Reactivity of intermetallic compounds: A solid state approach to direct reactions of silicon

Article abstract of DOI:10.1021/jp0130315

The present work is focused on a new approach to describe, quantify, and compare the reactivity of various transition metal silicide phases toward hydrogen chloride. Thermodynamic and kinetic parameters are obtained from isothermal calorimetric studies of these reactions. The reactivity of the silicide phases is discussed in terms of reaction start temperatures, rate constants, and apparent activation energies. Negative apparent activation energies are observed at low temperatures and are attributed to an initial stage of reaction where chlorine is chemisorbed and then incorporated into the silicide lattice near the surface. At a later time, a chlorine-containing reaction layer is formed having a composition and reactivity remarkably different from that of the bulk phase. On the basis of solid-state investigations, a diffusion model of the microscopic structure of these layers is presented, where a displacement of nickel atoms occurs followed by the occupation of nickel sites by chlorine. A model is suggested in which the electron level of the reaction layer is adjusted by the chlorine content of this layer, resulting in a electronic stabilization of silylenoide species at the surface. The model is applied to explain product distribution in the induction period during the direct reaction of silicon and methyl chloride.

Full text of DOI:10.1021/jp0130315

J. Phys. Chem. B 2002, 106, 5105-5117
5105
Reactivity of Intermetallic Compounds: A Solid State Approach to Direct Reactions of
Silicon
,†
‡
J o1 rg Acker* and Klaus Bohmhammel
Princeton UniVersity, Department of Chemistry, Princeton, New Jersey 08544, and Freiberg UniVersity of
Mining and Technology, Department of Physical Chemistry, D-09599 Freiberg, Germany
ReceiVed: August 7, 2001; In Final Form: February 16, 2002
The present work is focused on a new approach to describe, quantify, and compare the reactivity of various
transition metal silicide phases toward hydrogen chloride. Thermodynamic and kinetic parameters are obtained
from isothermal calorimetric studies of these reactions. The reactivity of the silicide phases is discussed in
terms of reaction start temperatures, rate constants, and apparent activation energies. Negative apparent
activation energies are observed at low temperatures and are attributed to an initial stage of reaction where
chlorine is chemisorbed and then incorporated into the silicide lattice near the surface. At a later time, a
chlorine-containing reaction layer is formed having a composition and reactivity remarkably different from
that of the bulk phase. On the basis of solid-state investigations, a diffusion model of the microscopic structure
of these layers is presented, where a displacement of nickel atoms occurs followed by the occupation of
nickel sites by chlorine. A model is suggested in which the electron level of the reaction layer is adjusted by
the chlorine content of this layer, resulting in a electronic stabilization of silylenoide species at the surface.
The model is applied to explain product distribution in the induction period during the direct reaction of
silicon and methyl chloride.
Introduction
in such a way that silicon is removed only to a small extent
from the studied silicide phases. Because of the small changes
The present work deals with the question of how a binary
alloy constituent A influences the reactivity of alloy constituent
B if B is selectively removed from the binary alloy by reaction
with a gaseous compound. In particular, this work is focused
on the heterogeneous reaction of hydrogen chloride and various
transition metal silicide phases where silicon is removed as
dichlorosilane, trichlorosilane, and silicon tetrachloride.
in composition during the reaction, the measured kinetic and
thermodynamic reaction parameters can be attributed to the pure
silicide phase. This serves as the necessary precondition for
correlation of the reaction parameters to thermodynamic func-
tions of pure phases and to the discussion of differences between
various metals or different metal-to-silicon ratios of pure phases
quantitatively.
Our approach to investigate the reactivity of transition metal
silicides is split into two levels. On a macroscopic level, the
reactivity of the studied silicide phases is described in terms of
reaction start temperatures, rate constants, activation energies,
and product distribution in relation to thermodynamic functions
of pure silicide phases. On a microscopic level, processes near
and on the silicide surface are discussed in light of diffusion
phenomena and surface reactions.
Heterogeneous reactions of intermetallic compounds show
characteristic features for heterogeneous bimetallic catalysis
where the concepts of surface segregation of one alloy constitu-
ent at the surface, spill-over processes, and altered adsorption
and reaction properties due to alloying the constituents were
1
introduced. Furthermore, features for oxidation of alloys are
also appearing, introduced as concepts of solid metal oxide layer
formation, ion conductivity of the reaction layer, internal
oxidation of the less noble alloy constituent, and solubility of
oxygen in one or both constituents.2
In the following section, the background and previous studies
are presented. The two major processes of industrial silicon
chemistry are the reactions of silicon with either methyl chloride
or hydrogen chloride, which are known as direct reactions of
When dealing with transition silicides, it is important to note
that silicon can be removed from the solid phase in the form of
various chlorosilanes. The silane chlorine content depends on
the transition metal that is used as well as on the metal-to-silicon
ratio of the reacting solid phase. The depletion of silicon changes
the solid-phase composition, which alters the reactivity of the
solid phase and the product distribution. As a consequence, a
thermodynamically defined starting point and a defined ending
point are urgent conditions for reliable and comparable results.
Pure silicide phases are used as staring materials to fulfill the
first demand. Also, the reaction conditions have to be chosen
3
-7
silicon.
They are unique examples of how the reactivity of
silicon can be controlled by the addition of various metals. In
direct synthesis, methyl chloride is passed over a silicon-copper
mixture at temperatures between 250 and 320 °C. The main
product is dimethyldichlorosilane, which is the basic compound
for silicone production. The second reaction is the hydrochlo-
rination reaction where technical grade silicon, containing
numerous metallic and nonmetallic impurities, reacts with
hydrogen chloride. Trichlorosilane is the major product and is
used for manufacturing solar grade and semiconductor grade
silicon.
*
Author to whom correspondence should be addressed. New address:
Leibniz Institute for Solid State and Materials Research Dresden, P.O. Box
2
7 01 16, D-01171 Dresden, Germany. Email: j.acker@ifw-dresden.de.
†
‡
A third reaction belongs to the same class of metal-controlled
reactions: the silicide-assisted conversion of silicon tetrachloride
Princeton University.
Freiberg University of Mining and Technology.
1
0.1021/jp0130315 CCC: $22.00 © 2002 American Chemical Society
Published on Web 04/18/2002

5106 J. Phys. Chem. B, Vol. 106, No. 19, 2002
Acker and Bohmhammel
TABLE 1: Direct Synthesis Metal Catalysts and Product
Distributionsa
TABLE 3: Free Energies of Formation (∆
f
G) of Metal
16
Chlorides
∆ G (kJ mol-1)
compound
CrCl
ZnCl2a
∆ G (kJ mol-1)
estimated
compound
f
f
product
product
M-Cl bond
main
product
distribution secondary distribution
energy
PtCl
2
-93.248
-173.810
-138.089
-259.151
-269.661
-286.233
-302.354
-314.122
2
-356.209
-369.362
-440.489
-592.086
-748.123
-927.754
-978.066
-995.389
catalyst
(mol %)
product
(mol %)
(kJ mol-1)
CuCl2
CuCl
MnCl
MgCl
2
Pb
Cu
Sn
Fe
Si
CH
(CH
3
HSiCl
SiCl
2
75
85
85
85
70
60
HSiCl
3
10
10
10
10
20
15
272
326
356
372
381
431
NiCl
CoCl
2
2
3
)
2
2
CH
SiCl
(CH
CH
(CH
3
SiCl
3
4
CH
CH
3
SiCl
SiCl
3
2
CaCl
YCl
CeCl
LaCl3
2
a
3
3
3
)
2
SiCl
HSiCl
SiCl
2
SnCl
FeCl
PbCl2a
2
3
HSiCl
(CH
3
3
2
2
3
Ca
3
)
3
SiCl
3
)
2
2
a
This table is taken from Falconer et al.12
a
No silicide phases formed.
TABLE 2: Metals and Product Distributions in the
Hydrodechlorination Reaction
only the metals that are able to form silicide phases, the results
in Tables 1 and 2 are similar in the manner of instigating the
attachment of hydrogen atoms or methyl groups to silicon.
Copper, the most effective metal to form products of industrial
need, is one exception. The free enthalpies of CuCl and CuCl2
formation are smaller than that for PtCl2 formation and larger
than that for NiCl2 formation. Platinum and nickel form
trichlorosilane, but copper forms dichlorosilane. The second
exception is found between magnesium silicide and calcium
silicide in both reactions. In direct synthesis, calcium does not
form Si(CH3)4; however, magnesium in the hydrodechlorination
reaction produces SiH4. The silicide phases of these metals
should exhibit a similar ionicity. The differences in the reactivity
between calcium and magnesium in the direct synthesis and in
the hydrodechlorination reaction, respectively, can be explained
by differences in acid-base strength between methyl chloride
and the silicon tetrachloride/hydrogen mixture. A comparable
scale of oxidation strength is known from the treatment of
calcium silicides with hydrochloric acid at different concentra-
tions and temperatures that produces different layered silicon
metals
solid products
gaseous products
HSiCl
SiCl
HSiCl
SiH and HSiCl
HSiCl
ref
Fe, Co, Ni silicide phases
Cu
Mn, Cr
Mg
La, Y
3
9, 10, 13
9, 10, 13
13
silicide phases and CuCl
H
2
2 3
and HSiCl
3
a
metal chlorides
MgCl
2
4
3
14
15
nanocrystalline Si
and metal chlorides
3
a
Volatile under reaction conditions.
into trichlorosilane (eqs 1 and 2). Here, a gaseous mixture of
silicon tetrachloride and hydrogen is passed over a desired
transition metal.8-10 During an induction period, silicon tetra-
chloride is almost completely consumed by the metal under the
formation of the silicide phase, which is thermodynamically
stable at given reaction temperature (eq 1).
xSiCl + 2xH + M f MSi + 4xHCl
(1)
4
2
x
1
7-19
compounds.
Once a silicide phase has been established, the conversion of
silicon tetrachloride into trichlorosilane starts (eq 2).
From the overwhelming number of papers dealing with direct
reactions of silicon, four topics of research on the mechanism
of direct reactions have to be considered for their importance
SiCl + H f HSiCl + HCl
(2)
4
2
3
2
0
to the present study. Lewis et al. found evidence of inter-
mediacy of a CH3SiCl silylene species in direct synthesis, a
The most remarkable feature of this reaction is that silicon atoms
of the silicide phase are continuously depleted by reaction with
hydrogen chloride (yielding trichlorosilane) and continuously
restored by silicon deposition from the gas phase (by the reaction
of silicon tetrachloride and hydrogen) at the same time. The
mechanism in principle is the same as that described for the
catalytic activity of RuO2(110) in a CO oxidation reaction in
which oxygen lattice atoms reacting with CO are restored by
the gas phase.11
9
,10,21,22
finding that was later confirmed by several groups.
23
24
Falconer et al. and Lewis et al. discovered silicon segregation
at the surface of Cu3Si samples after exposing the samples to
methyl chloride. Zinc, an effective promoter in direct synthesis
and an alloy in copper silicide samples, was found to enhance
silicon segregation. Therefore, the silicon-enriched surface was
24-26
considered to be the active phase in direct synthesis.
27-29
Falconer et al.
studied the induction period of direct
Tables 1 and 2 illustrate the wide range of product distribution
for a variety of metals during direct synthesis and hydrodechlo-
rination reactions, respectively. Table 1 is taken from Falconer’s
work of 1985.12
synthesis extensively and established a detailed view of the
reactive sites, which are always related to the presence of
chlorine at the surface. Auger electron spectroscopy studies on
copper-silicon alloy samples revealed the existence of Si-Cl
(and Si-C) bonds, which is coincidental to a weakening of the
As the most recent results, the influence of a metal on the
product distribution of the hydrodechlorination reaction should
1
2
Cu-Si bonds. Finally, Roewer and Walter discovered the
presence of chlorine in the bulk of transition metal silicide
phases, which were active in the conversion of silicon tetra-
12
be discussed in more detail. Similar to Falconer, who correlated
the obtained product distributions to metal-chlorine bond
strength, we correlate the product distribution of the hydro-
dechlorination reaction to the free enthalpy (Gibbs function) of
metal chloride formation (Table 3).
9
,10
chloride to trichlorosilane.
These results suggested that chlorine bound on the surface
or incorporated into a silicide bulk phase is necessary to make
the solid phase reactive. Therefore, the nickel-silicon-chlorine
system was extensively studied by means of a model reaction.
It was found that the reaction of a nickel silicide phase and
nickel(II) chloride (eq 3) having a water content of about 1 wt
% is a convenient model for the reaction of hydrogen chloride
With increasing negative free enthalpy of metal chloride
formation, the ability to substitute the chlorine atoms of silicon
tetrachloride by hydrogen atoms and even by silicon atoms
1
5
increases. La and Y were found to form silicon nanoclusters;
metals such as Cr or Mn form stable, but under our reaction
13
30
conditions, volatile chlorides; and Fe, Co, Ni, and Pt belong
and a nickel silicide phase. The reaction has to proceed in a
to one group9
,10,13
and form trichlorosilane. Taking into account
closed reaction system to retain all gaseous products involved

