Etching, insertion, and abstraction reactions of atomic deuterium with amorphous silicon hydride films

Article abstract of DOI:10.1021/jp963717a

We report structural and kinetic studies of the reactions of hydride species in Si thin films with atomic deuterium (D_at). Infrared (IR) spectroscopy is used to obtain Si-H bonding information, and direct recoiling methods are used to measure reaction rates. Two kinds of films are prepared by filament-assisted growth from Si₂H₆ and are characterized by IR spectroscopy. A film containing only monohydride hydrogen is grown at 200 °C, and a polymer containing tri-, di-, and monohydride is grown at -110°C. Rates of H abstraction by D_at, and of D_at insertion into Si-Si bonds, are reported. The abstraction rate of H by D_at in both films is similar to the abstraction rate on H-terminated crystal Si surfaces. The insertion rate into Si-Si bonds in both films is about one-tenth the rate of abstraction. A qualitative study of the etching reaction of D_at with the polymeric film is reported, and a strong temperature dependence is observed.

Full text of DOI:10.1021/jp963717a

Evaluation Only. Created with Aspose.PDF. Copyright 2002-2021 Aspose Pty Ltd.
J. Phys. Chem. B 1997, 101, 9537-9547
9537
Etching, Insertion, and Abstraction Reactions of Atomic Deuterium with Amorphous Silicon
Hydride Films
C.-M. Chiang and S. M. Gates*
IBM T. J. Watson Research Center, Yorktown Hts., New York 10598
Szetsen S. Lee, M. Kong, and Stacey F. Bent
Department of Chemistry, New York UniVersity, New York, New York, 10003
ReceiVed: NoVember 6, 1996; In Final Form: September 11, 1997X
We report structural and kinetic studies of the reactions of hydride species in Si thin films with atomic deuterium
(Dat). Infrared (IR) spectroscopy is used to obtain Si-H bonding information, and direct recoiling methods
are used to measure reaction rates. Two kinds of films are prepared by filament-assisted growth from Si2H6
and are characterized by IR spectroscopy. A film containing only monohydride hydrogen is grown at 200
°C, and a polymer containing tri-, di-, and monohydride is grown at -110 °C. Rates of H abstraction by Dat,
and of Dat insertion into Si-Si bonds, are reported. The abstraction rate of H by Dat in both films is similar
to the abstraction rate on H-terminated crystal Si surfaces. The insertion rate into Si-Si bonds in both films
is about one-tenth the rate of abstraction. A qualitative study of the etching reaction of Dat with the polymeric
film is reported, and a strong temperature dependence is observed.
I. Introduction
understand these effects. At the conventional T (300 ( 50 °C),
if the H2/SiH4 ratio is increased to the range of 30-100,
crystalline films are deposited, rather than a-Si:H,1,16,18 another
intriguing effect of hydrogen addition. The films consist of
small crystallites, 30-500 Å in diameter. A few atomic % of
hydrogen is bonded at the crystallite surfaces, and the material
is known as “microcrystalline Si” (µc-Si),1 simply a polycrys-
talline film with very small grain size, and with H termination
of these grains.
Chemistry of atoms on semiconductor surfaces is an area
where plasma processing and fundamental studies of surface
reactions converge. The silicon + atomic hydrogen system is
a well-studied example of this convergence. Thin films of
amorphous hydrogenated Si (a-Si:H) are deposited by plasma-
enhanced (PE) chemical vapor deposition (CVD) for several
applications, and many properties of this useful material depend
on the amount of hydrogen bonded in the bulk of the film.1,2
The surface reactions of atomic H (Hat) during PE CVD film
growth influence the concentration and bonding of H in the
a-Si:H film.1 Cleaning of Si surfaces using H2 plasmas has
also been investigated.3-5 The mechanisms and dynamics
involved in the reactions of Hat on metal6-8 and Si9-14 surfaces
are fundamentally intriguing. Mounting evidence for direct,
Eley-Rideal (ER), abstraction reactions of Hat is of interest.9-14
Here, we further define the kinetics of different reaction
pathways for Hat reactions with Si films. We study the
monohydride on the surface of a polycrystalline thin film and
a mixture of the higher hydrides SiH2 and SiH3 in a heavily
hydrogenated film, which we call “polysilane”. Use of these
two films prepared in situ allows the reactivity of the three
hydrides to be studied under identical conditions. This com-
parison is not possible using single-crystal Si surfaces or a-Si:H
films. We address the question, What are the mechanisms by
which Hat improves the film quality in low-temperature PE CVD
of a-Si:H?
