxygen atmosphere can be neglected
under the conditions of these experiments, with the
implication that the oxidizing species is water.
B. Dry Oxygen
In contrast to the case with wet oxygen, a detailed
plot of oxide thickness vs time for dry oxygen did no[
extrapolate to zero initial thickness at any of the ex-
perimental temperatures. The difference between the
wet- and dry-oxygen cases is illustrated in Fig. 5. Here
both dry- and wet-oxygen data at 920
D
C are plotted in
the same graph with the time scales adjusted to provide
similar curve' shapes. While the wet-oxygen data
extrapolate to zero, the plot for dry oxygen appears to
pass through the thickness axis at Xo= 250 A. This
was found to be the case for all temperatures from
700° to 1200°C, with the intercept being in all cases
230±30 A. A re-examination of Fig. 2 shows that the
differences in the variation of the exponent n in the
empirical power-law formula, xon=kt, between wet-
and dry-oxygen oxidation described in the Introduction
are in fact due to this phenomenon. Consideration of the
corresponding data of Fuller and Strieter
l4
and of Evitts

! 0.05

.§ 0.04

w
0.03
<.>
...... ,;
0.02
o
x
00.01
20 40 60 80 100 120
OXIDATION TIME (hotnI
FIG. 6. Oxidation of silicon in dry oxygen at 700°C.
et al.!3 lead to the same conclusion: an offset of about
200 A in the thickness data when extrapolated to t=O.
This finding could mean one of two things. Either an
oxide film of this order of thickness was originally
present before the oxidations were started, or a different
mechanism of oxidation prevailed at oxide thicknesses
less than 300 A. The latter was assumed to be the case,
since no initial oxide could be detected by experimental
techniques. In addition, the wet-oxygen data indicated
no initial oxide and the silicon preparation was identical
for the two types of oxidation. Further confirmation was
provided by precise and detailed measurements at
700°C. A plot of oxide thickness vs time at 700
D
C is
shown in Fig. 6. It can be observed that an initial rapid
rate of oxidation occurs, followed by a strictly linear
process which extrapolates back to Xo= 210 A at t= O.
Thus, for oxidation in dry oxygen, the initial condi-
tion Xi= 230 A was adopted. The corresponding values
of T= (X,2+AXi)IB were estimated graphically by
plotting Xo vs t and extrapolating the curve back through
-0.4
-o·so 20 40 60 80 100 120
(t+T)/x. (hrlll)
FIG. 7. Evaluation of rate constants for oxidation of silicon in
dry oxygen. Slopes of lines correspond to Bin Eq. (12), intercepts
at (t+T)/XO=O correspond to -A in Eq. (12).
Xi= 230 A to the time axis. The value of T varies with
temperature even though Xi is a constant since A and B
are a function of temperature.
In Fig. 7, Xo is plotted vs (t+ T) I Xo. (Values of
(t+T)lxo for goo°C are of the order of 400 hi}.'. For
this reason, only the line extrapolated from these data
is shown in Fig. 7.) The resulting rate constants for
dry oxygen are given in Table II. In the case of 700°,
the relationship was found to be completely linear and
therefore only the ratio BI A could be determined.
The precision of these constants is of the same order as
discussed for the wet-oxidation results. The significance
of the finite Xi is discussed later.
C. Pressure Dependence of Rate Constants
According to Eq. (12), the parabolic rate constant B
should be proportional to C*, which in turn should be
Downloaded 26 Jul 2010 to 203.110.243.22. Redistribution subject to AIP license or copyright; see http://jap.aip.org/jap/copyright.jsp
RELATIONSHIP FOR OXIDATION OF Si 3775
TABLE II. Rate constants for oxidation of silicon in
dry oxygen. xo2+Axo=B(t+,.).
Oxidation
temperature (·C)
A ("') B(",'/h) B/ A (",/h)
1200° 0.040 0.045 1.12
1100· 0.090 0.027 0.30
1000· 0.165 0.0117 0.071
920· 0.235 0.0049 0.0208
800· 0.370 0.0011 0.0030
700· 0.00026
T(h)
0.027
0.076
0.37
1.40
9.0
81.0
proportional to the partial pressure of the oxidizing
species in the gas if Henry's law is obeyed. Conversely,
the coefficient A should be independent of the partial
pressure. Data presented by Flint7 on oxidation of
silicon with various partial pressures of oxygen or
water vapor in argon were examined. It was found
through plots of Xo vs (t+r)/xo that A was a constant
over a range of partial pressures, while the value of B
was directly proportional to the partial pressure, at all
experimental temperatures. This dependence is illus-
trated in Fig. 8. Here the values of B, normalized to the
760-Torr value at each temperature are shown as a
function of the partial pressure of the oxidizing species-
O
2
or H
2
0, respectively-in the gaseous ambient. (The
760-Torr value was determined by correcting the wet-02
data of the present investigation from 640 to 760 Torr
using the linear relationship. Values obtained by this
extrapolation closely agreed with Flint's7 data based on
oxidation in atmospheric steam.) A very good linear
dependence is observed for temperatures ranging be-
tween 1000° and 1200°C.
D. Temperature Dependence of Rate Constants
In Fig. 9, the logarithm of the parabolic rate constant
B is plotted against the reciprocal of the absolute tem-
perature for both dry- and wet-oxygen oxidation. Also
shown are the wet-oxygen values correlated to 760-
Torr water vapor pressure using the linear pressure
dependence. As is evident, good straight lines are ob-
tained in all cases.
According to Eq. (12b), the temperature dependence
of B should be substantially the same as that of Deft.
For dry oxygen, the activation energy of B is found to
be 28.5 kcal/mole from Fig. 9. This can be compared to
the value of 27.0 kcal/mole reported by Norton
21
for
the diffusivity of oxygen through fused silica. (The
structure of the Si0
2
formed by thermal oxidation of
silicon corresponds to that of amorphous, fused silica.)
The activation energy for wet-oxygen oxidation is
16.3 kcal/mole, in reasonably good agreement with the
18.3-kcal/mole value found by Moulson and Roberts
22
for the diffusivitv of water in fused silica. These
diffusivity values shown in Fig. 10 for reference.
21 F. J. Norton, Nature 171, 701 (1961).
22 A. J. Moulson and J. P. Roberts, Trans. Faraday Soc. 57,
1208 (1961).
I. I ,------,-10'P0,,---,,2,;<00,,-,,3<>p<.-=;""--''''¥'''--'''fL--'-¥''-''¥''1
1.0
0.9
0.8
0.7