Reactivity of Intermetallic Compounds
J. Phys. Chem. B, Vol. 106, No. 19, 2002 5107
TABLE 4: Transition Metal Silicide Chlorides and Their Structural Featuresa
compound
ionic formulation
structural features
ref
33
33
34, 35
34, 35
34, 35
35
3
3
+
-
4-
-
-
Gd
Gd
4
I
I
5
Si
Si
(Gd
)
3
4
(I )
5
Si ‚3e
(Gd +)
(I )
-
Si ‚2e
4-
3
3
3
3
+
-
-
-
-
MISi, M ) La, Ce, Pr
(La )(I ) [Si(3b) ]‚1e
1.0 e /Si: Si
6
rings
3
3
3
+
+
+
-
2-
2-
-
-
-
La
5
La
4
La
3
La
6
I
I
3
Si
Si
5
(La
(La
(La
)
)
)
5
4
3
6
(I )
3
[Si(2b)
[Si(2b)
[Si(2b)
]
]
4
[Si(3b) ]‚3e
- -
1.80 e /Si: Si22 rings
-
-
3
4
(I )
3
2
[Si(3b) ]
2
‚3e
6
1.50 e /Si: Si - and Si14 rings
-
2-
-
-
Cl
Br
2
Si
3
(Cl )
2
]
]
3
‚1e
5
2.00 e /Si: Si zigzag chains
Si
7
(La3+)
(Br )
-
3
[Si(2b)
2-
[Si(3b) ]
-
2
‚3e-
1.71 e /Si: Si12 rings
-
35
3
e-/Si ) the number of electrons formally transferred to silicon.
for subsequent reactions.
mixed with silicon powder (99.9%) in certain ratios. To
compensate for a loss of silicon by evaporation, an extra 1-2%
of silicon was added (the values are based on our experience
of our furnace system). The mixtures were heated under a
NiSi + nNiCl f Ni1+nSim-0.5n(Cl) + 0.5nSiCl4 (3)
m
2
This reaction was studied in detail by thermokinetic measure-
ments, in situ X-ray powder diffraction measurements, and in
situ IR spectroscopy gas-phase analysis.13,30,31 By dehydratation
of nickel(II) chloride (containing 1 wt % water), a very low
hydrogen chloride partial pressure is built up, and the reaction
proceeds less rigorously than it does under a hydrogen chloride
partial pressure of 1 bar. At temperatures below 350 °C,
hydrogen chloride chemisorbs on the silicide phase. Above 350
-
1
-1
constant argon flow (25 L h ) at a rate of 10 K min to a
temperature of about 50 K above the melting point of the metal
(TF(Ni) ) 1453 °C, TF(Fe) ) 1535 °C, TF(Si) ) 1410 °C). After
holding samples at this temperature for 1 h, they were slowly
-
1
cooled to 900 °C (5 K min ) and held for 7 h. After cooling
to room temperature, the reguli were annealed in sealed,
-
5
evacuated quartz ampules (10 Torr) for 6 weeks at 900 °C.
Analysis by X-ray powder diffraction (Siemens D5000) gave a
diffraction pattern of the pure silicide phases. Slices were
prepared from all samples and then polished and investigated
by SEM-EDX. The deviations from the desired composition
were below 0.2 at. %. Cu3Si was prepared by arc melting of
semiconductor grade silicon (roughly crushed wafers) and
copper (99.999%). The sample was melted and crushed five
times and annealed in an evacuated ampule for 3 weeks at 770
°C. Phase purity was confirmed by X-ray powder diffraction.
A copper-silicon-zinc alloy with a composition of Cu3Si +
1.5 at. % Zn (Zn 99.98%) was prepared in the same manner.
°
C, the chemical reaction between the silicide phase and
chemisorbed hydrogen chloride starts, and silicon tetrachloride,
trichlorosilane, and hydrogen are formed. By a parallel process,
in the reaction system, hydrogen chloride is continuously
regenerated by the reduction of nickel(II) chloride by hydrogen
or trichlorosilane.3 Silicide phases were prepared by reaction
0,31
3
. In the solid bulk phase, chlorine was present in higher
concentrations throughout the reactions than it was in the
equilibrated solid products. The chlorine content in so-prepared
nickel silicide phases depends on the metal-to-silicon ratio of
the formed product silicide phase, on the silicon tetrachloride
partial pressure during their synthesis, and on the reaction
Calorimetric Measurements. Isothermal thermokinetic mea-
surements were carried out in a differential scanning calorimeter
30
temperature. With increasing chlorine content, an enlargement
30
of the silicide phase unit cell was observed. On the basis of
the analytically determined chlorine contents, a thermodynamic
model of the dilute solution of chlorine in nickel silicide phases
based on nonspecific chlorine-silicide interaction was proposed,
and the enthalpy and entropy of the chlorine intake processes
(DSC-111, Setaram), a Tian-Calvet-type calorimeter. The
reaction proceeds in a stainless steel flow cell having a gas inlet
and outlet, also made from stainless steel. The sample cell
(volume 0.125 mL) filled with a grained silicide sample (particle
sizes between 50 and 300 µm) and the empty reference cell are
both connected to a gas dosing system. A typical run starts by
passing a constant argon flow through the cells (flow rates vary
between 90 and 105 mL per hour) while heating them to the
desired temperature. After establishing a stable baseline under
isothermal conditions, a sample valve, taken from a gas
chromatograph, is opened, and the argon flow carries a constant
volume of dry hydrogen chloride gas (between 1 and 8 mL)
into the sample cell. For each run, a new silicide sample was
used.
3
2
were calculated.
Yet unclear is the state of chlorine bonding in these silicide
phases. It is definitely excluded that chlorine is present as a
metal chloride. Moessbauer spectroscopic studies of chlorine-
containing iron silicides showed that the presence of chlorine
in these compounds disturbs the crystal field and slightly disturbs
the lattice structure. But iron chlorides were never found. To
avoid any confusion about the bonding state of chlorine in these
silicide phases, they are referred to as “chlorine-containing
silicide phases”.
Their most remarkably chemical property is their thermal
decomposition starting above 200 °C, which forms breakdown
products of elemental nickel, highly disordered Ni3Si or Ni5Si2
phases, and silicon tetrachloride.30,31
The thermal decomposition reveals the totally changed
chemical properties of chlorine-containing silicide phases
compared to those of pure silicide phases. Therefore, chlorine-
containing silicide phases can be considered to be a distinct
class of compounds. Meanwhile, numerous compounds consist-
ing of a rare earth metal, silicon, and chlorine were synthesized
3
0
Blank tests were performed to determine the heat and rate of
passing hydrogen chloride into empty sample cells. The obtained
peaks for the blank tests have totally different shapes compared
to those for the silicide-filled cells. Depending on the choice of
silicide phase and the sample mass, the heat of reaction for blank
tests is between 1 and 6% of the heat of reaction for the silicides
(Figure 1). Furthermore, it was found that the rate constant for
the reaction of silicide and hydrogen chloride does not depend
on the volume of hydrogen chloride gas that is introduced.
Solid-State Investigations. Polished nickel silicide slices
reacted with hydrogen chloride under various partial pressures
and temperatures. To study the reaction on the sample surface
under mild conditions (i.e., under very low hydrogen chloride
partial pressure), silicide samples were embedded in NiCl2
(having 1 wt % H2O) and heated to a preset temperature. After
the reactions, the samples were broken into two pieces, and their
3
3-35
and designated as transition metal silicide chlorides.
summary of these compounds is given in Table 4.
A
Experimental Section
Preparation of the Silicides. For the preparation of nickel
and iron silicides, nickel (99.9%) or iron powders (99.8%) were

