Comparison of FTIR transmission spectra of thermally and LPCVD SiO 2 films grown by TEOS pyrolysis

Article abstract of DOI:10.1149/1.1676725

Dispersion analysis of Fourier transform infrared (FTIR) spectra taken on thermally grown SiO₂ films developed in dry oxygen and low pressure chemically vapor deposited (LPCVD) from tetraethylorthosilicate vapors was performed. LPCVD was carried out at temperatures of 635, 650, 710, and 820°C while thermally grown samples were prepared at 950, 1050, and 1150°C. The thickness of all films was approximately 100 nm. Transmission spectra within the range 900-1400 cm^-1 were analyzed using four Lorentzian oscillators located near 1060 (1), 1089 (2), 1165 (3), and 1220 (4) cm^-1, the last two being much weaker than the first two, so the study was limited to oscillators 1 and 2. It was found that the increase of the temperature of growth implies a shift of central frequencies of both oscillators toward higher wavenumbers and the oscillator strength of 1 increases while that of 2 decreases. The damping coefficient, γ, of both oscillators initially increases and above 950°C decreases and the Gaussian width of central frequencies, δ, for 1 increases toward 30 cm^-1 while that of 2, above 950°C, saturates near 17 cm^-1. It was suggested that oscillators 1 and 2 correspond to different Si-O-Si bridges in which angles distribute in different ways. Bridges associated with oscillator 1 are located in the bulk of the films and those with oscillator 2 appear near interfaces and grain boundaries.

Full text of DOI:10.1149/1.1676725

Journal of The Electrochemical Society, 151 ͑5͒ F93-F97 ͑2004͒
F93
0013-4651/2004/151͑5͒/F93/5/$7.00 © The Electrochemical Society, Inc.
Comparison of FTIR Transmission Spectra of Thermally
and LPCVD SiO2 Films Grown by TEOS Pyrolysis
*
**,z
Vassilis Em. Vamvakas and Dimitris Davazoglou
Institute of Microelectronics, NCSR ‘‘Demokritos’’, 1510 Athens, Greece
Dispersion analysis of Fourier transform infrared ͑FTIR͒ spectra taken on thermally grown SiO2 films developed in dry oxygen
and low pressure chemically vapor deposited ͑LPCVD͒ from tetraethylorthosilicate vapors was performed. LPCVD was carried
out at temperatures of 635, 650, 710, and 820°C while thermally grown samples were prepared at 950, 1050, and 1150°C. The
thickness of all films was approximately 100 nm. Transmission spectra within the range 900-1400 cmϪ1 were analyzed using four
Lorentzian oscillators located near 1060 ͑1͒, 1089 ͑2͒, 1165 ͑3͒, and 1220 ͑4͒ cmϪ1, the last two being much weaker than the first
two, so the study was limited to oscillators 1 and 2. It was found that the increase of the temperature of growth implies a shift of
central frequencies of both oscillators toward higher wavenumbers and the oscillator strength of 1 increases while that of 2
decreases. The damping coefficient, ␥, of both oscillators initially increases and above 950°C decreases and the Gaussian width of
central frequencies, ␦, for 1 increases toward 30 cmϪ1 while that of 2, above 950°C, saturates near 17 cmϪ1. It was suggested that
oscillators 1 and 2 correspond to different Si-O-Si bridges in which angles distribute in different ways. Bridges associated with
oscillator 1 are located in the bulk of the films and those with oscillator 2 appear near interfaces and grain boundaries.
© 2004 The Electrochemical Society. ͓DOI: 10.1149/1.1676725͔ All rights reserved.
Manuscript submitted March 28, 2003; revised manuscript received September 26, 2003. Available electronically March 15, 2004.