9
B(p,T)
B(760,Tl 0.6

0.5
0.4
0.3
0.2
0.1
o 1200'C, H
2
0
t; II OO'C, H
2
0
o 1000'C, H
2
0
• 1200'C, O
2
00 0.1 0.2 0.3 04 0.5 06 0.7 0.8 09 1.0 1.1
p (aIm)
FrG. 8. The effect of partial pressure of the oxidant on the
parabolic rate constant. Rate constants are normalized to the
760-Torr value at the same temperature. (Flint's7 data for wet and
dry oxygen, 1000· to 1200·C).
The temperature dependence of the linear rate con-
stant B/ A is shown in Fig. 11. Exponential temperature
dependence is obtained again for both the dry- and
wet-oxygen cases. The activation energies 46.0 and
45.3 kcaljmole for dry- and wet-oxygen oxidation,
respectively, are almost identical, indicating a similar
surface-control mechanism for the two oxidants.
E. Equilibrium Solubility Concentrations
The equilibrium concentration C* of the oxidants
oxygen and water in Si0
2
can be calculated using
Eq. (12b). Parabolic rate constants are taken from
Fig. 9, diffusivities from Fig. 10, Nl for oxygen is
2.25 X 10
22
cm-a, while the corresponding value for
<.>
.

0
b b
. .
b
0 0
0
0 0
0 0

0 on
1.0
11> (I) CD on
'ti:::
"
DRY 0. (760 Torr)
E.-28.5 keal/male

1000
"flIi(j'
FrG. 9. The effect of temperature on the parabolic rate constant Bl
Downloaded 26 Jul 2010 to 203.110.243.22. Redistribution subject to AIP license or copyright; see http://jap.aip.org/jap/copyright.jsp
3776
B. E. DEAL AND A. S. GROVE
water is twice this number. The resulting values of C*
at l000°C are listed in Table III.
Also shown in Table III are the equilibrium solu-
bilities of these two species in fused silica as determined
by Norton
21
through permeation experiments and by
Moulson and Roberts
22
through infrared-absorption
experiments, respectively. The agreement between
these values and the C* results obtained from the
oxidation data is indeed excellent. Since the tempera-
ture dependence of B and of Deff were found to be
substantially the same, similar values of C* would be
obtained at any other temperature within the experi-
mental range. Such a conclusion is substantiated by
Moulson and Roberts, who found an essentially constant
solubility of water in fused silica between 900° and
1200°C.
F. Surface-Reaction Rate Constants
The linear rate constants determined in Sec. 4(A)
and (B),
B kh (C*)
A = k+h Nl '
contain the effects of the phenomena taking place both
at the gas-oxide interface, through h, and at the oxide-
silicon interface, through k. If these two constants are
very different in magnitude, the quantity kh/(k+h)
will approximately equal k or h, whichever is smaller.
This quantity can be calculated from the experi-
mentally measured linear rate constants B/ A) the
known values of N 1, and the values of the equilibrium
solubility concentrations C*, as determined in the
previous section. Thus, the value of kh/(k+h) is
calculated to be 3.6X 10
4
,(L/h for dry O
2
and 1.8X 10
3
,(L/h for wet 02, at 1000°C. If it is now assumed that h
is determined solely by a gas-phase transport process, its
value can be estimated on the basis of standard bound-
ary-layer considerations
23
to be about 10
8
,(L/h for the
present flow conditions (Reynolds number"'" 25). Since
this is several orders of magnitude larger than the above
values of kh/(k+h), it follows that
kh {3.6XI04 ,(L/h for dry-02
--=k= oxidation
k+h 1.8X 103 ,(L/h for wet-02
at l000°C. Thus, the activation energies reported in
Sec. 4D, 46.0 and 45.3 kcal/mole, respectively, reflect
the temperature dependence of the interfacial reaction-
rate constant k. These values may be compared with
42.2 kcal/mole, the energy required to break an Si-Si
bond (as given by Pauling,24 p. 85).
23 H. Schlichting, Boundary Layer Theory (MCGraw-Hill
Book Company, Inc., New York, 1960), 4th ed., Chap. 14; and
R. B. Bird, W. E. Stewart, and E. N. Lightfoot, Transport
Phenomena (John Wiley & Sons, Inc., New York, 1960), Para-
graph 21.2.
24 L. Pauling, The Nature oj the Chemical Bond (Cornell Univer-
sity Press, Ithaca, New York, 1960), 3rd ed.
~
.
o
g
.
~
'\ EA :27 kcol/mole
'\
'\
H
2
0 '\
"\
E
A
a
18.3 kcal/mole
FIG. 10. The diffusivity of oxygen (Norton
21
) and of water
(Moulson and Roberts
22
) in fused silica as a function of
temperature.
The relatively large value of the gas-phase transport
coefficient h is confirmed by two other experiments. In
one, the carrier-gas flow rate was varied SO-fold without
any significant effect on the oxidation rate. In the other,
it has been noted
lO
that the backside of a slice lying flat
on the boat was oxidized to the same extent as its
topside. Both of these observations point to the
relatively small importance of the gas-phase transport
process in controlling the over-all oxidation rate.
5. DISCUSSION
In summary a simple model of the oxidation process
has been developed, and then compared to experimental
results obtained over a wide range of variation of
conditions: ambient pressure and temperature. It was
found that the general relationship, Eq. (12), is obeyed
by silicon oxidation throughout this range. Furthermore,
the coefficients of this relationship were shown to depend
in the predicted manner on pressure and temperature.
Thus, the parabolic rate constant B is proportional to
the partial pressure of the oxidant in the gas (0
2
and
H
2
0, respectively) while the coefficient A is independent
Species
O2
H
2
0
TABLE III. Solubility of oxidizing species
in Si0
2
at lO00°C.
Solubility (cm-
3
)
This work (C*) Other methods
5.2X10
10
3.0X10
19
5.5XI0
10
(permeation)21
3.4XIQl9
(infrared)"
Downloaded 26 Jul 2010 to 203.110.243.22. Redistribution subject to AIP license or copyright; see http://jap.aip.org/jap/copyright.jsp
RELATIONSHIP FOR OXIDATION OF Si 3777
of the pressure. The temperature dependence of the
parabolic rate constant is exponential, with the activa-
tion energies corresponding to the available activation
energies for the diffusivities of oxygen and of water
through fused silica. Finally, the equilibrium solubility
concentrations C* determined from the rate constants
are in good agreement with the solubility of oxygen
and water in fused silica as determined by other
methods. This general agreement serves as an indication
of the validity of the assumptions made in the deriva-
tion of Eq. (12).
The general relationship for the oxidation of silicon
is illustrated in Fig. 12. Large numbers of oxidation
data, including results shown in Figs. 1 and 2, earlier
measurements obtained at these laboratories,
7
and the
work of two other groups of investigators,13.14 were
reduced using the parabolic and linear rate constants
given in Figs. 9 and 11 and the pressure dependence
shown in Fig. 8. Then, these data were plotted following
the rearranged form of the general oxidation relation-
ship, Eq. (13). Included as a solid line (almost masked
out by the experimental data) is a calculated plot
corresponding to the general relationship X02+ Axo
=B(t+r). The excellent over-all agreement is quite
evident and speaks for the validity and usefulness of
the general relationship.
Figure 12 also shows the two asymptotic forms of the
general relationship: the parabolic law valid at "large"
times and the linear law valid at "small" times. It is
important to note that the criterion of "small" or
"large" times depends on the oxidation conditions, as
reflected in the Parameters A and B. Thus, this criterion
is that the time should be smaller (or larger) than the