5108 J. Phys. Chem. B, Vol. 106, No. 19, 2002
Acker and Bohmhammel
Figure 2. Arrhenius plot for the reactions of various metal silicide
phases and hydrogen chloride.
Figure 1. Fit of a heat-flow curve (Cu Si + HCl) using the Avrami-
3
Erofeyev equation (left) and rate constants and heat of reaction as
function of inverse temperature (right).
point, the applied thermokinetic model fails to evaluate the heat
flow curves. A deconvolution of the signals fails because of
the high interference and the lack of sufficient chemical
background to establish a model, which describes the proceeding
reactions or processes causing these structured signals.
cross sections were either polished or left as received. The
samples were studied by scanning microscopy and EDX analysis
(
(
JEOL JSM 6400, EDX: Noran) and X-ray powder diffraction
Siemens D5000). Because of handling in air, chlorine is
detected only by the penetration depth of the electron beam
when chlorine is located below the surface. The obtained values
are not considered to be representative and are omitted;
therefore, only nickel-to-silicon ratios are discussed. Rutherford
backscattering measurements were performed on a 4 MeV van
de Graff accelerator at the State University Utrecht, Utrecht,
The Netherlands.
Thermokinetic Model. The obtained heat-flow curves were
fitted by the Avrami-Erofeyev equation (eq 4), a common
thermokinetic approach that describes the formation and growth
of 3-D spherical nuclei.
Results
I. Thermokinetic and Thermodynamic Studies. The mea-
surements were carried out under the guidelines given in the
Introduction. In all experiments, the amount of the silicide phase
was, by a molar factor, 4 or 5 times higher than the amount of
hydrogen chloride to make sure that the silicide phase was
converted into the silicide phase having the adjacent higher metal
content according to the phase diagram. After each run, the
removed silicide sample was studied by X-ray powder diffrac-
tion. In all samples, only the adjacent silicide phase and the
unreacted initial phase were found. Metal chlorides were never
found. In the case of copper silicides, elemental copper was
detected.
From the Arrhenius plot in Figure 2, it can be seen that all
of the investigated silicide phases show a v-shaped activation
energy pattern. At low temperatures, a negative activation energy
is consistently found, which describes a decreasing reaction rate
with increasing temperature. After passing through a minimum,
the reaction rate increases with temperature, and a positive
activation energy is observed. From the slope of the line fits,
the activation energies were calculated. For a given silicide,
the point of intersection of the line fits of positive and negative
slopes corresponds to a temperature, referred to as the “reaction
start temperature”, on the y axis; this point corresponds to the
natural logarithm of a rate constant, which is called the
“minimum rate constant” ln kmin. Reaction start temperatures,
minimum rate constants, and activation energies are listed in
Table 5.
In all experiments, an induction period is observed in which
the incoming hydrogen chloride is adsorbed, corresponding to
a decrease and followed by an increase in pressure. This pressure
drop lasts from a few seconds to few minutes. In blank tests
with empty cells, this pressure change was never observed.
Figure 3 compares four heat flow curves of the reaction of
hydrogen chloride with NiSi2, NiSi, Ni2Si, and Ni3Si, respec-
tively. The curves are taken from the middle of the positive
apparent activation energy region and normalized to the start
of the reaction at time zero. A bar in each graph indicates how
long hydrogen chloride is present in the flow cells (calculated
from the flow rates). NiSi is the phase with the highest reaction
dq
dt
3
1
3
-(kt)3
2
rate )
) qtotal
k
t e
(4)
(
)
(
dq/dt ) heat flow, qtotal ) heat of reaction, t ) time, k ) rate
constant).
One should notice that only one rate constant k is used to
describe the whole process. Other than what is known from
molecular reaction dynamics or kinetics of elementary steps,
the rate constant cannot be attributed to a single reaction.
Usually, no conclusion about the reaction mechanism can be
drawn from thermokinetic models.
For the reaction of a silicide phase with hydrogen chloride,
the rate constants as well as corresponding heats of reaction
were obtained by fitting the Avrami-Erofeyev equation to the
experimental curve. For each curve fit, the considered time scale
is at least 6 times higher than the time constant of our DSC
apparatus of 30 s; a deconvolution of the curve (desmearing)
was not necessary (Figure 1). The error in the curve fitting of
eq 4 is less than 5%, which was determined from the standard
deviation. The reliability of the curve fits was tested by varying
the number of points taken into account for curve fitting. A
maximum deviation of the rate constant of 10% was obtained.
Normally, the values are less than 8% (from the standard
deviation).
An evaluation of heat flow curves by the Avrami-Erofeyev
equation requires a widely uniform signal (Figure 1). This
demand is consistently fulfilled for all thermokinetic data
presented in this work. At higher temperatures (above the
reaction temperatures discussed herein), the signal becomes too
structured, shoulders appear, and peak splitting occurs. At this

Reactivity of Intermetallic Compounds
J. Phys. Chem. B, Vol. 106, No. 19, 2002 5109
Figure 3. Heat-flow curves of the reaction of nickel silicide phases
and hydrogen chloride taken from the middle of the positive activation
energy branch. The bar indicates how long hydrogen chloride is present
in the flow cells.
Figure 4. SEM micrograph of the reaction layer of a hydrochlorinated
NiSi sample (left) and the silicon depth profile of the reaction layer
(right).
TABLE 5: Apparent Activation Energies, Reaction Start
Temperatures, and Minimum Rate Constants of the
Reaction of Transition Metal Silicides with Hydrogen
Chloride
apparent activation energy
absolute value/kJ mol
reaction start minimum rate
-
1
silicide temperature
constant
phase
Ni Si
Cu Si
T
start/°C
ln kmin at Tstart
T > Tstart
T < Tstart
2
270
310
313
355
375
435
513
-6.63
-6.29
-4.07
-4.42
-7.33
-4.82
-5.38
49.9 ( 1.8
69.8 ( 5.0
-35.7 ( 7.0
-44.3 ( 2.4
3
a
NiSi
FeSi
88.1 ( 13.0 -59.2 ( 5.1
a
41.6 ( 3.9
-59.4 ( 0.6
-19.6 ( 3.5
Ni
3
Si
186.1 ( 9.2
a
a
FeSi
NiSi
a
2
196.3 ( 22.1 -137.6 ( 17.1
100.9 ( 15.1 -115.8 ( 0.7
2
Apparent activation energies have been published in ref 36.
rate, and it shows a narrow signal; the peak dropped after HCl
was flushed out of the cell. Comparing NiSi2, Ni2Si, and Ni3Si,
the reaction time exceeds the residence time of hydrogen
chloride in the cell.
II. Solid-State Investigations. To follow how chlorine
interacts at the silicide surface, several studies of the surface
region were performed. All nickel silicide phases (polished
compact samples) were either directly treated with hydrogen
chloride or were reacted under milder condition with hydrogen
chloride by embedding the sample in NiCl2 (having a water
content lower than 1%), thus forming reaction layers with a
thickness in the micrometer range and a composition different
from that of the bulk. This process occurs within few minutes.
Chlorine was present in all of these studies throughout the entire
reaction layer.
Figure 5. RBS profile of the surface region of a reaction layer.36
The RBS study reveals details of the uppermost surface
region. Figure 5 proves the existence of a chlorine-rich surface
layer and also confirms an enrichment of silicon near the
2
3,24
surface.
The penetration depth of the helium atoms is too
low to detect bulk NiSi, but a thick layer of Ni2Si is detected
on which the chlorine- and silicon-rich layers are lying. The
chlorine content in the Ni2Si layer was proven by SEM-EDX
analysis of the same sample. The estimated thickness of the
chlorine-containing layer is approximately 0.9 µm, which was
Figure 4 shows the micrograph of a polished cross section
of a NiSi sample treated with 0.8 bar hydrogen chloride at 410
1
6
-2
estimated by assuming 10 atoms cm .
°
C for 8 min. The formed surface reaction layer consists of
several layers with different chemical compositions. The gray
tones depend on the nickel, silicon, and chlorine content. The
distinctly separated layers indicate perpendicular nickel and
silicon migration with respect to the surface plane. The drawing
in Figure 4 shows the molar fraction of silicon as a function of
depth and reveals an enrichment of nickel in the different
reaction layers. According to XRD studies, the reaction layers
on NiSi samples consist of Ni2Si.
Discussion
I. Thermokinetic and Thermodynamic Studies. Apparent
Activation Energies. Negative activation energies are opposite
to the general experience that rising temperature leads to an
increase of the rate constant. In the literature, many examples
of different origins are known in which this phenomena is
37
explicitly discussed (e.g., OH addition to substituted ethenes,
3
8
Previous XRD studies of the reaction of NiSi with hydrogen
chloride or NiCl2 for comparable time scales such as that of
the surface studies (i.e., reactions were terminated before
equilibrium) always gave Ni2Si (with more or less broadened
or slightly shifted peaks) along with the unreacted NiSi
hydrogen abstraction from aldehydes and substituted alde-
39
hydes by OH radicals, the silylene or germylene insertion into
4
0
41
Si-H bonds and Ge-H bonds, respectively, and killing of
4
2
mammalian cells and Drosophilia embryos (and references
therein for other biological systems)). The common explanation
in all these articles is that the kinetically followed process
1
3,30,31
phase.

