Electrochemical preparation of H2S and H2Se

Article abstract of DOI:10.1149/1.1852441

H₂S and H₂Se have been electrolytically prepared by electrolysis in aqueous H₂SO₄ solutions of composite cathodes made of S or Se and graphite. The efficiencies depended strongly on the electrolyte composition, in particular on the acid concentration and presence of K⁺. Faradaic efficiencies of 80% were obtained in dilute (0.05 M) acid, and this increased to 100% with added K₂SO₄. The efficiencies dropped drastically in concentrated acid (>a few moles). H ₂Te and AsH₃ generation were also briefly studied for comparison. The mechanisms of hydride formation are discussed. Both the reaction of nascent hydrogen with the free element and direct reduction of the element are considered. The latter is believed to be the dominant mechanism.

Full text of DOI:10.1149/1.1852441

Journal of The Electrochemical Society, 152 ͑3͒ D35-D41 ͑2005͒
D35
0
013-4651/2005/152͑3͒/D35/7/$7.00 © The Electrochemical Society, Inc.
Electrochemical Preparation of H S and H Se
2
2
a
a
a,
b, ,z
St e´ phane Bastide, Paul H u¨ gel, Claude L e´ vy-Cl e´ ment, * and Gary Hodes *
a
Laboratoire de Chimie M e´ tallurgique des Terres Rares, CNRS, 94320 Thiais, France
Weizmann Institute of Science, Rehovot, Israel
b
H S and H Se have been electrolytically prepared by electrolysis in aqueous H SO solutions of composite cathodes made of S
2
2
2
4
or Se and graphite. The efficiencies depended strongly on the electrolyte composition, in particular on the acid concentration and
ϩ
presence of K . Faradaic efficiencies of 80% were obtained in dilute ͑0.05 M͒ acid, and this increased to 100% with added
K SO . The efficiencies dropped drastically in concentrated acid ͑Ͼa few moles͒. H Te and AsH generation were also briefly
2
4
2
3
studied for comparison. The mechanisms of hydride formation are discussed. Both the reaction of nascent hydrogen with the free
element and direct reduction of the element are considered. The latter is believed to be the dominant mechanism.
©
2005 The Electrochemical Society. ͓DOI: 10.1149/1.1852441͔ All rights reserved.
Manuscript submitted March 31, 2004; revised manuscript received August 3, 2004. Available electronically January 24, 2005.
Gaseous hydrides of S, Se, Te, or As are commonly used in
semiconductor preparation, either as precursors or as a source of
dopant. All these hydrides are very poisonous, and care must be
have been generated in a two-step process using a graphite cathode
and a suspension of S or Se in a neutral or alkaline solution ͑no
hydride was produced in acid solutions͒, followed by acidification of
1
5,16
taken in their storage and use ͑for instance, metal corrosion by H S).
the catholyte.
The efficiency of selenide formation varied from
2
7% ͑neutral solution͒ to ca. 95% ͑alkaline solution͒ and for sulfide,
In addition, some, e.g., H Te, are so unstable under normal condi-
2
56% independent of pH.
tions that they cannot easily be stored.
H S has been found in the products of catholysis of cathodes
A few methods are available to synthesize these hydrides. Thus,
2
1
7-20
made from various sulfide minerals.
An early attempt to produce
for H Se, the action of acid on metal selenides ͑usually the Al salt͒,
2
H Se using a Se cathode was, not surprisingly, unsuccessful, al-
heating Se with paraffin or heating Se in a flow of H can be used.
2
2
though H Te evolution was ͑also not surprisingly͒ reported in the
Another method, which has been often described to prepare very
2
8
same paper. However, attempts to mix S or Se with a more con-
small amounts of H Se ͑and some other volatile hydrides͒ for use as
2
ductive phase were more successful. Fischer reported formation of
an analytical tool ͑the hydride is injected into a mass spectrometer͒
is electrolysis of solutions of Se compounds or compounds of some
other elements ͑Ref. 1-3 give many references to this method͒. In
principle, this technique could be used as a method to prepare larger
quantities of the hydrides, although, to the best of our knowledge,
this does not seem to have been seriously considered. It may be
limited by the tendency of oxidized forms of Se to react with se-
lenide and regenerate elemental Se.
2
1
H S using a composite copper sulfide-sulfur cathode and H Se
2
2
2
2
using composite Se-Cd or Se-Sn cathodes. In the case of H S, the
2
yield was reported to be quantitative based on the formation of CdS
in the electrolyte using a Cd anode. In 1978, a composite S/graphite
cathode was described to generate sulfide ions which then reacted in
situ with various organic compounds in the nonaqueous
2
3
electrolyte. Shortly thereafter, a report described H Se formation
2
from Se cathodes ͑made by three different methods, solid Se made
by melting Se ͑it is not clear how much current could have been
passed through such an electrode͒, a paste of Se and graphite and a
AsH has been generated by electrolysis using an As cathode. In
3
earlier studies, low yields of AsH were obtained ͑typically 1-2% in
3
4,5
acidic solution and ca. 10% in neutral solution, but much higher
yields ͑close to 100%͒ were later obtained in alkaline solutions, and
slurry of Se, in aqueous acidic solutions ͑the yield of H Se was not
2
1
4
6
given͒. More recently, an electrode made of 90% S and 10%
the method was used for the express purpose of generating AsH3 .
graphite has been studied with emphasis on the electrochemical re-
H Te can also be prepared in quantity by electrolysis of a Te cathode
2
7-10
action of S ͑in the form of H S) with metal cations in the
2
in strong aqueous acid.
2
4
electrolyte.
In principle, there is no reason why this electrochemical method
In this paper, we demonstrate that H S and H Se can be prepared
2
2
should not work for H S and H Se, except for the important differ-
2
2
as needed, easily, controllably, and safely, with high yield ͑100%͒ by
electrolysis using a cathode formed by melting S or Se together with
graphite powder, thereby rendering the composite conductive. We
compare the yield obtained under different electrolysis conditions.