FIG. 11. The effect of temperature on the linear rate constant B / A.
x.
A/2
• imiK
,REF 7
, REF 13
t REF 14
10'
FIG. 12. General relationship for thermal oxidation of silicon.
The solid line represents the general relationship, the dotted lines
its two limiting forms. Experimental data were reduced using
values of A and B determined from Figs. 8, 9, and 11. The values
of r correspond to Xj=O and 200 A for wet and dry oxygen,
respectively.
characteristic time of the oxidation process,
A2 DoffNl
4B 2C*(1/k+1/h)2'
or, alternatively, that the oxide thickness should be
smaller (or larger) than the characteristic distance of
the oxidation process,
A Doff
2 1/k+1/h
It should be noted that the general relationship, to-
gether with the rate constants from Figs. 9 and 11,
corrected to the proper partial pressure according to
Fig. 8, enables the construction of the correct oxidation-
rate equation for any set of experimental conditions
with confidence.
It is worthwhile to point out certain interesting
aspects implicit in the results of this work. While the
diffusivity of water in fused silica is lower than the
diffusivity of oxygen (Fig. 10), the parabolic rate
constant B is considerably larger for oxidation in wet
oxygen than in dry (Fig. 9). This seeming paradox is
immediately resolved when it is recalled that the flux of
oxidant, and therefore the rate constant B is propor-
tional to the equilibrium concentration C*. This con-
centration is three orders of magnitude larger for water
than for oxygen, thus accounting for the higher oxida-
tion rate.
The existence of a rapid initial-oxidation stage with
dry oxygen as well as the absence of such a phase with
wet-oxygen oxidation are of interest. If the transported
species is ionic, as has been shown to be the case for
dry-oxygen oxidation by Jorgensen,s such a phase is
Downloaded 26 Jul 2010 to 203.110.243.22. Redistribution subject to AIP license or copyright; see http://jap.aip.org/jap/copyright.jsp
3778 B. E. DEAL AND A. S. GROVE
to be expected
16
until such time that the oxide thickness
becQmes large in comparison to the extent of the space-
charge region within the oxide. This latter quantity is
of the order of the extrinsic Debye length of the oxide/
6
LD,o= [(kT / q) (Ko€o/2qC*)J.
Using the value of C*=5X10
16
cm-
3
as determined in
this work, and the dielectric constant Ko=4, the
oxide-Debye length is estimated to be of the order of
150 A for dry O
2
oxidation, at 1000°C. This indeed
corresponds well to the oxide thickness beyond which
the rapid initial-growth mechanism stops being effective,
In contrast, the extrinsic Debye length for oxidation in
wet O
2
, at 1000°C, using C*= 3 X 10
19
cm-
3
, is estimated
to be only around 6 A. Thus, even if an initial rapid
oxidation phase existed in the case of wet O
2
, it could
not be observed using the present techniques,
It is worthwhile to emphasize again that the good
agreement between the pressure dependence of the
coefficient B with the linear dependence expected on the
basis of the assumption of Henry's law implies the
absence of any dissociation effects at the gas-oxide
interface. This, in turn, would imply that, for oxidation
in wet oxygen, the diffusing species is undissociated
water while for oxidation in dry oxygen it is molecular
oxygen. The latter finding is in agreement with a similar
conclusion reached by N orton
21
from the pressure
dependence of permeation rate of oxygen through fused
silica. Jorgensen
8
showed through experiments relating
to the effect of electric fields on dry-oxygen oxidation
that the diffusing species is negatively charged. These
two observations, taken together, point to the conclus-
ion that the diffusing species in the case of dry-oxygen
oxidation is the superoxide ion O
2
-, (For the properties
of this ion, see Pauling,24 p. 351.)
The field created by the space charge of the negatively
charged oxygen ions should be directed toward the gas-
oxide interface. Recent studies
25
of the N a+ ion distri-
bution in thermally grown oxides show that most of the
Na+ ions are concentrated near the outer surface after
oxidation, in agreement with the above conclusion.
ACKNOWLEDGMENTS
The valuable assistance of Mrs. D. S. Sklar in the
experimental portion of this work, as well as helpful,
discussions with l\;I:. Dumesnil and E. H. Snow, are
very much appreciated. The authors also wish to
thank G. E. Moore for his valuable comments con-
cerning this manuscript.
APPENDIX
A simple criterion regarding the validity of the
steady-state condition can be derived on the basis of
an order-of-magnitude consideration, as follows. If the
concentration distribution shown in Fig. 3 were dis-
turbed, the return to equilibrium would take a period
of time of the order of
ttransient
total amount of oxidant needed to return to the steady-state distribution
flux of oxidant
Approximating the numerator by the total amount of
oxidant present within the oxide layer in excess of a
uniform concentration, i.e., !(Co-C;)xo, and the
denominator by the flux F2 as given by Eq. (4), the
relaxation time is found to be of the order of
in agreement with the ,result obtained by Fromhold
26
through a more detailed treatment.
For simplicity, taking X02= Bt,
ttransient / t "" C* / N 1 < 10-
3
in the parabolic region of oxidation-certainly a
negligible time period.
25 See abstracts of "Recent News" papers describing work
of E. Yon, A. B. Kuper, and W. H. Ko, and of H. G. Carlson,
C. R. Fuller, and J. Osborne, in J. Electrochern. Soc. 112, 259C
(1965).
26 A. T. Frornhold, Jr. J. Phys. Chern. Solids 24, 1081 (1963).
Downloaded 26 Jul 2010 to 203.110.243.22. Redistribution subject to AIP license or copyright; see http://jap.aip.org/jap/copyright.jsp

