Selective chemical vapor transport as a means of varying the composition of nonstoichiometric indium sulfides

Article abstract of DOI:10.1134/S0020168507110039

Investigation of halide vapor transport with the participation of indium and indium sulfides in a closed system indicates that applying a temperature gradient is insufficient for quantitative spontaneous indium transfer from the lower indium sulfides or for the transport of elemental indium. A major reason for this is that the vapor phase over indium and its lower sulfides is dominated by monohalides. Impossible under conventional experimental conditions, chemical vapor transport can be achieved by diluting indium with an inert substance, e.g., gold. Our results indicate that the vapor transport of indium is possible in systems of the form indium sulfide-indium chloride vapor-charge (Au-In) and that chloride vapor transport can be used to nondestructively control the composition of indium sulfides. The transport process is shown to be selective and reversible. Conditions are determined for nondestructive chemical transport control over the composition of indium sulfides.

Full text of DOI:10.1134/S0020168507110039

ISSN 0020-1685, Inorganic Materials, 2007, Vol. 43, No. 11, pp. 1167–1178. © Pleiades Publishing, Inc., 2007.
Original Russian Text © A.Yu. Zavrazhnov, A.V. Naumov, A.V. Sergeeva, V.I. Sidei, 2007, published in Neorganicheskie Materialy, 2007, Vol. 43, No. 11, pp. 1303–1315.
Selective Chemical Vapor Transport as a Means of Varying
the Composition of Nonstoichiometric Indium Sulfides
a
a
a
b
A. Yu. Zavrazhnov , A. V. Naumov , A. V. Sergeeva , and V. I. Sidei
a Voronezh State University, Universitetskaya pl. 1, Voronezh, 394006 Russia
b Uzhgorod State University, Pidhirna ul. 46, Uzhgorod, 88000 Ukraine
e-mail: zav@comch.ru
Received March 1, 2007
Abstract—Investigation of halide vapor transport with the participation of indium and indium sulfides in a
closed system indicates that applying a temperature gradient is insufficient for quantitative spontaneous indium
transfer from the lower indium sulfides or for the transport of elemental indium. A major reason for this is that
the vapor phase over indium and its lower sulfides is dominated by monohalides. Impossible under conventional
experimental conditions, chemical vapor transport can be achieved by diluting indium with an inert substance,
e.g., gold. Our results indicate that the vapor transport of indium is possible in systems of the form indium sul-
fide–indium chloride vapor–charge (Au–In) and that chloride vapor transport can be used to nondestructively
control the composition of indium sulfides. The transport process is shown to be selective and reversible. Con-
ditions are determined for nondestructive chemical transport control over the composition of indium sulfides.
DOI: 10.1134/S0020168507110039
INTRODUCTION
THEORETICAL ANALYSIS AND SELECTION
OF COMPONENTS
Selective chemical vapor transport (SCVT) is par-
ticularly attractive for finely tuning the composition
(nonstoichiometry) of phases with a low vapor pressure
[1, 2].
OF THE TRANSPORT SYSTEM
As pointed out earlier [3], indium chemical trans-
port in the vapor transport system under consideration
must be due to the selective reaction
The basic principle of the method is that a selective
transport reaction may proceed in the forward direction
(extraction of a component from the material) or
reverse direction (saturation of the material with the
component being transported), depending on the pro-
cess conditions. The process is run in a closed system
(ampule) containing the sample and charge. The latter
acts as a source or getter of the component to be trans-
ported (Fig. 1). The vapor phase in the ampule consists
of several vapor species, which insure selective trans-
port. According to earlier findings [1], the direction of
selective vapor transport is only dependent on the sam-
ple temperature T2, charge temperature T1, and charge
composition x1. The sample and charge compositions
stop varying when the system reaches a steady state, in
which no mass transport occurs. The sample composi-
tion in the final state is determined by the same vari-
ables: T2, T1, and x1.
'
2In'' + In2Hal4 = 4InHal',
(1)
p4InHal
---------------
K =
,
(2)
p
In2Hal4
where the primed and double-primed species are in the
vapor phase and condensed phase, respectively (In6S7,
1
2
T1
T2
3
4
In this work, we focus on approaches for controlling
the composition of solid phases in the In–S system
using SCVT without recrystallization of the starting
material.
Fig. 1. Schematic representation of the experimental geom-
etry for varying the composition of low-volatility binary
phases via SCVT: (1) two-zone furnace, (2) quartz ampule
filled with transport agent vapor, (3) sample, (4) charge act-
ing as a source or getter of the component being transported.
1167