5110 J. Phys. Chem. B, Vol. 106, No. 19, 2002
Acker and Bohmhammel
consists of a complex mechanism of at least two consecutive
reaction steps. In numerous other articles, negative activation
energies are shown in Arrhenius plots (e.g. silicon deposition
by low-pressure MOCVD43 or formation of formaldehyde in
constant k1 (k1,0 ) the preexponential factor, EA ) the activation
energy) gives eq 10:
d[I]
dt
-(∆ H-E )/RT
A
A
)
k A p e
(10)
4
4
1,0 a eq
the oxidation of methanol on polycrystalline platinum ). Two
examples of negative activation energies related to direct
reactions of silicon are known: the formation of various silanes
in direct synthesis on Si(100) surfaces4 and a study of the
reaction of cuprous chloride and silicon.46
From eq 10 are the following results:
1. The reaction rate is proportional to the equilibrium pressure
of the gaseous reactant.
5
2
. The activation energy derived from an Arrhenius plot is
To formulate a formalized kinetic model of an apparent
negative activation energy, the reaction between hydrogen
chloride and a silicide phase is thought of as a series of
consecutive reactions. Below the reaction start temperature,
hydrogen chloride is adsorbed on the surface and then incor-
porated into the silicide surface region in an exothermic reaction
an apparent activation energy, which consists of the adsorption
enthalpy and the activation energy of the adsorption rate
(
according to the model): Eapparent ) Ea + ∆aH.
3
. The relation between the exothermic adsorption enthalpy
and the activation energy of the intermediate formation deter-
mines the slope of the apparent activation energy for the total
process: |∆aH| > EA, which is a negative apparent activation
energy for T < Tstart
(
formation of an intermediate), but gaseous products are not
3
0,31
yet formed
or are formed only to a negligible extent. Here,
the conversion of the intermediate into reaction products (k2) is
completely hindered but becomes possible predominantly above
the reaction start temperature and results in a positive apparent
activation energy.
In agreement with the model, it was found in a previous
3
0,31
study
that a low hydrogen chloride partial pressure lowers
the rate and increases the reaction start temperature. The
reactions of two different NiSi to NiCl2 molar ratios, (a)
In particular, the model of the negative apparent activation
energy below the reaction start temperature consists of a fast
adsorption of the gaseous compound (A (g)) and the reaction
of the adsorbed molecules (A(ads)) with the underlying silicide
phase to form a reaction intermediate I as the rate-limiting step,
as shown in eq 5. For the sake of simplicity, only a single
adsorbate case is considered. To include the dissociative
chemisorption of hydrogen chloride, a model for the reaction
path of the resulting chemically nonequivalent species would
be required, which is not yet established.
1
:0.2857 and (b) 1:0.5, were studied by isothermal DSC
30,31
measurements.
The Avrami-Erofeyev equation was applied
to calculate the rate constants, and the Arrhenius plot gave the
same v-shaped pattern. Assuming that the same amount of
moisture adhered to NiCl2 because of the higher total amount
of NiCl2 in series (b), the hydrogen chloride partial pressure is
higher than that for series (a). For both series, a reaction start
temperature of 347 °C was found. For series (a) and (b), ln kmin
values of -8.6 and -6.5 were found, respectively, compared
to the corresponding 1 bar hydrogen chloride value of -4.07
and a characteristic temperature of 313 °C (Table 5). The
obtained reaction start temperatures of 347 °C corresponds to
the onset temperature when the chemical reaction between NiSi
and NiCl2 under the formation of gaseous products begins.3
Equation 10 shows, in accordance with the experimental
results, that the reaction rate depends on the hydrogen chloride
partial pressure. Therefore, this partial pressure must be kept
constant for a series of measurementsto obtain comparable
results.
k1
k2
A(g) a A(ads)
98 I (intermediate)
98
solid + gaseous products (5)
0,31
Hence, the rate of formation of I is proportional to the hydrogen
chloride coverage of the silicide surface (eq 6).
d[I]
dt
)
k1Θeq
(6)
The model of the apparent negative activation energy is
consistent with the results in Figure 1 that show that the
measured heat of reaction increases with increasing temperature
for all of the studied silicide phases. Although the heat of
adsorption should decrease with rising temperature, the incor-
poration of hydrogen chloride (intermediate formation) becomes
more important. Therefore, the thermal signal below the reaction
start temperature is the sum of the exothermic hydrogen chloride
chemisorption and the exothermic chlorine incorporation into
the silicide surface region.
The adsorption of the molecules is described by the Langmuir
equation (eq 7).
Kap
eq
Θ )
(7)
eq
1
+ Kapeq
Θeq ) coverage at equilibrium, Ka ) the adsorption equilibrium
constant, and peq ) the equilibrium partial pressure of A(g),
and by neglecting the denominator for low partial pressures
Reaction Start Temperature as a Thermodynamic Pa-
rameter. In Figure 6, the reaction start temperatures of nickel
silicides are plotted as a function of the molar fraction of silicon.
The plot shows that the obtained curve corresponds to the
(Kapeq , 1), eq 8 follows.
Θ ) K p
(8)
eq
a
eq
47
standard formation enthalpies of the silicides at 298.15 K, to
the heat of mixing of liquid nickel and liquid silicon (values
By substitution of eq 8 into eq 6, eq 9 results:
d[I]
47
shown here are at 1800 K ), and to the curve for the heat of
32
chlorine dissolution in nickel silicides. This general agreement
leads to the conclusion that the enthalpy of chlorine intake into
nickel silicides is proportional to the strength of interaction
between nickel and silicon in the solid or liquid phase.
Furthermore, these curves are in agreement with the reaction
start temperatures; therefore, the reaction start temperature is
considered to be a thermodynamically determined parameter.
)
k K p
1 a eq
(9)
dt
The Arrhenius-type temperature dependence of the adsorption
equilibrium constant Ka (Aa ) the preexponential factor of
adsorption, ∆aH ) the adsorption enthalpy) and of the rate

Reactivity of Intermetallic Compounds
J. Phys. Chem. B, Vol. 106, No. 19, 2002 5111
Figure 7. Reaction start temperatures and minimum rate constants
for the reaction of the various nickel silicide phases with hydrogen
chloride as a function of the molar fraction of silicon.
Figure 6. Correlation of the reaction start temperature for various
nickel silicide phases (a) to the enthalpy of chlorine dissolution in nickel
silicide phase (b) and to thermodynamic functions of the nickel-silicon
system (c) (b represents the standard enthalpy of formation of nickel
4
7
silicides at 298.15 K, and 0 represents the mixing enthalpy of nickel
4
7
and silicon at 1800 K ).
The question remains of what parameters can be used to
discuss the differences in the reactivity between various transi-
tion metal silicides or between various metal-to-silicon ratios
of one metal. Band structure calculations of transition metal
silicides show the presence of mainly covalent bonds and minor
48-53
ionic contributions for the late 3d transition metals
and a
higher ionic contribution for Cu3Si,54 but these calculations are
not able to reflect trends in the reactivity of transition metal
silicide phases. Rather, enthalpies and excess entropies of mixing
of silicon and transition metals more suitably reflect this
phenomenon. In the metal row from Mn to Ni, the enthalpy
and excess entropy of mixing increase, but they decrease for
-
1
Figure 8. Apparent activation energies for the reaction of various
silicide phases with hydrogen chloride as a function of the molar fraction
of silicon.
Cu: ∆mixHNi-Si ) (-56.0 ( 2.1) kJ mol at xSi ) 0.39,
-
1
∆
∆
∆
mixHCu-Si ) (-14.0 ( 0.6) kJ mol
at xSi ) 0.25,
-
1
-1
mixS(excess)Ni-Si ) (-17 ( 3) J K mol at xSi ) 0.38,
mixS(excess)Cu-Si ) (-4 ( 0.5) J K-1 mol at xSi ) 0.20.
-1
55,56
to divide the silicide surface/reaction region into a number of
chlorine adsorption sites, which is proportional to the molar
fraction of the metal, and into a number of silicon reaction sites,
which is proportional to the molar fraction of silicon. Disilicides
have a low number of chlorine adsorption sites and a high
number of silicon reaction sites, and hydrogen chlorine bound
to the surface is readily converted into silanes and hydrogen.
As a result, the heat-flow signal is quite narrow, and the reaction
rate is high. Because of its high metal content, the M3Si phase
allows high chlorine coverage, but the reaction proceeds at a
low rate because of the low number of silicon reaction sites.
Hence, a broadened heat-flow signal is observed. The “chlorine
absorption site” and “silicon reaction site” do not necessarily
imply that all of these sites are directly located on the surface
or that these sites can be considered to be surface species. It
also should not be concluded that chlorine is bonded to nickel.
Rather, the metal should be considered to be an intermediary
that transfers chlorine to silicon according to a general mech-
The large negative excess entropy of mixing arises from the
electron contribution by transfer of silicon valence electrons to
fill the metal 3d shell. The atomic interaction between 3d metals
and silicon is stronger as the 3d shell fills with electrons. The
decrease in the excess enthalpies and entropies of mixing from
10
nickel to copper is a result of the completely filled d shell (3d -
1
55,56
4
s ) of copper.
The thermodynamic approach is able to
explain similar chemical properties of chlorine-containing iron,
cobalt, and nickel silicide phases in the hydrodechlorination
reaction and the comparable chlorine contents as well as the
exception of copper silicide phases.
Kinetics and Apparent Activation Energies. As expected,
the largest minimum rate constants in Figure 7 go along with a
low positive apparent activation energy in Figure 8.
Reaction start temperatures (Figure 7) and positive apparent
activation energies (Figure 8) show a comparable dependence
on the molar fraction of silicon. A low reaction start temperature
and hence strong metal-silicon interactions and a high chlorine
solubility correspond to a low apparent activation energy,
indicating that chlorine intake is directly related to the reactivity
of the silicide phases. Over the whole composition range, the
thermodynamic parameter Tstart and the minimum rate constant
are inversely dependent on the molar fraction of silicon (Figure
2
0
anism (Scheme 1) suggested by Lewis et al.
The linear dependence of the negative activation energies on
the molar fraction of silicon is difficult to explain. It seems that
a high metal content lowers the activation barrier of chlorine
incorporation because of a higher number of chlorine adsorption
sites. High nickel mobility may also support the incorporation
of chlorine into the silicide phases (see later discussion). As in
the case of the positive apparent activation energies, no
dependence on the silicide phases crystal structure exists.
7
). A low rate is observed for low and high silicon content,
and the maximum seems to be an optimum value between two
opposite tendencies. To explain the kinetic behavior, we propose

