Reactivity of Cu3Si of different genesis towards copper(I) chloride

Article abstract of DOI:10.1016/S0040-6031(00)00365-8

A comparative study of the reactivity between copper(I) chloride and three types of Cu₃Si obtained in a molten medium (Cu₃Si-Ref) and from mechanical activation following an annealing process (Cu₃Si-M2AP) or a self-propagating high-temperature synthesis (Cu₃Si-MASHS) was performed by thermogravimetry under vacuum using non-isothermal and isothermal methods of kinetic measurement. It was established that for the three Cu₃Si/CuCl systems, the acceleration and decay stages in the temperature range 145-215°C are very closely approximated by an equation of the Prout-Tompkins type where an autocatalytic process was proposed. The lower apparent activation energy obtained for the Cu₃Si-MASHS/CuCl system (63 kJ mol^-1 against 68 and 78 kJ mol^-1 for Cu₃Si-M2AP and Cu₃Si-Ref, respectively) has been attributed to a small grain size which induces nanoscale contacts between reactants and impedes CuCl to sublime. (C) 2000 Elsevier Science B.V.

Full text of DOI:10.1016/S0040-6031(00)00365-8

Thermochimica Acta 351 (2000) 71±77
Reactivity of Cu3Si of different genesis
towards copper(I) chloride
H. Souhaa,b, F. Bernarda, E. Gaffetc, B. Gillota,*
a
  Â
Laboratoire de Recherches sur la Reactivite des Solides, UMR 5613, Universite de Bourgogne,
9, Avenue Alain Savary, BP 47870 F-21078 Dijon, Cedex, France
 Á
Laboratoire de Chimie Physique, Faculte des Sciences Dhar El Mehraz, BP 1716 Fes, Morocco
b
c
  Â
UPR A806 CNRS, Groupe `Nanomateriaux', Universite de Technologie de Belfort-Montbeliard,
F-90010 Belfort, Cedex, France
Received 26 October 1999; received in revised form 12 December 1999; accepted 6 January 2000
Abstract
A comparative study of the reactivity between copper(I) chloride and three types of Cu3Si obtained in a molten medium
(Cu3Si-Ref) and from mechanical activation following an annealing process (Cu3Si-M2AP) or a self-propagating high-
temperature synthesis (Cu3Si-MASHS) was performed by thermogravimetry under vacuum using non-isothermal and
isothermal methods of kinetic measurement. It was established that for the three Cu3Si/CuCl systems, the acceleration and
decay stages in the temperature range 145±2158C are very closely approximated by an equation of the Prout±Tompkins type
where an autocatalytic process was proposed. The lower apparent activation energy obtained for the Cu3Si-MASHS/CuCl
system (63 kJ mol 1 against 68 and 78 kJ mol 1 for Cu3Si-M2AP and Cu3Si-Ref, respectively) has been attributed to a small
grain size which induces nanoscale contacts between reactants and impedes CuCl to sublime. # 2000 Elsevier Science B.V.
All rights reserved.
Keywords: Cu3Si/CuCl system; Mechanically activated powders; Thermogravimetry; Nucleation-growth; Autocatalytic process
1. Introduction
silicon is initially reacted in the solid phase with a
small quantity of CuCl (8±10%) to prepare the contact
mass for the main reaction. The contact mass consists
of silicon and catalytic quantities of the Cu3Si phase
and its selectivity is de®ned by the ratio CH3SiCl3/
(CH3)2SiCl2. Most researchers consider that the for-
mation of Cu3Si is accomplished by the reaction
between Cu and CuCl to give ®nely divided Cu and
gaseous SiCl4, Cu then diffuses into the remaining Si
matrix to form Cu3Si [5].
Methylchlorosilanes, in particular, dimethyldi-
chlorosilane, which are the starting materials for the
production of a variety of silicones, are synthesized by
a gas±solid reaction between CH3Cl and excess silicon
in the presence of the alloy Cu3Si (the so-called Z-
phase) as catalyst [1±4]. In practice, a large quantity of
We have previously established [6] that the reaction
between Si and CuCl can form the Cu3Si, Cu5Si and
Cu phases through successive reactions according to
* Corresponding author. Tel.: ‡33-3-80-39-61-42;
fax: ‡33-3-80-39-61-67.
E-mail address: bgillot@satie.u.bourgogne.fr (B. Gillot)
0040-6031/00/$ ± see front matter # 2000 Elsevier Science B.V. All rights reserved.
PII: S 0 0 4 0 - 6 0 3 1 ( 0 0 ) 0 0 3 6 5 - 8