1168
ZAVRAZHNOV et al.
1090
lated as the too high potential of indium (µ1 ) in the
In
metallic indium charge:
1270
µ1 = 1/2RT lnK+F(T).
(3)
L
In
Here, F(T) is related to standard potentials and depends
852
780
683
on temperature only.
1070
870
670
470
751
Two approaches have been proposed to reduce the
752
2
chemical potential µIn:
L1 + L2
660
638
to dilute the charge with an inert component R such
that a homogeneous solution can be obtained over the
entire concentration range of interest;
to bind the component to be transported (In) in a dis-
tinct chemical compound.
659
In2.8S4
593
414
L + InS
DILUTION OF THE COMPONENT
TO BE TRANSPORTED (CONTINUOUS SERIES
OF SOLUTIONS)
156
Consider the first approach. It is reasonable to
expect that indium dilution will be accompanied by a
In + InS
decrease in activity and will reduce µ1 according to
In
the well-known relation
0
10
20
30
40
50
60
70
80
S
In
mol %
µ1 = µ01 + RT lnaIn,
(4)
In
In
Fig. 2. T–x phase diagram of the In–S system [4].
where µ1 is the chemical potential of the indium in the
In
charge, and µ01 is the chemical potential of pure metal-
lic indium.
In3 – xS4 is indium in the condensed phase, e.g., in an
indium sulfide).
In
By analogy with previous studies of SCVT with the
participation of gallium selenides [1, 2], we attempted
to use metallic indium as a controlling charge, but we
failed to control the sample composition in the system
sample (In–S)–vapor (indium halides)–charge (In
metal): independent of the process conditions, any of
the known indium sulfides (InS, In6S7, In3 – xS4, and
According to Eq. (4), the decrease in activity will
cause a reduction in K and, hence, a rise in the content
of In2Hal4, necessary for indium chemical transport.
Transport selectivity imposes the following require-
ments on the diluent R:
(1) It must have a negligible vapor pressure.
(2) It must be nonreactive with the transport agent.
(3) The forming solution should not exhibit large
positive deviations from ideality.
As the diluent, we used gold, which meets all of the
requirements above [3]. The Au–In system contains
several low-melting-point intermetallic phases (Fig. 3)
[12]. The use of melts appears more attractive com-
pared to intermetallic-based solid solutions from the
viewpoint of the rate of indium exchange in the bulk of
the charge.
As the transport agent, we used chlorine (in the form
of InCl), first, because indium chlorides have higher
vapor pressures than do bromides and iodides [3] and,
second, because, by analogy with halides of some ele-
ments (Ga, Ge, Si, and others), we assumed that the
1
In2S3 [4–7] (Fig. 2) ) converted to the most indium-rich
phase, InS. Transport was impossible even in a simpli-
fied system, where the indium sulfide in the sample was
replaced with pure indium.
The lack of transport can be interpreted as follows:
According toYasuo Kunia et al. [8–10], at temperatures
above 400°ë equilibrium (1) with the participation of
pure liquid indium is shifted almost fully to the right
(K
∞). In2Hal4, which returns the halogen to the
charge, is essentially missing in the vapor phase, and
indium transport is impossible. Given the relation
between K and the chemical potential of the indium
present in the condensed phase in equilibrium with the
vapor phase [11], the problem in hand can be formu-
1
Various designations have been proposed in the literature for
2
phases close in composition to In S . The In S phase is often
2
3
3–x 4
The problem might possibly be resolved by using another trans-
designated β-In S [3], whereas in [4, 5] β-In S is used to
port agent. We, however, failed to find substances that would
selectively and reversibly react with crystalline indium sulfides
without dissolving in them.
2
3
2 3
denote the low-temperature phase with a sulfur content close to
60 mol %.
INORGANIC MATERIALS Vol. 43 No. 11 2007

SELECTIVE CHEMICAL VAPOR TRANSPORT
1169
1064
1270
1070
870
L
649.2
α
641
540.7
487
457.5
ζ
γ
ψ
ε
670
β
α'
224.3
470
156.6
~156
ε'
γ'
β1
Au
20
40
60
80
In
mol %
Fig. 3. T–x phase diagram of the In–Au system [12].
oxidized form of indium halides (In2Hal4) would be tial of indium in the charge, we used an indium-free
more stable in the chlorine compound.
substance, pure gold (µ1
–∞), as the charge and
In
To verify and refine these considerations in experi-
ment, the problem of controlling the composition of
indium sulfides via SCVT using a gold-diluted indium
charge was divided into the following steps:
metallic indium as the sample. Indium transport, i.e.,
mass transfer between the sample and the charge, was
examined using static gravimetry, a method used earlier
in studies of similar systems [1, 3].
1. First, the feasibility of the chloride transport of
indium was assessed using a simplified, sulfur-free sys-
tem: sample (In metal)–vapor (indium chlorides)–
charge (Au metal).
2. We examined the possible relationships between
the compositions of the In–S sample and indium–dilu-
ent (R) charge (during the process and in the final state,
in equilibrium with respect to indium).
In our preparations, about 1 g of metallic indium
(In-00) was placed at one end of an ampule, and a plate
of refined gold (~0.1 g) was placed at the other end. The
ampule also contained indium monochloride in an open
capillary. InCl was synthesized by melting a weighed
portion of AgCl and an excess of indium metal [3]. The
amount of InCl was chosen so as to ensure an unsatur-
ated vapor pressure pInCl = 2.5 × 104 Pa.
3. Next, we experimentally determined the interre-
lated changes in the compositions of the sample and
charge in the closed isothermal system sample (In–S)–
vapor (indium chlorides)–charge (In–Au melt) and
compared them with the prediction results (particular
attention was paid to the correspondence between the
sample and charge compositions in equilibrium with
respect to In).
Indium transport and dissolution in gold were easily
observable: the Au plate fully melted at a temperature
well below the melting point of gold, and the In–Au
melt changed in color from yellowish to silvery white.
According to the rate data (Fig. 4), indium was trans-
ported most rapidly in the initial stages of the process,
when the indium activity in the charge was lowest. As
the gold was diluted with indium, the indium concen-
4. Finally, we examined the possibility of revers- tration and activity in the melt increased; at 55 mol %
ible changes in the sample and charge compositions In, the transport almost ceased. The rate data
upon variations in their temperatures in the above (0.1 mol % accuracy in the compositions of In–Au
system.
Selective chemical transport of indium in the sys-
tem sample (In metal)–vapor (indium chlorides)–
charge (Au metal). In the context of the first problem,
we examined chloride transport of indium in a simpli-
alloys) were checked by an independent technique:
after the experiment, the In–Al alloy was weighed. The
compositions evaluated by the two methods coincided
to within 1 mol %.
Experiments under nonisothermal conditions
fied, sulfur-free system. To reduce the chemical poten- showed that temperature variations did not reverse the
INORGANIC MATERIALS Vol. 43 No. 11 2007

