surface hydride composition of plasma deposited hydrogenated amorphous silicon: in situ infrared...

Surface hydride composition of plasmadeposited hydrogenated amorphous silicon: in situ

infrared study of ion flux and temperature dependence

D.C. Marra a, W.M.M. Kessels b,*, M.C.M. van de Sanden b,K. Kashefizadeh a, E.S. Aydil a,*

a Department of Chemical Engineering, University of California Santa Barbara, Santa Barbara, CA 93106, USAb Department of Applied Physics, Eindhoven University of Technology, P.O. Box 513, 5600 MB Eindhoven, The Netherlands

Received 10 September 2002; accepted for publication 10 March 2003

Abstract

The surface silicon hydride composition of plasma deposited hydrogenated amorphous silicon (a-Si:H) films has

been investigated through surface sensitive in situ attenuated total reflection infrared spectroscopy. The fraction of SiHx

ðx ¼ 1; 2; 3Þ on the surface is reported for films deposited at substrate temperatures in the range 40–370 �C and a series

decomposition reaction set in which higher hydrides decompose into lower hydrides (SiH3 !SiH2 !SiH) for increasing

substrate temperature is proposed. Surface dangling bonds promote the decomposition reactions on a-Si:H as con-

cluded from experiments in which the incident ion flux during deposition is enhanced. A comparison is made with

results reported for hydrogenated crystalline silicon surfaces and the hydrogen coverage of the a-Si:H surface is dis-

cussed.

� 2003 Elsevier Science B.V. All rights reserved.

Keywords: Amorphous surfaces; Infrared absorption spectroscopy; Plasma processing; Silicon; Surface chemical reaction; Ion

bombardment; Growth

1. Introduction

Hydrogen plays an essential role in the prepa-

ration and performance of silicon-based devices in

the microelectronics and photovoltaics industry.

For example, in thin films of hydrogenated amor-

phous silicon (a-Si:H), which are used in thin film

transistors and thin film solar cells, the extent of

hydrogenation is a critical factor that determines

the material�s quality and stability. Furthermore,

atomic hydrogen and silane radicals SinHm play a

crucial role in low temperature a-Si:H film growth

by techniques such as plasma enhanced chemical

vapor deposition and hot wire chemical vapor de-position (HWCVD). In this respect, the chemical

state of the a-Si:H surface in terms of silicon

hydrides is of fundamental interest. First, the

variation of surface hydride composition with de-

position conditions can give information about the

*Corresponding authors. Tel.: +31-40-247-3477; fax: +31-

40-245-6442 (W.M.M. Kessels), Tel.: +1-805-893-8205; fax: +1-

805-893-4731 (E.S. Aydil).

E-mail addresses: w.m.m.kessels@tue.nl (W.M.M. Kessels),

aydil@engineering.ucsb.edu (E.S. Aydil).

0039-6028/03/$ - see front matter � 2003 Elsevier Science B.V. All rights reserved.

doi:10.1016/S0039-6028(03)00396-0

Surface Science 530 (2003) 1–16

www.elsevier.com/locate/susc

gas phase-surface interactions that result in film

growth and the surface chemical reactions that

convert the surface species into bulk film. Second,

the chemical state of the surface can affect and even

determine the interaction of gas phase species with

the surface.The formation and chemical stability of the

surface hydrides (SiHx) has been studied exten-

sively on crystalline silicon (c-Si) surfaces [1–9].

The stability of the di- and tri-hydrides depends on

the substrate temperature and on the presence of

surface dangling bonds [3,7]. A similar dependence

on dangling bond coverage [3,10] has been pro-

posed for the stability of silicon hydrides on po-rous silicon, polysilane silicon, and polycrystalline

silicon films [10–12]. For plasma deposited a-Si:H

films, the surface hydride coverage has been stud-

ied by Toyoshima et al. [13] and Aydil and co-

workers [14]. The results of Toyoshima et al.

showed that at low temperatures, the a-Si:H sur-

face is predominantly covered with the higher hy-

drides SiH3 and SiH2 while, as the temperature isincreased, the surface mono-hydride SiH becomes

increasingly more dominant [13]. Aydil and co-

workers have investigated the hydride coverage at

a substrate temperature of 230 �C and have ad-

dressed the influence of dangling bonds on the

stability of higher hydrides [14]. In this present

work, both the substrate temperature dependence

(in the range 40–370 �C) and the influence ofdangling bonds on the surface hydrides of plasma

deposited a-Si:H is addressed.

To investigate the surface hydrides on c-Si,

mass spectrometry-based techniques such as tem-

perature programmed desorption have been em-

ployed. However, the in situ investigations on

porous, polycrystalline, and amorphous silicon

have mostly relied on infrared spectroscopy. Thesilicon hydride composition on the surfaces and

near-surfaces of these silicon thin films have been

studied using techniques such as infrared phase

modulated ellipsometry [15,16], infrared reflection

absorption spectroscopy (IR-RAS) [13,17] and

IR-RAS combined with optical cavity substrates

[18,19]. Recently a number of research groups in-

volved in thin film growth have also adopted totalinternal reflection spectroscopy [11,14,20–24], a

method which has thoroughly been explored in the

field of surface science [25,26]. In this technique,

multiple reflections in an internal reflection ele-

ment (IRE) are used to greatly enhance the sensi-

tivity of the measurements [27]. Consequently

relatively small infrared absorptions can be mea-

sured by Fourier transform infrared spectroscopyand sub-monolayer sensitivity can be achieved.

The mode in which the absorptions in a film de-

posited on top of an IRE are measured depends on

the ratio of the refractive indices of the IRE and

vacuum and/or IRE and film: the absorptions are

either measured in attenuated total reflection

(ATR-FTIR) mode [25,28,29] or in multiple total

internal reflection (MTIR-FTIR) mode [27,30]. Byselecting an IRE with a refractive index close to

that of the film to be deposited (e.g., a c-Si or

GaAs IRE for a-Si:H) [30] in situ MTIR-FTIR is

perfectly suited to monitor the silicon hydride

bonding in the growing a-Si:H film in real time as

well as to delineate the hydrogen depth profile in

the as-deposited film. The corresponding surface

composition can be identified in the ATR-FTIRmode [14,31].

To identify the surface composition of a thin

film in the ATR-FTIR mode, the technique should

be made surface-specific. The reason for this is that

the absorption due to the surface hydrides is much

smaller than the absorptions due to the hydrides

in the bulk film. Toyoshima et al. have employed

isotope exchange techniques to gain informationon the surface hydrides in their IR-RAS experi-

ments [13]. An alternative method, which has been

employed for the results described in this article, is

the application of a brief Ar plasma pulse to de-

sorb the surface hydrides [14,23,24,31]. Informa-

tion on the fractional coverage of the hydrides can

then be obtained by comparing an infrared spec-

trum before and after this ion-induced desorption.In a previous publication, the validity of this

technique has been proven and it has also been

pointed out that the technique of ion-induced de-

sorption has some advantages over the isotope

exchange experiments [21].