Silicon dioxide films are very important in microelectronics since
they are used as gate dielectrics in metal oxide semiconductors
͑MOS͒ transistors, as insulators between conductive lines, as protec-
tive coatings, etc.1,2 There are several methods for growing SiO2
films, such as the thermal oxidation of silicon, low pressure chemi-
cal vapor deposition ͑LPCVD͒, or plasma enhanced chemical vapor
deposition ͑PECVD͒. Depending on the method and input gases, the
growth temperatures may vary from less than 300°C to more than
1000°C.
were treated by LPCVD from pyrolysis of tetraethyl orthosilicate
͑TEOS͒ vapors, while those between 950 and 1150°C were ther-
mally grown in dry oxygen. TEOS SiO2 films were chosen because,
as shown in the past,7 they exhibit a granular form, i.e., an enhanced
internal surface relative to thermally grown films. The grain size in
TEOS SiO2 films was found to increase with the deposition
temperature7 leading to a corresponding decrease of internal surface.
The evolution of the FTIR transmission spectra of TEOS SiO2 films
with the temperature of growth was analyzed and compared with
spectra taken on thermally grown films with the hope of observing
the magnitude of the absorption components to vary with tempera-
ture of growth, i.e., the internal surface. It was found that for all
films the transmission minimum, shown on the FTIR spectra near
1100 cmϪ1, could be analyzed using two oscillators centered near
1090 and 1060 cmϪ1. The relative magnitude of these components
was indeed found to be temperature dependent. For films grown at
temperatures up to 650°C the two components were almost equal
while as the temperature increased that at 1090 cmϪ1 decreased in
favor of the other at 1060 cmϪ1, corroborating the assumption made
before.
In a previous work3 we compared Fourier transform infrared
͑FTIR͒ transmission spectra within the domain 700 to 1400 cmϪ1 of
bulk fused silica with those of thermally grown, gate quality, SiO2
films with thicknesses varying between 10 and 100 nm. It was
shown3 that while for the bulk material a single Lorenzian oscillator
was sufficient to describe each of the minima located near 800,
1090, and 1200 cmϪ1,4 for the SiO2 films each one of the above
minima would have to be analyzed using two ͑or, more generally,
multiple͒ oscillators causing two absorption components. Tests
showed that some of these oscillators, either due to oxygen diluted
in the substrate or because the SiO2 phases which were thermody-
namically favored at the oxidation temperature, did not have enough
cooling time to transform to the room temperature stable phase and
thus were excluded from further study.3 Moreover, these oscillators
did not seem to be connected to the transition zone between Si and
SiO2 only, at least for the range of film thicknesses examined.3 It
was concluded then that the multiplicity of the FTIR minima was
due to the fact that in SiO2 films the Si-O-Si angles do not distribute
following a simple gaussian scheme as in the bulk material, but the
distribution is a superposition of two gaussians and that this is an
inherent property of thermally grown SiO2 films.3 A similar conclu-
sion has been drawn by other workers,5 which have described the IR
spectrum of thermally grown SiO2 films using asymmetric Gaussian
oscillators with different high and low distribution widths.5 In pre-
vious work3 the dominant of the two Gaussian oscillators ͑i.e., the
one with the larger area͒ was attributed to Si-O-Si bridges located at
the bulk of films while the second, weaker one was attributed to
Si-O-Si bridges located near the two interfaces with the Si substrate
and air. In this work we further investigate the validity of this as-
sumption analyzing the FTIR spectra within the domain 900 to 1400
cmϪ1 of SiO2 films grown at temperatures between 635 and 1150°C
using the method described before.3,6 Films between 635 and 820°C
Experimental
All thermally grown films were grown in dry oxygen at tempera-
tures of 950, 1050, and 1150°C, on pieces cut from 3 in. Czochralski
͑CZ͒ ͗100͘ 5.5 ⍀ cm Si wafers. Oxidation time was the controlling
parameter of the thickness of the films which in all cases was about
100 nm.
Depositions were carried out in a Tempress Systems, Inc. ͑model
Omega Junior͒ horizontal, hot-wall reactor on pieces cut from 3 in.
CZ ͗100͘ 5.5 ⍀ cm Si wafer at 300 mTorr pressure and a constant
TEOS flow equal to 40 sccm. Deposition temperatures were 635,
650, 710, and 820°C, which is the upper limit for our equipment.