1170
ZAVRAZHNOV et al.
Indeed, if indium sulfide samples had a significant
(a)
indium vapor pressure (pIn, pIn ), Eq. (5) might be ful-
0.06
2
filled through physical transport of indium, that is, with
no transport agent. Our purpose is to predict how the
final sample composition (equilibrium In content) then
varies at a fixed temperature. Let us plot the sample
composition against the charge composition. It follows
from Eqs. (4) and (5) that to each equilibrium (with
respect to In) charge composition corresponds a partic-
ular sample composition, i.e., an indium sulfide of cer-
tain stoichiometry and structure. Let us follow the vari-
ation in the indium content of the sample as the charge
is diluted with R (Fig. 5a). Assume that selective
indium transport is conducted at a fixed temperature,
e.g., at 600°ë, where the In–S system contains five
phases: LIn (indium-based melt), InS, In6S7, In3 – dS4,
and LS (sulfur-based melt) (Fig. 5b).
In the initial stages of the process, a gradual varia-
tion in charge composition will be accompanied by a
gradual variation in the composition of the LIn solution
(Fig. 5a, segment AB), and, at a higher diluent (R) con-
tent, the equilibrium (with respect to indium) sample
will contain more sulfur (nontransportable component).
At a certain charge composition, the equilibrium chem-
ical potential of indium in the sample will reach the
value characteristic of the phase boundary of LIn. The
same potential of indium and, hence, the same charge
composition will correspond to the In-rich phase
boundary of the neighboring phase, InS (Fig. 5a, point C).
Moreover, this charge composition will correspond to a
heterogeneous mixture of these phases, independent of
their contents (Fig. 5a, vertical segment BC).
0.04
0.02
0
100
90
80
70
60
50
(b)
200
400
600
Time, min
Fig. 4. (a) Amount of indium transferred to the gold getter
and (b) getter composition as functions of time; sample and
getter temperatures, 823 and 773 K, respectively; C
=
InCl
3
7 mol/m .
Further raising the diluent concentration in the
charge will gradually increase the content of the non-
volatile component (S) within the homogeneity range
of the InS phase (Fig. 5a, segment CD). At the sulfur-
rich phase boundary of indium monosulfide, the same
charge composition will correspond to the composi-
tions of two coexisting phases: InS and In6S7. In this
manner, we obtain the broken line A–J in Fig. 5a. Thus,
to nonstoichiometric compositions of each phase corre-
sponds a certain composition range of the In–R solu-
tion. The use of such a solution as a charge must enable
the preparation of each intermediate phase (any sulfide)
with controlled stoichiometry, while maintaining the
selectivity of the process. The only obstacle to imple-
menting this approach is that it is desirable to use the
maximum possible amount of the charge. Otherwise,
both the sample and charge compositions will vary
markedly during the evolution of the system. This may
reduce the accuracy in predicting the final sample com-
position.
direction of mass transport, which implies that, in the
system under consideration, the gold and the forming
In–Au solution act solely as an indium getter, rather
than as a donor. This seems to be due to the large differ-
ence in chemical potential, µ2 – µ1 , especially in the
In
In
case of Au-rich In–Au solutions. Thus, our results dem-
onstrate that indium dilution with an inert (in the vapor
phase) substance ensures indium vapor transport in at
least one direction.
2. Possible relationships between the composi-
tions of the In–S sample and indium–diluent (R)
charge. Consider mass transport conditions in a closed
isothermal vapor transport system of the form sample
(In–S)–vapor (indium chlorides)–charge (In–R). Note
that, in this system, the transport agent has no effect on
the final state (equilibrium with respect to indium) of
the sample and charge, which is defined by the well-
known equality
The above considerations must apply to both physi-
cal and selective chemical transport of indium. The
only role of the transport agent is to drastically acceler-
ate indium transfer between the sample and charge.
This statement, however, only applies to isothermal
SCVT systems.
µ1 = µ2 ,
(5)
In
In
where subscripts 1 and 2 refer to the charge and sample,
respectively.
INORGANIC MATERIALS Vol. 43 No. 11 2007

SELECTIVE CHEMICAL VAPOR TRANSPORT
1171
(a)
(b)
J
L(S)
L(S)
In2S3
I
In3–δS4
60
55
50
0
H
In3–δS4
G
F
In6S7
InS
In6S7
InS
E
D
C
B
L(In)
L(In)
A
In
R
620
720
820
920
T, K
Fig. 5. (a) Diagram illustrating the predicted relationship between the compositions of In–S samples and In–R charges (isothermal
section at T ꢀ 870 K; R is a diluent forming a continuous series of solutions with In). (b) Partial T–x phase diagram of the In–S
system; the homogeneity ranges of the phases are represented only schematically.
3. Isothermal mass transport in the system sam- These results confirm that the transport process is selec-
ple (In–S)–vapor (indium chlorides)–charge (In–Au tive.
melt) and sample and charge compositions at equi-
Figure 6 illustrates the variations in the sample and
librium (with respect to In). In all our experiments,
charge compositions and presents the final (equilib-
samples (~0.1 to ~1 g) consisted of a particular In–S
rium) compositions. The plot in Fig. 6 is analogous to
phase. To raise the reaction rate, the samples were
3
the composition relationship diagram in Fig. 5. The
compositions may refer to the materials in the initial,
intermediate, and final (equilibrium with respect to In)
states of the vapor transport system. This allows one to
follow compositional changes and to correlate the final
compositions of the condensed phases in the system.
ground to a particle size no greater than 1 mm. The
charges used were weighed portions (~0.2 to 0.5 g) of
In–Au alloys containing 0 to 80 mol % Au. The process
was run in quartz ampules at temperatures from 550 to
650°ë and a chlorine content corresponding to pInCl
=
104 Pa. After annealing for a certain time, the ampules
were quenched. To this end, cold water was poured onto
them so that the indium chlorides condensed in the cen-
tral part of the ampule and did not contaminate the sam-
ple or charge, which were located at the ends of the
ampule. Next, the substances were withdrawn from the
ampule and weighed on a VLM-2000 analytical bal-
ance to determine the weight change. This procedure
allowed us to evaluate the compositional changes
caused by indium transport. After the measurements,
the experiment was continued and the above procedure
was repeated. The compositions of the heat-treated
materials were correlated with x-ray diffraction and,
occasionally, static gravimetry data.
The data in Fig. 6 were obtained at 650°C. The
arrows indicate the variations in the sample and charge
compositions during the evolution of the system to the
equilibrium (with respect to In) state. These data lead
us to the following conclusions:
1. The transport direction at T2 = T1 is governed by
the initial sample and charge compositions. At high
diluent (gold) contents, indium is transported to the
charge, and the indium content of the sample decreases.
At lower gold contents of the charge, indium is trans-
ported to the sample and dissolves in it.
3
Clearly, the equilibrium (with respect to In) state of the isother-
After 200 h of annealing, the sample and charge
weights stopped varying, suggesting that an equilib-
rium (with respect to In) state had been reached.
According to x-ray microanalysis and roentgenolumi-
nescence data for the condensed phases, the charge
contained no sulfur, and the sample contained no gold.
mal vapor transport system under consideration is also deter-
mined by the temperature and, to a far lesser extent, by the pres-
sure. The diagram in question must then be represented in four
dimensions (sample composition–charge composition–tempera-
ture–pressure). Thus, the two-dimensional diagrams below are
isobaric–isothermal sections of the composition relationship dia-
grams.
INORGANIC MATERIALS Vol. 43 No. 11 2007