In this study, the surface silicon hydride com-

position is studied for different substrate temper-

atures. In addition we investigated the influence ofdangling bonds on the decomposition of higher

hydrides. In situ ATR-FTIR in conjunction with

2 D.C. Marra et al. / Surface Science 530 (2003) 1–16

ion-induced desorption is employed to probe the

surface species on a-Si:H films deposited by an

inductively coupled plasma (ICP) from SiH4

diluted in Ar. Specifically, we determined the

fractional surface coverage in terms of silicon

mono-, di-, and tri-hydrides as a function of thesubstrate temperature and ion bombardment

during deposition. The results are compared to

those reported for crystalline and non-crystalline

silicon surfaces and the reactions that are likely to

be responsible for the observed surface composi-

tion are discussed. Furthermore, under select op-

erating conditions, knowledge of the surface

species provides insight into the primary growthprecursor(s).

2. Experiment

The experiments were conducted in situ on a-

Si:H films deposited by an ICP reactor equipped

with ATR-FTIR and spectroscopic ellipsometry(SE). A complete description of the infrared ap-

paratus and the ICP deposition chamber (base

pressure �10�8 mTorr) has been given in a previ-

ous publication [14]. The plasma is excited by ap-

plying radio frequency (rf) power at 13.56 MHz to

a 6 inch diameter planar coil placed on a quartz

window 8 inch above the substrate platen. The

chamber pressure is regulated using a throttlevalve and is independent of the gas flow. During

deposition, the reactor pressure was maintained at

40 mTorr with 50 sccm of SiH4 (1% in Ar) fed

from an injection ring surrounding the substrate

holder and an additional 50 sccm of Ar fed from

a gas injection ring directly below the coil. The

temperature of the stainless steel substrate elec-

trode is regulated by a feedback controller with a300 W ring heater and a thermocouple placed

immediately below the sample. The substrate

temperature ranged from 40 to 370 �C, and the rf

power to the ICP source was 25, 50, and 100 W at

each temperature studied. The electrode was left

floating for these experiments. Based on ion energy

distribution function measurements in a similar

reactor, we expect the ion energies to range from10–20 eV with a peak at approximately 15 eV

under these conditions [32,33].

The substrates were undoped GaAs IREs which

were double-side polished and 0.7 mm thick. The

IREs were 50 mm long and 10 mm wide and had

45� bevels at each of the short sides. The infrared

radiation, normally incident on one of the bevels,

is reflected approximately 35 times from the IRE/vacuum interface. The infrared radiation was un-

polarized but for the 45� internal incidence angle

used the Fresnel equations reveal that all compo-

nents of the surface hydrides are probed with ap-

proximately equal sensitivity [27]. The selection of

GaAs enables detection of the low frequency de-

formation modes of the higher hydrides. Further-

more, since the index of refraction of GaAs issimilar to that of the a-Si:H, the beam passes

through the growing film approximately 70 times

enhancing the IR signal. Although not reported

here, the bulk deposition was monitored in real

time using MTIR-FTIR. By probing the surface

at different deposition times, we determined an

adequate thickness of film such that the sur-

face species were independent of the underlyingsubstrate. In other words, these data are not

representative of the surface composition during

nucleation, but rather at steady state. To reach

steady state, films as thin as �50–100 nm are

needed, and we chose 6 min of deposition (�200

nm) for convenience.

Multiple passes of the infrared beam through

the IRE and through the growing film greatlyenhance the signal due to the surface hydrides,

however, it is still necessary to obtain surface

specificity, i.e., to decouple the surface modes from

the strong bulk signal. To isolate the surface sig-

nal, we expose the deposited film to a 100 W Ar

plasma for 10 s in 2 s intervals to remove the

surface hydrides by ion-bombardment induced

desorption [14,21,23,24,31]. The validity of thismethod of studying the surface coverage was pre-

viously established, and the effects of this brief

plasma pulse were investigated and the conditions

selected such that sputtering and evolution of H

from the bulk film was minimal [21]. As further

support, in a study of hydrogen and disilane ad-

sorption on ion-roughened Si(1 0 0), Gong et al.

showed that low energy ions (50 eV Arþ) could beused to sputter the silicon surface incurring little

damage to the bulk film [34].

D.C. Marra et al. / Surface Science 530 (2003) 1–16 3

The film thickness and index of refraction were

measured using in situ SE [35–37]. Atomic force

microscopy (AFM) was used to determine the

surface roughness of films grown under select

conditions to understand how the morphology

might affect the IR absorbance intensity. The ionflux to the surface was determined from Langmuir

probe measurements in Ar discharges without si-

lane to avoid film deposition on the probe. Since

we typically use 0.5 sccm of SiH4 in 99.5 sccm of

Ar, the measured ion flux in pure Ar will be ap-

proximately the same as that during deposition.

3. Experimental observations

3.1. Temperature dependence of surface composition

The infrared spectra displaying the SiH stretch-

ing modes of SiHx species on surfaces of a-Si:H

films deposited at 50 W and at several substrate

temperatures are shown in Fig. 1. Since the infra-red spectrum of the deposited film is used as the

reference, the surface species removed by ion-

bombardment induced desorption appear as a de-

crease in absorbance. The infrared assignments of

the surface silicon hydride stretching vibrational

frequencies are made based on those of H on c-Si

as described in previous publications [14,21]. Ac-

cording to the literature for Si–H on various c-Si

surface reconstructions [38–41], the surface mono-

hydride in different bonding environments is as-

sociated with vibrational frequencies ranging from2069 to 2100 cm�1. Since the amorphous surface is

microscopically rough without a well-defined pre-

ferred orientation, it is not surprising to find a

distribution of peaks associated with hydrogen in

various bonding environments. The wide span of

frequencies for the higher hydrides also reflects the

complexity of the amorphous surface, and ab-

sorption peaks in the range from 2101 to 2129cm�1 are ascribed to SiH2 on the surface, while

SiH3 is responsible for peaks appearing from 2130

to 2150 cm�1. The broad shape increasing in ab-

sorbance and centered at 1970 cm�1 is attributed

to vibrations of bond-centered hydrogen (Si–H–Si)

[42–44] in the bulk. The formation of these species

during Ar plasma treatment was consistently de-

tected and is interesting in its own right. Bond-centered H may be formed as Ar ions collide with

surface species. However, appearance of bond-

centered H is outside the scope of this article and

does not alter the conclusions of this work.

In Fig. 1, a clear shift of the SiHx absorption

band towards lower frequencies with increasing

deposition temperature is evident and indicative of

a shift from tri- and di-hydride dominated cover-age to mono-hydride coverage. The relative con-

centration of the various surface silicon hydrides

can be extracted by deconvoluting the various

stretching mode contributions to this band [14,21].

For example, the stretching region of the infrared

spectrum of the film deposited at 230 �C has been

deconvoluted using multiple narrow Gaussian

peaks as shown in Fig. 2. The individual absorp-tion peaks used to fit this band are shown as

dotted lines. The frequencies of these peaks were

carefully selected based on their consistent ap-

pearance in the spectra of films deposited under

various conditions and at different Ar sputtering

times. Although the spectra in the stretching re-

gion can also be fit reasonably well using several

broader peaks, we have chosen narrow (6–14cm�1) peaks because it is expected that the surface

hydride peaks have narrow absorption line shapes.