Alternatively, exposure of an a-Si:H film to Hat after film
growth causes related changes. Smaller Hat exposures decrease
the bulk H content.19 Larger Hat exposures convert the
amorphous film to µc-Si.20,21 Wronski, Collins, and co-workers
demonstrated that etching and crystallization both occur, with
the film thickness reduced by about 10 times in the µc-Si product
film compared to the initial a-Si:H film.20,21 The term “chemical
annealing” has been used to describe these Hat-induced effects
on a-Si:H.19,22 Thicker films of µc-Si are made by alternating
a-Si:H growth with Hat exposure using pulsed values.18
The fundamental reactions of Hat on crystal Si surfaces have
been investigated by many groups and reviewed by Boland.12
Chabal pioneered the use of multiple internal reflection Fourier
transform infrared absorption spectroscopy (hereafter called
“MIR-FTIR”) and has studied the H + Si system extensively.23-25
Much of our knowledge of the details of the mono-, di-, and
trihydride species comes from MIR-FTIR.23-29 Scanning tun-
neling microscopy (STM)12 and mass spectrometry30 have
revealed other details. Figure 1 shows four classes of reaction
that are known for atomic hydrogen or deuterium (Dat) interact-
ing with a Si surface or film. Using the symbol -Si for a
dangling bond, we write reactions 1-3 for addition, abstraction,
and insertion (see Figure 1).
Electronic quality a-Si:H is typically deposited using a pure
SiH4 plasma, on an amorphous substrate (glass) held at 270-
420 °C. The bandgap, photoconductivity, dark conductivity,
and stability of a-Si:H depend on the bulk H content, with
optimum properties found at 15-25 atomic % H.1,2,15 However,
a pure SiH4 plasma and 100 °C growth yields highly defective
material.15 Low growth temperatures (≈100 °C) require the
addition of H2 for high-quality material.16,17 Near 100 °C, the
dangling bond density (measured by electron spin resonance)
is decreased with an H2/SiH4 ratio of 1-10.17 We want to
Dat + -Si f D-Si
(1)
Dat + H-Si f H-D(g) + -Si
Dat + Si-Si f D-Si + -Si
(2)
(3)
* Corresponding author.
X Abstract published in AdVance ACS Abstracts, November 1, 1997.
In Figure 1, we show only one example of an etching reaction
S1089-5647(96)03717-0 CCC: $14.00 © 1997 American Chemical Society

Evaluation Only. Created with Aspose.PDF. Copyright 2002-2021 Aspose Pty Ltd.
9538 J. Phys. Chem. B, Vol. 101, No. 46, 1997
Chiang et al.
rate expression should be written for each. Reaction 2 has been
previously studied on monohydride-terminated Si(100)-(2×1)
and Si(111)-(7×7) by direct recoiling using thermally generated
atomic H and D9 and by other methods using hyperthermal H
atoms.11 Atomic H reactions with the disilane molecule have
been investigated both experimentally33 and using computational
methods.34 These results give information on the surface
abstraction and etching reactions (see Discussion section).