Deposition time was the parameter controlling film thickness, and it
was set so as to obtain approximately 100 nm thick films. Thick-
nesses of both deposited and thermally grown films were measured
using a Gaertner rotating analyzer ellipsometer model L116B-85B
with a He-Ne laser at 632.8 nm.
Before oxidation or deposition all substrates were given an RCA
cleaning, dried in a compressed jet of nitrogen, and inserted directly
into the oxidation furnace. Oxidation or deposition furnaces used
ultrapure N2 as the purge gas. After oxidation or deposition the SiO2
film was removed from the back of samples in a reactive ion etcher
using SF6 chemistry.
FTIR transmission spectra were recorded with a Magna single-
beam IR 550 Nicolet spectrometer connected to a personal computer
using the OMNIC software version 1.2a. Each of the spectra is the
* Electrochemical Society Student Member.
** Electrochemical Society Active Member.
z E-mail: D.Davazoglou@imel.demokritos.gr
Downloaded on 2015-03-13 to IP 129.128.216.34 address. Redistribution subject to ECS terms of use (see ecsdl.org/site/terms_use) unless CC License in place (see abstract).

F94
Journal of The Electrochemical Society, 151 ͑5͒ F93-F97 ͑2004͒
average of 512 passes in order to obtain the best possible signal to
noise ratio. Before recording a spectrum the background was taken
using a piece of silicon cut from the same wafer used to grow the
sample. This piece was oxidized or used as a substrate for SiO2
deposition under the same conditions as the sample under examina-
tion and, at the end of the process, striped. This was done to elimi-
nate absorption caused from the interstitial oxygen of the substrate
at 1136 cmϪ18 from the Si-Si vibration at 611 cmϪ19 and from the
possible precipitation of oxygen in the substrate.3
Calculations
Our analysis used the classical dispersion theory of crystals10
modified so as to account for the amorphous nature of SiO2 films.6
Following this approach ␧ and ␧2 , ͑the real and imaginary parts of
1
dielectric constant, ␧ ϭ ␧ ϩ ␧2) related to n and k ͑real and imagi-
1
˜
nary part of the refractive index, N ϭ n Ϫ ik), vary with frequency
(␯) as6
␧ ϭ n2 Ϫ k2 ϭ ␧
1
ϱ
2
␯j Ϫ ␯
0i
2
2
exp Ϫ
␯ Ϫ ␯ ͒
j
͑
ͫ
ͩ
ͪ
ͬ
4ͱ␲
1.201␦i
1.201␦i
2
0i
ϩ
␳ ␯
i
d␯
d␯
͵
j
͚
2
2
2
j
2
2
2
i
␯ Ϫ ␯ ͒ ϩ ␥ ␯ ␯
͑
j
i
2
␯ Ϫ ␯
j
0i
exp Ϫ
␥ ␯ ␯
ͫ
ͩ
ͪ
ͬ
i
j
4ͱ␲
1.201␦i
1.201␦i
2
0i
␧ ϭ 2nk ϭ
␳ ␯
i
͵
2
j
͚
2
2
i
2
2
j
2
2
i
␯ Ϫ ␯ ͒ ϩ ␥ ␯ ␯
͑
j
͓1͔
where the summation is made over the lattice oscillators and each of
them is characterized by its polarization ͑strength͒, ␳i ; width ͑damp-
ing͒, ␥i ; and central frequency, ␯0i . The quantity ␳i is related to the
2
oscillator strength, fi , by ␳i ϭ Ne2/4␲2m␯0i fi where N is the os-
cillator concentration and e and m are the charge and mass of the
oscillating charge. To account for the amorphous nature of the films
the central frequency, ␯0i , for the ith oscillator was considered to
distribute following a gaussian frequency with dispersion ␦i .