1172
ZAVRAZHNOV et al.
61
59
57
55
53
51
49
1
2
3
4
In3–δS4
Initial state 3
In6S7
Initial state 2
Initial state 1
InS
34
39
44
49
54
59
64
69
74
mol % Au (In–Au)
Fig. 6. Diagram illustrating the relationship between the compositions of In–S samples and In–Au charges, as inferred from the
gravimetric data for the system sample (indium sulfides)–vapor (indium chlorides)–charge (In–Au melt); (1, 3, 4) evaluated from
the weight change of the charge (In–Au alloy), (2) evaluated from the weight change of the In–S sample. The arrows indicate the
evolution of the sample and charge compositions from the initial to final (200 h of annealing) state. The heavy segments represent
the sample and charge compositions in the equilibrium (with respect to In) state of the system at 650°ë.
2. Further evidence that the vapor transport system the monotonic segments of the final state indicate that
under consideration can reach the equilibrium (with the sulfide sample consists of a single phase with a con-
respect to In) state is provided by the fact that a given siderable homogeneity range.
composition can be obtained at higher and lower dilu-
ent contents, corresponding to the opposite transport
directions (different directions of the arrows in Fig. 6).
3. The final sample composition can be obtained using
different initial states (see the points near 57 mol % S and
55 mol % Au). Since the charge is homogeneous, the
evolution trajectory depends not only on the initial
compositions but also on the initial amounts of the sam-
ple and charge.
4. The sample and charge compositions are interre-
lated: increasing the diluent content of the charge
reduces the indium content of the sample. Each point
corresponding to a final, equilibrium state must be only
determined by the temperature (at slight variations in
total pressure) and is independent of the other factors
(the nature of the transport agent, the amount of solids,
etc.).
In addition, the data in Fig. 6 lead us to some con-
clusions about the homogeneity ranges of the interme-
diate phases in the In–S system. In particular, it seems
likely that the homogeneity range of the In3 – δS4 phase
is several percent in width and extends from ~58.3 to
60.3 mol % S at 650°C, that is, it includes the stoichio-
metric composition In2S3. The inflection in the compo-
sition relationship line in the In3 – δS4 phase region pro-
vides indirect evidence that different point defects
dominate in different composition regions of this
phase.
The results of similar experiments performed at
other temperatures (700 and 750°ë) indicate that mod-
erate changes in temperature (on the order of 100°ë)
have little effect on the relationship between the sample
and charge compositions.
In addition, it was important to ascertain whether
the indium transport between the sample and charge
was indeed due to an increase in the In2Cl4 (oxidized
5. As would be expected, the composition ranges in
which the sample is heterogeneous (two-phase) are rep-
resented by vertical segments, which reflect the rela- form) concentration in the vapor phase. We assumed
tionship between the equilibrium (with respect to In) that, at lower chemical potentials of indium in the con-
sample and charge compositions. On the other hand, densed phases (sample and charge) at partial equilib-
INORGANIC MATERIALS Vol. 43 No. 11 2007