Fig. 1. Infrared spectra of the surface of a-Si:H films deposited

at varying substrate temperatures. The range of frequencies

corresponding to SiH, SiH2 and SiH3 stretching vibrations are

indicated with arrows. A shift from higher to lower hydrides

with increasing temperature is evident. The collection time for

each spectrum was approximately 7 min with a spectral reso-

lution of 4 cm�1.

4 D.C. Marra et al. / Surface Science 530 (2003) 1–16

Furthermore, only the narrow peaks in Fig. 2 can

capture the fine features which are sometimes on

the level of noise but which are consistently ob-

served for different sputtering times, plasma con-

ditions, and substrate temperatures. The hydride

coverage data can be extracted from either fittingprocedure with some loss in accuracy when the

broad peak method is employed. We report the

surface composition in terms of the fraction of

SiHx (x ¼ 1, 2, or 3) by summing the integrated

absorption intensities of the individual peaks that

have been assigned to a particular hydride species.

We divide by the number of Si–H bonds per silicon

hydride species such that we report the fraction ofSiHx bound as SiH, SiH2 or SiH3. In the absence

of data to the contrary, we assume that the

absorption cross sections of the surface mono-,

di- and tri-hydride species are the same and inde-

pendent of coverage and substrate temperature.

Although this assumption has important implica-

tions for the conclusions derived from the data in

this paper, the assumption has been made plausi-ble in Refs. [10,21] while the assumption is fur-

thermore supported by the analysis presented in

Section 4.3.

Using the above quantification procedure, the

surface coverage of silicon hydrides as a function

of substrate temperature is displayed in Fig. 3. As

can be concluded from the raw spectra of Fig. 1,

the fraction of SiH3 on the surface decreasesmonotonically with increasing temperature, while

the fraction of SiH on the surface increases. These

data are consistent with a thermally activated, se-

ries decomposition reaction from an SiH3 precur-

sor (as shown below), where SiH3 ! SiH2 ! SiH.

For such a process, one would expect a maximum

in the intermediate concentration, and, in fact, the

fraction of surface SiH2 undergoes a maximumwith increasing substrate temperature as seen in

Fig. 3. Similar temperature dependence was re-

ported by Toyoshima et al. on a-Si:H surfaces [13]

and has also been reported in the literature for

both SiH3 and disilane on c-Si [1–7]. In c-Si liter-

ature, the temperature range at which SiH3 and

SiH2 on c-Si became unstable was found to be

highly dependent on the hydrogen surface cover-age [3,7]. Thus, one would expect a similar effect

on the amorphous surface depending on the

prevalence of dangling bonds. In fact, Chiang et al.

prepared a-Si:H films that were mono-hydride-

terminated at temperatures as low as 200 �C [10],

while Marra et al. found SiH2 not only stable, but

dominant on the surface during plasma deposition

of a-Si:H at 230 �C [14]. This disparity is likelydue to the availability of dangling bonds during

Fig. 2. The surface infrared spectrum of the a-Si:H film de-

posited at 230 �C. The film was exposed to 10 s of a 100 W Ar

plasma to remove the surface hydrides and the as-deposited film

was used as the reference spectrum. The features of the

stretching region have been fit with 8 narrow Gaussian peaks

that can be attributed to SiH, SiH2, and SiH3 on the surface.

Fig. 3. The fraction of SiH (�), SiH2 (�) and SiH3 (M) on the

surface as a function of substrate temperature as determined

from Gaussian fitting of the infrared data. Values of the ab-

sorption cross section for the Si–H stretching mode are assumed

to be the same for the silicon hydrides. For clarity, only one

error bar is indicated and equal to one standard deviation of

three separate experiments; others are similar.

D.C. Marra et al. / Surface Science 530 (2003) 1–16 5

growth and is studied in more detail in the next

section.

3.2. Effect of ion bombardment

To test the hypothesis that the stability ofhigher hydrides depends on the dangling bond

density [3,7,14] and that the presence of dangling

bonds affects the surface coverage during plasma

deposition, we investigated the effect of plasma

power on the surface composition. At each tem-

perature of interest, the surface spectra were re-

corded for films deposited using different plasma

powers, i.e., 25, 50 and 100 W. Based on Langmuirprobe measurements, at 40 mTorr, we found that

the ion density increases linearly with increasing

plasma power as shown in Fig. 4. The corre-

sponding ion flux to the surface, shown on the

right ordinate of Fig. 4, was computed assuming

an electron temperature of 2 eV and assuming that

Arþ and ArHþ are the dominant ions [45]. One of

the principal effects of the ion bombardment dur-ing deposition is the physical sputtering of hy-

drogen and silicon hydrides to form dangling

bonds (see Section 4.1). Therefore, we varied the

ion flux during deposition by adjusting the plasma

power to study the influence of surface dangling

bonds on the surface composition. In Sections

4.1.1 and 4.1.3 we provide evidence that the prin-

cipal precursor to deposition is not substantiallyaffected by adjusting the plasma power.

The surface infrared spectra were collected and

deconvoluted as described in the preceding section.

Based on these assignments, the power dependence

of the surface hydride coverage for films deposited

at several different substrate temperatures is shown

in Fig. 5. At low temperature, Fig. 5(a), the surfacecomposition is independent of the plasma power,

hence the ion flux. However, as the temperature

increases, the effect of ion bombardment becomes

increasingly more significant until very high tem-

perature (370 �C) upon which the effect of ion flux

again becomes negligible. This clearly shows that

the temperature dependent decomposition reac-

tions of the higher hydrides are affected by thepresence of surface dangling bonds.

4. Discussion

4.1. Surface hydride composition and surface reac-

tions

4.1.1. Low temperature regime and growth precur-

sors

Since the gas phase composition in plasmas is

not significantly affected by the substrate temper-

ature, 1 the temperature dependence of the surface

composition, as shown in Fig. 3, can provide in-

formation about the reactions that occur on the

surface during a-Si:H growth and may even aidin identification of the dominant precursor. The

surface coverage at low temperature is most indi-

cative of precursors from the gas phase since the

thermal energy is insufficient to activate many of

the surface reactions that lead to dissociation and

H expulsion from the film. In fact, a-Si:H films

grown at low temperature typically contain more

SiH2 and SiH3 relative to films grown at highersubstrate temperatures. On the surface at 40 �C,we find predominantly SiH3 for all rf powers and

this is consistent with the belief that SiH3 is the

principal precursor to deposition irrespective of

the rf power used in the present work [46–50].

Fig. 4. Increase in ion density as a function of the plasma

power as determined by Langmuir probe measurements. The

corresponding ion flux is shown on the right ordinate.

1 The gas phase composition in plasmas is predominantly

determined by gas phase reactions which depend on the

geometry and type of plasma, gases used, flow ratios, etc.

6 D.C. Marra et al. / Surface Science 530 (2003) 1–16

Explicitly, the infrared spectrum of the surface of

the film deposited at 40 �C is shown in Fig. 6 with

the SiH3 stretching vibration centered at 2143

cm�1, and the corresponding symmetric and de-

generate deformation modes at 870 and 915 cm�1.

The presence of SiH2 is also evidenced by defor-

mation modes at 850 and 893 cm�1 and stretching

motions appearing as a shoulder at 2122 cm�1.