In order to reproduce the asymmetric stretch ͑AS͒ region of the
FTIR spectrum ͑900-1400 cmϪ1͒ of SiO2 films with the above
model we supposed the existence of four oscillators within it. This is
justified because it is well known11 that the AS stretch of the Si-O-Si
bridge, in which the oxygen atoms moves back and forth along a
line parallel to the axis through the two silicon atoms, gives two
absorption peaks located near 1100 and 1200 cmϪ1. The stronger of
these peaks, near 1100 cmϪ1, corresponds to a vibration in which
adjacent oxygen atoms execute the AS motion in phase, while the
other, near 1200 cmϪ1, corresponds to a vibration in which two
adjacent oxygen atoms vibrate with a phase difference of 180°.
Therefore, since two absorption components are present near 1100
cmϪ1, clearly shown on our second derivative spectra of FTIR
transmission7 and by other workers,5,12 it is expected that two also
are present near 1200 cmϪ1. The same holds for the case of the
symmetric stretch mode ͑near 800 cmϪ1͒ where, as shown before in
Ref. 3 two absorption components are also present. Comparing this
model with that of Ref. 5 one observes that the latter contains fewer
adjustable parameters, which is a desirable mathematical feature
when fitting experimental data, especially when the number of ex-
perimental points is comparable to the number of adjustable param-
eters. On the other hand, the present model is self-consistent since it
explains the presence of two absorption components at the transmis-
sion minima at 1200, 1100, and 800 cmϪ1, gives information with
respect to the nature of the oscillators ͑frequency, strength, number,
and damping͒, the way oscillators interact with each other, and pro-
vides information with respect to the positioning of the two oscilla-
tors within the SiO2 film ͑see below͒.
Figure 1. ͑a͒ Experimental and simulated FTIR transmission spectra within
the range 900-1400 cmϪ1 of a TEOS-deposited SiO2 film at 710°C, 300
mTorr with a thickness of 103.9 nm. ͑b͒ Analysis of the spectra presented in
͑a͒ with four Lorentzian oscillators located at 1, 1040.5; 2, 1075.8; 3, 1138.8;
and 4, 1200.0 cmϪ1 causing the absorption components T1, T2, T3, and T4,
respectively.
1 Ϫ R RЈ exp Ϫ2␣ d ͒
͑
Tf
Tf
Ts
s
s s
s
Tcalc
ϭ
Х
͓2͔
Ts 1 Ϫ RЈRЈ exp Ϫ2␣ d ͒
͑
s
s
f
s
where Tf is the transmission of a film ϩ infinite substrate and Ts is
the transmission of a finite substrate.13
Theoretically calculated spectra were fit to those experimentally
recorded (Texp), minimizing the quantity ͑unbiased estimator͒
f ϭ ͚(Texp(k) Ϫ Tcalc(k)/␴(k))2. The summation was made at all
measured wavenumbers ͑k͒. ␴(k) represents the error of the mea-
surement at every wavenumber which, since it was not specified by
the instrument, was taken to be equal to 0.001 for all of them. The
minimization was made with a program taken from the CERN li-
brary ͑MINUIT͒ which was also able to give the confidence inter-
vals and the errors in various mathematical senses. For all fitting
parameters the 95% confidence intervals were calculated ͑as re-
ported in Ref. 3͒ and were found to be at lease one order of magni-
tude smaller than the minimum values.
Results and Discussion
In Fig. 1a we see the experimental transmission spectrum within
the range 900-1400 cmϪ1 of a TEOS deposited oxide at 710°C hav-
ing a thickness equal to 103.9 nm together with the simulated one
using four Lorentzian oscillators.3 It can be seen that the agreement
between experimental and simulated spectra is excellent; similar re-
As shown in detail previously, the transmission Tcalc of a film
ϩ finite substrate can be calculated from the expression13
Downloaded on 2015-03-13 to IP 129.128.216.34 address. Redistribution subject to ECS terms of use (see ecsdl.org/site/terms_use) unless CC License in place (see abstract).