5112 J. Phys. Chem. B, Vol. 106, No. 19, 2002
Acker and Bohmhammel
Figure 9. Trichlorosilane-to-silicon tetrachloride ratio calculated for
various iron, nickel, and copper silicide phases.
Figure 10. Heat-flow curves of the reaction of Cu
3
Si and Cu
3
Si +
1
.5 at. % Zn with hydrogen chloride at various temperatures. Because
SCHEME 1. General Mechanism of the Direct Reaction
of Silicon and Hydrogen Chloride According to Lewis et
of the structures in the curves of Cu Si + 1.5 at. % Zn, only the tails
3
of the curves were fitted by the Avrami-Erofeyev equation.
24
al.
Reaction Products. In the previous sections, the formed
Figure 11. Arrhenius plot of the reactions of Cu Si and Cu Si + 1.5
3
3
gaseous products were neglected, but here, the product distribu-
tion of formed silanes is discussed on the basis of thermo-
dynamic considerations. The product gas-phase composition was
thermodynamically calculated for the reactions of hydrogen
chloride and various silicide phases in a molar ratio of 1:100,
respectively. The obtained trichlorosilane-to-silicon tetrachloride
ratio is plotted as a function of temperature in Figure 9.
Pure silicon gives the highest trichlorosilane yield, followed
by the disilicides of nickel and iron. Generally, with increasing
metal content, the trichlorosilane yield decreases, and Ni3Si,
which is not indicated in the diagram, gives no appreciable
at. % Zn with hydrogen chloride.
(i) To increase the reaction rate of metal-rich silicide phases,
the number of silicon reaction sites has to be increased by a
higher diffusion rate of silicon to the surface. Alternatively,
chlorine transfer from the chlorine adsorption sites to the silicon
reaction sites can be enhanced. We studied the reaction of a
Cu3Si alloyed with 1.5 at. % Zn with hydrogen chloride in the
same manner as that described above. As can be seen from
Figure 10, zinc addition causes a stronger interaction with
hydrogen chlorine. We have found that the rate is decreased
dramatically with a parallel shift in the rate and with a conserved
apparent activation energy (Figure 11). The optimum amount
-
5
amount of trichlorsilane (ratios at 200 and 600 °C are 5 × 10
and 0.0015, respectively). Having a high number of chlorine
adsorption sites favors silicon tetrachloride or, generally, a high
chlorine content in the gaseous product phase. Furthermore, iron
silicides always produce more silicon tetrachloride than do nickel
silicide phases of the same metal-to-silicon ratio. Therefore, the
chlorine content of the product silanes also correlates with the
metal-chlorine bond strength of Fe > Ni > Cu (Table 3).
Copper is a remarkable exception in two cases: (i) despite its
high metal content, it gives a high trichlorosilane yield (Figure
2
4
of a zinc promoter is around 0.4 at. % Zn. From the parallel
shift, we conclude that an excess of zinc (more than the
optimum) decreases the reaction rate by lowering silicon
diffusion to the surface without changing the reaction mecha-
nism.
(ii) To decrease the reaction start temperature of silicon-rich
silicide phases, a small amount of a metal could be introduced
that strongly interacts with silicon and therefore enhances the
ability of chlorine intake. Side products may play a role here.
Furthermore, silicide phases having a high silicon content have
a smaller number of chlorine adsorption sites compared to their
high number of silicon reaction sites. A metal chloride (or its
pure metal, as a precursor) that sticks on the surface and hence
causes a higher chlorine content on the surface than the initial
silicide phase would be able to form may be introduced. Ideally,
the metal chloride should give its chlorine easily to the silicon
9
) and (ii) it is the only metal that also produces dichlorosilane
in a reasonable yield (here calculated and experimentally proven
for copper-silicon mixtures in ref 57).
Direct Synthesis Promoters. On the basis of the presented
results, some general principles about structural promoters in
direct reactions of silicon can be derived. According to Lewis
et al., structural promoters cause a change in chemical composi-
tion in the solid phases.2
0,24

Reactivity of Intermetallic Compounds
J. Phys. Chem. B, Vol. 106, No. 19, 2002 5113
reaction sites, and the metal chloride should easily be regener-
ated by the hydrogen chloride from the gas phase.
SCHEME 2. Formation of a Ni2Si Layer on NiSi as a
Result of Nickel Diffusion.
Introducing additional compounds may also cause synergistic
effects between the reaction start temperature and reaction
kinetics and may explain why different promoters show
unexpected behaviors in different concentration ranges and in
mixtures with other promoters.7
was found to be the mobile atom, and the silicon atom mobility
Summary
60
is about 10% of that of the nickel atoms. In Ni2Si, the majority
of the atoms constitute a network where vacancies can easily
migrate. In the case of NiSi, the nickel and silicon atoms are
located in geometrically different sites, and nickel migrates via
vacancies in the nickel sublattice. In NiSi2, the nickel motion
is assumed to proceed via the empty cube centers of the unit
cell. To increase the relative diffusion rate of silicon, a larger
activation energy is required compared to that of the mobiliza-
The reactivity of transition metal silicides can be described
by thermodynamic and kinetic parameters that are clearly
separated from each other:
(1) The characteristic temperature correlates to the silicide
phases thermodynamic functions as standard enthalpies of
formation, mixing enthalpies of the liquid elements, and
enthalpies of chlorine dissolution. The characteristic temperature
is an expression of metal-silicon interactions with respect to
the ability to incorporate chlorine into the lattice structure. As
the chlorine intake increases, the characteristic temperature
decreases. The silicide phase crystal structures do not seem to
play a role in reactivity.
59,62
tion of nickel atoms.
On the basis of these results, the phase formation sequence
for the nickel silicide phases Ni2Si, NiSi, and NiSi2 can be
described by the following equation where Ni3Si2 formation is
omitted:
(2) Chlorine adsorption sites and silicon reaction sites are
introduced to discuss the minimum rate constants as a function
of the silicide phase composition and to explain the shape of
the calorimetric signals. Reaction start temperatures, positive
apparent activation energies, and minimum rate constants show
the same dependence on the molar fraction of silicon. Therefore,
the presence of chlorine in the silicides is the necessary condition
for their reactivity. The negative apparent activation energies,
increasing linearly with silicon content, are attributed to
hydrogen chloride chemisorption and chlorine intake into silicide
phases below the reaction start temperature.
+Ni
+Ni
NiSi2 { } 2NiSi { } 2Ni Si
(11)
2
-Ni
-Ni
SEM-EDX Study. The formation of a Ni2Si reaction layer
on bulk NiSi as found by the SEM-EDX study (Figure 4) is
easily explained as a diffusion phenomenon (eqs 12 and 13).
NiSi + 4HCl f Ni + SiCl + 2H
(12)
(13)
4
2
NiSi + Ni f Ni Si
2
(
3) A high metal content of the reacting silicide phase favors
silicon tetrachloride formation. Silicides of metals known to
form a strong metal-chlorine bond yield more silicon tetra-
chloride than do metal silicides whose metal forms a metal-
chlorine bond of lower strength. Copper is an exception and
forms dichlorosilane.
Silicon is removed from the silicide surface by reacting with
hydrogen chloride. The excess nickel atoms diffuse into the
underlying NiSi layer to form Ni2Si (Scheme 2).
According to this simple model, chlorine incorporated into
the silicide bulk phase does not seem to play an important role.
Chlorine becomes the only matter of importance to the question
of whether NiCl2 is found as a reaction product or not. The
absence of NiCl2 can be explained by the following thermo-
dynamic consideration (eq 14). Metal chloride formation does
not occur if the metal chloride can be reduced by silicon (or
another silicide phase) under the formation of a (or another)
silicide phase. As a rule of thumb, the free enthalpy of chloride
(4) A high metal concentration correlates to strong hydrogen
chloride chemisorption (Figure 3, chlorine adsorption sites) and
to a high silicon tetrachloride yield, although metal chloride
formation was never observed, which points to the role of the
metal as a transmitter of chlorine to silicon in a spill-over
process. This result is consistent with Lewis’ general mechanism
of direct reaction in which chlorine transfer from metal to
silicon is the key step of silylene species formation (Scheme
).
20
-
1
formation must be less than -320 kJ mol , as in the case of
iron, cobalt, nickel, and copper (exact values for any metal and
1
3
6
temperature can be calculated).
II. Solid-State Investigations. To introduce the following
discussion, a brief review of diffusion phenomena in the nickel-
silicon system is necessary. By the reaction of thin films of
nickel and silicon at around 400 °C, Ni2Si is growing as the
first phase by a diffusion mechanism. If all nickel is consumed
in the formation of Ni2Si, NiSi, which is formed by diffusion,
appears at around 500 °C. Finally, NiSi2 is formed by nucleation
4
/xMCl + (4y/x + 1)Si f 4/xMSi + SiCl (g) (14)
x
y
4
When the reaction conditions are mild, silicon is always present
at or near the surface. The possibly formed nickel-chlorine
bonds can be reduced by silicon, and chlorine is transferred
from nickel to silicon by the same process. Hence, the absence
or presence of NiCl2 is a clear indication that the reaction
conditions are dominated by thermodynamics or kinetics,
respectively. If silicon is removed under aggressive reaction
conditions, NiCl2 is immediately formed despite the thermo-
dynamic capability of silicon (or a present silicide phase) to
reduce NiCl2 according to eq 14. The lower free enthalpy of
formation of copper chloride explains the higher activity of
copper chloride at low temperatures compared to that of the
chlorides of iron, cobalt, and nickel.
5
8,59
around 800 °C.
Because of a nucleation hindrance, Ni3Si2
is formed only when the temperature is equal to or higher than
6
0
6
00 °C at the interface in the Ni2Si/NiSi diffusion couple. An
interesting phase formation pathway was found in the case of
Ni2Si and NiSi. NiSi is found to grow between a Si/Ni2Si
interface via decomposition of Ni2Si into NiSi and Ni. The
nickel atoms move backward to the silicon interface and react
there with silicon atoms, forming NiSi.61 The reaction of nickel
with silicon was determined to be the rate-limiting step, not
the nickel diffusion in NiSi. In all nickel silicide phases, nickel