For comparison, we also give some results for other hydrides which
were previously electrogenerated (H Te, AsH ) using both compos-
ence that S or Se ͑in whatever allotropic form͒ are much less con-
ductive than is Te and will not allow any meaningful current to flow
in their bulk form. There is an old paper which describes polychal-
cogenide, S, Se, Te, formation at cathodes in alkaline solution using
electrodes of Pt dipped in molten S or Se and of elemental Te.
2
3
However it is likely that such a process would give H S͑e) in acid
2
ite and pure elemental electrodes. The possible mechanisms for hy-
dride formation are discussed.
solution for a short time until the presumably thin layer of chalcogen
was used up.11
As early as 1868, electrolysis of a suspension of S in water was
12
Experimental
reported to give a small amount of H S. A later study of electroly-
2
sis of a thick suspension ͑or paste͒ of S in dilute H SO using
various cathode materials reported a large and irreproducible varia-
2
4
Electrode preparation.—S and Se electrodes were made by melt-
ing the respective element with graphite powder ͑synthetic graphite,
TIMREX KS44, Lonza, now Timcal AG, average grain size 45 ␮m,
tion in the yield of H S, between 0.7 and 28%, which was appar-
2
13
2
ently dependent on the nature of the S suspension. The best yields
were obtained on smooth Pt cathodes and poorer ones on metals
with high hydrogen overpotential. Se suspensions were also reduced
Brunauer-Emmett-Teller ͑BET͒ surface area 9 m /g͒ in a nearly ver-
tical position. The S ͑99%, Prolabo 28256.296, particle size Ͻ160
␮m͒ or the Se ͑Riedel-de-Haen 10112͒ and the graphite powders
¨
to H Se on Pb and Hg cathodes ͑on Hg, HgSe was formed which
were first mixed in a mortar in the desired ratio ͑see below͒, then
2
1
4
was then reduced by the catholysis to give H Se). H S and H Se
transferred to a quartz tube which was then sealed under vacuum
2
2
2
Ϫ1
(
10 mbar͒. The sealed tube was heated to ϳ40°C above the chal-
cogen melting point ͑115.2 and 217°C for S and Se, respectively͒ for
a couple of hours. The mixture obtained at these temperatures is
very viscous and, rather than flow to fill the bottom part of the tube,
*
Electrochemical Society Active Member.
E-mail: gary.hodes@weizmann.ac.il
z
Downloaded on 2014-10-20 to IP 130.63.180.147 address. Redistribution subject to ECS terms of use (see ecsdl.org/site/terms_use) unless CC License in place (see abstract).

D36
Journal of The Electrochemical Society, 152 ͑3͒ D35-D41 ͑2005͒
results in rods which contract from the walls of the tube and are
smaller in diameter than that of the tube itself. In some cases, we
have tapped the quartz tubes with the molten mixture ͑at the heating
temperature͒ which brings all the mixture to the bottom of the tube.
For S electrodes, there was no difference in performance; however
for Se electrodes, the more porous, nontapped electrodes were su-
perior to the smoother tapped ones. All the results given for Se
electrodes are for nontapped ones. While the density of graphite is
3
close to that of S ͑both ϳ2.1 g/cm ͒, it differs significantly from the
3
density of Se ͑ϳ4.8 g/cm ͒, and this could lead to nonhomogenous
mixing during the tapping of Se/graphite rods.
Figure 1. SEM back-scattered electron images of the cross section of a
G37%-S rod. White and black areas are S and carbon, respectively, as con-
firmed by EDX analysis.
Several different graphite/chalcogen volume ratios were tested.
In the end, we used a graphite volume percentage of 37%, since this
percentage appears to be the best trade off between maximizing
conductivity ͑0.6 and 1.2 ⍀ cm for S and Se, respectively͒, and also
maximizing the amount of chalcogen, and thus the amount of hy-
drogen chalcogenide that can be prepared.
mum͒. Third, after the electrolysis has finished, the Ar flow is con-
tinued for several minutes to remove dissolved hydride in the
catholyte.
For G37%-S rods ͑37% graphite and 63% S volume ratio͒, the
3
densities have been calculated and found to be ϳ1.2 g/cm for the
3
On the anolyte side, a platinum wire counter electrode was used.
The separated compartments are necessary to prevent the oxygen
evolving at the counter electrode from reacting with the hydrides
generated at the working electrode. The whole cell can be immersed
in an ice bath. This is more important for the unstable hydrides, in
particular H Te, and much less so for H S: the Joule heating in the
rods obtained without tapping and ϳ1.9 g/cm for the tapped ones
3
͑
the latter is much closer to the expected density of 2.1 g/cm ͒. Thus
the untapped rods are much more porous than the tapped ones.
XRD of the electrodes shows no change in the graphite lattice
spacing, i.e., no large-scale intercalation of S into the graphite.
Tellurium rods 8 mm in diameter and 5 cm long, were prepared
by melting tellurium ͑purum, Fluka 86362͒ in a sealed quartz tube at
2
2
cell can increase the temperature considerably, particularly if high
currents are used. For voltammetry, a saturated calomel reference
electrode ͑SCE͒ connected via a bridge to the cathode compartment
was used.
650°C for ca. 1 h in a nearly vertical position. The tube was then
removed from the furnace and allowed to cool.
