You are on page 1of 6

210 Energy & Fuels 2003, 17, 210-215

Instability of Sulfate and Selenate Solid Acids in Fuel


Cell Environments
Ryan B. Merle, Calum R. I. Chisholm, Dane A. Boysen, and Sossina M. Haile*
California Institute of Technology, Materials Science, Mail Code 138-78,
Pasadena, California 91125
Received May 28, 2002

The chemical and thermal stability of several solid acid compounds under fuel cell operating
conditions has been investigated, primarily by thermogravimetric methods. Thermal decomposi-
tion of CsHSO4, a material which has shown promise as an alternative electrolyte for proton
exchange membrane (PEM) fuel cells, initiates at ∼175°C under inert conditions. The overall
decomposition process can be expressed as 2CsHSO4 f Cs2SO4 + H2O + SO3 with Cs2S2O7
appearing as an intermediate byproduct at slow heating rates. Under reducing conditions,
chemical decomposition can occur via reaction with hydrogen according to 2CsHSO4 + 4H2 f
Cs2SO4 + 4H2O + H2S. In the absence of fuel cell catalysts, this reduction reaction is slow;
however, materials such as Pt, Pd, and WC are highly effective in catalyzing the reduction of
sulfur and the generation of H2S. In the case of M3H(XO4)2 compounds, where M ) Cs, NH4, or
Rb and X ) S or Se, a similar reduction reaction occurs: 2M3H(XO4)2 + 4H2 f 3M2XO4 + 4H2O
+ H2X. In an operational fuel cell based on CsHSO4, performance degraded with time, presumably
as a result of H2S poisoning of the anode catalyst. The performance loss was recoverable by
exposure of the fuel cell to air at 160 °C.

Introduction form of steam, which is harmless to the otherwise water-


soluble electrolyte.
It has recently been demonstrated that CsHSO4, a
While numerous advantageous features of solid acid
prototypical “superprotonic” solid acid, can be utilized
electrolytes have been noted, it was also reported that
as a fuel cell electrolyte, despite its solubility in water
fuel cell degradation occurred during operation.1 In that
and unusual mechanical properties.1 Solid acids can be
first work, the focus was on the possible negative impact
regarded as chemical intermediates between normal
of steam in the environment. Accordingly, fuel cell
acids, such as H2SO4, and normal salts, such as Cs2SO4.
performance was measured both prior to and after 18
Those with superprotonic properties undergo a struc-
h of exposure of the cell to humidified air, but not after
tural transformation at slightly elevated temperatures
prolonged exposure to standard fuel cell gases (humidi-
to a highly disordered state. Accompanying this trans-
fied H2 at the anode and humidified O2 at the cathode).
formation is an increase in protonic conductivity by
Degradation under these conditions was slight, but
several orders of magnitude.2-9 Such materials offer
measurable. It was suggested that the cause was the
several advantages over polymeric electrolytes in fuel
reduction of sulfur in the solid acid at the anode to form
cell applications such as operability at slightly elevated
dihydrogen sulfide, rather than detrimental interactions
temperatures, no humidification requirements, and,
with water. In the present work, we examine the
given their solid nature, presumably impermeability to
thermal stability of CsHSO4 and explore the possibility
methanol. The key to successful implementation of
that reduction of superprotonic solid acid sulfates and
superprotonic solid acids in a fuel cell, as has been
selenates occurs under hydrogen gas and can cause
demonstrated for CsHSO4, is simply operation above
degradation in fuel cell performance.
100 °C. At these temperatures, H2O is present in the
* Corresponding author. Experimental Section
(1) Haile, S. M.; Boysen, D. A.; Chisholm, C. R. I.; Merle, R. B.
Nature 2001, 410, 910-913. The chemical stability of several known superprotonic solid
(2) Baranov, A. I.; Shuvalov, L. A.; Shchagina, N. M. JETP Lett. acid compounds was examined. These were CsHSO4, (NH4)3H-
1982, 36, 459-462.
(3) Moskvich, Yu. N.; Sukhovskii, A. A.; Rozanov, O. V. Sov. Phys. (SO4)2, (NH4)3H(SeO4)2, Rb3H(SeO4)2, and the mixed sulfate
Solid State 1984, 26 (1), 21-25. phosphate, Cs2(HSO4)(H2PO4). In addition, the salt Cs2SO4 was
(4) Pawlowski, A.; Pawlaczyk, Cz.; Hilczer, B. Solid State Ionics also studied for comparison. The solid acid sulfates were
1990, 44, 17-19. synthesized in-house according to standard recipes, see refs 6
(5) Ramasastry, C.; Ramaiah, A. S. J. Mater. Sci. Lett. 1981, 16,
2011-2016. and 10, for example, whereas the solid acid selenates were
(6) Gesi, K. Phys. Status Solidi A 1976, 33, 479-482. provided by the Russian Institute of Crystallography. The
(7) Chisholm, C. R. I.; Haile, S. M. Solid State Ionics 2000, 136- mixed sulfate phosphate was prepared from a solution of Cs2-
137, 229-241. CO3, H2SO4, and H3PO4, in which the Cs:S:P ratio was fixed
(8) Haile, S. M.; Calkins, P. M.; Boysen, D. Solid State Ionics 1997,
97, 145-151.
(9) Haile, S. M.; Lentz, G.; Kreuer, K.-D.; Maier, J. Solid State Ionics (10) Boysen, D. A.; Chisholm, C. R. I.; Haile, S. M.; Narayanan, S.
1995, 77, 128-134. R. J. Electrochem. Soc. 2000, 147, 3610-3613.