SELECTIVE CHEMICAL VAPOR TRANSPORT
1173
rium, the equilibrium vapor phase was richer in In2Cl4
(a)
0.2
0.1
0
1
2
3
4
compared to InCl. In addition, we expected that
quenching had an insignificant effect on the relative
amounts of these vapor species, which, therefore, can
be assessed by analyzing the condensed phases.
1
Qualitative information can be gained from the
color of the condensed phases. It is well known that,
with increasing Cl content, solid In–Cl phases change
from colorless (In2Cl4 and InCl3) to yellow-orange
(InCl) [13, p. 56]. The observed color changes in the
condensed chlorides were consistent with the above
assumptions. The highest indium content (maximum
µIn) of the equilibrium charge (>55 mol % In) and sam-
ple (InS phase) corresponds to a bright orange color of
the condensed phases. When such condensed phases
are in equilibrium with indium chloride vapors, the
fraction of molecules with the lowest chlorine content
(InCl) must be the highest. With decreasing indium
concentration in the sample, the color of the condensed
chlorides became less intense. For example, a pale
orange color corresponded to the In–S sample with a
final sulfur content of 58.5 mol %, and a pale yellow
color corresponded to sulfur contents above 59 mol %.
Thus, our results demonstrate that a reduction in the
indium content of the sample and charge is accompa-
nied by an increase in the In2Cl4 (oxidized form) con-
centration in the vapor phase.
Reversible changes in the composition of phases
during nonisothermal SCVT in the system sample
(In–S)–vapor (indium chlorides)–charge (In–Au
melt). One would expect that, in a nonisothermal
InmSn–vapor (In–Cl)–In1 – xAux system, the chemical
transport direction can be reversed by adjusting the
chemical potential of indium in the sample or charge
via temperature variations. As shown earlier [1, 2], the
direction of mass transport under nonisothermal condi-
tions depends not only on the charge composition but
also on the sample and charge temperatures (T2 and T1,
respectively). (It is convenient to use the difference
∆T = T2 – T1 instead of T2 or T1: at constant T2 or T1,
increasing ∆T reduces the concentration of the compo-
nent being transported [1, 2]. The reversibility of
nonisothermal indium transport was examined using
static gravimetry.
2
3
4
–0.1
0
200
400
600
800
1000
(b)
50
49
48
1
2
3
4
0
200
400
(c)
600
70
65
60
55
50
1
2
3
4
0
200
400
600
Time, min
Fig. 7. (a) Amount of indium transferred, (b) sample com-
position, and (c) charge composition as functions of time
for nonisothermal vapor transport at a sample temperature
of 823 K: (1) In + InS sample (∆T = 40 K), (2) further indium
transport owing to InS conversion to In S (∆T = 55 K),
(3) insignificant transport (∆T = –110 K), (4) reverse
indium transport (∆T = –240 K). The error of determination
is represented by the size of the data points.
First, a ground In–S alloy containing 48.0 mol % S
was used as the sample. The alloy consisted of In and
InS, as determined by x-ray diffraction. As the charge,
we used a weighed amount of pure gold. The sample
temperature was 823 K, and the charge temperature
was 783 K (∆T = 40 K). Curves 1 in Figs. 7a–7c show
the time dependences of the sample weight, sample
composition, and charge composition. It follows from
these data that indium extraction from the sample
ceases at 50 mol % S (InS). The corresponding charge
composition in this steady state is 55 mol % Au. It
seems likely that the chemical potential of indium in
this melt is too high for further indium extraction via
InS conversion to In6S7.
6
7
At the same time, InS conversion to In6S7 proved
possible when the charge temperature was reduced by
15 K (T2 = 823 K, T1 = 768 K, ∆T = 55 K). As a result,
we observed further indium transport from the sample
to the charge, which ceased when the charge composi-
INORGANIC MATERIALS Vol. 43 No. 11 2007

1174
ZAVRAZHNOV et al.
tion was 49.6 mol % Au (Fig. 7, curves 2). The final basic principle of this approach is to bind this compo-
sample composition was 50.5 mol % S, which corre- nent in distinct chemical compounds with significant,
sponds to a mixture of InS and In6S7. As the indium but limited homogeneity ranges. Component Q, which
forms a number of compounds (AQ, A2Q3, and others)
increased, which probably prevented full conversion of with the component to be transported (A), must meet
content of the sample decreased, that of the charge
InS to In6S7.
To reverse the direction of indium transport, we
almost the same requirements as the diluent R in the
approach involving the formation of a continuous series
of solutions:
increased T1 (compared to the previous experiment),
without changing the temperature T2 of the InS + In6S7
sample. After a relatively small increase in T2, the rate
of indium transport was rather slow (Fig. 7, curves 3;
T2 = 823 K, T1 = 933 K, ∆T = –110 K). At the same
time, after the charge temperature T1 was raised to 1063 K
(Fig. 7, curves 4; ∆T = –240 K), the rate of indium
transport to the In–S sample was initially relatively fast.
Under those conditions, mass transport ceased when a
steady state with 49.9 mol % S (which corresponds to
InS within the present experimental accuracy) was
reached.
It must have a negligible vapor pressure.
It must be nonreactive with the transport agent.
The enthalpy of formation of the compounds must
be significant: ∆f H(T) ꢀ 0.
Consider the system sample (A–Q)–vapor–charge
(A–Q) under isothermal conditions. Assume that there
is a vapor-phase indium exchange between the sample
(AxBy) and a phase (AwQz) of the charge.
Assume that, at temperature Ti, the A–Q system
contains the phases α (A-based phase), β (intermediate
compound), and γ (Q-based phase) (Fig. 8c). At this
temperature, the A–B system also contains three con-
densed phases: α' (A-based solution), β', and γ'
(B-based solution) (Fig. 8b).
Therefore, our results demonstrate that the direction
of mass transport in the system under consideration
depends on the thermal conditions of the process.
Thus, the charge dilution approach offers the fol-
lowing advantages:
Let us examine how the content of component A in
the sample varies with increasing content of component
Q in the charge (Fig. 8a). Like in the case of charge
dilution, we expect that, at a low content of component
Q in the charge, the equilibrium (with respect to com-
ponent A) phases are α (charge) and α' (sample); the
dependence for partial equilibrium must be continuous
and monotonic (Fig. 8a, segments DE1, DE2). At a
higher content of the binding component Q in the
charge, the equilibrium sample must contain more
component B (nontransportable).
1. Varying the concentration component of the
chemical potential, one can control the equilibrium
state of the system, which offers the possibility of using
those phase compositions of the sample and transport
agent which are commonly believed to be unsuitable
for reversible mass transport.
2. The proposed approach notably extends the capa-
bilities of isothermal selective transport. It was pointed
out earlier [2] that, if the charge belongs to the same
binary system as the sample, a single-zone process
allows one to reproducibly obtain only samples with
compositions at the limits of the homogeneity range. In
As the content of component Q in the charge
increases, two situations are possible.
1. The mutual chemical affinity of the interacting
the approach in question, control over the activity of the components (Q and A) in the α phase of the charge is
component being transported allows one to obtain com- high, and increasing the concentration of the binding
positions with the desired nonstoichiometry.
component Q sharply reduces the chemical potential µA
in the charge. Even at a relatively low concentration of
component Q in the charge (which also corresponds to
the α phase), the chemical potential µA in the sample
will then take the value corresponding to the phase
boundary of the α' phase (Fig. 8a, point E1). The neigh-
boring phase, β', will have the same value of the chem-
ical potential. Thus, the composition relationship line
will have a vertical segment, E1F1, representing the
selective equilibrium between the homogeneous charge
of particular composition and a mixture of arbitrary
amounts of the α' and β' phases. Further increasing the
content of component Q in the α phase will change the
composition of the β' phase within its homogeneity
range (segment F1G1).
3. To control the sample composition in the course
of isothermal selective chemical transport, it is neces-
sary to scan the composition relationship diagram of
the system. Such diagrams are similar to classical phase
diagrams in that they represent phase assemblages
under equilibrium conditions.
4. In nonisothermal SCVT, a variety of sample com-
positions can be obtained by varying only three param-
eters: the sample and charge temperatures and charge
composition.
A WAY OF BINDING THE ACTIVE COMPONENT
OF THE CHARGE IN DISTINCT CHEMICAL
COMPOUNDS: EXPERIMENTAL AND RESULTS
2. A different situation occurs when the mutual
Theoretical analysis. Consider the latter of the chemical affinity of the interacting components Q and
above approaches to reducing the chemical potential of A in the α phase of the charge is low throughout the
the component being transported in the charge. The homogeneity range of this phase. Even significant com-
INORGANIC MATERIALS Vol. 43 No. 11 2007