5114 J. Phys. Chem. B, Vol. 106, No. 19, 2002
Acker and Bohmhammel
SCHEME 3. Formation of a Chlorine-Containing
Reaction Layer.
TABLE 6: Adiabadic Electron Affinities of Gaseous
Silylenes
electron
electron
species
affinity/eV
ref
species
affinity/eV
ref
a
a
SiH
SiH
SiH
SiH
SiH
SiH
SiH
SiH
SiH
2
2
2
2
2
2
2
2
2
1.124 ( 0.020
65 SiCl
66 SiCl
67 SiCl
68 HSiCH
69 Si(CH
70
2
2
2
0.64; 0.90
0.79; 1.25
74
75
76
69
69
a
b
1.2
1.003b
1.26; 1.50
c
c
b
1.37; 1.00
3
-0.018b
b
1.029
1.17
3
)
2
-0.064
b
1.12a
1.16b
71 CH
72 CH
0.08 ( 0.03
a
77
78
3
0.07b
(a) Hydrogen-chloride induced motion of nickel and silicon atoms
3
1.03; 1.08; 1.20c 73
yields the idealized silicide phase sequence in (b). For the sake of
simplicity, stoichiometric considerations between the layers are omitted,
and a uniform layer thickness is drawn. (c) Expected layer composition
if nickel sites are occupied by chlorine. (d) Results from the RBS study.
a
Experimental. b Calculated. c Calculated by various models.
consists of various layers having different compositions. The
diffusion of nickel, silicon, and chlorine can explain the observed
sublayer sequence.
RBS Study. A more detailed understanding of the reaction
surface layer and the role of chlorine is obtained by Rutherford
backscattering studies on hydrochlorinated NiSi samples (Figure
The main consequence of the formed reaction layer is a local
separation of the proceeding reaction steps. The sites where the
dissociative chemisorption of hydrogen chloride occurs, where
chlorine and hydrogen are attached to silicon, yielding silyle-
noide surface species, and where surface species react to form
product molecules are not identical. Away from the surface,
silicon atoms at the interface between the reaction layer and
the unreacted silicide phase are removed from the pure silicide
phase and forced into the reaction layer. The mechanism of this
interface reaction is unclear; silicon might be removed from
the pure silicide phase by a direct reaction with chlorine or might
be loosened from the pure phase by reacting with the backward-
diffusing nickel atoms.
5). To explain the formed layers of different compositions, we
refer to eq 11. Applying this equation, the phase sequence as
shown in Scheme 3b should be obtained. As seen from Figure
5, a high chlorine content corresponds to a low nickel content
and vice versa. The conclusion is that chlorine displaces nickel
by occupation of nickel sites. Occupancy of nickel sites by
chlorine forces nickel atoms to move into the bulk, and silicon
atoms diffuse to the surface by counter diffusion. Recalculating
the nickel content as nickel ) [nickel + chlorine] and the nickel-
to-silicon ratio, a layer sequence is obtained that matches the
composition of the layer (Scheme 3c and d). It should be noted
that the diffusion model describes a system of constant mass,
and the RBS study is taken at an early stage of the reaction in
which silicon is already removed from the bulk.
Results from the gas-phase silicidation of nickel by a SiCl4/
H2 gas mixture point to an associated mobility of silicon-
chlorine clusters within a nickel matrix.
1
3
The assumption of the occupation of nickel sites by chlorine
is indirectly supported by in situ XRD studies of the reaction
III. Suggestion of a Model for the Reactive Surface Layers.
At the present point of research, the bonding character of
chlorine incorporated into the silicide phases is yet unknown.
But it had been shown that the chemical properties of chlorine-
containing silicide phases are different from the chemical
properties of pure silicide phases. Furthermore, the chlorine-
silicon interaction in the chlorine-containing phases is much
stronger than the metal-silicon interactions (thermal decom-
position). Therefore, it is at least reasonable to assume that
chlorine intake changes the electronic structure with respect to
that of the initial silicide phase.
In the following section, the role of the chlorine-containing
reaction layer is discussed using the model of electronegativity
equalization. This model says that different species have
different electronic chemical potentials and that at equilibrium
electrons flow from one species to another to equalize the
13
of NiSi and NiCl2. It was found that the silicide phase
degradation always proceeds through ordered structures, either
through Ni2Si at 410 °C or through an unknown intermediate
ordered structure at 390 °C. Amorphization was never observed
in any of our studies, but numerous presumably kinetically
determined ordered structures were identified.3
0
Some questions arising from the presented model should be
discussed here. The formation of silicon tetrachloride is the
thermodynamic driving force of the hydochlorination reaction.
Chlorine incorporation into the lattice can be thought of as a
series of states in which a silicon atom makes a transition from
a lattice atom to a silicon tetrachloride molecule. Therefore,
chlorine incorporation is thermodynamically driven. The fact
that chlorine is diluted in a silicide phase can be understood as
6
3
an analogue to oxygen solubility in titanium or zirconium or
6
4
chemical potential. To determine the stability of silylenoids
on the silicide surface, we consider the adiabatic electron
affinities of gaseous silylenes and the work function of the
surface reaction layer. As seen from Table 6, electropositive
and electronegative silylenes are known.
2
to the oxidation of metals or alloys by diluted oxygen.
On the microscopic level, chlorine incorporation is assumed
to proceed by the displacement of nickel atoms, followed by
the occupation of nickel atoms sites with chlorine. Nickel atoms
are forced to move toward the bulk, but silicon atoms seem to
diffuse to the surface. The high mobility of silicon might be a
combination of (i) silicon diffusion along the chlorine concen-
tration gradient using defects created by chlorine and nickel
motion; and (ii) fast diffusion of nickel atoms into the bulk and
exposing the less mobile silicon atoms to the surface. This result
might explain that the reaction of nickel silicides with hydrogen
chloride is a very selective reaction. Not all silicon atoms seem
to be immediately attacked by hydrogen chloride or incorporated
chlorine so that silicon atoms remain, which constitutes an
ordered structure of the reaction layer. The reaction layer itself
SiCl > (HSiCl) > SiH > CH > (CH SiCl) > HSiCH >
2
2
3
3
3
Si(CH ) (15)
3 2
The sequence in eq 15 represents the known adiabatic electron
affinities of gaseous silylenes. The electronegative character
increases with the substituents and their number bound to silicon
as Cl > H > CH3. For HSiCl and CH3SiCl, there is no data
available, but the substituents have been ordered according to
their character. The work functions of chlorine-containing