the rate of diffusion of either the oxidizing species or the metal across the oxide film determines the oxidation kinetics.-l-L-J-LL. In the other.~I. This general relationship is shown to be in excellent agreement with data obtained by various groups of investigators over a wide range of conditions.) Note that when Co=C*. Redistribution subject to AIP license or copyright.aip. a general relationship can be obtained with parameters which are e~:plicitly related to the physico-chemical constants of the oxidation system. It is shown that the parameters of this relationship follow the predictions of the model. It is shown that when the reactions occurring at the two boundaries of the oxide layer are taken into account.. where h is a gas-phase transport coefficient. Xo a: tt. Oxidation of silicon in dry oxygen (760 Torr). Gnall.2.------1 C C· --- x F1=h(C*-C o). These two approaches have been emphasized to such an extent that deviations from parabolic oxidation have been attributed incorrectly to space-charge effects. The steady-state fluxes are approximated as follows. 2..1OO OXIDATION TIME (hours) FIG.IO!. C*=Kp.§ . . 872 (1964). it is assumed that oxidation proceeds by the inward movement of a species of oxidant rather than by the outward movement of silicon. 11 (1) IN STEADY STATE. Thus. '" !:! :I: " .. In addition.110. A.---L-J. (The flux of oxidant is the number of oxidant molecules crossing a unit surface area in a unit time. The experimental evidence will be seen to justify this assumption.O. GAS OXIDE SILICON . and C* is the equilibrium concentration of the oxidant in the oxide. (2) It is transported across the oxide film towards the silicon.I#. ~ ~O. the oxidant is assumed to be molecular O2.---'--'-J. Downloaded 26 Jul 2010 to 203.. (3) It reacts at the silicon surface to form a new layer of Si02 • It is assumed in this paper that the oxidation process is beyond an initial transient period with the consequence that the fluxes of oxidant in each of the above three steps are. The equilibrium concentration of the oxidant is assumed to be related to the partial pressure of the oxidant in the gas by Henry's law. Model for the oxidation of silicon.org/jap/copyright. FIG. . Electrochem. The flux of the oxidant from the gas to the vicinity of the outer surface is taken to be ~ . GENERAL OXIDATION EQUATION Consider silicon covered by an oxide layer of thickness Xo..243..-.22. (2) Henry's law only holds in the absence of dissociation or association of the oxidant at the outer surface..-J. and the oxidation kinetics is strongly influenced by the space charge or by a voltage drop across the oxide film due to contact potential differences..3. This latter condition leads to a parabolic relationship...L\-'.. Ill.and wet-oxygen oxidation. as indicated in Fig. In accordance with the experimental evidence for silicon. W. and H 20 in the cases of dry.O z '" .·Fz·F. respectively.) The validity of this steady-state assumption is shown in the Appendix. at all times. F.I o O'O'O"". In this paper the oxidation kinetics are examined in greater detail.I..1.17 The transported species must go through the following stages: (1) It is transported from the bulk of the oxidizing gas to the outer surface where it reacts or is adsorbed.RELATIONSHIP FOR OXIDATION OF Si 3771 oxide thickness is small in comparison to the extent of possible space-charge regions within the oxide.8. 3.J. F1=O. Soc. further information is obtained regarding the mechanism of oxidation: the nature of the transported species. see http://jap. 2 . (This linearized approximation to the transport rate is analogous to Newton's law of cooling. and the role of space-charge effects.--'-'-'-'. identical.jsp . Co is the concentration of the oxidant at the outer surface of the oxide at any given time. 2.LU. Pliskin and R. P. . J. 3.

k and h (i. R. If the transported species are ionic. Electrochem.22. the general relation. The form of this mixed parabolic relationship was first proposed by Evans. At relatively large times. Phys. 247 (1924). pp. and W.by Cabrera and Mott. (10). 3. the rate of growth 18 J. The effective diffusion coefficient Deft incorporates the effect of such space charges in enhancing the rate of transport. Brattain. see http://jap. The Corrosion and Oxidation of M etaJs (Edward Arnold and Company. Evans. and Shockley. the total oxide thickness Xo is taken to consist of two parts: an initial layer of oxide Xi that might have been present on the silicon prior to the oxidation step under consideration. the flux corresponding to the oxidation reaction is expressed by the first-order relation (11) which can be rewritten in the form xo2+Axo=B(t+T). and the flux F 2 is given by Nl l+k/h+kxo/Deff F 2= Deff(CO-C i ) / XO. (12) (12a) (12b) (12c) where A = 2Deff (l/k+ l/h). subject to the initial condition. Soc. Finally. Xo ( t A/2'. 19 . obtained by a straightforward integration. one obtains 1 and T= (X?+AXi)/B. The extent of the spacecharge region in Si0 2 films is shown later to be of the order of 200 A. (4) To arrive at a general initial condition to this equation. Thus the initial condition for this step is: (10) XO=Xi at t=O.I8 that if the oxide thickness is large enough in comparison to the thickness of the space-charge regions. Eq. reduces to the well-known parabolic oxidation law for relatively long 19 U. (13).3772 B. . an initial oxide layer may be formed by mechanisms involving fields and space charges within the oxide layer. 20 U. Redistribution subject to AIP license or copyright.20 Solving the quadratic equation (12) yields the form Xo [ A/2 = 1+ A2/4B t+r J' )! -1. H. and solving for Ci and Co between these two equations. 819-859. is X02+ AxO=Bt+x?+Axi. 1960). 16 The solution of the differential equation.243. GROVE The flux of the oxidant across the oxide layer is assumed to be given by Fick's law. (13).110.16 as well as by Bardeen. Evans. Such division of the oxide layer is not only important because it permits consideration of multiple oxidations. DEAL AND A. (13) Note that in the limit where the diffusivity becomes very small relative to the rate constants associated with surfaces. S.. It follows from the assumption of steady-state oxidation that F2 must be the same at any point x within the oxide film. (6) and C* Co C* l+ k/h+ kxo/Deff' l+kxo/Deff (7) l+k/h+kxo/Deff' The quantity T corresponds to a shift in the time coordinate which corrects for the presence of the initial oxide layer Xi.e. they set up a space charge within the oxide film. London..jsp . i. It was shown . It is interesting to examine two limiting forms of Eq. t»A2/4B.org/jap/copyright. Eq.e. 14. so that this condition is met in most practical cases. If Nl i~ the number of oxidant molecules incorporated mto a umt volume of the oxide layer. B=2DeffC*/N l . Shockley. Thus. (3) of the oxide layer is described by the differential equation dxo dt F kC*/Nl (9) at any point x within the oxide layer. Trans. A2/4B (8) or (14) Thus. J. the concentration of the oxidant within the oxide layer is linear as indicated in Fig. where Ci is the concentration of the· oxidant near the oxide-silicon interface. where Deff is the effective diffusion coefficient and dC/dx is the concentration gradient of the oxidizing species in the oxide. (9)." With C and Co eliminated. R. and also t»T. the flux is given by kC* F=Fl=F2=Fa=-----l+k/h+k x o/Deff which gives the oxide thickness as a function of time. Bardeen. Deff is approximately twice the actual diffusion coefficient. but also because Xi can be regarded as the thickness of the layer grown before the approximations of the above treatment become valid. E. (5) Noting t?~t Fl=F2 and F2=Fa due to the steadystate conditlon. Brattam. Consequently. Eq.::::. Ci ~ 0 and Co~ C*. 714 (1946). and the additional thickness grown during the step under consideration. This condition is commonly referred to as "diffusion controlled.aip. or less. Downloaded 26 Jul 2010 to 203. Chern. Deu/kxo ~ 0). 46. W. or that dFddX =0.