Arsenic ͑As͒ electrodes were made by two techniques. In the
first, 37% graphite—As ͑v/v͒ electrodes were prepared by pressing
pellets from the mixed powders of As and graphite. These pellets
were glued ͑Araldite͒ vertically to a glass rod ͑8 mm external diam-
eter͒, and contact made by pressing graphite powder in the tube onto
the electrode and putting a wire into the tube and the graphite pow-
der. These As/graphite electrodes were much more brittle than the
melted ones of S and Se and had the tendency to slowly disintegrate
during electrolytic hydride evolution. However, reasonable quanti-
We passed the hydrides into a wash bottle containing an aqueous
solution ͑0.25 M͒ of cadmium acetate. The hydride reacts to give a
precipitate of the corresponding Cd compound ͑CdS, CdSe, CdTe,
or Cd As ). The output from this bottle is passed into another wash
3
2
bottle containing concentrated KOH solution and then into a third
one containing an aqueous solution of CuSO . The KOH solution
4
absorbs any hydride not reacted with the Cd solution while the
CuSO acts as an indicator; a black precipitate would immediately
4
form if any hydride reached this bottle. These precautions were
taken because of the toxicity of the hydrides. Sometimes two cad-
mium acetate bottles were connected in series, as an additional pre-
ties of AsH could still be prepared by this method. Electrodes of
3
pure As were also prepared. Since As melts only under high tem-
perature and pressure and otherwise sublimes around 500°C, a
simple melting technique could not be used. As powder was placed
in a vertical sealed quartz tube and heated with a flame at Ϸ700°C
so that the As deposited on the top of the tube. Since the resulting
deposit was not uniform ͑it was in the form of a hollow tube͒, it was
reheated so that the As redeposited at the bottom of the tube. This
procedure was repeated a few times until a relatively homogeneous
rod was obtained. The resulting As rod was rather brittle and tended
to break up with extended use. However, like the As/graphite com-
posite electrodes, it could be used for some time to investigate the
formation of arsine. It is probable that more controlled heating in a
furnace with a lower temperature gradient would produce a better
rod, but we did not investigate this.
caution and to check that all the H S͑e) was absorbed in the first
2
bottle. Of course, the entire experiment was carried out in a hood.
Experimental procedure and measurement of the faradaic
efficiency.—An experiment corresponds to the following step se-
quence: ͑i͒ degassing of the catholyte by the Ar flow for 15 min; (ii)
electrochemical reduction at the required current density for a fixed
time ͑a defined amount of charge passed͒; (iii) when the electroly-
sis is finished, the Ar flow is maintained for 15 min in order to
remove all the dissolved hydride.
The corresponding Cd-compound precipitates during the experi-
ment and at the end, is filtered out, dried, and weighed. The faradaic
efficiency of the chalcogen or As reduction into the hydride form is
calculated as the ratio between the actual and theroretical quantity of
cadmium compound that would be obtained for the total charge
passed, assuming a 100% efficiency of hydride generation.
Electrochemical setup.—The electrochemical hydride evolution
was performed using a two-electrode arrangement, with the cell
separated into two compartments by a frit ͑no. 2͒ and both compart-
ments filled with the same H SO electrolyte. On the catholyte side,
Results
2
4
Since a large part of our results were obtained for H S genera-
the working electrode was immersed ca. 1 cm into the electrolyte
and Ar gas bubbled through the electrolyte to degas the electrolyte
both before and after the electrolysis. The Ar flow serves several
purposes. First, it deaerates the catholyte to remove oxygen which
could otherwise react with the generated hydride and thus reduce the
efficiency. Second, it removes the hydride more rapidly from the cell
and carries it to where it is needed. This is particularly important for
very unstable hydrides like H Te or even H Se, both of which react
2
tion, we first describe results for this gas. Then, we extend the re-
sults to the other hydrides
Sulfur/graphite composite electrodes.—We noted above that tap-
ping the hot, molten mixture resulted in denser electrodes, although
no difference in performance was found whether tapped or not ͑for
S electrodes, not for Se ones as noted in the experimental͒.
We made a SEM investigation of the electrodes to see how the
graphite and S were distributed. Figure 1 shows micrographs of an
electrode at two different size scales. The images were made in the
2
2
rapidly with any polymer tubing in the gas flow stream ͑for this
reason, any rubber tubing used should be kept to an absolute mini-
Downloaded on 2014-10-20 to IP 130.63.180.147 address. Redistribution subject to ECS terms of use (see ecsdl.org/site/terms_use) unless CC License in place (see abstract).

Journal of The Electrochemical Society, 152 ͑3͒ D35-D41 ͑2005͒
D37
Figure 2. Current-voltage plots of electrodes of pure graphite (G100%) and
Figure 3. H S formation efficiency as a function of the electrolysis current.
2
composite S/graphite (G37%-S) in 0.5 M and 10 M H SO solutions.
Electrode: G37%-S electrode; 0.5 M H SO .
2
4
2
4
backscattered mode; this mode is sensitive to differences in atomic
weight and thus more accurately shows the elemental distribution.
The two elements were distributed on a scale of several mi-
crometers, i.e., most parts of the graphite were within some mi-
crometers ͑rarely greater than 10 ␮m͒ of the S͑a very similar dis-
persion was found for Se/graphite electrodes͒. We return to this
point later when we consider the mechanism of the hydride forma-
tion.
sured range ͑20-500 mA͒, there is very little dependence of the
efficiency on this parameter ͑if any, a slight increase with increasing
current͒. Efficiencies are typically between 70 and 80%. The im-
2
mersed surface area of the cathode is ca. 2.2-3.0 cm . Since this area
was not well-defined ͑due to changes in liquid level and due to the
porous nature of the electrodes͒, it could vary very appreciably from
experiment to experiment and also during any one experiment.
However, as seen from the virtual independence of efficiency on
current, this variability was not important.
Synthesis of CdS to quantify H S production.—Reaction of the
2
H S with the CdAc solution resulted in a yellow precipitate of CdS.
2
2
The identity of this precipitate was confirmed by XRD, which
showed the spectrum of sphalerite CdS with broad peaks ͑full width
at half maximum ϭ3.5° corresponding to a crystal size of ca. 3.2
nm͒.
Acid concentration.—The efficiency was strongly dependent on the
concentration of H2SO4 . The original concentrations of H2SO4 used
in the early H Te generation studies were 50%, ca. 15 M. We started
2
off with this concentration, but found that more dilute solutions gave
much higher efficiencies. The efficiency appears to be approxi-
mately inversely proportional to the acid concentration, with the
main change between 0.5 and 4 M ͑Fig. 4͒. It is clear that, in this
intermediate concentration range, the efficiency is highly dependent
on the acid concentration. It should be noted that the activity coef-
ficient of H SO does not vary strongly in the range between 0.5 and
Voltammetry of the electrodes.—Current-voltage plots of pure
graphite ͑pressed pellet͒ and G37%-S electrodes in 0.5 M and 10 M
H SO are shown in Fig. 2. The current onset in all cases is in the
2
4
region of Ϫ0.8 V ͑a large overpotential for both S reduction and H
evolution͒.