II. Experimental Details
MIR-FTIR studies were performed at N.Y.U. to obtain Si-H
bonding information. Kinetic studies were performed at IBM
using time-of-flight (TOF) direct recoiling (DR) spectroscopy,
or TOF DR, with mass separation of recoiled H+ and D+ ions
as described previously.9 IR characterization and kinetic results
are compared for two kinds of samples. One is a thin film of
polycrystalline Si prepared by growth at 200 °C on top of a
crystallized seed layer, which we call “polycrystalline mono-
hydride” (PMH). This designation is based on the IR spectrum,
which resembles that of the monohydride on single-crystal
surfaces (see Results section). The second sample is a film of
amorphous SiHx polymer prepared at -110 °C, which we
designate “polysilane” (PS). Both films are grown in situ on
top of the native oxide on a Si(100) sample or on top of Si
deposited on the native oxide. Before each experiment, the
sample was annealed at 600 °C, which makes a crystallized
seed layer. A fresh film was then grown for each experiment.
All films used here were prepared from Si2H6 by filament-
assisted (FA) growth using a Si2H6 pressure of 10-5 Torr with
a hot filament (temperature of ∼1550 °C, measured by an optical
pyrometer) near the sample.
Figure 1. Schematic illustration of four principal reaction pathways
for the interaction of D atoms with a silicon hydride surface: addition
(sticking), abstraction, insertion, and etching. Also shown are char-
acteristic expressions for the reaction rates of each process. See text
for discussion.
The growth flux resulting from Si2H6 plus the hot filament
has not been characterized directly. We believe that some Si2H6
is decomposed on the filament, producing a stable metal silicide
and Hat desorbing from the silicide surface. At 10-5 Torr
pressure, the mean free path of Hat exceeds 100 cm, so that
gas-phase reaction of Hat with Si2H6 is excluded. According
to this model, the growth flux consists primarily of Hat impinging
on intact, physisorbed Si2H6 molecules. For the case of FA PS
growth at -110 °C, we have demonstrated that the rate of PS
deposition increases with added D2 and that the growth rate is
roughly proportional to the D2 pressure.35 These data support
the model that at least one growth mechanism at -110 °C
involves D atoms initiating reaction of physisorbed Si2H6. At
200 °C substrate temperature (growth of the PMH film), the
surface residence time of Si2H6 is on the order of 30 µs,36 so
this same mechanism may be active for the PMH film case.
The growth mechanism will be discussed in more detail
elsewhere.35 In both chambers, Dat was generated by flowing
D2 through the chamber into a turbomolecular pump and
dissociating the gas on a tungsten filament at ∼1550 °C. The
absolute flux of Dat is unknown but is estimated as roughly 1%
of the D2 flux (ref 9, and references therein). The D2 pressure
was always 1 × 10-5 Torr, an ionization gauge reading, not
corrected for the relative ionization probability of D2, which is
0.42 with respect to N2.37
(ER mechanism) for simplicity, although both ER and Lang-
muir-Hinshelwood (LH) mechanisms are possible for etching
processes. During exposure of Si films and surfaces to Hat,
more than one of these reactions can occur simultaneously.
Also shown in Figure 1 are rate expressions. Each reaction
rate depends on the surface coverage or local concentration of
the reacting species which are -Si, H-Si, and Si-Si in
reactions 1, 2, and 3, respectively. These reactions may have
different rates in the “bulk” of a film compared to the film
surface, due to differences in the rate coefficient or the local
concentration (coverage). The rate coefficient of (1) is the
“sticking probability”, S, and is generally assumed to be near
unity. Schulze and Henzler measured S on the Si(111)-(7×7)
surface and obtained a value of 1, within 30%.31 The rate of
(2) is written in Figure 1 assuming a reaction with monohydride
(MH) species. We write the rate of (3) as kIFD{Si-Si}, where
{Si-Si} represents the coverage or bulk density of strained Si-
Si bonds, and FD the D atom flux. The rate coefficients kI will
be larger for strained Si-Si bonds compared to unstrained
bonds. In a-Si:H films, the disorder results in a distribution
(range) of Si-Si bond strengths with some weaker and some
stronger, as described by Street and co-workers.32 Weak Si-
Si bonds in a-Si:H are identified with the optical absorption at
energies just below the bandgap (the Urbach tail).15 These may
control the electronic properties of the a-Si:H. Another example
is a “backbond” between first- and second-layer Si atoms on a
reconstructed crystal Si surface.12 Modification of a-Si:H using
Hat preferentially removes the weakest Si-Si bonds, and our
experiments are most sensitive to the most reactive Si-Si bonds
(those with largest kI).