Quantitative analysis of the stretching region

shows that approximately 30% of the surface hy-

drides are bound as SiH2. A surface composed of

70% SiH3 with the balance SiH2 could conceivably

be produced by deposition from lower silane rad-

icals SiHn (n6 2) with sequential insertion of

atomic H. However, in a previous publication, we

showed that the rate of H insertion was much

Fig. 5. Fraction of silicon hydrides on the a-Si:H surface as function of power for several substrate temperatures, (a) 40 �C, (b) 160 �C,(c) 230 �C, and (d) 370 �C. Symbols: SiH (�), SiH2 (�), and SiH3 (M).

Fig. 6. Infrared spectrum of the surface of the a-Si:H film deposited at 40 �C. The (a) stretching and the (b) deformation regions

indicate SiH3 species in close proximity on the surface.

D.C. Marra et al. / Surface Science 530 (2003) 1–16 7

slower than the rate of SiH3 adsorption [14]. Spe-

cifically, H insertion could not account for the

overwhelming SiH3 signal at 40 �C. Thus, we

conclude that at least 70% of the Si brought onto

the surface is by SiH3 and SiH3 is the dominant

precursor to a-Si:H deposition in our reactor. In-cident SiH3 can adsorb onto surface dangling

bonds as follows, 2

SiH3ðgÞ þ dbðsÞ ! SiH3ðsÞ; ð1Þ

where db represents a dangling bond and the

subscripts (g) and (s) denote gas phase and ad-sorbed surface species, respectively. In addition,

SiH3 radicals have been shown to insert into

strained Si–Si bonds [19,51],

SiH3ðgÞ þ SiHx � SiHxðsÞ ! SiHx � SiH3 � SiHxðsÞ;

ð2Þwhere the surface SiH3 produced via reaction (2)

can remain over coordinated or create a dangling

bond by reducing its Si bonds. In this way, both

reactions (1) and (2) can be responsible for the

overwhelming presence of SiH3 on the surface at

room temperature. The fraction of SiH2 found on

the a-Si:H surface at 40 �C can be produced by

several reactions: the SiH2 can either originatefrom surface reactions involving adsorption of

SiH3 radicals (as shown below) or by adsorption

reactions of SiH2 radicals similar as reactions (1)

and (2) for SiH3 adsorption.

The low temperature a-Si:H surface that is

dominantly covered by SiH3 groups has not a

straightforward c-Si counterpart although SiH3

groups have frequently been observed on atomichydrogen or HF-prepared c-Si surfaces (both on

Si(1 0 0) and Si(1 1 1)) [4,39]. Si2H6 adsorption ex-

periments on Si(1 0 0) lead also directly to SiH3

groups by dissociative adsorption of the Si2H6

molecule [3,5,52,53] and Wang et al. [5] and Lub-

ben et al. [53] found even substantial fractions of

SiH3 groups on Si(1 0 0). Wang et al. found a

maximum surface SiH3 coverage of 33% of thesurface sites [5] while Lubben et al. reported a

maximal surface coverage of 43% in an atomic

layer epitaxy experiment [53]. This saturation in

coverage has been attributed to steric hindrance of

the SiH3 groups [53]. For Si(1 1 1) surfaces, how-

ever, higher SiH3 coverages have been reported:

Uram and Janssons found SiH3 as dominant spe-cies on Si(1 1 1)-(7 7) at low temperatures [41]

while Morita et al. reported a fully SiH3 termi-

nated Si(1 1 1) surface after HF preparation [54].

The absence of steric hindrance on the Si(1 1 1)

surface has been explained by minimization of the

hydrogen repulsion by rotation of the SiH3 groups

[54,55]. Consequently, a fractional SiH3 surface

coverage of 70% on low temperature a-Si:H mightcertainly be possible considering also the fact that

a-Si:H is disordered and less dense than c-Si.

4.1.2. Temperature dependence

As the substrate temperature increases, addi-

tional surface reactions become activated which

lead to the change in surface hydrides coverage as

shown in Fig. 3. Regarding these surface reactions,we can in principle distinguish between two types

of decomposition reactions. First, there might be

surface reactions that are initiated by the growth

process itself (i.e., by the SiH3 adsorption reac-

tions, etc.) and therefore rely on the dynamics of

the deposition process. Such reactions have been

hypothesized in previous publications [56–58] in

which it was suggested that the released energyfrom chemisorption of SiH3 radicals on surface

dangling bonds can be crucial for hydrogen elim-

ination from a-Si:H. Other reactions that do not

necessarily need the dynamic process of adsorption

during film growth have been studied on e.g.,

Si(1 0 0) surfaces [7] and ‘‘Si(1 0 0)-like’’ porous

silicon [59] by heating hydride covered surfaces to

different temperatures. Only these kind of reac-tions will be considered here. According to the

literature concerning the stability of higher hy-

drides on these surfaces, it has been found that tri-

hydride SiH3 surface groups can be stable up to

230 �C, the di-hydride SiH2 groups up to 370–430

�C, while the mono-hydride SiH groups are stable

up to 450–530 �C [7,59]. Our data for the SiH3 and

SiH2 groups on the a-Si:H surface show fairlygood agreement with this general behavior. For

the a-Si:H, the SiH3 decomposition starts already

2 This adsorption can either be direct adsorption process

from the gas phase onto the dangling bond or involve a weakly

absorbed state.

8 D.C. Marra et al. / Surface Science 530 (2003) 1–16

at lower temperatures but this is not inconsistent

with the c-Si data which has been obtained in

absence of dangling bonds [7]. It has been pro-

posed that when active sites are present on the

surface at elevated temperatures [3,5,7], decom-

position of SiH3 can proceed by the followingreaction pathway,

SiH3ðsÞ þ SiHxdbðsÞ ! SiH2dbðsÞ þ SiHxþ1ðsÞ ð3Þ

with x ¼ 0; 1. Reaction (3) can consequently lead

either to two SiH2 species or to one SiH2 and one

SiH.In a similar manner, the di-hydride is known to

decompose at high temperatures in the presence of

dangling bonds by [7]

SiH2ðsÞ þ SidbðsÞ ! SiHdbðsÞ þ SiHðsÞ: ð4Þ

Apart from these reactions involving surface

dangling bonds, Olander et al. proposed anotherreaction that can lead to loss of surface SiH3

groups [1]

SiH3ðsÞ þ SiHxðsÞ ! SiH4ðgÞ þ SiHx�1dbðsÞ þ dbðsÞ:

ð5ÞThis disproportion reaction, which leads to SiH4

desorption, has been observed by several groups

although the reported temperatures for the onset of

this reaction differ considerably. Gates et al. and

Lutterloh et al. reported an onset temperature of

375 and 330 �C, respectively, for Si(1 0 0) surfaces[2,7]. Cheng and Yates, however, reported an onset

temperature of approximately )73 �C for the sametype of surface [4], whereas Chiang et al. and Glass

et al. proposed a similar low temperature, albeit for

porous silicon films [10,12]. Chiang et al. however

also mentioned that they found SiH3 to be stable

up to 200 �C under some circumstances which they

attributed to the fact that the SiH3 was isolated

from its coreactants in this specific case.