2
4
Of particular note is the fact that the currents at the S/graphite
electrode are greater than at the pure graphite electrode at the same
potential. ͓The relatively small cathodic current at pure graphite
begins positive of 0 V ͑SCE͒ and is clearly not hydrogen evolution;
it may be a capacitative current.͔ Also notable is the large difference
between the graphite and S/graphite in 10 M acid; the difference is
much less in 0.5 M acid as is the difference between graphite in the
two acid concentrations. Since the difference is due to the slope
rather than the onset ͑in fact, for the S/graphite electrode, the onset
in 0.5 M acid is very slightly positive of that in 10 M acid͒, this
would suggest that the difference is due to a differing resistance of
the electrode in the 10 M acid ͑if it were due to the electrolyte, then
we would expect a similar difference between pure graphite in the
two solutions͒. It is very possible that these differences are due to
differences in porosity ͑real surface area͒ of the electrodes and
therefore we prefer not to emphasize these results.
ϩ
4
M an increase of ca. 35% over this range͒, so the H activity does
not increase greatly beyond the concentration increase.
We carried out voltammetric studies of the other electrodes.
While we will not show them here, we note that the onsets ͑V/SCE͒
vary somewhat in the following order:
Te ͑Ϫ0.65 V͒;
S ͑Ϫ0.75 V͒;
As ͑Ϫ0.9 V͒
Se and graphite ͑Ϫ0.85 V͒;
Figure 4. H S formation efficiency as a function of the H SO concentra-
2
2
4
tion. Electrode: G37%-S electrode; current ϭ 500 mA. Each symbol shape
designates one particular electrode and the numbers given in one case ͑black-
filled squares͒ gives the order in which the experiments were carried out. The
It is interesting that Te, with the most negative standard reduction
potential, exhibits the most positive onset.
star ͑upper left corner͒ represents
a
separate experiment using
Influence of the (electro)chemical parameters on the hydride
formation efficiency.—Current density.—Figure 3 shows the effi-
0
.05 M H SO ϩ 0.25 M K SO ͑0.998 efficiency͒. Another identical ex-
2
4
2
4
periment, only using an Se electrode, gave an efficiency of 0.993 ͑almost the
same point on the scale of the figure͒.
ciency of H S generation as a function of current. Over the mea-
2
Downloaded on 2014-10-20 to IP 130.63.180.147 address. Redistribution subject to ECS terms of use (see ecsdl.org/site/terms_use) unless CC License in place (see abstract).

D38
Journal of The Electrochemical Society, 152 ͑3͒ D35-D41 ͑2005͒
In Fig. 4, we show measurements made using different electrodes
as different shaped/filled points. In particular, a relatively large set
of measurements using the same electrode is shown as filled squares
and the numbers above the squares represent the order in which the
measurement was carried out ͑1 was the first, 8 the last in the set͒.
This information is given in view of the limited lifetime of each
electrode ͑see the next section͒ and shows that, with the possibility
of some small underestimation of the efficiency values of numbers 7
and, more so, 8 the variations are due to the changes in acid con-
centration and not to aging of the electrode.
Effect of added K SO .—We wanted to extend the pH range to
2
4
higher values to see if the efficiencies were even higher at lower
acid concentrations. In order to maintain a reasonable conductivity
in the electrolyte, we added K SO to the dilute acid.
2
4
In an electrolyte of 0.05 M H SO ϩ 0.25 M K SO , we ob-
2
4
2
4
tained efficiencies of essentially 100%. ͑The same was found also
Figure 5. H S formation efficiency as a function of the cumulative charge
passed through the electrodes for several consecutive experiments. G37%-S
2
for H Se generation using Se electrodes͒.
2
Furthermore, the presence of the K SO caused two very strong
2
4
electrode; 0.5 M H SO ; current ϭ 500 mA.
2
4
and visible effects during electrolysis. Gas evolution from the elec-
trode was noticeably reduced and a white coloration developed in
the cathode chamber. These effects increased with increasing current
and with decreasing acid concentration. Additionally, a noticeable
yellow coloration appeared very close to the cathode during elec-
Other hydrides.—Using the above method, or variations of it, we
can generate the hydrides of Se, Te, and As.
Se, although considerably more conductive than S, still cannot be
prepared as a pure bulk conductive electrode ͑films of Se may allow
trolysis in the dilute H SO solutions, becoming more noticeable as
2
4
the acid concentration decreased. The yellow color is characteristic
of polysulfide and the white color is probably locally concentrated
K S or KHS ͑which can react with S at the electrode to give
2
reasonable currents to be drawn͒ and therefore H Se can only be
2
polysulfide͒. Continued bubbling of the carrier gas after cessation of
electrolysis removed this white coloration ͑by eventual conversion
made in this manner by using a composite electrode. The efficien-
cies are similar to those of H S, and, unlike H Te, they depend on
2
2
to H S).
the acid concentration in a very similar manner to H S, with values
2
2
While this effect requires a research effort of its own, we did
carry out some preliminary experiments.
of only several percent at high acid concentration and up to 80% at
ϩ
the lower concentrations ͑100% in the presence of K ).
To see if the salt affected the results in concentrated acid, we
Te has previously been generated using pure Te, which is suffi-
ciently conductive, as noted in the introduction. The efficiencies of
measured the efficiency in 5 M H SO ϩ saturated K SO . The ef-
2
4
2
4
ficiency ͑average of three experiments͒ was 17%, appreciably
greater than the typical efficiency for the same concentration with-
out salt, but still low.