Journal of The Electrochemical Society, 151 ͑5͒ F93-F97 ͑2004͒
F95
Figure 2. Analysis of the FTIR transmission spectrum of a film thermally
Figure 4. Variation of the central frequency, ␯0i , with the growing tempera-
ture for oscillators 1 and 2. Films grown at temperatures up to 850°C are
LPCVD grown from TEOS vapors and thermally grown above this tempera-
ture.
grown at 1050°C SiO2 with a thickness equal to 94 nm, with four Lorentzian
oscillators located at 1, 1062.5; 2, 1090.8; 3, 1174.7; and 4, 1230.0 cmϪ1
.
sults were obtained for all samples in this study. In Fig. 1b we see
the analysis of the same spectrum to the corresponding ͑four͒ ab-
sorption components noted as T1, T2, T3, and T4. In Fig. 2 the
transmission spectrum of a thermally grown 94 nm thick oxide at
1050°C and its decomposition into four components corresponding
to four Lorentzian oscillators3 is shown. As seen in Fig. 1b and 2 the
oscillators 3 and 4 are very weak compared to the other two, and this
leads to a limited accuracy in the determination of their parameters
from the fit. Due to the limited accuracy of these parameters, in the
following we now focus our study on the other two stronger oscil-
lators giving the absorption components T1 and T2. In Fig. 3 the
evolution of the absorption components T1 and T2, normalized in
order that all T1 to have the same minimum, with the temperature of
growth as shown. It is seen that the magnitudes of T1 and T2 are
comparable for the TEOS SiO2 film grown at 635°C and as the
growing temperature increases T2 becomes weaker. It is also ob-
served that as the temperature of growth increases, both T1 and T2
are shifted towards higher wavenumbers. This shift implies that of
the resonance peaks observed on the FTIR spectrum moves toward
higher wavenumbers, and this effect has been observed in the
past14,15 and related to the relaxation of stress in the bulk of the
oxide by a viscoelastic relaxation process. Microscopically, this
means that in moving away from the interface SiO2 /Si towards the
bulk of the film the way the Si-O-Si angles distribute changes
abruptly.15 Moreover, it is expected that the way these angles dis-
tribute near the interface of the SiO2 films with air is different from
that in the bulk of the film. The above discussion justifies our as-
sumption that the total distribution of the Si-O-Si angles in SiO2
films is not simple, but a superposition of two gaussian frequencies.
In Fig. 4-7 the temperature of growth evolution of the parameters of
oscillators 1 and 2 are shown. It can be seen that for the TEOS
LPCVD films the increase of deposition temperature implies a con-
tinuous shift of the central frequencies of both oscillators 1 and 2
towards higher wavenumbers. For the thermally grown films the
central frequency of oscillator 2 saturates at 1090 cmϪ1, which is the
wavenumber of bulk SiO2 ,16 while that of oscillator 1 shifts slowly
toward higher wavenumbers. In Fig. 5 the evolution of the oscillator
strength for both oscillators with the same temperature is shown. It
is seen that the oscillator strength of 1 increases while that of 2
decreases. Since the oscillator strength is proportional to the number
of oscillators, it is concluded that as the growth temperature in-
creases the number of oscillators 2 and decreases in favor of oscil-
lators 1. The growth temperature evolution of the damping factor, ␥,
is shown in Fig. 6. As seen in this figure, for temperatures up to
Figure 3. Evolution of the absorption components T1 and T2 for the TEOS-
deposited and the thermally grown SiO2 films. All peaks have been normal-
ized so that the T1 components to have the same minimum. T1 and T2
components of the TEOS deposited films at 650 and 710°C and of the ther-
mally grown SiO2 film at 1050°C are located at the expected positions and
are not presented in this figure for clarity.
Figure 5. The temperature of the oscillator strength growth evolution, ␳ ␯2 ,
for oscillators 1 and 2.
i
0i
Downloaded on 2015-03-13 to IP 129.128.216.34 address. Redistribution subject to ECS terms of use (see ecsdl.org/site/terms_use) unless CC License in place (see abstract).