We note the difference in sampling depth of the MIR-FTIR
and TOF DR measurements. On a well-defined surface, the
two methods probe the same species. In both the PMH and PS
films, TOF DR probes the surface hydrides. The TOF DR
signal is primarily from the first layer, with some signal from
the second and third layers (due to the small incident angle of
3°). In the present MIR-FTIR experiments, the entire thin films
is probed rather than the surface layer, and the difference
between MIR-FTIR and TOF DR probe depth is most significant
for the PS film.
For simplicity, we write one example rate expression for an
ER etching reaction as kEFD{Si-SiH3}. If there are other
surface precursors to etching in addition to Si-SiH3, a similar

Evaluation Only. Created with Aspose.PDF. Copyright 2002-2021 Aspose Pty Ltd.
Etching, Insertion, and Abstraction Reactions
J. Phys. Chem. B, Vol. 101, No. 46, 1997 9539
Figure 3. Infrared absorbance spectra of silicon monohydride species.
Spectrum b from the polycrystalline monohydride (PMH) thin film is
characterized by two peaks, a symmetric stretch at 2083 cm-1 and an
antisymmetric stretch at 2102 cm-1; the fwhm of the antisymmetric
Figure 2. Top: Time-of-flight (TOF) recoiled ion data from the PMH
film. Bottom: Example TOF recoiled ion data from the PS thin film.
stretch mode is 15 cm-1
.
For comparison, spectrum a is from a
monolayer of monohydride species and has a fwhm of the antisymmetric
stretch mode of 7 cm-1
.
The MIR-FTIR instrument and the data collection procedure
have been described previously.29 Some of the present experi-
ments were performed with base pressures up to 10-7 Torr with
no discernable effects on the data. Although both sides of the
sample are probed by the IR, the PMH and PS films were grown
and reacted with D atoms only on the front face. Any IR
absorbance from the back face is removed by referencing to a
background spectrum acquired immediately prior to growth of
the film.
In the TOF DR studies, a 10 keV K+ ion beam is pulsed
at 50 kHz and impinged at a grazing incident angle (R) of
3° to the surface. Direct recoil (DR) particles are formed by a
binary collision of K+ with a surface atom. A fraction of these
are recoiled ions (RI), which are mass analyzed by TOF
using a “reflection” energy/time focusing electrostatic analyzer.9
The analyzer is located at a recoil angle, Φ, of 60°, measured
from the incident ion beam. The H+ and D+ recoil ion sig-
nals are well resolved, as shown in Figure 2. Each spectrum is
signal averaged for 30 s. The K+ beam current measured at
the crystal is 0.2 nA, which in 900 s corresponds to an ion dose
of ∼1-2% ML. To study isotopic exchange kinetics, the PMH
or PS film is exposed to Dat, and simultaneously TOF DR
spectra are acquired and stored until a set (20-40 spectra) is
complete. The relatiVe coverages of H (θH) and D (θD) are
determined by integrating the H+ and D+ ion signals (see Figure
2). Studies on well-defined surfaces show a linear relation
between coverage and the H+ or D+ signal.9 Here, we are
analyzing thin films (not well-defined surfaces), so the coverage
units are simply the H+ or D+ RI signals, with no absolute
calibration.
method, detailed in ref 29. Briefly, the film shown in Figure
3a was prepared by annealing a polysilane (PS) film to 600 °C
and using disilane exposure (without a hot filament) during
cooling from 600 °C to terminate the surface with monohydride.