SELECTIVE CHEMICAL VAPOR TRANSPORT
1175
(a)
(c)
γ'
β' + γ'
β'
I3
H3
F1
α' + β'
E1
G3
G2
I2
H2
1
E2
F2
2
α'
D
α
α + β
β
γ
γ
A
B
B
Ti T
(b)
T
Ti
α
α + β
β
A
Fig. 8. Possible types of diagrams illustrating the relationship between the compositions of A–B samples and A–Q charges, where
Q is an inert binder forming distinct condensed phases (α, β, and γ) with component A: (a) isothermal section of the composition
relationship diagram, (b, c) partial T–x phase diagrams of the A–B and A–Q systems, respectively.
positional changes then have little effect on µA in the concentration of component Q, increasing the concen-
tration of this component may bring the sample compo-
sition to the α'–β' phase boundary (Fig. 8a, vertical seg-
ment G3H3) and then change the composition of the β'
charge. As a result, the corresponding change in the
sample composition will also be relatively small, and
the α phase in the charge will reach its phase boundary
before the α' phase in the sample (Fig. 8a, point E2). In phase within its homogeneity range (curved segment
H3I3).
It is, therefore, reasonable to assume that, if a com-
bination of distinct phases is used as the charge, the
composition relationship line may consist of three
types of segments:
this instance, to the same composition of the α' solution
corresponds the boundary of the β phase in the charge,
which is adjacent to the α phase. Thus, a horizontal seg-
ment, E2F2, appears in the composition relationship
diagram.
As the concentration of the binding component Q in
the β phase of the charge increases further, two scenar-
ios are again possible.
curved segments, corresponding to gradual varia-
tions across the homogeneity ranges of phases close in
the chemical potential of the component being trans-
ported;
vertical segments (homogeneous, single-phase
charge and two-phase sample);
horizontal segments (two-phase charge and sample
composition within the homogeneity range).
It is easily seen that the last situation is best suited
for accurate control of the sample composition within
the homogeneity range of a given phase in the sample.
(1) The µA in the charge is also a weak function of
concentration. Even “traversing” the entire homogene-
ity range of the β phase is then incapable of reducing
the chemical potential of component A to an extent suf-
ficient for reaching, in equilibrium with the sample,
another phase, poorer in component A (β'). As a result,
we obtain the curved segment F2G2 and then the hori-
zontal segment G2H2 in the composition relationship
line (Fig. 8a).
Selection of the transport agent and charge
(2) If the chemical potential of indium in the homo- material. In the proposed approach for controlling the
geneity range of the β phase is a strong function of the composition of indium sulfides, hydrogen is used as the
INORGANIC MATERIALS Vol. 43 No. 11 2007