F96
Journal of The Electrochemical Society, 151 ͑5͒ F93-F97 ͑2004͒
with the angle. This result is coherent with the electrical properties
of the TEOS films, which are used as insulators in metal oxide
semiconductor ͑MOS͒ structures and have been found excessively
charged.20 Returning to Fig. 4 it is seen that the increase of growth
temperature shifts the central frequencies of oscillator 1 towards
higher values while the central frequency of oscillator 2 saturates at
1090 cmϪ1 corresponding to that for fused silica. Extrapolating
curve 1 in Fig. 4 to 1800°C where fused silica is produced, one
obtains a wavenumber of approximately 1090 cmϪ1. It can be con-
cluded then that Si and O atoms located near grain boundaries and
the film interface with air, related to oscillator 2, have an enhanced
mobility similar to atoms in fused SiO2 where no stresses are
present. Hence they vibrate in a similar manner as the latter. On the
contrary, atoms near the bulk of the films or the core of the grains
need to reach the temperature of fusion of SiO2 to start vibrating in
a way similar to that of the bulk material. LPCVD of TEOS SiO2
films mostly proceeds through gas-phase reactions.7 Therefore, the
atomic arrangement at the core of the clusters, which form in the gas
phase, is different from that on their surface. For TEOS films grown
at temperatures up to 650°C, the grain size is of the order of several
nanometers, there is no distinction between Si-O-Si bridges within
the core of grains and on their surface, and the ␥s for oscillators 1
and 2 coincide as seen in Fig. 6. At higher temperatures the grain
size increases and the difference between the ways Si-O-Si bridges
located at the core of grains and near the grain boundaries vibrate
becomes evident ͑Fig. 6͒. Atoms near the surface are more mobile,
and this explains the low values of ␥ for oscillator 2 associated with
Si-O-Si bridges near the surface. Above 950°C films were grown
thermally and due to the viscous flow, which occurs near this
temperature,14,15,17 the way Si and O atoms distribute near the inter-
face remains unaffected by the oxidation temperature. This explains
the saturation of ␥ for oscillator 2 whose low value is related to the
absence of an internal surface in films. A stress release near the
interface with the substrate is expected to cause a corresponding
release within the bulk of thermally grown films, which explains the
decrease of ␥ for oscillator 1 at temperatures above 950°C, seen in
Fig. 6. The Gaussian width of the central frequency, ␦, represents the
influence of a vibration at a certain frequency to vibrations at other
frequencies. This is related to the number of Si-O-Si bridges in-
volved in the vibration ͑the more the oscillators the more likely they
are to influence each other͒ and explains the similarity between the
growth temperature evolutions of ␦ ͑Fig. 7͒ and of oscillator
strength ͑Fig. 4͒.
Figure 6. Damping factor, ␥i , variation vs. growing temperature for oscil-
lators 1 and 2.
650°C the ␥s for both oscillators coincide. Above this temperature ␥
for oscillator 1 takes higher values than for 2. For temperatures
above 950°C the ␥ for oscillator 1 decreases while for 2 the ␥
saturates at low values. This behavior is probably related to the
stress release that occurs, due to the viscous flow,14,15,17 in SiO2
thermally grown films above 950°C. In Fig. 7 the evolution of the
Gaussian width, ␦, with the temperature of film growth is shown. It
is seen that, as for the oscillator strength ͑Fig. 5͒, the increase of
growth temperature implies an increase of ␦ for oscillator 1 and a
decrease for 2.
As mentioned in the introduction, contrary to thermally grown
SiO2 films TEOS LPCVD films exhibit an internal surface, which
decreases with the increase of deposition temperature. Therefore, the
number of Si-O-Si bridges near grain boundaries, which plausibly
vibrate in a way different than that of bridges within the core of the
grains, decreases with growth temperature. Hence, if oscillator 2 is
related to Si-O-Si bridges located near the grain boundaries, for the
TEOS LPCVD films, and the two interfaces, for the thermal SiO2
films ͑with the substrate and air͒, it is expected that a corresponding
decrease of its oscillator strength in favor of bridges within grains,
oscillator 1, as shown in Fig. 5. Additionally, the Si-O-Si angles near
grain boundaries and interfaces are expected to be larger than those
within grains and this has been shown previously ͑Fig. 8 of Ref. 3͒.