VXo (hrIILI FIG. since A itself TABLE 1. At the other extreme. (12). Straight lines are indeed obtained for all four temperatures with the absolute values of A increasing with decreasing temperatures. Silicon slices were placed flat on a quartz boat during oxidation. Then the oxide thickness Xo was plotted vs the quantity t/xo.110. the general relationship reduces to a linear law. The partial pressure of water in the wet-oxygen ambient was estimated to be 640 Torr. Values of the constants A and B determined from this plot for the four temperatures are tabulated in Table I. the accuracy was ±2S A.4SXlO16 cm-3). (12). EXPERIMENTAL METHODS The oxidation apparatus used in this program was similar to that used in previous investigations. EXPERIMENTAL RESULTS A. 20 First. The general relationship and its two limiting forms are shown in Fig. 3. 4. Oxide thicknesses were determined using multiplebeam interferometric techniques.203 14. the precision of A is ± 12%.0 3.3 Q·cm. 12. Slopes of lines correspond to B in Eq.720 0. ~~~(~) A/2-2 A2/4B B or xO::=::-(t+T). if that relationship holds and T=O.226 0. the following procedure was employed. Cleaning procedures prior to oxidation included both organic and inorganic rinses using ultrasonic agitation to insure noncontaminated surfaces.27 0. Oxidation temperature (OC) 1200° 1100° 1000° 920° A (I') 0.. 4.A 2/4B. As can be seen from Eq. For larger values of A (low temperatures) the precision is ±2%. Silicon used was in the form of circular slices 22 mm in diameter and mechanically polished to 200 Il thickness. and slope equal to B.15 Either purified dry oxygen (watercontentlessthanSppm) or wet oxygen (oxygen bubbled through water at 95°C) could be supplied to the quartz oxidation tube. xo2+Axo=B(t+r). For this reason.0 2. it appears that many of the contradictions referred to in the Introduction may well have been due to forcing the parabolic law in regions where it could not provide a meaningful fit.64 1.243.jsp . Thus. Analysis of the data has indicated that the precision of the constant B is ±2% over the entire range of measurements.406 Downloaded 26 Jul 2010 to 203.11 0. see http://jap. Evaluation of rate constants for oxidation of silicon in wet oxygen (95°C H 20).50 B~2/h) B/A~/h) r(h) 0 0 0 0 0. intercepts at t/xo=O correspond to -A in Eq. is identical with that derived by Cabrera and MotU 6 It is referred to as the parabolic rate constant. (12b). for relatively small oxidation times.A. Czochralski pulled. Redistribution subject to AIP license or copyright. 1 are plotted in this manner in Fig. For thinner oxides.ide thickness vs oxidation time was made to determine what initial oxide thickness Xi the data extrapolated to at t=O. In the case of high temperatures. Accuracy of the measurement for oxides thicker than 0. 4.RELATIONSHIP FOR OXIDATION OF Si 3773 times. A (15) Thus.287 0.I°. Wet Oxygen To check how well the general oxidation equation. that for most of the range the parabolic limit does not provide a good approximation. Rate constants for oxidation of silicon in wet oxygen (95°C H 20).05 0.e. At the same time.40 4. (12). The coefficient B. The coefficient -0. Wet oxygen data from Fig. such a plot should yield straight lines with intercept equal to . simply dividing xo2 by t does not necessarily yield the true parabolic rate constant B.0 B kh (C*) A = k+h Nt is referred to as the linear rate constant. Oxidation temperature was controlled to ± 1°C over the range 700° to 1200°C. (12).0 4.aip. fits the data. but rather it gives some quantity which mayor may not approximate B depending on the oxidation conditions and the time. t<<.22.0 6.20 Il was ±40 A. p-type boron doped (CB = 1.org/jap/copyright. together with experimental points which are discussed later.0 5. however. i. In the case of wet-oxygen oxidation Xi was found to be zero at all temperatures. It should be noted at this point. with an initial dislocation count of less than 100 cm-2• Silicon resistivity was 1.60 1. It was (111) surface oriented.510 0. values of the slope B decrease with decreasing temperature. a detailed plot of ox. Eq. It should be noted that this larger deviation at higher temperatures is not especially critical. as given by Eq.