1176
ZAVRAZHNOV et al.
transport agent, which makes it possible to introduce
The sulfur vapor pressure is well known for many
sulfur into sulfides or to extract it. Hydrogen has many low-volatility sulfides, which facilitates the selection of
advantages over other transport agents, the most impor- the charge material. An additional requirement for the
tant of which are the insignificant solubility of hydro- charge material is fully incongruent vaporization: only
gen in most semiconductors (including all of the sulfur molecules must vaporize.
indium sulfides) and the high selectivity of the reac-
tions involved.
The phases of the Cu–S system meet all of these
requirements.
At temperatures above 300°ë, free hydrogen readily
reacts with any solid sulfides according to the follow-
ing, well known reaction, reversible to a significant
degree:
In the temperature range 430–708 K, the Cu–S sys-
tem contains three phases:
The “high-temperature” phase Cu2S (chalcosine),
which decomposes peritectoidally at 708 K to form
anilite and copper;
anilite, a phase with a broad homogeneity range,
from Cu1.96S to Cu1.75S, which melts congruently at
1378 K;
covellite, CuS, which decomposes peritectically at
780 K to form liquid sulfur and anilite.
Earlier data [14, 15] and our mass spectrometry
results for the In–S system can be used to compare the
partial sulfur vapor pressures over In–S and Cu–S
phases at different temperatures (table).
The data in the table lead us to a number of assump-
tions.
Under conditions that insure a constant sulfur vapor
pressure over the charge (a large amount of the sub-
stance or the use of a mixture of two neighboring
phases), an In–S sample, independent of its initial com-
position,
must convert to In2S3 if the charge consists of CuS,
CuS + Cu1.75S, or Cu1.75S (clearly, the composition of
In2S3 will vary within its homogeneity range, depend-
ing on the nature of the charge);
'
''
'
H2 + S = H2S .
(6)
(Here, like in the above schemes, the primed and dou-
ble-primed species are in the vapor phase and con-
densed phase (combined sulfur), respectively.) In the
case of indium sulfides, the selectivity of reaction (6) is
due to the fact that there are no stable indium hydrides
or sulfohydrides.
Recall that experiments on controlling the composi-
tion of indium sulfides were to be conducted in a closed
isothermal system (ampule). The sample and charge
were placed at the opposite ends of the ampule. The
charge (sulfur source or getter) must consist of a sub-
stance which also reacts with hydrogen according to
scheme (6) under the experimental conditions of
interest.
The question that now arises is how the charge com-
position must be selected in order to vary the sample
composition in the desired direction. Selective trans-
port in an isothermal system is governed by the rela-
tionship between the chemical potentials of the compo-
nent being transported (sulfur) in the sample and
charge. The chemical potential of sulfur, in turn, deter-
must convert to In6S7 if the charge consists of Cu2S
and Cu.
mines the vapor pressure of this element ( pS ):
2
Controlling the composition of indium sulfides
using hydrogen and a Cu–S charge. Hydride vapor
transport of sulfur was conducted as follows: First, a
quartz ampule and containers for sample and charge
powders were prepared. To this end, a bulbous joint
with a neck was made at one end of a cylindrical quartz
tube ~3 cm in diameter, and then containers with the
powdered sample and charge (1–2 g) were inserted into
the ampule. Next, a second neck with a bulbous joint
was made, and the ampule was ready to fill with hydro-
gen. The distance between the containers was about
1 cm, the ampule length was approximately 12 cm, and
the ampule volume was about 75 ml.
The ampule was then filled with dry hydrogen pre-
pared by the aluminate process (aqueous KOH + Al).
The amounts of the reagents were chosen so as to pass
30 l of hydrogen (~400 ampule volumes) through the
system. After filling the ampule with hydrogen, the
joints were sealed off, and the ampule was placed in the
isothermal zone of a horizontal cylindrical furnace.
µS(solid) = µS (vapor);
2
µS (vapor) = µ0S (p) + RT ln pS
(7)
2
2
2
(at xS(solid)= const).
The charge can then be selected according to the
equilibrium sulfur vapor pressure over it, even if this
vapor pressure is extremely low. Namely, if the pS
2
over the charge is higher than that over the sample,
vapor transport will be toward the sample, which will
become richer in sulfur. Otherwise, the charge will act
as a sulfur getter, and the sample will lose sulfur. The
final state (equilibrium with respect to sulfur) will be
reached when the sulfur vapor pressure over the sample
will be equal to that over the charge as a result of com-
positional changes. In the above reasoning, only one
assumption was made, that changes in the total pressure
in the system upon the formation of an atmosphere con-
taining the transport agent (H2 + H2S) have little or no
effect on the chemical potential of sulfur.
The samples consisted of an In–S alloy containing
56 mol % sulfur. According to our phase-diagram and
INORGANIC MATERIALS Vol. 43 No. 11 2007

SELECTIVE CHEMICAL VAPOR TRANSPORT
S2 vapor pressures over phases in the Cu–S and In–S systems
1177
p(S2), Pa
In6S7–In3S4–V In2S3(60%)–V CuS–Cu1.75S–V Cu1.75S–Cu2S–V Cu2S–Cu–V
t, °C
InS–In6S7–V
360
380
400
420
440
460
480
9 × 10–26
5 × 10–24
2 × 10–22
1 × 10–20
3 × 10–19
1 × 10–17
2 × 10–16
9 × 10–16
1 × 10–14
9 × 10–14
7 × 10–13
5 × 10–12
3 × 10–11
2 × 10–10
1 × 10–11
7 × 10–11
4 × 10–10
2 × 10–9
9 × 10–9
4 × 10–8
1 × 10–7
16
40
2 × 10–5
8 × 10–5
2 × 10–4
8 × 10–4
2 × 1003
6 × 10–3
2 × 10–2
8 × 10–15
4 × 10–14
2 × 10–13
8 × 10–13
3 × 10–12
1 × 10–11
3 × 10–11
96
218
471
978
1953
x-ray diffraction data, the alloy consisted of roughly our experimental conditions, the composition ëu2S
equal amounts of two phases, In6S7 and In3S4. The
charge consisted of a Cu–S alloy: ëu2S, Cu1.76S, or
CuS. Accordingly, three series of experiments were
planned, with each type of charge.
Experiments were conducted in steps, for 50, 100,
and 200 h at 410°ë. The temperature was chosen so as,
on the one hand, to achieve favorable kinetics of selec-
corresponded to a mixture of chalcosine and a small
amount of copper (ëu2 – xS + Cu), Cu1.76S corresponded
to anilite with a stoichiometry close to its sulfur-rich
phase boundary, and CuS corresponded to covellite
(sulfur removal during vapor transport might lead to the
formation of a mixture with anilite, Cu1.76S + CuS). The
vapor pressure in the system was evaluated from the
tive transport and, on the other, to maintain a sufficient ideal gas law, for room temperature and atmospheric
phase diversity in the Cu–S system. In particular, under pressure in the initial state and for a constant number of
mol % S (In–S)
61
*
60
*
59
58
57
Initial states of Cu–S and In–S alloys
56
*
!
55
54
γ
CuS
(Cu1.76S)
γ + CuS
β-Cu2S
53
β-Cu2S + Cu
30
34
38
42
46
50
mol % S (Cu–S)
Fig. 9. Diagram illustrating the relationship between the compositions of In–S samples and Cu–S charges, as inferred from exper-
imental data for the system sample (In–S)–vapor (H + H S)–charge (Cu–S) (isobaric–isothermal section); the dashed line repre-
2
2
sents the sample and charge compositions in the equilibrium (with respect to S) state at 680 K.
INORGANIC MATERIALS Vol. 43 No. 11 2007