72
H. Souha et al. / Thermochimica Acta 351 (2000) 71±77
the scheme:
was used to activate an annealing process (AP) or a
self-propagating high-temperature synthesis (SHS)
reaction. The kinetics of these three Cu3Si/CuCl reac-
tions which results in a change in mass due to evolu-
tion of SiCl4 gas following the global reaction
7Si ‡ 12CuCl ! 3SiCl4 ‡ 4Cu3Si
9Cu3Si ‡ 8CuCl ! 2SiCl4 ‡ 7Cu5Si
Cu5Si ‡ 4CuCl ! SiCl4 ‡ 9Cu
(1)
(2)
(3)
Cu3Si ‡ 4CuCl ! 7Cu ‡ SiCl4 (4)
has been investigated by thermogravimetry under
vacuum.
In such a scheme, Cu is the ®nal product of the
reaction, whereas Cu3Si and Cu5Si are intermediates
which react with CuCl. The above reactions are the
overall reactions, and one of the elementary steps
comprises the reaction between Cu and CH3Cl to give
CuCl which in turn reacts with Si to regenerate the
active Cu3Si alloy according to reaction (1).
2. Reactants and analytical methods
Three different powder samples of Cu3Si were used:
One prepared from Cu and Si in a molten medium at
8808C. This powder with a grain size close to 300 nm
is referred to hereafter as Cu3Si-Ref. The X-ray
analysis reveals in addition to the Cu3Si phase, the
presence of a large proportion of copper (Fig. 1c).
Two other Cu3Si powder referred as Cu3Si-M2AP
and Cu3Si-MASHS were prepared by mechanical
high-energy ball milling. Pure elemental powders
(3Cu‡Si) were sealed into 45 ml stainless steel vials
The formation of Cu3Si and its reactivity with CuCl
have already been studied by different methods such
as those using solid±solid reactions, solid±gas reac-
tions [7,8] or melting environment [9]. Nevertheless,
due to the complexity of the reaction mechanism and
the dif®culty of obtaining a pure Cu3Si, in this paper,
we report a comparative study of the reactivities of
three samples of Cu3Si of which two are prepared by
mechanical activation (MA) of Cu and Si powder
mixture. This mechanical high-energy ball milling
Fig. 1. X-ray diffraction patterns of Cu3Si: (a) Cu3Si-MASHS; (b) Cu3Si-M2AP; (c) Cu3Si-Ref.