H Te generation using pure Te ͑and also Te/graphite mixtures͒ were
found to be typically Ϸ50% ͑values in excess of 65% have been
2
obtained͒. In contrast to the situation for H S ͑and H Se) genera-
2
2
A few experiments were carried out in neutral solutions ͑satu-
tion, the efficiencies appeared to be independent of the acid concen-
tration: moderately high efficiencies were obtained at 10 M acid
rated K SO , ca. 0.6 M, pH 6-7͒ using S and Se cathodes. Despite
2
4
the absence of acid, strong cathodic currents could be drawn from
the electrodes ͑400 mA/cm at Ϫ3V/SCE in the case of a graphite-S
concentration. The important issue in H Te generation is to mini-
2
2
mize its decomposition ͑it is much more unstable than H S and
2
electrode͒. A yellow ͑polysulfide͒ and a red ͑polyselenide͒ colora-
tion appeared at the S and the Se cathode, respectively, and no gas
evolution was observed. In the case of S, the yellow color was
confined to the volume close to the electrode and became white
farther away from the electrode. We note that these colorations were
earlier reported at S and Se cathodes, respectively, in 1 M KOH
electrolyte.1
H Se) by using high currents and an inert carrier gas to minimize
the residence time of the gas in the cell and cooling the cell in ice
2
water. Like H Te, AsH generation from As cathodes has been pre-
2
3
viously reported. As mentioned in the Experimental section, As elec-
trodes, either in composite form with graphite or as the pure ele-
ment, tended to disintegrate with use. However, some reasonable
amount ͑Ϸ20% faradaic efficiency͒ of arsine could be generated and
detected by precipitation of Cd As . There was only a small differ-
1
3
2
Total charge passed.—It is clear that these electrodes have a finite
ence in efficiency for electrolysis in 0.5 and 10 M acid ͑ca. 25% less
in the concentrated acid, but at the same time, a pronounced brown
color formed in the catholyte chamber with the concentrated acid,
probably due to decomposition of electrolytically generated arsine͒.
In the relative independence of faradaic efficiency on acid strength,
the As behaves like Te and unlike S and Se. We made a few attempts
lifetime. As the S is used up, eventually the efficiency of H S gen-
2
eration will fall. Figure 5 shows the efficiency of two different elec-
trodes after various amounts of charge have been passed. The charge
referred to here is that part of the total charge used in the H2S
generation only ͑the rest is assumed to be consumed for H evolu-
2
tion͒. It is calculated using the faradic efficiency determined after
to generate stibine (SbH ) based on the early literature where yields
3
the experiment (Q • ␩) and summing up after each experiment
T
up to 2.3% have been reported using Sb cathodes ͑see, for example,
Ref. 25͒ and a later investigation with maximum yields Ͻ1%.26 We
used both elemental Sb rods as well as Sb/graphite composite elec-
trodes. Unfortunately, we never found evidence of stibine generation
under the conditions used by us. In particular, taking into account
the instability of this gas, we used conditions similar to those for
(
͑
͚ Q • ␩) with the same electrode. The efficiency remains stable
T
even increasing slightly͒ until close to 1000 C of charge has been
passed, and then decreases more or less linearly with increasing
charge. The slight initial increase might be due to increasing poros-
ity of the electrode and increasing accessibility of the S to the elec-
trolyte. After formation of 1000 C of H S, the volume of S is re-
H Te generation: cooled cell, high carrier gas flow, and high cur-
2
2
duced by only 21%, which represents a S-depleted layer of 0.5 mm
in thickness for an initial rod diameter of 8 mm. Therefore, the
limiting factor for efficiency seems not to be the quantity of S re-
maining ͑see Discussion͒.
rents. Additionally, we tried several electrolytes, including slightly
acid ones of salts of organic acids ͑such as citrates͒ to complex Sb
and slightly alkaline ones, based on previous literature.25 However,
in view of the various earlier studies which have described stibine
Downloaded on 2014-10-20 to IP 130.63.180.147 address. Redistribution subject to ECS terms of use (see ecsdl.org/site/terms_use) unless CC License in place (see abstract).

Journal of The Electrochemical Society, 152 ͑3͒ D35-D41 ͑2005͒
D39
generation, it is reasonable to assume that with the right conditions,
stibine can be generated at some detectable level.
atomic H, but we should also include freshly generated molecular
H , which is formed in a vibrationally excited state. This is the
2
28
We were also unsuccessful in generating phosphine. Electrodes
could only be made by pressing pellets of graphite mixed with red P.
These were physically very unstable as electrodes and readily disin-
tegrated. Even before this happened and when some current was
passed, no evidence of phosphine generation was observed.
same as saying that the freshly formed H is locally at a high tem-
perature. In this form, it may react with the element while still in
this excited state.
2
The typical average separation between graphite and S is of the
order of micrometers ͑Fig. 1͒. Either this is a typical migration dis-
tance for the excited hydrogen ͑based on the high efficiencies of
hydride formation͒ or else a large fraction of the active hydrogen
must be generated much closer to S. The latter would necessitate
most of the hydrogen evolution occuring at the graphite grain
boundaries and at the S/graphite interface. As noted in a previous
section there is still plenty of S left in the electrode when the elec-
trode efficiency has degraded considerably. The limiting factor for
efficiency of the depleted electrodes, in the above scenario, is there-
fore not the quantity of S remaining but rather its accessibility to H
Discussion
We now discuss our results from two separate viewpoints, one is
the mechanistic one, the other the practical one.
Mechanistic considerations.—There are two fundamentally dif-
ferent pathways for hydride formation, which we illustrate, by way
of example, for H S.
2
ϩ
The first is reduction of water (H in acid solution͒ to hydrogen
atoms and reaction of the reactive ͑nascent͒ hydrogen atoms with S
atoms which can preferentially recombine to H molecules if there is
2
no S nearby.
ϩ
Ϫ
H
ϩ e → H;
S ϩ 2H → H2S
͓1͔
Since no H evolution is observed at neutral pH ͑point 2͒ but
͑from the results of other studies described in the introduction͒ sul-
The second ͑mechanism 2͒ is direct reduction of S ͑actually S ).
There are a number of different pathways for this reaction, involving
electron transfer only to S, to an HS species or to both S and H .
The major difference between this mechanism and the previous one
is that at least one electron transfer occurs directly to S.
We do not distinguish between S and HS in the various re-
actions; S is present at very low concentrations at all pH values
discussed here. Also, except where it is specifically discussed, we do
8
fide ions are produced in at least a moderately high yield, hydrogen
does not appear to be an intermediate under these conditions. ͑It is
conceivable that water is reduced to hydrogen which reacts at 100%
yield with S. However, even a few percent drop in yield would lead
ϩ
to an expected visible H evolution. Therefore it is much more likely
2
2
Ϫ
Ϫ
that hydrogen is not involved under these conditions͒. No less im-
portant, is the formation of sulfide at neutral pH and the resulting
deduction that current can pass through the S undermines the main
reason to favor the excited H/S reaction.