05 . The corresponding values of T= (X. a detailed plot of oxide thickness vs time for dry oxygen did no[ extrapolate to zero initial thickness at any of the experimental temperatures.02 ~~~~-*2-L-4~~S~~8~-'~0-L~1~2~ DRY-OXIDATION TIME (hours) FIG. (12). Comparison of initial oxide growth at 920°C for oxidation in wet-(95°C H 20) and dry-oxygen ambients.§ 0.10 ~ . For this reason. . the effect of the oxygen in the wet-oxygen atmosphere can be neglected under the conditions of these experiments.22.org/jap/copyright.jsp . Dry Oxygen et al. While the wet-oxygen data extrapolate to zero. B.01 20 40 60 80 100 120 x Xi= 230 A to the time axis. In addition. Further confirmation was provided by precise and detailed measurements at 700°C.!3 lead to the same conclusion: an offset of about 200 A in the thickness data when extrapolated to t=O. since no initial oxide could be detected by experimental techniques. 00. see http://jap. Xo is plotted vs (t+ T) I Xo.aip.06''-~r--. the wet-oxygen data indicated no initial oxide and the silicon preparation was identical for the two types of oxidation.243. 2 shows that the differences in the variation of the exponent n in the empirical power-law formula.04 ~ ~ 0.4 -o·so 20 40 60 80 100 120 (t+T)/x.S~ ! ~ . for oxidation in dry oxygen.----r---r---. (12). Thus. (hrlll) FIG. with the intercept being in all cases 230±30 A.) The resulting rate constants for dry oxygen are given in Table II. This was found to be the case for all temperatures from 700° to 1200°C. ! 0.18 o :. between wetand dry-oxygen oxidation described in the Introduction are in fact due to this phenomenon. the plot for dry oxygen appears to pass through the thickness axis at Xo= 250 A. the initial condition Xi= 230 A was adopted. A re-examination of Fig. The value of T varies with temperature even though Xi is a constant since A and B are a function of temperature.and wet-oxygen data at 920 D C are plotted in the same graph with the time scales adjusted to provide similar curve' shapes. Consideration of the corresponding data of Fuller and Strieterl4 and of Evitts 0.110. Either an oxide film of this order of thickness was originally present before the oxidations were started.OS ~.12 ~ . Evaluation of rate constants for oxidation of silicon in dry oxygen. with the implication that the oxidizing species is water. DEAL AND A..2~~. It can be observed that an initial rapid rate of oxidation occurs. The latter was assumed to be the case. The significance of the finite Xi is discussed later. 6.08 ~ .3774 B. 5..04 § . Special experiments demonstrated that the values of A and B are essentially the same when the carrier-gas oxygen is replaced by argon. is very small and hence its influence on the oxidation rate becomes less important. In the case of 700°.----'--' -0.03 <. The difference between the wet. or a different mechanism of oxidation prevailed at oxide thicknesses less than 300 A.. 6.. which in turn should be Downloaded 26 Jul 2010 to 203.3~~Q~4__~Q~5~0~.> ~ w o ~ 0. intercepts at (t+T)/XO=O correspond to -A in Eq. Pressure Dependence of Rate Constants OXIDATION TIME (hotnI FIG. (Values of (t+T)lxo for goo°C are of the order of 400 hi}.and dry-oxygen cases is illustrated in Fig. C.2+AXi)IB were estimated graphically by plotting Xo vs t and extrapolating the curve back through In contrast to the case with wet oxygen. According to Eq.'. the parabolic rate constant B should be proportional to C*.. 7.14 ~ . GROVE WET-OXIDATltlN TIME (hours) . E. only the line extrapolated from these data is shown in Fig. 7.IS . 7. Oxidation of silicon in dry oxygen at 700°C. . 5. Here both dry. Thus. The precision of these constants is of the same order as discussed for the wet-oxidation results.02 . Redistribution subject to AIP license or copyright.~r-~0~~0'LI~0~. S. This finding could mean one of two things. Slopes of lines correspond to Bin Eq.~ . In Fig. (12). A plot of oxide thickness vs time at 700 D C is shown in Fig.. the relationship was found to be completely linear and therefore only the ratio BI A could be determined. xon=kt. followed by a strictly linear process which extrapolates back to Xo= 210 A at t= O.

40 9. Data presented by Flint7 on oxidation of silicon with various partial pressures of oxygen or water vapor in argon were examined.-.). Nature 171.jsp . This can be compared to the value of 27. 701 (1961).0030 0.and wet-oxygen oxidation.5 kcal/mole from Fig. 10. Parabolic rate constants are taken from Fig.Tl 0.0 81.------.) The activation energy for wet-oxygen oxidation is 16.~I---I~. 1000· to 1200·C). Exponential temperature dependence is obtained again for both the dry. 0 (I) ~ 0 0 CD b on on 'ti::: " DRY 0. are almost identical.. B(760. fused silica. Rate constants for oxidation of silicon in dry oxygen.3 kcaljmole for dry. 21 F.7 B(p. These diffusivity values ~re shown in Fig.0117 0.8 0.076 0. Rate constants are normalized to the 760. H2 0 0.1 p (aIm) FrG. xo2+Axo=B(t+. indicating a similar surface-control mechanism for the two oxidants. Here the values of B. the activation energy of B is found to be 28.22. (Flint's7 data for wet and dry oxygen. The effect of temperature on the parabolic rate constant Bl Downloaded 26 Jul 2010 to 203. This dependence is illustrated in Fig..---.1 00 II OO'C.0 B(".25 X 1022 cm-a.0011 B/A (".2 0..5 06 0. 0.0 kcal/mole reported by Norton21 for the diffusivity of oxygen through fused silica.3 04 0..8~~0~~--~1. According to Eq.) A very good linear dependence is observed for temperatures ranging between 1000° and 1200°C. (12b).027 0.165 0. 11. Also shown are the wet-oxygen values correlated to 760Torr water vapor pressure using the linear pressure dependence.071 0. 10 for reference. diffusivities from Fig. 22 A.4 0. while the corresponding value for <. 0 0 0 ~ ~ b 0 1. good straight lines are obtained in all cases.7 0. 8. the coefficient A should be independent of the partial pressure.-28.<00. H2 0 o 1000'C.'/h) 0.9 0. The effect of partial pressure of the oxidant on the parabolic rate constant. the temperature dependence of B should be substantially the same as that of Deft.5 0. (The structure of the Si0 2 formed by thermal oxidation of silicon corresponds to that of amorphous.235 0. 9. H2 0 • 1200'C. Moulson and J. As is evident.3<>p<. Values obtained by this extrapolation closely agreed with Flint's7 data based on oxidation in atmospheric steam. Redistribution subject to AIP license or copyright.045 0. at all experimental temperatures.3 proportional to the partial pressure of the oxidizing species in the gas if Henry's law is obeyed.. Trans. respectively. FrG. Conversely. I .and wet-oxygen cases.370 OF Si 3775 I.6 0.3-kcal/mole value found by Moulson and Roberts22 for the diffusivitv of water in fused silica.090 0.110.0~--~I. Temperature Dependence of Rate Constants In Fig.2 0.30 0..0 11> b 0 .0 1.1 0.""--''''¥'''--'''fL--'-¥''-''¥''1 1.and wet-oxygen oxidation. O2 o 1200'C.2~ 1000 "flIi(j' J.5 keal/male I04~0~~'--'0~. 9. The temperature dependence of the linear rate constant B/ A is shown in Fig./h) 1. E.3 kcal/mole. (12b). Equilibrium Solubility Concentrations The equilibrium concentration C* of the oxidants oxygen and water in Si0 2 can be calculated using Eq.RELATIONSHIP FOR OXIDATION TABLE II.0 0.2.37 1.0208 0.org/jap/copyright.243.0 and 45.027 0. 57.12 0. (The 760-Torr value was determined by correcting the wet-02 data of the present investigation from 640 to 760 Torr using the linear relationship.0049 0. (760 Torr) E. Nl for oxygen is 2. normalized to the 760-Torr value at each temperature are shown as a function of the partial pressure of the oxidizing speciesO2 or H 20. Faraday Soc. Norton.040 0.8 09 1.-=.00026 T(h) 0. J. while the value of B was directly proportional to the partial pressure.aip. 1208 (1961). P.> 0 ~ ~ .T) 9 • • t. For dry oxygen. .-10'P0. It was found through plots of Xo vs (t+r)/xo that A was a constant over a range of partial pressures. 9. The activation energies 46. Roberts. see http://jap. 8. D.Torr value at the same temperature. respectively-in the gaseous ambient. in reasonably good agreement with the 18. the logarithm of the parabolic rate constant B is plotted against the reciprocal of the absolute temperature for both dry. Oxidation temperature (·C) A ("') 1200° 1100· 1000· 920· 800· 700· 0. 9.