H. Souha et al. / Thermochimica Acta 351 (2000) 71±77
73
with four stainless steel balls (15 mm in diameter, 14 g
in weight) under enclosed air. The ball to powder mass
ratio was 6/1. Mechanical activation was performed
using a planetary ball mill, hereafter described as the
G5 machine, which allows shock frequency and shock
energy to be independently selected [10]. The vials
were ®xed onto a rotating disc (rotation speed O) and
rotated in the opposite direction to the disc with a
speed o. The milling duration was equivalent to 2 h to
avoid the formation of some intermetallic fractions,
but to form a chemical gradient at a nanoscale. G5/
350/50/2 h ball milling condition was selected to study
the reactivity under extreme thermal conditions of an
3Cu‡Si mechanically activated powders. In order to
obtain the pure Cu3Si compound, the mechanical
activation was followed by a second step which con-
sists of a high temperature annealing (5008C for 24 h)
for Cu3Si-MA2P and of a self-propagating low-tem-
perature synthesis reaction (1808C for 10 s) for Cu3Si-
MASHS [11,12]. X-ray diffraction patterns of the
solid products formed show that only the mechanical
activation step added to the SHS process leads to the
formation of Cu3Si without copper formation (Fig.
1a). Small amounts of copper were observed in Cu3Si
prepared using the M2AP process (Fig. 1b). The grain
size determined by XRD pro®le analysis and scanning
electron microscopy were 150 and 80 nm, respec-
tively.
Thermogravimetric analysis (TG) was used ®rst to
determine the main features of the reaction by inves-
tigating the mass loss from the Cu3Si sample during
reduction. The total mass change due to evolution of
SiCl4 was measured as a function of time by means
of a Setaram B70 thermobalance with a sensitivity of
0.02 mg. Cu3Si and CuCl (300 mesh, 99.999% pure,
lot 400151, Alfa Products) powders were mixed inti-
mately and manually ground at room temperature
under an inert atmosphere in an agate mortar for
15 min in a reactant ratio Cu3Si:CuClˆ1.50:1.00 cor-
responding to an excess of Cu3Si. 50 mg of the
mixture was subsequently evacuated, outgassed in
vacuo (1 Pa) for 1 h at room temperature and then
for 1 h at 708C before the sample was heated to the
temperature of the experiment. The preheating at 708C
is useful to remove any moisture present and to
shorten the time required for the mixture to reach
the selected temperature. The SiCl4(g) liberated dur-
ing the reaction was condensed in a liquid nitrogen
trap and the mass loss was calculated per 100 mg of
CuCl.
X-ray diffraction patterns were recorded on an
`INEL CPS 120' linear counter (curved position sen-
sitive) equipped with monochromatized Cu Ka radia-
tion and calibrated by a quartz standard. The
resolution was 0.028 (2y) at 2 deg min 1. The mor-
phological analysis was carried out with a scanning
electron microscopy (SEM `Cambridge' 250 MK2).
3. Results
3.1. Non-isothermal reduction
Fig. 2 shows the thermogravimetric curves Dmˆ
f(T) at a heating rate of 38C min 1 for the sublimation
of CuCl, when it is placed alone in the scoop (curve 1)
or mixed with an inert oxide such as Al2O3 (curve 2).
In the absence of an inert oxide, the sublimation
temperature of CuCl is lowered by ca. 408C. As a
result, the temperature at which CuCl begins to react
with Cu3Si can serve as a criterion for the reactivity of
the mixture. For comparison, curves 3, 4 and 5 give the
variation of mass loss with temperature for Cu3Si-Ref,
Cu3Si-M2AP and Cu3Si-MASHS, respectively. The
theoretical maximum mass loss of the system is
calculated (Dmcalˆ42.8 mg, dashed line) for the com-
plete reduction of CuCl by Cu3Si according to reaction
(4). It can be seen from Fig. 2 that the difference
between the experimental (Dmexp) and the theoretical
maximum mass loss (Dmcal) decreases as the grain size
of Cu3Si is diminished, that is, when the reactive
surface is larger. This implies that the mass fraction
of CuCl which sublimes without reaction decreases as
the grain size is decreased.
We can also postulate, that because the vaporization
of CuCl starts at about 1608C, the mass loss from the
mixture below this temperature is only due to a solid±
solid reaction between CuCl and Cu3Si. Above this
temperature, we have also to consider that the solid±
gas reaction prevails.
3.2. Isothermal reduction
The isothermal TG curves Dmˆf(t) for the reaction
between CuCl and the three types of Cu3Si are shown
in Figs. 3±5 (curves (a)). These curves have an

74
H. Souha et al. / Thermochimica Acta 351 (2000) 71±77
Fig. 2. Curves Dmˆf(T) obtained in vacuum for the reaction between Cu3Si and CuCl powders: (1) CuCl alone in the scoop; (2) CuCl mixed
with an inert oxide; (3) Cu3Si-Ref; (4) Cu3Si-M2AP; (5) Cu3Si-MASHS.
Fig. 3. Cu3Si-Ref/CuCl: (a) mass loss Dm vs. time and (b) rate Dm/dt vs. Dm.
Fig. 4. Cu3Si-M2AP/CuCl: (a) mass loss Dm vs. time and (b) rate Dm/dt vs. Dm.