The studies on disilane decomposition on c-Si inliterature have not cited reaction (5) [3,5], although

in these cases higher hydride decomposition might

be overwhelmed by the reactions (3) and (4) in-

volving dangling bonds. At this point, we are

therefore unable to say whether or not reaction (5)

is important for the decomposition reaction set in

which higher hydrides decompose into lower hy-

drides for increasing substrate temperatures as

shown in Fig. 3. 3 However, that dangling bonds

have an important role in the decomposition re-

actions can be concluded from the power depen-

dence experiments in Section 3.2 and this will

be discussed next.

4.1.3. Power dependence

Surface dangling bonds are assumed to be es-

sential for a-Si:H film growth and they are gener-

ally created during the deposition process by H

abstraction reactions from the surface. In the c-Si

literature, it has been shown that gas phase atomic

hydrogen H can abstract surface hydrogen by theEley–Rideal type of reaction

HðgÞ þ SiHxðsÞ ! SiHx�1dbðsÞ þH2ðgÞ; ð6Þ

in which the corresponding abstraction probability

is temperature independent [60]. Recently, the

same reaction scheme with almost zero activationenergy has been observed in experiments on a-Si:H

[61]. Apart from atomic hydrogen H, also gas

phase SiH3 radicals have been shown to abstract

surface hydrogen by

SiH3ðgÞ þ SiHxðsÞ ! SiHx�1dbðsÞ þ SiH4ðgÞ: ð7Þ

This reaction has been observed in moleculardynamics simulations of SiH3 radicals impinging

on an a-Si:H surface at substrate temperatures of

500 and 773 K [51]. The computed abstraction

probability is on the order of 5% while from the

relatively low activation barrier it is expected that

reaction (7) takes place at all temperatures.

Moreover, the simulations showed that reaction

(7) is the dominant mechanism of H removal/dangling bond formation in the absence of ion

bombardment, although also reaction (5) has been

observed during the high-temperature a-Si:H

growth simulations.

3 The occurrence of the reaction (5) with a temperature

dependent rate would lead to Si desorption and would therefore

have implications for the mass growth flux (deposition rate in

terms of Si atoms deposited) of plasma deposited a-Si:H. This

mass growth flux is generally observed to be temperature

independent which can be explained fairly simple when Si

desorption is absent as discussed in Ref. [58].

D.C. Marra et al. / Surface Science 530 (2003) 1–16 9

Apart from these abstraction reactions which

are most probably dominant during HWCVD of

a-Si:H, ions can play an important role in the

creation of surface dangling bonds in plasma de-

position. Plasma ions can create dangling bonds

by physical sputtering of the surface hydrogen via

ionðgÞ þ SiHxðsÞ ! SiHx�1dbðsÞ þHðgÞ; ð8Þ

or the silicon atom itself may be dislodged by the

ion,

ionðgÞ þ SiHxðsÞ ! dbðsÞ þ SiHxðgÞ; ð9Þ

where the majority of ions in our discharge are

Arþ and ArHþ [45].

By increasing the ion flux, hence the rate of

dangling bond generation, we were able to pro-

mote the hydride decomposition reactions and

accelerate the transition from higher hydride- to

mono-hydride-dominated surface coverage, asshown in Fig. 5. The surface coverage data at 160

�C (Fig. 5(b)), suggest that reaction (3) and (4)

require the high dangling bond density associated

with high ion exposure at 100 W. When the tem-

perature is increased to 230 �C (Fig. 5(c)), there is

some SiH on the surface even at 25 W indicating

that decomposition via reaction (4) has just begun

to proceed. However, as the power is increased to100 W, the surface composition changes from

primarily SiH2 coverage to mainly SiH indicating

the progression of the decomposition reactions. At

370 �C, the thermal energy is sufficient to enable

both reactions to proceed even at 25 W.

For all films, independent of surface tempera-

ture, more dangling bonds are created in the

presence of increased ion bombardment. Above160 �C any increase in the ion flux is sufficient to

create the dangling bonds required to promote the

decomposition reactions. These reactions result in

SiH dominance on the surface as shown in Fig.

5(c) and (d). In contrast, when the thermal energy

is insufficient, the surface composition remains

constant despite substantial increases in ion bom-

bardment and the corresponding dangling bonddensity on the surface. For example, at 40 �C, theSiH3-rich surface is preserved even when the ion

flux is increased by a factor of seven. Unfortu-

nately we were unable to measure the surface

composition at higher plasma power since further

increase in ion bombardment caused substantial

heating of the ATR crystal.

Fig. 5 also clearly establishes the fact that en-

hanced dissociation in the gas phase can not be

responsible for the increase in lower hydrides on

the surface at high power and high temperatures.Any change in dissociation in the discharge would

be independent of the substrate temperature, and

as such, the surface coverage would reflect this

change at all temperatures. Instead, the low tem-

perature SiH3 surface coverage is preserved over

the entire range of plasma power verifying that the

shift to lower hydrides at high temperature and

high power is a result of surface reactions.

4.2. Hydrogen and dangling bond surface coverage

In an earlier work we proposed that the stability

of higher hydrides on our a-Si:H surface at tem-

peratures as high as 230 �C was due to a lack of

dangling bonds on the surface [14]. In this present

work, higher hydrides are also found to be stableup to high temperatures, especially at low plasma

powers. Because surface dangling bonds are as-

sumed to be essential for a-Si:H film growth [58],

we will discuss two other observations that sup-

port a low dangling bond surface coverage for the

a-Si:H deposited in our reactor.

First of all, recent molecular dynamic simula-

tions of a-Si:H deposition from SiH3 precursor byRamalingam et al. showed that most surface Si

atoms are either fourfold or over-coordinated with

a few under-coordinated Si atoms scattered on the

surface [51]. However, these dangling bonds deter-

mined the reactivity of the surface and the reaction

mechanism of SiH3 adsorption and decomposition

on the surface. Furthermore, the silicon hydride

composition as a function of substrate temperaturefor these simulated films showed good agreement

with our experimental results for surfaces grown at

25 W with limited ion bombardment [51]: the film

growth simulated at 500 K has predominantly SiH2

on the surface while the film at 773 K has mainly

SiH. There did not seem to be a significant differ-

ence between the dangling bond coverage of the

films grown at the two temperatures despite the factthat the silicon hydride composition was not the

same.

10 D.C. Marra et al. / Surface Science 530 (2003) 1–16

Second, an experiment was performed in which

an a-Si:H film was deposited at 40 �C followed by

heating to 300 �C in an attempt to promote the

decomposition reactions (3) and (4). After heating

to 300 �C, we expected to find a surface composed

of fewer SiH3 but with SiH2 and especially manySiH. Instead, we found that the surface after heat-

ing was nearly identical to that of the deposited film

indicating that reactions (3) and (4) did not occur

on the a-Si:H surface. This strongly suggests that

the surface dangling bond density was very low.