.. 4D. the activation energies reported in Sec.4XIQl9 3. Inc. Pauling. In the other. Redistribution subject to AIP license or copyright. and at the oxidesilicon interface. whichever is smaller. reflect the temperature dependence of the interfacial reactionrate constant k. N.3 kcal/mole. Surface-Reaction Rate Constants The linear rate constants determined in Sec. If these two constants are very different in magnitude. Bird. Both of these observations point to the relatively small importance of the gas-phase transport process in controlling the over-all oxidation rate. The diffusivity of oxygen (Norton21 ) and of water (Moulson and Roberts22) in fused silica as a function of temperature.2 kcal/mole. Lightfoot. E. Also shown in Table III are the equilibrium solubilities of these two species in fused silica as determined by Norton21 through permeation experiments and by Moulson and Roberts22 through infrared-absorption experiments. 23 Species O2 H 20 Downloaded 26 Jul 2010 to 203. New York. 1960). Solubility (cm-3) This work (C*) Other methods 5. The Nature oj the Chemical Bond (Cornell University Press. Since the temperature dependence of B and of Deff were found to be substantially the same. Inc. 3rd ed. and then compared to experimental results obtained over a wide range of variation of conditions: ambient pressure and temperature. and E. the parabolic rate constant B is proportional to the partial pressure of the oxidant in the gas (0 2 and H 2 respectively) while the coefficient A is independent 0. who found an essentially constant solubility of water in fused silica between 900° and 1200°C.org/jap/copyright. the quantity kh/(k+h) will approximately equal k or h.(L/h for the present flow conditions (Reynolds number"'" 25). W. Such a conclusion is substantiated by Moulson and Roberts.jsp . 1960). Chap.6X 104 . S. The relatively large value of the gas-phase transport coefficient h is confirmed by two other experiments. Thus. Solubility of oxidizing species in Si0 2 at lO00°C. 5.6XI04 . 85). This quantity can be calculated from the experimentally measured linear rate constants B/A) the known values of N 1. DISCUSSION In summary a simple model of the oxidation process has been developed. and R.3776 B. 4th ed. through k.(L/h for wet 02. The agreement between these values and the C* results obtained from the oxidation data is indeed excellent. E.(L/h for wet-02 at l000°C. 46. (12).2X1010 5. see http://jap. 14.22. DEAL AND A. Since this is several orders of magnitude larger than the above values of kh/(k+h). it has been notedlO that the backside of a slice lying flat on the boat was oxidized to the same extent as its topside. .(L/h for dry O 2 and 1. Eq. respectively. 10. The resulting values of C* at l000°C are listed in Table III. is obeyed by silicon oxidation throughout this range. Boundary Layer Theory (MCGraw-Hill Book Company. through h. These values may be compared with 42. it follows that C*.8X 103 . 1960). at 1000°C.0 and 45.24 p. It was found that the general relationship. In one.0X1019 (infrared)" --=k= k+h kh {3.243.110. B.3 kcal/mole B A kh (C*) = k+h Nl ' FIG. the coefficients of this relationship were shown to depend in the predicted manner on pressure and temperature. 4(A) and (B). If it is now assumed that h is determined solely by a gas-phase transport process. H.2. the carrier-gas flow rate was varied SO-fold without any significant effect on the oxidation rate. respectively. similar values of C* would be obtained at any other temperature within the experimental range. New York. g o . ~ ~ H2 0 '\ EA :27 kcol/mole '\ '\ '\ "\ EAa 18.(L/h for dry-02 oxidation 1. New York. Paragraph 21. Thus. the energy required to break an Si-Si bond (as given by Pauling. Schlichting. the value of kh/(k+h) is calculated to be 3. Thus. TABLE III.8X 103 . its value can be estimated on the basis of standard boundary-layer considerations23 to be about 108 . Transport Phenomena (John Wiley & Sons.5XI010 (permeation)21 3. 24 L.aip. GROVE water is twice this number. F. contain the effects of the phenomena taking place both at the gas-oxide interface. and the values of the equilibrium as determined in the solubility concentrations previous section. Furthermore. Stewart.. Ithaca.