2
Ϫ
ϩ
ϩ
not differentiate between reduction of H and H O (H is used in
2
the above pathways; as pH increases, direct water reduction be-
comes increasingly more probable͒.
The following various data can give clues as to which mecha-
nism is more likely.
It is more difficult to explain the sharp dependence on acid con-
centration of yield ͑point 1͒ based on this mechanism. The yield of
stibine evolution ͑always low͒ from antimony cathodes has been
found to be much higher at a pH of 1.3 compared to Ϫ0.3 ͑it de-
2
6
1
. For H S and H Se, the yield decreases sharply in a narrow
creases again at very high values of pH͒, and this was explained
2
2
by the mechanism proceeding through reduction of water and not
acid concentration range ͑between 1 and 4 M H SO concentration͒.
2
4
ϩ
H , which then reacts with the Sb. It would seem unlikely that the
This sharp dependence is not seen for H Te and AsH ͑although
2
3
very large differences we observe between 0.5 and 4 M acid ͑Fig. 4͒
are due solely to such a reason, although we cannot offer a better
reason. An important point on reactions involving H atoms. The
standard potential for hydrogen evolution is based on molecular H2 .
The potential for a single H atom is much more negative ͑Ϫ2.1 V͒.
AsH3 evolution efficiency does increase for more alkaline
6
solutions ͒. A comparable decrease has been reported previously for
stibine evolution ͑ca. 20-fold decrease between pH of 1.3 and Ϫ0.3,
26
although the yields of stibine are always very low ͒.
2
. In neutral K SO4 solution, sulfide is produced with no gas
2
This potential will be reduced depending on the interaction ͑stabili-
zation͒ of the H atom with the electrode surface. For graphite, this
stabilization will be relatively weak. This argues against the forma-
tion of isolated H atoms which can diffuse to the S͑e͒. It does not
influence the possibility of a reaction by nascent hydrogen in the
form of vibrationally excited molecules rather than in the more com-
monly understood form of atoms.
evolution. This means either hydrogen is not involved in the reaction
or, if it is, it reacts with S at essentially 100% yield. For a low acid
concentration ͑up to ca. 0.5 M͒ where K SO is also present, the gas
2
4
evolution is considerably reduced compared to a K SO -free solu-
2
4
tion and most of the sulfide is formed as a concentrated solution of
K S or KHS͒.
2
3. S is a very good insulator; it seems unlikely that electrons
We note that, purely from a solubility viewpoint, the H S should
2
would flow directly through the S; it is conceivable that electrons
injected at the graphite/S interface penetrate a small distance into the
S.
be less soluble in the electrolyte and therefore the in situ hydride
evolution efficiency ͑in contrast to hydride formation͒ should in-
crease with increasing acid concentration, based on the equilibrium
in Eq. 2
We now consider the two basic mechanisms separately.
Generation of hydrogen and subsequent chemical reaction with
S.—S is an excellent insulator ͑Se much less so, although a poor
conductor; Te and As are good conductors͒. We originally assumed
that, for the S/graphite electrodes, all the current flows through the
graphite ͑point 3 above͒. In this case, mechanism 1 is the one which
would best explain the hydride generation. According to this sce-
nario, hydrogen is formed at the graphite and migrates to a neigh-
boring S where it reacts, either as atomic hydrogen or as vibra-
tionally excited H2 .
Ϫ
ϩ
H S͑e) ϩ H O ꢀ HS͑e) ϩ H O
͓2͔
2
2
3
This equilibrium explains why gas evolution predominates in
strongly acid solution while sulfide ion formation occurs to an in-
creasing extent as the pH increases.
The reaction between gaseous H and S at room temperature is
Direct reduction of S.—The apparent lack of hydrogen formation in
neutral solution ͑point B͒ is the main support in favor of this path-
way. While it could be argued that the mechanism may be different
at different values of pH, the fact that sulfide is apparently generated
without hydrogen implies that the high resistivity of S, the main
factor in support of the H-mediated reaction, is not, in fact, a prob-
lem. Also, the fact that at 0.05 M acid ϩ K SO , sulfide is formed
2
essentially negligible. It is, however, much more efficient at inter-
mediate temperatures ͑Ϸ300-400°C͒. There are a number of studies
which do claim or suggest direct reaction at room temperature be-
tween nascent hydrogen ͑whether prepared electrochemically or
chemically͒ and elemental sulfur ͑or other hydride-forming
2
,12,27
element͒.
Nascent hydrogen is usually understood as meaning
2
4
Downloaded on 2014-10-20 to IP 130.63.180.147 address. Redistribution subject to ECS terms of use (see ecsdl.org/site/terms_use) unless CC License in place (see abstract).

D40
Journal of The Electrochemical Society, 152 ͑3͒ D35-D41 ͑2005͒
at high efficiency but with reduced gas generation at the electrode
point 2͒ together with the probable formation of KHS (K S) im-
local changes in pH possible at the cathode, the balance need not be
so delicate.
͑
2
plies that sulfide is formed directly as a first step ͑see also the sec-
We should mention another possible ͑but probably unlikely͒ rea-
son for the concentration effect, based on the assertion by Mon-
blanova and Kobozev that atomic hydrogen cannot be involved
͑rather they believed excited, molecular hydrogen to be the reactive
species͒ since hydrogen atoms would be expected to decompose
tion on the effect of K SO below͒.
2
4
In view of the high resistance of S, it is probable that reduction
takes place at the graphite/S interface ͑this is also true for the pre-
vious mechanism, but much less stringently so, since it is more
conceivable that nascent hydrogen can diffuse relatively large dis-
tances than for electrons to move through the S͒. Earlier studies
where a S/graphite cathode was operated in a solution containing
Cu͑II͒ ions indeed showed that CuS formation occurred only at the
S/graphite interface.24 This goes to confirm the conclusion that the
reaction occurs at or near this interface.