Reactivity of Intermetallic Compounds
J. Phys. Chem. B, Vol. 106, No. 19, 2002 5115
silicides are not known. On the basis of the concept of
electronegativity equalization, a higher electron density is
necessary to stabilize the electronegative SiCl2, whereas HSiCH3
is already stabilized by the lower electron density of the reaction
layer.
and the gradual differences in the reactivity between various
metals on the basis of a correlation of mixing enthalpies and
entropies of transition metal-silicon systems. From the results
in Tables 1 and 2, one might conclude that the product
distribution of direct synthesis and the hydrochlorination reaction
follows the polarity of metal-silicon bonds. Ranging from late
Instead of the unknown work functions of the chlorine-
containing silicide phases, we refer to the observed steady-state
product distributions in the direct synthesis and hydrochlorina-
tion (Tables 1 and 2). It is assumed that these products are
transition metals, which form mainly covalent silicides,4
8-53
to
ionic magnesium and calcium silicides, chlorine is attached to
the positively charged metal constituent. The remaining sub-
stituents (either hydrogen or a methyl group) are bound to the
negatively charged silicon. We do not follow this consideration,
formed by the reaction of a silylenoide species with either
hydrogen chloride or methyl chloride.20 From the product
7
9
which is similar to Voorhoeve’s model, because the presence
of chlorine changes the bonding state in the silicide phases.
Our model can be understood as an extension of Roewer’s
distributions (Tables 1 and 2), it can be concluded that the
copper-based, chlorine-containing reaction layers have a lower
electron density than do nickel- or iron-based reaction layers,
which is in accordance with the lower interaction of copper and
1
0
model of the hydrodechlorination reaction, which introduces
an electron-transfer process from an adsorbed silicon tetrachlo-
ride molecule to a silicide surface under the formation of a
silylenoide surface species.
55,56
silicon compared to that of iron, cobalt, or nickel and silicon.
According to the product distributions in Tables 1 and 2, reaction
layers or chlorine-containing silicide phases formed by reaction
with hydrogen chloride or a silicon tetrachloride/ hydrogen
mixture prefer more electronegative silylenoide species than
does the same chlorine-containing silicide phase formed in the
reaction with methyl chloride. Here, the difference is the result
of the acid-base strengths of methyl chloride and hydrogen
chloride. The reactivity seems not to depend on the possible
maximum intake of chlorine but rather on the amount of
accessible chlorine. Hence, a silicide phase can exhibit a wide
range of reactivity, and its reactivity can be dramatically changed
by the addition of small amounts of promoters.
IV. Some Comments about the Role of Hydrogen. We have
to emphasize that the role of hydrogen is still unknown in the
direct reactions and in the hydrodechlorination reaction as well.
All analytical methods employed failed to detect hydrogen in
the solid phases. Therefore, hydrogen is not included in our
consideration.
But, the silicon-hydrogen system exhibits a large variety of
8
0
81
surface chemistry on crystalline and on porous silicon.
Certainly, hydrogen plays an important role in direct reactions,
but as bonded hydrogen in hydrogen chloride or trichlorosilane
rather than as molecular hydrogen. Few examples should
underline this result. In the hydrodechlorination reaction of
silicon tetrachloride, it was found that only a sufficient pulse
of hydrogen chloride reactivates an inactive catalyst, whereas
On the basis of these considerations, a model of the induction
period of the direct synthesis is proposed. At the first exposure
of a copper silicide sample, methyl chloride decomposes on the
surface into carbon, hydrogen, and chlorine. Methane, hydrogen,
and hydrocarbons are formed, and chlorine incorporation starts.
Because of the low chlorine content of the copper silicide surface
layer, the electron level is quite low. Therefore, the preferred
electropositive silylenoids HSiCH3 and Si(CH3)2 are stabilized.
By inserting methyl chloride molecules into these silylenoids,
9
,10
5
molecular hydrogen is inactive.
Hydrogen chloride, chlo-
5
29
rine, and trichlorosilane were found to activate the copper-
silicon mass in direct synthesis by forming silicon-chlorine
surface species. It seems that a certain hydrogen concentration
bound to the surface is required to start and to maintain the
reaction propagation. In the direct synthesis of phenylchlorosi-
lane using phenyl chloride and a copper-silicon mixture, an
addition of hydrogen chloride increased the rate of trichlor-
phenylsilane and diphenyldi- chlorosilane, although hydrogen
(CH3)2SiHCl and (CH3)3SiCl are formed. Over the time of the
reaction, chlorine intake proceeds and the surface region
incorporates chlorine as long as the steady-state chlorine
concentration, according to the steady-state methyl chloride
partial pressure, is not reached. When the steady-state chlorine
uptake is reached, the higher electron level is higher than that
in the previous period. Hence, more electronegative silylenoids,
like CH3SiCl, can be stabilized, and dichlorodimethylsilane is
formed by inserting methyl chloride molecules. In combination
with methyl chloride, copper-based solid phases are not able to
stabilize SiCl2 silylenoides. Generally, this model always implies
the formation of silanes with a higher hydrogen or methyl
content than that under later steady-state conditions.
8
2,83
chloride is formally not involved.
Conclusion
The present study gives a comprehensive and quantitative
view of the chemical reactivity of intermetallic compounds,
particularly for the reactions of transition metal silicides and
hydrogen chloride. Macroscopic and microscopic approaches
are presented and unified in a model of chlorine dissolution in
silicide phases.
Temperature and surface chlorine concentration have ad-
ditional effects on the product distribution. In hydrochlorination,
silicon tetrachloride is the thermodynamically stable product
and what is preferably formed at high temperatures. Trichlo-
rosilane is considered to be a kinetically preferred product at
low temperatures. It is assumed that electropositive hydrogen-
rich surface species (silylenoides) are more stable at low
temperatures.3 Furthermore, the increasing metal content of the
solid phase (e.g., by silicon depletion) leads to an increase of
the silicon tetrachloride yield (Figure 9). Hence, a higher number
of chlorine adsorption sites are formed, and the chlorine
On the basis of calorimetric measurements, reaction start
temperatures, rate constants, and apparent activation energies
were determined. Reaction start temperatures and positive
apparent activation energies are related to the model of chlorine
solubility and can be understood as expressions of the interac-
tions between metal and silicon in the silicide phases. High
chlorine solubility is correlated to a low reaction start temper-
ature, to a low positive apparent activation energy, and to a
high reaction rate. Negative apparent activation energies are
considered to describe the chemisorption of hydrogen chloride
and the incorporation of chlorine into the surface region of the
silicide phases at temperatures too low to form gaseous products.
Solid-state studies proved the formation of a chlorine-containing
reaction layer by exposing the silicide phases to hydrogen
6
36
concentration at the surface increases remarkably.
Our model explains the differences in product distribution
through changes in the electronic structure of the reaction layer