It is known that18,19 the excess charge ͑ionicity͒ on each atom in
SiO2 is dependent on the Si-O-Si angle and that the charge increases
Conclusions
This work is an attempt to investigate the double nature of FTIR
transmission minima of SiO2 films near 1100 cmϪ1. Evidence is
presented that the double nature of this minimum is due to the fact
that Si-O-Si bridges located within the bulk of the films vibrate
differently than those located near the surfaces ͑internal or external͒
and interfaces. This differentiation in the way bridges oscillate is
related to differentiations in the way Si-O-Si angles distribute, i.e.,
the way Si and O atoms are arranged near the bulk and near surfaces
and interfaces in SiO2 films.
NCSR Demokritos assisted in meeting the publication costs of this
article.
References
1. L. E. Katz, in VLSI Technology, 2nd ed., S. M. Sze, Editor, McGraw-Hill Int. Ed.,
New York ͑1988͒.
2. R. B. Fair, in ULSI Technology, C. Y. Chang and S. M. Sze, Editors, McGraw-Hill
Int. Ed., New York ͑1996͒.
3. D. Davazoglou and V. Em. Vamvakas, J. Electrochem. Soc., 150, F90 ͑2003͒.
4. G. Spitzer and D. A. Kleinman, Phys. Rev., 121, 1324 ͑1961͒.
5. K. Ishikawa, K. Suzuki, and S. Okamura, J. Appl. Phys., 88, 7150 ͑2000͒.
6. M. L. Naiman, C. T. Kirk, R. J. Aucoin, F. L. Terry, P. W. Wyatt, and S. D. Senturia,
J. Electrochem. Soc., 131, 637 ͑1984͒.
Figure 7. Evolution of the gaussian width ␦i with the growing temperature
for oscillators 1 and 2.
7. D. Davazoglou, V. Em. Vamvakas, and C. Vahlas, J. Electrochem. Soc., 145, 1310
͑1998͒.
Downloaded on 2015-03-13 to IP 129.128.216.34 address. Redistribution subject to ECS terms of use (see ecsdl.org/site/terms_use) unless CC License in place (see abstract).

Journal of The Electrochemical Society, 151 ͑5͒ F93-F97 ͑2004͒
F97
8. I. P. Herman, in Optical Diagnostics for Thin Film Processing, Academic Press
Inc., New York ͑1996͒.
15. J. T. Fitch, G. Lucovsky, E. Kobeda, and A. Irene, J. Vac. Sci. Technol. B, 7, 153
͑1989͒.
9. H. R. Philipp, in Properties of Silicon, EMIS Datareview RN ϭ 16133, p. 1019,
INSPEC, The Institution of Electrical Engineers, London ͑1987͒.
10. F. Seitz, Modern Theory of Solids, Chap. 17, McGraw-Hill, New York ͑1940͒.
11. T. Kirk, Phys. Rev. B, 38, 1255 ͑1988͒.
16. H. R. Philipp, in Properties of Silicon, EMIS Datareview RN ϭ 16133, p. 1019,
INSPEC, The Institution of Electrical Engineers, London ͑1987͒.
17. E. P. EerNisse, Appl. Phys. Lett., 30, 290 ͑1977͒.
18. S. T. Pantelides and W. A. Harrison, Phys. Rev. B, 13, 2667 ͑1976͒.
19. R. N. Noucho and A. Madhukar, Phys. Rev. B, 21, 1576 ͑1980͒.
20. V. Vamvakas, D. Davazoglou, and C. Vahlas, Microelectron. Reliab., 38, 265
͑1998͒.
12. I. Boyd, Appl. Phys. Lett., 51, 418 ͑1987͒.
13. D. Davazoglou, Thin Solid Films, 302, 204 ͑1997͒.
14. J. T. Fitch, C. H. Bjorkman, G. Lucovsky, F. H. Pollak, and X. Yin, J. Vac. Sci.
Technol. B, 7, 775 ͑1989͒.
Downloaded on 2015-03-13 to IP 129.128.216.34 address. Redistribution subject to ECS terms of use (see ecsdl.org/site/terms_use) unless CC License in place (see abstract).