An important implication of this mechanism is that S is removed
from the electrode at or very close to the S/graphite interface. This
means that, after a small amount of S has been reduced, there should
be decreasing contact between the S and graphite. However, we
1
3
H S back into hydrogen and elemental S. This introduces the pos-
2
sibility that the low efficiencies at low pH may not be purely intrin-
sic but rather a moderately high yield of H S forming initially would
2
be redecomposed by atomic H to the elements, i.e., the net yield is
low. According to the scheme given by Monblanova and Kobozev,
one H atom could decompose one H S molecule. The lower yield at
2
ϩ
higher acid concentrations could be explained by more H reduc-
tion, resulting in a greater H atom concentration as well as reduced
direct S reduction. However, in view of our expectation, discussed
earlier, that free H atoms are not involved, this scenario does not
appear very likely.
What is the origin of the difference ͑in dependence on acid con-
centration͒ between Te and As on one hand and S and Se ͑also Sb͒
on the other is a question which we cannot reliably answer. The
have shown that the efficiency of H S generation is stable until ca.
2
20%, a relatively large fraction, of the S has been consumed. This
same consideration is also valid, although to a lesser extent, for the
hydrogen-mediated route. One might speculate that electrons do not
necessarily react at the nearest S atoms but may move on the surface
of the S at the interface with the electrolyte ͑e.g., charge hopping on
the surface via electron traps͒. The distance that an electron could
travel by such a mechanism would depend both on the trapping
characteristics of the interface as well as on the competing pathways
for electron injection into the electrolyte.
simplest answer would be that competitive H evolution occurs with
2
a larger overpotential at Te and As compared with the other ele-
ments. This problem would require a study in its own right. Finally,
it should be kept in mind that different dominant mechanisms may
be applicable to different systems or to the same system under dif-
ferent conditions. Having said this, for H S and H Se at least, the
To explain the strong dependence of sulfide ͑selenide͒ reduction
on acid concentration, we propose three hypotheses, the first of
which is the most likely.
2
2
evidence is more in favor of a chalcogen reduction as the dominant
mechanism for hydride formation.
All possible reactions at the cathode will be accompanied by a
ϩ
local increase in pH ͑depletion of H either by direct discharge or
Practical considerations and projected uses.—The details of the
methods we have described have only been partly optimized and
then only for laboratory use. We envisage that these methods may
find their greatest applications for such small-scale uses, although it
is possible that they could be scaled-up for larger scale use. Im-
provements ͑or simplifications, as mentioned in the Experimental
section͒ in the fabrication of the electrodes can surely be made
͑larger area, more optimized graphite/active element distribution,
better mechanical properties of some of the electrodes͒. For ex-
ample, towards the end of this study, we found that simply dipping
a metal substrate into a mixture of 10% ͑by volume͒ graphite stirred
into molten S in a beaker gave electrodes which were sufficiently
by reaction with sulfide͒. This can be seen visually by the yellow
coloration localized near the cathode in the 0.05 M H SO
2
4
ϩ 0.25 M K SO electrolyte ͑polysulfide is not stable in very acid
2
4
solutions͒. This will clearly reduce hydrogen evolution vis a vis S
ϩ
reduction. In concentrated acid, the concentration of H may be so
high that the pH difference on electrolysis is small. However, as the
ϩ
solution pH (͓H ͔) is diminished, the local change in pH at the
cathode will increase, thereby reducing hydrogen evolution to a
greater extent and favoring S reduction. There are several factors
which can cause this, such as, thermodynamic negative shift in H
evolution potential, decreased concentration of H ͑concentration
2
ϩ
polarization͒, and a change in the mechanism of H discharge from
conductive to allow a good H S generation rate. Thicker electrodes
could be formed by successive dipping and solidifying.
2
ϩ
H
to water reduction. The fact that increasing current only in-
creases H S efficiency slightly ͑Fig. 3, and also only a similarly
weak increase in efficiency was found in a separate experiment in 2
While higher values of pH appear to lead to higher faradaic
efficiencies of reduction, the efficiency of gas evolution ͑for the
chalcogens͒ will decrease if the pH becomes too high ͑Eq. 2͒. Ad-
ditionally, electrogenerated chalcogenide ions can dissolve elemen-
tal chalcogen from the electrodes at higher pH.
2
M H SO where the difference might be expected to be larger ac-
2
4
cording to the above hypothesis͒ argues somewhat against this hy-
pothesis.
The standard potential for S reduction ͑ϩ0.14 V͒ is more posi-
tive than for hydrogen evolution in the acidic solutions involved
here. Therefore thermodynamically, direct S reduction is favored
over hydrogen evolution. The strong concentration effect could then
Some extensions of the method can be envisaged. Gas mixtures
can be generated with good control over the composition, and used
for the formation of a mixed compound. As an example, we have
generated a mixture of H S and H Se. Two electrochemical cells
2
2
ϩ
be explained by changes in H concentration ͑activity͒ between the
were connected in series to produce the same amount ͑equal cur-
rents͒ of the two hydrides ͑both being generated at ca. 80% effi-
ciency in 0.5 M H SO ). The carrier gas ͑Ar͒ flowed through the
1
M and у4 M acid ͑a lower pH would favor hydrogen evolution͒.
The sharp change with pH ͑over less than one pH unit͒ would then
imply a fine balance between hydrogen evolution and S reduction.
2
4
H S cell and then the H Se cell and finally into a solution of CdAc .
2
2
2
This purely thermodynamic argument would be valid for H S gen-
2
XRD of the resulting precipitate gave peak positions corresponding
to a stoichiometry close to CdS0.5Se0.5 , based on Vegard’s law. More
generally, using a parallel configuration of two ͑or more͒ cells, any
composition of mixed gas can be generated by applying the desired
currents to each electrode.