However, the uncertainty arises from the pos-

sibility of the a-Si:H bulk providing an infinite

source of H. In this scenario, surface danglingbonds can be passivated by subsurface hydrogen

and removed surface modes are replenished by

reactions by subsurface H. The aforementioned

experiment would not be able to detect this H

transfer since the reference spectrum is collected

just prior to Ar sputtering, i.e., after H motion out

of the bulk. Bulk multiple internal reflection in-

frared spectra collected during the high-tempera-ture annealing step, reveal that H bound as SiHx

ðx ¼ 1; 2; 3Þ at internal surfaces preferentially de-

sorb upon heating. The signal from the surface

species is overwhelmed by this strong signal and

therefore we can not determine whether surface

has been desorbed as well. The vibrational fre-

quencies of H at internal surfaces are also surface-

like, adding to the difficulty. In addition, if thesurface modes were replenished as H released from

the subsurface diffused out of the film, the infrared

spectra would not detect any change in surface

bonding. In this experiment, also no evidence is

found for the disproportion reaction (5) which in

principle could also have taken place. A possible

explanation for the fact that reaction (5) has not

been observed is that the temperature for the onsetof reaction (5) is indeed over 300 �C [2,7].

As a final remark with respect to this experi-

ment, we mention that Fig. 5 suggests that at least

some dangling bonds are present on the a-Si:H

surface during plasma deposition at 25 W. We

want to stress here that we might need to distin-

guish the ‘‘static’’ situation of this particular ex-

periment from the ‘‘dynamic’’ situation of filmgrowth in Fig. 5. During the deposition process at

the different temperatures and at different plasma

powers there is a continuous production (and

passivation) of surface dangling bonds which can

play a role in the decomposition reactions (3) and

(4) during the film growth process itself.

Another approach to obtain information on the

dangling bond surface coverage is to determine thesurface coverage of hydrogen. This can be done by

considering the values of the total absorption in-

tensity of the surface hydrides for the different

temperatures. This procedure has been employed

by Toyoshima et al. in a related study [13], in

which they used IR-RAS and D/H isotope ex-

change technique to sample the surface species on

plasma deposited a-Si:H films from 25 to 500 �C.They reported a constant value for the total

surface hydride absorption intensity for tem-

peratures below 380 �C and attribute this obser-

vation to a fully hydrogen-terminated surface (i.e.,

full monolayer coverage). At higher temperature

(>380 �C), the integrated absorption intensity falls

off which they attributed to hydrogen desorption.

In a previous publication, we have commentedon these isotope exchange experiments and on the

specific applied procedure of determining the hy-

drogen surface coverage in their work [21]. We

have applied a similar procedure for our data and

the total integrated absorption intensity for the

surface as a function of temperature is shown in

Fig. 7. On the right ordinate, we have converted

the total integrated absorbance into concentrationusing the absorbance (

RAðvÞdv ¼ 2:27 10�3

cm�1 per reflection at the surface) for 1 monolayer

Fig. 7. The temperature dependence of the integrated absorp-

tion intensity of all surface silicon hydrides. The absolute sur-

face H concentration is shown on the right ordinate. All films

were deposited for 6 min.

D.C. Marra et al. / Surface Science 530 (2003) 1–16 11

(ML) of H in the form of silicon mono-hydride on

Si (1 1 1) reported by Jakob et al. [62] and as de-

scribed in a separate publication [21]. The total

hydrogen coverage decreases as a function of

substrate temperature and appears to be approxi-

mately constant above 230 �C.To address the question whether the obtained H

surface coverage corresponds to a full monolayer

on the a-Si:H surface, we need information on how

much hydrogen a fully covered a-Si:H surface

contains at the different substrate temperatures.

This information is not available for a-Si:H and

the only thing we can do is use information on c-Si

surfaces. For a fully SiH3 terminated surface wecan use the data available for the SiH3 covered

Si(1 1 1) surface [54] which yields a total hydrogen

concentration of 23.4 1014 cm�2. For a fully SiH2

covered surface we can use the data for the ideal

1 1 surface of Si(1 0 0) [63], yielding a total hy-

drogen concentration of 13.6 1014 cm�2. For a

mono-hydride covered surface we can choose

between the data for Si(1 0 0)-(2 1) and theSi(1 1 1)-(1 1) surface, giving 6.8 1014 cm�2 and

7.8 1014 cm�2, respectively. We can also assume

the 7 7 reconstructed Si(1 1 1) surface which

would give a hydrogen concentration of only

3 1014 cm�2. Although this information in com-

bination with Fig. 7 shows that the a-Si:H surface

is most probably considerably covered by hydro-

gen, it also reveals immediately the complicationof this comparison: the calculated surface coverage

in terms of monolayers depends fully on the as-

sumptions made with respect to the corresponding

c-Si surface. The assumption of, e.g., a Si(1 1 1)-

(1 1)-like surface for 370 �C would yield an

equivalent a-Si:H coverage of 0.72 ML, while the

assumption of a Si(1 1 1)-(7 7)-like surface would

yield 1.87 ML. The assumption of a SiH3 termi-nated Si(1 1 1)-like surface for 40 �C would yield

1.03 ML. Furthermore, we want to stress that it is

very plausible that the site density of the a-Si:H

surface is lower than that for the c-Si surface be-

cause the a-Si:H material is generally less dense.

Accurate quantitative information on the relative

hydrogen and dangling surface coverage (as ex-

tracted in Ref. [13]) can therefore not be obtainedfrom this kind of experiments [21].

4.3. Hydrogen surface coverage and surface rough-

ness

Finally, we want to address another parameter,

the surface roughness, that complicates the deter-mination of the hydrogen surface coverage in

terms of monolayer coverage. At the low temper-

atures the relatively high integrated Si–H stretch-

ing absorbance intensities in Fig. 7 are inflated by

the double and triple counting due to the di- and

tri-hydrides. Therefore, in Fig. 8 we display the

absorbance data of Fig. 7 corrected for the number

of Si–H bonds per SiHx ðx ¼ 1; 2; 3Þ (i.e., by divi-sion by the number of bonds). On the right ordi-

nate of the figure, the corresponding SiHx density

is shown. The figure shows that when the absor-

bance is corrected for the number of Si–H bonds

per SiHx, the surface silicon site density is con-

stant, within experimental error, as a function of

temperature in the range 160–370 �C. The larger

SiHx density at 40 �C could in principle be due toeither a higher hydrogen coverage at this temper-

ature or due to a larger absorption cross section

for the SiH3 stretch as compared with SiH and

SiH2. Another explanation is that the high Si–H

absorption intensity is due to a relatively high

surface roughness at 40 �C which would effectively

increase the surface area.

To obtain insight into this issue, we have deter-mined whether the surface roughness can account

for the enhanced absorbance at low temperature.

Fig. 8. The integrated absorbance from Fig. 7 has been cor-

rected for the number of Si–H bonds in each SiHx ðx ¼ 1; 2; 3Þ.The corresponding SiHx coverage is reported on the right or-

dinate.

12 D.C. Marra et al. / Surface Science 530 (2003) 1–16

To do so, we measured the surface roughness using

AFM. These measurements indicate that the film

deposited at 40 �C is indeed rougher than the films

grown at higher temperatures. The height images

obtained by AFM of the surface of films deposited

for 5min at 40 �C (a) and at 230 �C (b) are displayedin Fig. 9. According to spectroscopic ellipsometric

measurements, the film thickness is 190 and 163 �AAfor the films grown at 40 and 230 �C, respectively.Based on the AFMdata in Fig. 9, the film deposited

at 40 �C is rougher than the film deposited at 230 �Cand the corresponding RMS roughness values are

4.9 and 2.0 �AA respectively measured over an area of

1 lm 1 lm. The respective maximum featureheights for the images shown are 50 and 28 �AA.