5116 J. Phys. Chem. B, Vol. 106, No. 19, 2002
Acker and Bohmhammel
chloride. The formation, renewal, and microscopic structure of
such a layer can be explained by diffusion.
(19) Yamanaka, S.; Matsuura, H.; Ishikawa, M. Mater. Res. Bull. 1996,
1, 307.
3
(20) Lewis, K. M.; McLeod, D.; Kanner, B.; Falconer, J. L.; Frank, T.
The chlorine-containing reaction layer acts as a transmitter
between the pure underlying silicide phase and the gas phase
and is attributed to a local separation of elemental reaction steps.
The level of chlorine intake adjusts the electronic properties of
the reaction layer and determines what silylenoide surface
species are stabilized.
C. In Catalyzed Direct Reactions of Silicon; Lewis, K. M., Rethwisch, D.
G., Eds.; Studies in Organic Chemistry Series 49; Elsevier: Amsterdam,
1
993; p 333.
21) Clarke, M. P.; Davidson, I. M. T. J. Organomet. Chem. 1991, 408,
49.
(22) Okamoto, M.; Onodera, S.; Okano, T.; Suzuki, E.; Ono, Y. J.
Organomet. Chem. 1997, 531, 67.
(
1
(
(
23) Frank, T. C.; Falconer, J. L. Appl. Surf. Sci. 1982, 14, 359.
24) Lewis, K. M.; McLeod, D.; Kanner, B. In Catalysis 1987: Studies
Primarily, the reactivity of the silicon atoms in the silicide
phases is determined by the metal. The strength of metal-silicon
interactions correlates to the chlorine intake ability. Thermo-
dynamic parameters such as enthalpies of mixing or formation
are found to be more useful to explain metal-silicon interactions
rather than band structures of silicide phases. As a second
parameter, the acid-base strength of the reaction gas phase
determines the chlorine intake as well. Different product
distributions between reactions with methyl chloride or hydrogen
chloride can be explained in this manner. Furthermore, the
change in reactivity from metal-rich to silicon-rich silicide
phases can be explained only when the existence of chlorine
adsorption sites (related to the metal content) and silicon reaction
sites (related to the silicon content) is assumed. The essential
step here is a spill-over process where chlorine is transferred
from chlorine adsorption sites to silicon reaction sites.
Yet unclear is how hydrogen is involved in the reaction
mechanism of chlorosilane formation.
in Surface Science and Catalysis; Ward, J. W., Ed.; Elsevier: Amsterdam,
1988; Vol 38, p 415.
(25) Banholzer, W. F.; Burrell, M. C. Surf. Sci. 1986, 176, 125.
(26) Banholzer, W. F.; Burrell, M. C. J. Catal. 1988, 114, 259.
(27) Frank, T. C.; Kester, K. B.; Falconer, J. L. J. Catal. 1985, 95, 396.
(28) Magrini, K. A.; Falconer, J. L.; Koel, B. E. In Catalyzed Direct
Reactions of Silicon; Lewis, K. M., Rethwisch, D. G., Eds.; Studies in
Organic Chemistry Series 49; Elsevier: Amsterdam, 1993; p 249.
(29) Magrini, K. A.; Falconer, J. L. In Catalyzed Direct Reactions of
Silicon; Lewis, K. M., Rethwisch, D. G., Eds.; Studies in Organic Chemistry
Series 49; Elsevier: Amsterdam, 1993; p 265.
(
30) Acker, J. Ph.D. Thesis, Freiberg University of Mining and Technol-
ogy, Freiberg, Germany, 1999.
31) Acker, J.; Bohmhammel, K.; Roewer, G. In Silicon for the Chemical
(
Industry IV; Øye, H. A., Rong, H. M., Nygaard, L., Sch u¨ ssler, G., Tuset,
J. Kr., Eds.; Norwegian University of Science and Technology: Trondheim,
Norway, 1998; p 133.
(32) Acker, J.; Bohmhammel, K.; Roewer, G. In Organosilicon Chem-
istry IV: From Molecules to Materials; Auner, N., Weis, J., Eds.; Wiley-
VCH: Weinheim, Germany, 2000; p 818.
(
33) Nagaki, H.; Simon, A.; Borrmann, H. J. Less-Common Met. 1989,
1
56, 193.
(
(
34) Mattausch, H.; Simon, A. Angew. Chem. 1998, 110, 498.
35) Mattausch, H.; Oeckler, O.; Simon, A. Z. Anorg. Allg. Chem. 1999,
Acknowledgment. We gratefully acknowledge Dr. Kenrick
M. Lewis, Osi Specialties, Crompton Corporation, Tarrytown,
NY, and Professor Gerhard Roewer, Freiberg University, for
fruitful discussions. We thank Dr. S. Roosendaal and Dr. A. J.
Vredenberg, Department of Atomic and Interface Physics, State
University Utrecht, The Netherlands, for RBS investigations and
Dr. N. Mora-Diez, Department of Chemistry, Dalhousie Uni-
versity, Halifax, Nova Scotia, Canada, for manuscripts of
submitted papers. J.A. thanks the Alexander von Humboldt
Foundation for funding provided by a Feodor Lynen Fellowship.
625, 297.
(36) Acker, J.; R o¨ ver, I.; Roewer, G.; Bohmhammel, K. In Silicon for
the Chemical Industry IV; Øye, H. A., Rong, H. M., Nygaard, L., Sch u¨ ssler,
G., Tuset, J. Kr., Eds.; Norwegian University of Science and Technology:
Trondheim, Norway, 2000; p 121.
(
37) Alvarez-Idaboy, J. R.; Mora-Diez, N.; Vivier-Bunge, A. J. Am.
Chem. Soc. 2000, 122, 3715.
38) Alvarez-Idaboy, J. R.; Mora-Diez, N.; Boyd, R. J.; Vivier-Bunge,
(
A. J. Am. Chem. Soc. 2001, 123, 2018.
(39) Mora-Diez, N.; Alvarez-Idaboy, J. R.; Boyd, R. J. J. Phys. Chem.
001, 105, 9034.
2
(40) Becerra, R.; Frey, H. M.; Mason, B. P.; Walsh, R.; Gordon, M. S.
J. Chem. Soc., Faraday Trans. 1995, 91, 2723.
(41) Becerra, R.; Boganov, S. E.; Egorov, M. P.; Faustov, V. I.; Nefedov,
O. M.; Walsh, R. J. Am. Chem. Soc. 1998, 120, 12657.
References and Notes
(1) Sinfelt, J. H. Prog. Solid State Chem. 1975, 10, 55.
(
42) Muench, J. L.; Kruuv, J.; Lepock, J. R. Cryobiology 1996, 33, 253.
(2) Schmalzried, H. Chemical Kinetics of Solids; VCH: Weinheim,
(43) Luo, L.; Zybill, C. E.; Ang, H. G.; Lim, S. F.; Chua, D. H. C.; Lin,
Germany, 1995.
J.; Wee, A. T. S.; Tan, K. L. Thin Solid Films 1998, 325, 87.
(3) Voorhoeve, R. J. H. Organohalosilanes: Precursors to Silicones;
(
(
(
44) McCabe, R. W.; McCready, D. F. J. Phys. Chem. 1986, 90, 1428.
45) Frank, T. C.; Falconer, J. L. Langmuir 1985, 1, 104.
46) Tamhankar, S. S.; Gorkan, A. N.; Doraiswamy, L. K. Chem. Eng.
Elsevier: Amsterdam, 1967.
(4) Bazant, V. Pure Appl. Chem. 1969, 19, 473.
(5) Turetskaya, R. A.; Andrianov, K. A.; Trofimova, I. V.; Chernyshev,
Sci. 1981, 36, 1365.
E. A. Russ. Chem. ReV. (Engl. Transl.) 1975, 44, 212.
(
(
(
47) Acker, J.; Bohmhammel, K. Thermochim. Acta 1999, 337, 187.
48) Xiao, C.; Hagelberg, F. J. Mol. Struct. 2000, 529, 241.
49) Kim, K. H., Park, T. S.; Sohn, K. S.; Kim, S. C. J. Korean Phys.
(6) Clarke, M. P. J. Organomet. Chem. 1989, 376, 165.
(7) Catalyzed Direct Reactions of Silicon; Lewis, K. M., Rethwisch,
D. G., Eds.; Studies in Organic Chemistry Series 49; Elsevier: Amsterdam,
993.
8) Bohmhammel, K.; Roewer, G.; Walter, H. J. Chem. Soc., Faraday
Trans. 1995, 91, 3879.
9) Walter, H. Ph.D. Thesis, Freiberg University of Mining and
Technology, Freiberg, Germany, 1995.
10) Walter, H.; Roewer, G.; Bohmhammel, K. J. Chem. Soc., Faraday
Trans. 1996, 92, 4605.
11) Over, H.; Kim, Y. D.; Seitsonen, A. P.; Wendt, S.; Lundgren, E.;
Schmid, M.; Varga, P.; Morgante, A.; Ertl, G. 2000, 287, 1474.
Soc. 1987, 20, 82.
50) Tanaka, K.; Saito, T.; Suzuki, K.; Hasegawa, R. Phys. ReV. B:
Condens. Matter 1985, 32, 6853.
51) Peterson, K. L.; Hsiao, J. S.; Chopra, D. R.; Dillingham, T. R. Phys.
ReV. B: Condens. Matter 1988, 38, 9511.
52) Nishitani, S. R.; Fujii, S.; Mizuno, M.; Tanaka, I.; Adachi, H. Phys.
1
(
(
(
(
(
(
ReV. B: Condens. Matter 1998, 58, 9714.
(53) Peterson, K. L.; Hsiao, J. S. Phys. ReV. B: Condens. Matter 1988,
(
38, 10911.
(54) Larsson, R.; Folkesson, B. Acta Chem. Scand. 1996, 50, 1060.
(55) Witusiewicz, V. T. J. Alloys Compd. 1994, 203, 103.
(56) Witusiewicz, V. T. J. Alloys Compd. 1995, 221, 74.
(57) Belyi, A. P.; Gorbunov, A. I.; Flid, R. M.; Golubtsov, S. A. Russ.
J. Phys. Chem. (Engl. Transl.) 1969, 43, 637.
(58) Calandra, C.; Bisi, O.; Ottaviani, G. Surf. Sci. Rep. 1985, 4, 271.
(59) d’Heurle, F. M.; Gas, P. J. Mater. Res. 1986, 1, 205.
(60) d’Heurle, F. M. J. Mater. Res. 1988, 3, 167.
(61) Finstad, T. G. Phys. Status Solidi A 1981, 63, 223.
(62) d’Heurle, F. M.; Peterson, C. S.; Baglin, J. E. E.; La Placa, S. L.;
Wong, C. Y. J. Appl. Phys. 1984, 55, 4208.
(
(
12) Frank, T. C.; Kester, K. B.; Falconer, J. L. J. Catal. 1985, 95, 396.
13) Acker, J.; R o¨ ver, I.; Otto, R.; Roewer, G.; Bohmhammel, K. Solid
State Ionics 2001, 141-142, 583.
14) Mulla, I. S.; Choube, A. C.; Dongare, M. K.; Sinha, A. P. B. Indian
J. Chem., Sect. A 1988, 27, 756.
15) Acker, J.; Bohmhammel, K.; Henneberg, E.; Irmer, G.; R o¨ ver, I.;
Roewer, G. AdV. Mater. (Weinheim, Ger.) 2000, 12, 1605.
16) Barin, I. Thermochemical Data of Pure Substances; VCH Verlags
Gesellschaft: Weinheim, Germany, 1993; Parts I and II.
(
(
(
(
(
17) Kautsky, H.; Pfleger, H. Z. Allg. Anorg. Chem. 1958, 295, 206.
18) Weiss, A.; Beil G.; Meyer, H. Z. Naturforsch., B: Chem. Sci. 1980,
(63) Massalski, T. B. Binary Alloy Phase Diagrams; ASM Int.: Materials
Park, OH, 1990.
3
5, 25.

Reactivity of Intermetallic Compounds
J. Phys. Chem. B, Vol. 106, No. 19, 2002 5117
(
64) Masel, R. I. Principles of Adsorption and Reaction on Solid
Surfaces; Wiley & Sons: New York, 1996.
65) Kasdan, A.; Herbst, E.; Lineberger, W. C. J. Chem. Phys. 1975,
2, 541.
(75) Gubtsev, G. L. J. Phys. Chem. 1994, 98, 1570.
(76) Sicilia, E.; Toscano, M.; Mineva, T.; Russo, N. Int. J. Quantum
Chem. 1997, 61, 571.
(77) Ellison, G. B.; Engelking, P. C.; Lineberger, W. C. J. Am. Chem.
Soc. 1978, 100, 2556.
(78) Dixon, D. A.; Feller, D.; Peterson, K. A. J. Phys. Chem. A 1997,
101, 9405.
(79) Voorhoeve, R. J. H.; Vlugter, J. C. J. Catal. 1965, 4, 220.
(80) Waltenburg, H. N.; Yates, J. T. Chem. ReV. 1995, 95, 1589.
(81) Street, R. A. Hydrogenated Amorphous Silicon; Cambridge Uni-
versity Press: New York, 1991.
(
6
(
(
(
(
(
66) Berkowitz, J.; Greene, J. P.; Cho, H. J. Chem. Phys. 1987, 86, 1235.
67) Kalcher, J. Chem. Phys. 1987, 118, 273.
68) Nguyen, M. T. J. Mol Struct. 1988, 164, 391.
69) Kalcher, J.; Sax, A. F. J. Mol Struct. 1992, 253, 287.
70) Mineva, T.; Russo, N.; Sicilia, E.; Toscano, M. Int. J. Quantum
Chem. 1995, 56, 669.
71) Perrin, J.; Leroy, O.; Bordage, M. C. Contrib. Plasma Phys. 1996,
6, 3.
(
3
(82) Golubtsov, S. A.; Andrianov, K. A.; Ivanova, N. T.; Turetskaya,
R. A.; Podgornyi, I. M.; Feldshtein, N. S. Zh. Obshch. Khim. 1972, 43,
2000.
(
(
72) Swihart, M. T. J. Phys. Chem. A 2000, 104, 6083.
73) Pak, Ch.; Rienstra-Kiracofe, J. C.; Schaefer H. F. J. Phys. Chem.
A 2000, 104, 11232.
74) Pabst, R. E.; Margrave, J. L.; Franklin, J. L. J. Mass Spectrom.
Ion Phys. 1977, 25, 361.
(83) Turetskaya, R. A.; Golubtsov, S. A.; Andrianov, K. A.; Eszerets,
M. A.; Dzvonar, V. G. IzV. Akad. Nauk SSSR, Ser. Khim. 1973,
394.
(