1178
ZAVRAZHNOV et al.
moles of vapor species in reaction (6) (~0.25 MPa at
410°ë).
REFERENCES
1. Zavrazhnov, A.Yu., Turchen, D.N., Goncharov, E.G.,
and Zlomanov, V.P., Chemical Transport Reactions as a
Means of Controlling the Composition of Nonstoichio-
metric Crystals, Zh. Neorg. Khim., 2002, vol. 47, no. 3,
pp. 463–467.
2. Zavrazhnov, A.Yu., Turchen, D.N., Naumov, A.V., and
Zlomanov, V.P., Chemical Transport ReactionsAs a New
Variant of the Phase Composition Control, J. Phase
Equilib., 2003, vol. 24, no. 4, pp. 330–339.
3. Zavrazhnov, A.Yu., Turchen, D.N., Naumov, A.V., and
Zlomanov, V.P., Controlling the Composition of Indium
Sulfides Using In1 – xAux Melts As Indium Sources or
Getters, Kondens. Sredy Mezhfaznye Granitsy, 2005,
vol. 7, no. 1, pp. 39–48.
After each experiment, the ampule was broken, and
the containers were weighed (VLM-200 balance, 5 ×
10–5 g accuracy). The weighing results were compared
with those obtained before annealing to assess the
annealing-induced weight and composition changes.
After weighing, the containers were again placed in an
ampule, and the hydrogen annealing cycle was
repeated. The results are presented in Fig. 9.
In combination with x-ray diffraction data, these
results lead us to the following conclusions:
1. Long-term (above 100 h) annealing transfers the
system into a state which seems to be close to partial
equilibrium (compositional changes decrease with
increasing annealing time and become insignificant
after 200 h of annealing).
2. The mass transport direction depends on the
nature of the charge. In particular, the use of the CuS
and Cu1.76S alloys causes the indium sulfide sample to
transform into In2S3 of variable stoichiometry: after
annealing with the CuS alloy, the In2S3 sample contains
more sulfur (61.0 mol % S) than it does after annealing
with the Cu1.76S alloy (60.1 mol % S in In2S3). Note that
in each of those experiments, the sample changed color
to bright red, characteristic of β-In2S3.
3. In the presence of a charge with an initial sulfur
content of 30 mol % (a mixture of Cu and Cu2S), sulfur
is transported toward the charge, but the compositional
changes are not very large: the selective transport, most
likely, ceases as the copper-based phase in the charge
disappears, and the composition of anilite varies within
its homogeneity range. (In3S4 + In6S7 mixtures can then
be used as charges for controlling the composition of
the Cu2 – xS phase.)
The high selectivity of SCVT in this implementa-
tion was confirmed by roentgenoluminescence mea-
surements (VRA-30 spectrometer): in the annealing
products, the indium sulfides contained no copper, and
the Cu–S alloys contained no indium.
The above results lead us to the following conclu-
sion: for nondestructive control over the composition of
solids via SCVT in the case of binding the component
being transported in distinct condensed compounds, the
most promising approach is to fix particular nonstoichi-
ometric compositions. This can be achieved by select-
ing two-phase mixtures such that the appropriate chem-
ical potential of the sample is equal to that of the
charge. For example, near 400°ë a mixture of Cu1.76 – δS
and CuS can be used to enrich the β-In2S3 phase with
sulfur (to ~61 mol % S).
4. Gödeke, T. and Schubert, K., On the Phase Diagram
InSM, Z. Metallkd., 1985, vol. 76, no. 5, pp. 358–364.
5. Likforman, A., Guittard, M., Tomas, A., et al., Mise en
évidence d’une solution solide de type spinelle dans le
diagramme de phase du système In–S, J. Solid State
Chem., 1980, vol. 34, no. 3, pp. 353–359.
6. Ansell, H.G. and Boorman, R.S., Phase Relationship in
the In–S System, J. Electrochem. Soc., 1971, vol. 118,
pp. 133–136.
7. Duffin, W.J. and Hogg, J.H., Crystalline Phase in the
System In–In2S3, Acta Crystallogr., 1966, vol. 20,
pp. 566–569.
8. Yasuo Kunia and Masahiro Hosuka, Vapor Phase Equi-
libria in the Systems In–InCl3 and Ga–GaCl3, J. Cryst.
Growth, 1975, vol. 28, pp. 385–391.
9. Yasuo Kunia and Masahiro Hosuka, High Temperature
Spectroscopic Studies in Visible and Near Ultraviolet
Regions on the Vapor Phase Reactions in the System In–
Cl2, Denki Kagaku, 1973, vol. 41, pp. 616–621.
10. Yasuo Kunia, Shozo Hosada, and Masahiro Hosuka,
Studies on the Vapor Phase Reactions in the System In–
Cl2, Denki Kagaku, 1974, vol. 42, pp. 20–25.
11. Zavrazhnov, A.Yu., p–T–x Phase Diagrams of Gallium
Chalcogenides Studied Using anAdditional Component,
Zh. Neorg. Khim., 2003, vol. 48, no. 10, pp. 1722–1736.
12. Okamoto, H. and Massalski, T.B., The Au–In System,
Phase Diagrams of Binary Gold Alloys, Metals Park:
ASM International, 1987, pp. 142–153.
13. Fedorov, P.I. and Akchurin, R.Kh., Indii (Indium), Mos-
cow: Nauka, 2000.
14. Abrikosov, N.Kh., Bankina, V.F., Poretskaya, et al.,
Poluprovodnikovye khal’kogenidy i splavy na ikh osnove
(Semiconducting Chalcogenides and Related Alloys),
Moscow: Nauka, 1975.
15. Brunetti, B., Piacente, V., and Scardala, P., Study on Sul-
fur Vaporization from Covellite (CuS) and Anilite
(Cu1.75S), J. Alloys Compd., 1994, vol. 206, no. 1,
pp. 113–119.
INORGANIC MATERIALS Vol. 43 No. 11 2007