To determine whether the measured variance in

the surface area is commensurate with the strong

absorbance at 40 �C, we prepared films with iden-

tical surface compositions but having varying de-

grees of roughness. To do so, we exploited the fact

that the surface roughness, as determined using

AFM, grows as the film thickness increases. The

AFM height image of a film grown at 230 �C for

15 min is shown in Fig. 9(c). This image should be

compared with that of the film grown at 230 �C for

5 min as shown in Fig. 9(b). Though visual inter-

pretation of the AFM images is quite adequate inthis case, the RMS roughness is plotted on the

right ordinate of Fig. 10 and shown to increase

with deposition time. The corresponding inte-

grated absorption intensity for the total hydrogen

coverage on the surface is shown on the left ordi-

nate. Films deposited in the range of 6–20 min

(240–640 �AA thick) had identical surface composi-

tions and only the total absorption intensitychanged. Thus, the total hydride removal (detected

as integrated absorption intensity) is found to

increase with increasing film thickness/surface

roughness.

It is useful to consider the integrated absorption

intensity and roughness of the film deposited at 40

�C in relation to those properties of the films

Fig. 9. AFM height images of the surface of a-Si:H deposited for (a) 5 min at 40 �C, (b) 5 min at 230 �C, and (c) 15 min at 230 �C. Thez scale of (c) has been expanded relative to the scales of the x and y directions to enhance clarity of the surface features. The scanned

area is 1 lm 1 lm.

D.C. Marra et al. / Surface Science 530 (2003) 1–16 13

grown for 5 and 15 min at 230 �C. Because the

fractional silicon hydride coverage of the two films

grown at the same temperature (230 �C) is identi-cal, the increase in integrated absorbance can be

attributed to the increase in surface area due toroughening during growth. To deposit a film at

230 �C having an RMS roughness on the order of

that of the surface deposited at 40 �C (4.9 �AA)

would require almost 10 min deposition. The in-

tegrated absorbance (corrected for the number of

Si–H bonds per SiHx ðx ¼ 1; 2; 3Þ) corresponding

to �10 min growth at 230 �C would be �0.055 as

indicated by the dotted line in Fig. 10. This value isalmost identical to the integrated absorbance (also

corrected for the number of Si–H bonds) measured

for the surface of the film grown for 5 min at 40 �C(0.054) and having the RMS roughness of 4.9 �AA.

Thus, we can conclude that the enhanced inte-

grated absorbance for the film deposited at 40 �Ccan be ascribed to the increased surface roughness.

This analysis also supports the assumption of ap-proximately equal absorption cross sections for

the various silicon hydrides.

It should be noted that the oscillator strengths

of the silicon hydride species on a-Si:H surfaces

are not known, and we have used values reported

by Jakob et al. for H on c-Si surfaces to determine

absolute values for surface H-coverage [62]. Fur-

thermore, although we have tested the sputteringmethod carefully for validity and reproducibility

[21], we can not be certain that we have removed

exactly the surface layer, no more, no less. These

uncertainties are inherent to detecting surface ad-

sorbates on hydrogenated amorphous silicon sur-

faces in situ. Assuming the oscillator strengths are

independent of temperature and coverage, we cancompare the relative hydride coverage as a func-

tion of temperature and deduce the reactions as

was done in this article.

5. Conclusions

Using in situ ATR-FTIR we report on the sili-

con hydride composition of plasma deposited

a-Si:H films over a range of substrate temperature

and ion flux. At low temperature and in the range

of ion flux investigated, the thermal energy is in-

sufficient to activate silicon hydride decomposition

reactions. At low temperature, the surface is pri-marily covered by SiH3, most likely in close

proximity to each other. Predominance of SiH3 on

the surface at low temperatures, combined with a

relatively slow H insertion rate [14], supports the

hypothesis that SiH3 is the dominant precursor for

a-Si:H deposition [46–50]. At higher temperatures

and in the presence of dangling bonds, decompo-

sition of SiH3 and SiH2 proceeds via reactions (3)and (4) resulting in fewer higher hydrides on the

surface. These decomposition reactions can be

accelerated by increasing the ion flux and gener-

ating more dangling bonds on the surface. In both

the low and high-temperature limits, however, the

dangling bond density, and hence, the ion bom-

bardment, had negligible effect on the surface

composition.We have addressed the hydrogen and dangling

bond surface coverage of the a-Si:H and concluded

that the dangling bond surface density is relatively

low for our films. This conclusion is not based on

the calculation of hydrogen surface coverage in

terms of fractional monolayer coverage and we

discussed the complications arising from this

method. Furthermore, it has been shown that theintegrated absorption intensity of the surface

hydrides increases with increasing deposition time,

hence with film roughness. The silicon hydride

composition, however, is independent of the sur-

Fig. 10. Total integrated absorption intensity for all surface

hydrides as a function of deposition time for films deposited at

230 �C. The RMS roughness as measured by AFM is shown on

the right ordinate and increases throughout the deposition. The

solid lines are included as a guide for the eye.

14 D.C. Marra et al. / Surface Science 530 (2003) 1–16

face roughness within the parameter range inves-

tigated in this study.

Acknowledgements

This research was supported by the NSF/DOE

Partnership for Basic Plasma Science and Engi-

neering (Award No. DMR 97-13280) and the

Camille and Henry Dreyfus Foundation. D.M.

acknowledges support from the National Science

Foundation pre-doctoral fellowship program. The

research of W.K. has been made possible by a

fellowship of the Royal Netherlands Academy ofArts and Sciences (KNAW). Many thanks are

due Prof. D. Maroudas, Dr. S. Ramalingam,

S. Agarwal, and S. Sriraman for insightful dis-

cussions.

References

[1] D.R. Olander, M. Balooch, J. Abrefah, W.J. Siekhaus,

J. Vac. Sci. Technol. B 5 (1987) 1404.

[2] S.M. Gates, R.R. Kunz, C.M. Greenlief, Surf. Sci. 207

(1989) 364.

[3] S.M. Gates, C.M. Greenlief, D.B. Beach, J. Chem. Phys. 93

(1990) 7493.

[4] C.C. Cheng, J.T. Yates, Phys. Rev. B 43 (1991) 4041.

[5] Y. Wang, M.J. Bronikowski, R.J. Hamers, Surf. Sci. 311

(1994) 64.

[6] S. Ramalingam, D. Maroudas, E.S. Aydil, S.P. Walch,

Surf. Sci. 418 (1998) L8.

[7] C. Lutterloh, M. Wicklein, A. Dinger, J. Biener, J.

K€uuppers, Surf. Sci. 498 (2002) 123.

[8] J.J. Boland, Phys. Rev. Lett. 65 (1990) 3325.

[9] A. Vittadini, A. Selloni, R. Car, M. Casarin, Phys. Rev. B

46 (1992) 4348.

[10] C.-M. Chiang, S.M. Gates, S.S. Lee, M. Kong, S.F. Bent,

J. Phys. Chem. B 101 (1997) 9537.

[11] S.S. Lee, M.J. Kong, S.F. Bent, C.-M. Chiang, S.M. Gates,

J. Phys. Chem. 100 (1996) 20015.

[12] J.A. Glass Jr., E.A. Wovchko, J.T. Yates Jr., Surf. Sci. 348

(1996) 325.