The resulting material was first labeled an amorphous mono-
hydride film29 but is more correctly described as polycrystalline.
The assignment is based on comparison of the two peaks at
2098 and 2082 cm-1 to crystalline silicon surfaces. The peak
at 2098 cm-1 was assigned to the symmetric stretch of dimerized
monohydride species, similar to the Si(100)-(2×1):H surface,
while the peak at 2082 cm-1 corresponds to the Si-H stretching
frequency reported for the Si(111) surface.29 Figure 3 suggests
that most of the resulting surface is Si(100)-like, with a small
population of Si(111) facets.29 Recent work by Chabal and
co-workers on H-implanted silicon is in agreement with this
assignment.25 The material prepared by H-implant shows an
IR spectrum with similar frequencies, and transmission electron
microscopy data were used to identify the (100) and (111)
facets.25 The coverage of hydrogen at the surface of the film
in Figure 3a was estimated at one monolayer, in comparison
with infrared and thermal desorption measurements on a
Si(100)-(2×1):H surface.29
The spectrum from the thin film measured in Figure 3b has
analogous features at 2102 and 2083 cm-1, similar to spectrum
3a. This suggests that this film also contains both Si(100) and
Si(111) facets and is the basis for its designation as polycrys-
talline monohydride, PMH. The peak width (fwhm) of the PMH
film is about 15 cm-1, compared to 7 cm-1 for the 1 ML
monohydride. The integrated intensity of Figure 3b is about 4
times that of 3a, suggesting that the PMH film contains the
equivalent of several monolayers of hydrogen. We do not know
the absolute H content of the film, so an accurate thickness
cannot be calculated from the data of Figure 3.
Figure 2 illustrates examples of the TOF DR data used to
measure the relative coverage of H and D (θH and θD) in real
time. The H+ and D+ signals are well resolved, so θH and θD
can be measured from the peak areas without fitting or
deconvolution. The top panel of Figure 2 is for a PMH thin
film reacting with Dat. The lower curve was acquired from the
PMH film as prepared, and the upper curve was acquired after
900 s exposure to Dat at 200 °C.
III. Results
A. Polycrystalline Monohydride: Preparation and Reac-
tion. The monohydride films were grown by heating the sample
to 600 °C to crystallize the Si seed layer and then using FA
CVD from Si2H6 for 15-30 min at 200 °C. An MIR-FTIR
spectrum appears as spectrum in Figure 3, and the designation
“polycrystalline monohydride” (PMH) is based on this spectrum,
which shows a narrow linewidth and no features due to higher
hydrides. Also shown in Figure 3 is the IR absorbance spectrum
a of a monohydride-terminated silicon film grown by a different

Evaluation Only. Created with Aspose.PDF. Copyright 2002-2021 Aspose Pty Ltd.
9540 J. Phys. Chem. B, Vol. 101, No. 46, 1997
Chiang et al.
Figure 5. Infrared absorbance spectra of the silicon hydride stretching
region during growth of the polysilane film. Spectra are shown after
10, 20, and 30 min of filament-assisted growth from Si2H6 at 1 × 10-5
Torr pressure and a sample temperature of -110 °C. Also shown is
the spectrum obtained after 30 min Si2H6 exposure at -110 °C in the
absence of a hot filament, with a peak maximum at 2150 cm-1
.
Figure 4. Surface coverage of H, D, and H+D on the PMH thin film
versus D atom exposure (in seconds) measured by TOF recoiling. At
zero exposure, the film is terminated only with monohydride. Reaction
temperatures are 200, -10, and -50 °C.
substantial concentration of both di- and trihydride species, as
well as some monohydride.
The assignments for the silicon hydride stretching modes are
as follows. Trihydride absorption occurs near 2143 cm-1 both
in porous Si38 and in thick a-Si:H films.39 On single-crystal Si
The kinetics of isotopic exchange on the PMH thin film were
studied as a function of TS using Dat as the reactive atom.