Chemical vapor deposition of semiconductor films can conceiv-
ably be carried out by producing two gases which can react with
each other. For example, H2S and AsH3 should react to give As
sulfide͑s͒ under suitable conditions.
eration but not for H Se ͑Se reduction potential is Ϫ0.4 V͒. Since
2
gas generation occurs in all our experiments ͑in acid solution͒ under
conditions of large overpotential, where both chalcogen and
ϩ
H /water reduction can occur, simple thermodynamic arguments are
clearly not sufficient, and it is possible, if somewhat coincidental,
that this fine balance between S or Se reduction and hydrogen evo-
lution is the reason for the strong acid concentration effect. Actually,
the previous explanation also is based on a balance between hydro-
gen and S reduction, but in that case, because of the relatively large
Downloaded on 2014-10-20 to IP 130.63.180.147 address. Redistribution subject to ECS terms of use (see ecsdl.org/site/terms_use) unless CC License in place (see abstract).

Journal of The Electrochemical Society, 152 ͑3͒ D35-D41 ͑2005͒
Conclusions
D41
References
We have demonstrated the electrochemical generation of H2S
1. A. Brockmann, C. Nonn, and A. Golloch, J. Anal. At. Spectrom., 8, 397 ͑1993͒.
2
3
. F. Laborda, E. Bolea, and J. R. Castillo, J. Anal. At. Spectrom., 15, 103 ͑2000͒.
. E. Bolea, F. Laborda, M. A. Belarra, and J. R. Castillo, Spectrochim. Acta, Part B,
and H Se from composite S͑e͒/graphite electrodes at high faradaic
2
efficiencies ͑up to 100%͒ in dilute ͑0.05 M͒ sulfuric acid with added
5
6, 2347 ͑2001͒.
4. G. Grube, Electrochem. Solid-State Lett., 30, 517 ͑1924͒.
K SO . Hydride generation is believed to occur by direct reduction
2
4
5
6
7
8
. H. W. Salzberg and B. Goldschmidt, J. Electrochem. Soc., 107, 348 ͑1960͒.
. J. L. Valdes, G. Cadet, and J. W. Mitchell, J. Electrochem. Soc., 138, 1654 ͑1991͒.
. E. Ernyei, Z. Anorg. Chem., 25, 313 ͑1900͒.
of S or Se. Much lower efficiencies were obtained in the concen-
trated acid ͑уseveral molars͒. The difference was attributed to the
balance between hydrogen evolution and S͑e͒ reduction and its
strong dependence on the local pH at the cathode surface. Compari-
son with H Te and AsH generation was also made. It is probable
. W. Hempel and M. G. Weber, Z. Anorg. Chem., 77, 48 ͑1913͒.
9. L. M. Dennis and R. P. Anderson, J. Am. Chem. Soc., 36, 882 ͑1914͒.
0. T. Engelhard, E. D. Jones, I. Viney, Y. Mastai, and G. Hodes, Thin Solid Films, 370,
1
2
3
1
01 ͑2000͒.
that direct reduction of the elements occurs in these cases, too. How-
ever, in sharp contrast to the hydrides of S and Se, the evolution of
those of Te and As are much less dependent on acid concentration.
At the same time, efficiencies ͑in acid solutions͒ are lower than
those for S and Se in dilute acid and higher than in concentrated
acid. The techniques described here should make in situ preparation
of these poisonous and often unstable hydrides much simpler on the
laboratory scale, and may even be useful on a larger scale. The
accurate control of generation rate via the current allows just the
quantity of hydride needed to be produced. Using more than one
electrode, mixtures of these hydrides can be generated with fine
control over the composition.
1
1. M. Le Blanc, Z. Elektrochem. 11, 813 ͑1905͒.
12. A. Baeyer, Ber. Deutsch. Chem. Ges., 1, 117 ͑1868͒.
13. V. V. Monblanova and N. I. Kobozev, Zh. Fiz. Khim., 29, 480 ͑1955͒.
1
4. V. P. Gladyshev, L. K. Dubinina, and S. V. Kovaleva, Izv. Vyssh. Uchebn. Zaved.
Khim. Khim. Tekhnol., 23, 259 ͑1980͒.
5. L. V. Kaabak, A. P. Tomilov, and S. L. Varshavskii, Zh. Vses. Khim. O. im D.I.
Mendeleeva, 9, 700 ͑1964͒.
1
16. A. P. Tomilov, L. V. Kaabak, and S. L. Varshavskii, Zh. Vses. Khim. O. im D.I.
Mendeleeva, 8, 703 ͑1963͒.
1
1
19. T. Biegler and D. A. Swift, J. Appl. Electrochem., 6, 229 ͑1976͒.
20. G. V. Makarov, Y. A. Batrakova, S. K. Nasipkalieva, and V. I. Fedorchenko, J.
Appl. Chem., USSR, 53, 1192 ͑1980͒.
2
2
2
7. I. Bernfeld, Z. Physik. Chem., 25, 46 ͑1898͒.
8. T. Tatebe, Jpn Pat., p. 8410 ͑1954͒.
1. P. Fischer, Z. Elektrochem., 31, 285 ͑1925͒.
2. P. Fischer, Z. Elektrochem., 31, 286 ͑1925͒.
3. H. Berge, B. Str u¨ bing, S. Strasen, M. Aquila, and K. Peseke, Wiss. Z. Wilhelm-
Pieck-Univ. Rostock, 27, 639 ͑1978͒.
Acknowledgments
2
2
2
2
4. L. Edwards, R. I. Holliday, and D. Honey, J. Electroanal. Chem., 297, 425 ͑1991͒.
5. H. J. S. Sand, J. Grant, and W. V. Lloyd, J. Am. Chem. Soc., 129, 378 ͑1927͒.
6. H. W. Salzberg and A. J. Andreatch, J. Electrochem. Soc., 101, 528 ͑1954͒.
7. A. K. Dmitriev, Zh. Obshch. Khim., 18, 1266 ͑1948͒.
We thank A. J. Bard for fruitful discussions. G. H. thanks Centre
National de la Recherche Scientifique ͑CNRS͒ for support of a one
year position ͑Poste Rouge͒ at the Laboratoire de Chimie M e´ tal-
lurgique des Terres Rares.
28. M. Rutigliano, M. Cacciatore, and G. Billing, Chem. Phys. Lett., 340, 13 ͑2001͒.
Downloaded on 2014-10-20 to IP 130.63.180.147 address. Redistribution subject to ECS terms of use (see ecsdl.org/site/terms_use) unless CC License in place (see abstract).