H. Souha et al. / Thermochimica Acta 351 (2000) 71±77
75
Fig. 5. Cu3Si-MASHS/CuCl: (a) mass loss Dm vs. time and (b) rate Dm/dt vs. Dm.
S-shape, indicating a nucleation growth mechanism,
that is, the rate initially increases, reaches a maximum
and then decreases. The corresponding rate curves
d (Dm/dt)ˆf(Dm) (Figs. 3±5, curves (b)) allow us to
determine the maximum rate aiˆDmi/Dmexp for each
Cu3Si, where Dmi corresponds to the Dm value at
which the rate is maximum. As already observed from
the Dmˆf(T) curves, the mass loss for Cu3Si-MASHS
best approaches the calculated value. It is also evident
from Fig. 5b that Dmi (aiˆ0.48) can be considered as
constant in the range 145±2108C.
It is worth mentioning that for the Cu3Si-Ref/CuCl
system where the mass fraction of CuCl which sub-
limes without reaction with CuCl is the highest, and
thereby, affects the kinetic behavior, Dmi strongly
changes with temperature (Fig. 3b).
3.3. Solid products of the reaction
X-ray diffraction analyses for the Cu3Si-MASHS/
CuCl system during the course of reaction indicates
only the formation of copper (Fig. 6a) which con®rms
the reduction of CuCl by Cu3Si in a single step
according to reaction (4). In the case of Cu3Si-Ref
and Cu3Si-M2AP, the reaction with CuCl leads to
Cu15Si4, Cu5Si and Cu (Fig. 6, curves (b) and (c))
through successive reactions as follows:
31Cu3Si ‡ 12CuCl ! 7Cu15Si4 ‡ 3SiCl4 (5)
9Cu15Si4 ‡ 20CuCl ! 31Cu5Si ‡ 5SiCl4 (6)
Cu3Si ‡ 4CuCl ! 9Cu ‡ SiCl4
(7)
4. Discussion and conclusion
From the studies just presented, it is clear that for
the three types of Cu3Si, the isothermal curves of
È
fraction reduced versus time were typically sigmoõd.
However, in the light of the above results, it appears
from Fig. 5b that ai can be only considered as constant
with a value close of 0.50 for Cu3Si-MASHS/CuCl
Fig. 6. X-ray diffraction patterns obtained after the reaction
between CuCl and: (a) Cu3Si-MASHS at 1808C; (b) Cu3Si-
M2AP at 1908C; (c) Cu3Si-Ref at 2308C.

76
H. Souha et al. / Thermochimica Acta 351 (2000) 71±77
system. Moreover, for this system, the isothermal
curves aˆf(t) can be superimposed for 0.10<a<0.85,
when a is the fractional mass change at time t de®ned
by aˆDmt/Dmexp. By the use of scaling factors, an
written as:
ꢀ
ꢁ
da
dt
a
2ai
ˆ k1a 1
1
activation energy of 63Æ3 kJ mol was obtained.
From the temperature dependence of k1, an apparent
1
The af®ne character of the curves indicates that the
states traversed by the system during the reaction are
dependent on intensive variables such as temperature
and pressure [13] (in our case, pressure of 1 Pa). The
rate equation for the process can be written in terms of
separated variables as:
activation energy of 61Æ3 kJ mol is deduced. This
value is very close to that calculated from the af®nity
ratio.
X-ray diffraction results and morphological
information which show that ®nely divided copper
is one of the reaction products predicted that we
must consider an autocatalytic process rather than a
branched chain nucleation mechanism, that is the
Prout±Tompkins model. Reactions involving in situ
formation of metallic copper are known to be auto-
catalytic in nature [15]. Thus, reduction of CuO and
NiO by H2 as well as by formaldehyde are both
autocatalytic reactions [16].
For the Cu3Si-Ref and Cu3Si-M2AP/CuCl systems,
the Dmi (or ai) value was found to decrease with
decreasing temperature. This decrease is particularly
noticeable for Cu3Si/CuCl system (Fig. 3b) where the
mass fraction of CuCl which sublimes is very impor-
tant. The presence of some intermediate Cu-Si com-
pounds, even after complete reaction, presumably also
re¯ects the loss CuCl by sublimation.
da
dt
v ˆ
ˆ kT kp e E=kT fꢀa†
(8)
where f(a) usually represents the kinetic rate expres-
sion. At constant temperature and pressure, the shape
of the curves is determined only by f(a). We ®nd a
very good linear representation of the sigmoid curves
(Fig. 7) if f(a) is an equation of type:
ꢀ
ꢁ
a
fꢀa† ˆ log
ˆ k1t ‡ C
(9)
1
ꢀa=2ai†
which is the integrated form of the Prout±Tompkins
equation [14]. The rate for the process can be thus
A change in ai with temperature indicates a shift in
the controlling mechanism from one to another in a
series of successive elementary steps [17]. The present
reaction being a consecutive step in the series of
reactions (5), (6) and (7), such a shift in controlling
mechanism is possible. For these two types of Cu3Si,
the maximum rate take place, depending of the tem-
perature, at values of 0.30<ai<0.40 for Cu3Si-Ref and
0.39<ai<0.44 for Cu3Si-M2AP. When the maximum
rate occurs at a value different from aiˆ0.50, it is
necessary to integrate Prout and Tompkins original
equation with different boundary conditions.
When this is done, Eq. (11) results [18]
da
dt 2ai
k1
ˆ
ꢀ1 2ai ‡ a†ꢀ1 a†
(11)
Eq. (11) is valid for a maximum rate occurring
at values of 0<a<0.50. It is obvious that the equa-
tions have no physical reality above a>0.50 because
da/dt values become negative for the onset of the
reaction.
Fig. 7. Conversion as a function of time plotted in terms of Eq. (9).