earlier measurements obtained at these laboratories. as has been shown to be the case for dry-oxygen oxidation by Jorgensen. these data were plotted following the rearranged form of the general oxidation relationship. It should be noted that the general relationship. the dotted lines its two limiting forms. General relationship for thermal oxidation of silicon. If the transported species is ionic. 12.~~~ ~ FIG. The existence of a rapid initial-oxidation stage with dry oxygen as well as the absence of such a phase with wet-oxygen oxidation are of interest. A2 4B DoffNl 2C*(1/k+1/h)2' or. see http://jap.s such a phase is Downloaded 26 Jul 2010 to 203. Redistribution subject to AIP license or copyright.jsp . thus accounting for the higher oxidation rate. 9 and 11 and the pressure dependence shown in Fig. 1 and 2. Finally. alternatively. t REF 13 REF 14 10' FIG. The temperature dependence of the parabolic rate constant is exponential. respectively. (13). Figure 12 also shows the two asymptotic forms of the general relationship: the parabolic law valid at "large" times and the linear law valid at "small" times.org/jap/copyright. 8. 8. characteristic time of the oxidation process. While the diffusivity of water in fused silica is lower than the diffusivity of oxygen (Fig. This concentration is three orders of magnitude larger for water than for oxygen. Experimental data were reduced using values of A and B determined from Figs. the parabolic rate constant B is considerably larger for oxidation in wet oxygen than in dry (Fig. The values of r correspond to Xj=O and 200 A for wet and dry oxygen.14 were reduced using the parabolic and linear rate constants given in Figs.13.243. The excellent over-all agreement is quite evident and speaks for the validity and usefulness of the general relationship. The general relationship for the oxidation of silicon is illustrated in Fig. Thus. corrected to the proper partial pressure according to Fig. 8. It is important to note that the criterion of "small" or "large" times depends on the oxidation conditions. and 11.aip. 9. The effect of temperature on the linear rate constant B / A. with the activation energies corresponding to the available activation energies for the diffusivities of oxygen and of water through fused silica.REF . This general agreement serves as an indication of the validity of the assumptions made in the derivation of Eq. 10). that the oxide thickness should be smaller (or larger) than the characteristic distance of the oxidation process. A/2 • imiK7 . as reflected in the Parameters A and B. The solid line represents the general relationship. 7 and the work of two other groups of investigators. and therefore the rate constant B is proportional to the equilibrium concentration C*. the equilibrium solubility concentrations C* determined from the rate constants are in good agreement with the solubility of oxygen and water in fused silica as determined by other methods. enables the construction of the correct oxidationrate equation for any set of experimental conditions with confidence.~--r.22. Then. 9). This seeming paradox is immediately resolved when it is recalled that the flux of oxidant. 11. Large numbers of oxidation data. Eq. (12).RELATIONSHIP FOR OXIDATION OF Si 3777 of the pressure. including results shown in Figs. 9 and 11. together with the rate constants from Figs. Included as a solid line (almost masked out by the experimental data) is a calculated plot corresponding to the general relationship X02+ Axo =B(t+r). this criterion is that the time should be smaller (or larger) than the x. It is worthwhile to point out certain interesting aspects implicit in the results of this work. 12. A Doff 1/k+1/h 2 IO&·~~--~'-~~--~I.110.

e. This. and the dielectric constant Ko=4. Redistribution subject to AIP license or copyright.result obtained by Fromhold26 through a more detailed treatment. The authors also wish to thank G. using C*= 3 X 1019 cm-3 . ACKNOWLEDGMENTS The valuable assistance of Mrs.)xo. discussions with l\. DEAL AND A. in J. see http://jap.jsp .110.aip. is estimated to be only around 6 A. Sklar in the experimental portion of this work.24 p. H. and W. Downloaded 26 Jul 2010 to 203. S. E. Electrochern.o= [(kT/ q) (K o€o/2qC*)J. the oxide-Debye length is estimated to be of the order of 150 A for dry O2 oxidation. t "" C*/ N 1 < 10-3 in the parabolic region of oxidation-certainly a negligible time period. and the denominator by the flux F2 as given by Eq. Ko. taking ttransient / X02= Bt. This latter quantity is of the order of the extrinsic Debye length of the oxide/ 6 LD. it could not be observed using the present techniques. Kuper. A. Osborne. Frornhold. 26 A.) The field created by the space charge of the negatively charged oxygen ions should be directed toward the gasoxide interface. D. Phys. (For the properties of this ion. 259C (1965). The latter finding is in agreement with a similar conclusion reached by N orton21 from the pressure dependence of permeation rate of oxygen through fused silica. C. H. the relaxation time is found to be of the order of For simplicity. !(Co-C. even if an initial rapid oxidation phase existed in the case of wet O2. the return to equilibrium would take a period of time of the order of total amount of oxidant needed to return to the steady-state distribution ttransient flux of oxidant Approximating the numerator by the total amount of oxidant present within the oxide layer in excess of a uniform concentration. Carlson. Fuller.243. Chern. at 1000°C. Jr. Soc.22.. T. E. Solids 24. are very much appreciated. 1081 (1963). If the concentration distribution shown in Fig. i. would imply that. 3 were disturbed. Moore for his valuable comments concerning this manuscript. as well as helpful. see Pauling. in agreement with the . APPENDIX A simple criterion regarding the validity of the steady-state condition can be derived on the basis of an order-of-magnitude consideration. Using the value of C*=5X1016 cm-3 as determined in this work. Snow. S.3778 B. G. 25 See abstracts of "Recent News" papers describing work of E. Dumesnil and E. Jorgensen 8 showed through experiments relating to the effect of electric fields on dry-oxygen oxidation that the diffusing species is negatively charged. as follows. 112. the extrinsic Debye length for oxidation in wet O2 . This indeed corresponds well to the oxide thickness beyond which the rapid initial-growth mechanism stops being effective. GROVE to be expected16 until such time that the oxide thickness becQmes large in comparison to the extent of the spacecharge region within the oxide. the diffusing species is undissociated water while for oxidation in dry oxygen it is molecular oxygen. in turn. These two observations. taken together. It is worthwhile to emphasize again that the good agreement between the pressure dependence of the coefficient B with the linear dependence expected on the basis of the assumption of Henry's law implies the absence of any dissociation effects at the gas-oxide interface. In contrast. Thus. (4). 351. for oxidation in wet oxygen. point to the conclusion that the diffusing species in the case of dry-oxygen oxidation is the superoxide ion O2-. Yon. and J. R.I:. J. in agreement with the above conclusion. at 1000°C. Recent studies25 of the N a+ ion distribution in thermally grown oxides show that most of the Na+ ions are concentrated near the outer surface after oxidation. and of H. B.org/jap/copyright.

Sign up to vote on this title
UsefulNot useful