Summaries of selected TOF DR experiments are shown in
Figure 4 for three different film temperatures. Here, θH (open
circles), θD (crosses), and the sum of (θH + θD) (diamonds)
are all plotted versus Dat exposure. Hereafter, the sum of (θH
+ θD) is refered to as the “total hydride coverage”. Data taken
at 100 and 27 °C are not shown, but are very similar to the
data collected at 200 °C. At 200 °C, the total hydride coverage
remains approximately constant throughout the experiment, to
within (10% (the scatter in the data). This observation is
important because it implies that higher deuterated species (such
as SiHD, SiD2, or SiD3) are unstable at 200 °C, and this is
consistent with IR results on monohydride-terminated Si films,
where no stable insertion products are seen at 200 °C.29 Below
room temperature the total hydride coverage increases, indicating
that stable insertion products are formed. At all temperatures
studied, θH decreases rapidly in the initial 200-500 s and then
remains about constant for the duration of the exposure to Dat.
At all temperatures, θD increases rapidly in the initial 200-
300 s. At high T (27-200 °C), θD stays constant after the initial
rapid increase. At lower T, a slowly changing component is
seen in θD, from roughly 300 to 800 s. We attribute this to
reaction 3 and the formation of stable insertion products, SiHDx
(x ) 1, 2). Analysis of the data of Figure 4 is discussed in
section IV.
surfaces the SiH3 stretch can be observed at 2154 cm-1 26,27
.
This SiH3 absorption feature is evident in our infrared spectrum
of disilane (Figure 5a). Silicon mono- and dihydride stretching
frequencies have not been consistently assigned, as discussed
by Glass et al.,38 but occur within the range 2000-2120 cm-1 39
.
Some of the discrepancy in assignment can be attributed to
different materials used in the various studies: single crystalline,
porous, or amorphous silicon. The consensus is that the lower
frequencies (2000-2090 cm-1) are due to monohydride species,
while the higher frequencies (2090-2120 cm-1) are assigned
to dihydride.23,38-41 This assignment is consistent with our
thermal annealing data (see Figure 6), where we see a shift to
lower frequencies as the surface temperature is increased, in
agreement with a shift to lower hydrides at higher tempera-
tures.29 Although the peak shape is unchanged with increasing
exposure, the total integrated absorbance over the Si-H
stretching region does not increase linearly with exposure time.
The integrated area of the IR absorbance after 30 min of growth
qualitatively suggests that the flims are 5-10 layers thick.
We compare the PS film of Figure 5 with films grown by
Hirose and co-workers using a SiH4 plasma at a similar
temperature.41,42 At -110 °C, Hirose deposited a polymer
consisting primarily of polysilane chains H3Si-(SiH2)n-SiH3,
based on the IR spectrum.41,42 Comparison of the IR spectra
of the Hirose polysilane film and the present PS reveals that
our filament-assisted process results in higher concentrations
of isolated monohydride species. We also compare the data at
the top of Figure 5 with IR spectra resulting from Si2H6 exposure
at -110 °C in the absence of a hot filament. The IR spectrum
shown at the bottom of Figure 5 is assigned to disilane
multilayers. The peak position is shifted to 2150 wavenumbers,
characteristic of SiH3 species, and is much less intense than
that produced in the presence of the hot filament. At still lower
temperatures (∼-173 °C), thick films of disilane could be
B. Polysilane: Preparation and Stability. To study the
reaction of Dat with the di- and trihydride species, we prepared
“polysilane” (PS). This polymeric material contains a mixture
of tri-, di-, and monohydride species and is grown at -110 °C.
Infrared spectra of a polysilane film deposited using a Si2H6
pressure of 1 × 10-5 Torr and the hot filament are shown as a
function of deposition time in Figure 5b-d. The broad,
asymmetric peak of the silicon hydride stretch has a maximum
at 2120 cm-1 with a fwhm of 75 cm-1
distribution of the SiH stretch indicates the presence of
. The frequency