H. Souha et al. / Thermochimica Acta 351 (2000) 71±77
77
Integration of Eq. (11) gives
References
1
2ai ‡ a
1
ai
log
ˆ
ꢀk1t ‡ C†
(12)
[1] R.J.H. Voorhoeve, Organohalosilanes: Precursors to Sili-
cones, Elsevier, Amsterdam, 1967, pp. 253.
[2] E. Rowkow, Comprehensive Inorganic Chemistry, Silicon,
Pergamon, Oxford, 1973, pp. 1338.
1
a
ai
For example, substituting aiˆ0.35, which is a good
average of ai for Cu3Si-Ref (an average value of ai was
used rather than individual values because of the
¯atness of the d Dmi/dt versus Dm curves near each
maximum), and rearranging gives:
[3] N. Floquet, S. Yilmaz, J.L. Falconer, J. Catal. 143 (1994)
348.
[4] D.H. Sun, B.E. Bent, A.P. Wright, B.M. Naasz, Catal. Lett. 46
(1997) 127.
0:3 ‡ a
[5] R.J.H. Voorhoeve, J.A. Lips, J.C. Vlugter, J. Catal. 3 (1964)
414.
log
ˆ ꢀk10 t ‡ C†
(13)
1
a
[6] G. Weber, D. Viale, H. Souha, B. Gillot, P. Barret, Solid State
Ionics 32/33 (1989) 250.
In plotting the kinetic data according to Eq. (13), it
was noticed that for a>0.77 for Cu3Si-Ref and a>0.82
for Cu3Si-M2AP, the data points deviate signi®cantly
from the linear part of the extrapolated model plot.
Arrhenius plots within the restricted ranges of
0.25ꢁaꢁ0.77 and of 0.20ꢁaꢁ0.82 gave an apparent
activation energy of 78Æ5 kJ mol 1 for Cu3Si-Ref and
[7] G. Weber, D. Viale, H. Souha, B. Gillot, C.R. Acad. Sci.,
Â
Paris 307, Serie II (1988) 1161.
[8] H. Souha, G. Weber, B. Gillot, Thermochim. Acta 171 (1990)
215.
[9] G. Weber, N. Gourgouillon, D. Viale, B. Gillot, P. Barret,
React. Solids 6 (1988) 75.
[10] M. Abdellaoui, E. Gaffet, Acta Mater. 44 (1996) 725.
[11] F. Bernard, F. Charlot, E. Gaffet, J.C. Niepce, Int. J. SHS 7
(1998) 233.
1
68Æ3 kJ mol for Cu3Si-M2AP.
In conclusion, this study clearly shows that the
MASHS process leads to the production of a pure
Cu3Si compound with a low grain size which allows
the reduction of CuCl in a single step where the
kinetics is governed by a nucleation-growth mechan-
ism and the maximum rate is observed at 50% reduc-
tion. Thereby, this reactant is more reactive than either
Cu3Si-Ref and Cu3Si-M2AP where the particle size is
more important and a uniform mixing of the powders
could be achieved, probably yielding intermediate
intermetallic compounds.
[12] F. Bernard, H. Souha, E. Gaffet, M. Zenkouar, B. Gillot, J.C.
Niepce, Int. J. SHS, (2000) in press.
[13] N. Lyakhov, React. Solids 8 (1990) 313.
[14] E.G. Prout, F.C. Tompkins, Trans. Faraday Soc. 40 (1994)
488.
[15] S.S. Tamhankar, A.N. Gokarn, L.K. Doraiswamy, Chem. Eng.
Sci. 36 (1981) 1365.
[16] J. Bandrowski, C.R. Bickling, K.H. Yang, O. Hougen, Chem.
Eng. Sci. 17 (1962) 379.
   Á
[17] P. Barret, Cinetique Heterogene, Gauthier-Villars, Paris,
1973, pp. 270.
[18] W.D. Bond, J. Phys. Chem. 66 (1962) 1573.