[13] Y. Toyoshima, K. Arai, A. Matsuda, K. Tanaka, J. Non-

Cryst. Solids 137–138 (1991) 765.

[14] D.C. Marra, E.A. Edelberg, R.L. Naone, E.S. Aydil,

J. Vac. Sci. Technol. A 16 (1998) 3199.

[15] N. Blayo, B. Drevillon, Appl. Phys. Lett. 59 (1991) 950.

[16] N. Blayo, B. Drevillon, J. Non-Cryst. Solids 137&138

(1991) 775.

[17] R. Nozawa, H. Takeda, M. Ito, M. Hori, T. Goto, J. Appl.

Phys. 85 (1999) 1172.

[18] M. Katiyar, Y.H. Yang, J.R. Abelson, J. Appl. Phys. 77

(1995) 6247.

[19] A. von Keudell, J.R. Abelson, Phys. Rev. B 59 (1999) 5791.

[20] S. Miyazaki, H. Shin, Y. Miyoshi, M. Hirose, Jpn. J. Appl.

Phys. 34 (1995) 787.

[21] W.M.M. Kessels, D.C. Marra, M.C.M. van de Sanden,

E.S. Aydil, J. Vac. Sci. Technol. A 20 (2002) 781.

[22] H. Fujiwara, Y. Toyoshima, M. Kondo, A. Matsuda,

Phys. Rev. B 60 (1999) 13598.

[23] H. Fujiwara, Y. Toyoshima, M. Kondo, A. Matsuda,

J. Non-Cryst. Solids 266 (2000) 38.

[24] H. Fujiwara, M. Kondo, A. Matsuda, Surf. Sci. 497 (2001)

333.

[25] Y.J. Chabal, Surf. Sci. Rep. 8 (1988) 211.

[26] Y.J. Chabal, in: J. Francis, M. Mirabella (Eds.), Internal

Reflection Spectroscopy: Theory and Applications, Marcel

Dekker, NY, 1993, p. 191.

[27] N.J. Harrick, Internal Reflection Spectroscopy, Wiley, NY,

1967.

[28] E.S. Aydil, R.A. Gottscho, Y.J. Chabal, Pure & Appl.

Chem. 66 (1994) 1381.

[29] E.S. Aydil, R.A. Gottscho, Solid State Technol. 40 (1997)

181.

[30] A.R. Godfrey, S.J. Ullal, L.B. Braly, E.A. Edelberg,

V. Vahedi, E.S. Aydil, Rev. Sci. Instrum. 72 (2001) 3260.

[31] D.C. Marra, E.A. Edelberg, R.L. Naone, E.S. Aydil, Appl.

Surf. Sci. 133 (1998) 148.

[32] E.A. Edelberg, A. Perry, N. Benjamin, E.S. Aydil, Rev. Sci.

Instrum. 70 (1999) 2689.

[33] E.A. Edelberg, A. Perry, N. Benjamin, E.S. Aydil, J. Vac.

Sci. Technol. A 17 (1999) 506.

[34] B. Gong, S. Jo, G. Hess, P. Parkinson, J.G. Ekerdt, J. Vac.

Sci. Technol. A 16 (1998) 1473.

[35] D.E. Aspnes, J. Vac. Sci. Technol. 18 (1981) 289.

[36] K. Vedam, J. McMarr, J. Narayan, Appl. Phys. Lett. 47

(1985) 339.

[37] B. Johs, D. Meyer, G. Cooney, H. Yao, P.G. Snyder, J.A.

Woollam, J. Edwards, G. Maracas, Mater. Res. Soc.

Symp. Proc. 216 (1991) 5989.

[38] Y.J. Chabal, K. Raghavachari, Phys. Rev. Lett. 53 (1984)

282.

[39] Y.J. Chabal, G.S. Higashi, K. Raghavachari, V.A. Bur-

rows, J. Vac. Sci. Technol. A 7 (1989) 2104.

[40] U. Jansson, K.J. Uram, J. Chem. Phys. 91 (1989) 7978.

[41] K.J. Uram, U. Jansson, J. Vac. Sci. Technol. B 7 (1989)

1176.

[42] L.C. Snyder, J.W. Moskowitz, S. Topiol, Phys. Rev. B 26

(1982) 6727.

[43] S. Agarwal, B. Hoex, M.C.M. van de Sanden, D. Marou-

das, E.S. Aydil, submitted for publication.

[44] S. Sriraman, S. Agarwal, E.S. Aydil, D. Maroudas, Nature

418 (2002) 62.

[45] S. Agarwal, private communication, 2001.

[46] R. Robertson, A. Gallagher, J. Appl. Phys. 59 (1986) 3402.

[47] N. Itabashi, N. Nishiwaki, M. Magane, S. Naito, T. Goto,

A. Matsuda, C. Yamada, E. Horita, Jpn. J. Appl. Phys. 29

(1990) L505.

D.C. Marra et al. / Surface Science 530 (2003) 1–16 15

[48] J.R. Abelson, Appl. Phys. A 56 (1993) 493.

[49] A. Matsuda, J. Vac. Sci. Technol. A 16 (1998) 365.

[50] W.M.M. Kessels, M.G.H. Boogaarts, J.P.M. Hoefnagels,

D.C. Schram, M.C.M. van de Sanden, J. Vac. Sci. Technol.

A 19 (2001) 1027.

[51] S. Ramalingam, S. Sriraman, E.S. Aydil, D. Maroudas,

J. Appl. Phys. 86 (1999) 2872.

[52] J.J. Boland, Phys. Rev. B 44 (1991) 1383.

[53] D. Lubben, T. Tsu, T.R. Bramblett, J.E. Greene, J. Vac.

Sci. Technol. A 9 (1991) 3003.

[54] Y. Morita, K. Miki, H. Tokumoto, Appl. Phys. Lett. 59

(1991) 1347.

[55] K.C. Pandey, T. Sakurai, H.D. Hagstrum, Phys. Rev. Lett.

35 (1975) 1728.

[56] G.H. Lin, J.R. Doyle, M. He, A. Gallagher, J. Appl. Phys.

64 (1988) 188.

[57] W.M.M. Kessels, R.J. Severens, M.C.M. van de San-

den, D.C. Schram, J. Non-Cryst. Solids 227–230 (1998)

133.

[58] W.M.M. Kessels, A.H.M. Smets, D.C. Marra, E.S. Aydil,

D.C. Schram, M.C.M. van de Sanden, Thin Solid Films

383 (2001) 154.

[59] P. Gupta, V.L. Colvin, S.M. George, Phys. Rev. B 37

(1988) 8234.

[60] W. Widdra, S.I. Li, R. Maboudian, G.A.D. Briggs, W.H.

Weinberg, Phys. Rev. Lett. 74 (1995) 2074.

[61] S. Agarwal, S. Sriraman, A. Takano, M.C.M. van de

Sanden, E.S. Aydil, D. Maroudas, Surf. Sci. 515 (2002)

L469.

[62] P. Jakob, P. Dumas, Y.J. Chabal, Appl. Phys. Lett. 59

(1991) 2968.

[63] J.E. Nothrup, Phys. Rev. B 44 (1991) 1419.

16 D.C. Marra et al. / Surface Science 530 (2003) 1–16