10.1021/ef0201174 CCC: $25.00 © 2003 American Chemical Society


Published on Web 11/28/2002
Sulfate and Selenate Solid Acids in Fuel Cell Environments Energy & Fuels, Vol. 17, No. 1, 2003 211

Table 1. Physical Characteristics of Catalyst Materials


surface
catalyst source particle size area, m2/g
Pt black Alfa Aesar N/A 27
RuO2 [Ru] Alfa Aesar <1 µm 45-65
Pd black Aldrich N/A 40
Au Aldrich 100 nma N/A
Ir black Alfa Aesar N/A 30-60
Ni/Al “Raney Ni” Alfa Aesar 200 mesh (<74 µm) N/A
WC Alfa Aesar <1 µm 1.2-1.4
a With agglomerates of ∼1-2 µm.

at 2:1:1. The solution was heated to 50 °C and continuously


stirred until a precipitate formed, which was then filtered and
rinsed with methanol. The sulfate salt was obtained com-
mercially and used as-received (Alfa Aesar, 99.9%).
To establish the thermal dehydration/decomposition tem-
perature of CsHSO4, this material was examined by thermal
gravimetric analysis (TGA) under flowing argon (40 cm3/min)
using a Netzsch STA 449. Samples were used in the form of
lightly crushed single crystals so as to minimize effects due to
surface water. Heating rates were varied from 0.5 to 40 °C/
min, and data collected up to 400 ( 2 °C. The sample mass
was kept constant at 16 mg. The impact of Pt catalyst material
on the dehydration process was examined by performing TGA
analysis under similar conditions, but for CsHSO4 powders
containing 35 wt % platinum powder, the properties of which
are detailed in Table 1.
To establish chemical stability under reducing conditions,
TGA measurements were carried out under flowing H2. The
majority of these experiments were performed isothermally Figure 1. (a) Weight change in CsHSO4 as a function of
at 160 ( 10 °C (the temperature at which the earlier fuel cell temperature at heating rates as indicated and under flowing
results were obtained 1) using a Perkin-Elmer TGA 7. The argon (40 cm3/min); and (b) weight change in CsHSO4, with
hydrogen flow rate was 20 cm3/min. For temperature-depend- and without 35 wt % Pt catalyst, as a function of temperature
ent measurements (1 °C/min), a Netzsch STA 449 was utilized at at 1 °C/min and under flowing argon (40 cm3/min). Note
and the gas was 4% H2 in argon (40 cm3/min). The evolved the difference in scales. Inset in (a) shows the temperature at
gases were analyzed using a Balzers Thermostar AMU200 which 0.05% of weight is lost as a function of heating rate. A
quadrupole mass spectrometer (MS). Prior to exposure to H2, total weight loss of 3.92% corresponds to the loss of 1/2 mole of
powders were equilibrated under flowing nitrogen to eliminate H2O and the formation of Cs2S2O7.
weight changes due to loss of surface water. Sample quantities
were on the order of 20-30 mg. Cesium hydrogen sulfate was Results and Discussion
examined both as-synthesized and mixed with a variety of The thermal behavior of CsHSO4 under inert atmo-
typical fuel cell catalyst materials (35 wt %). Weight changes sphere is shown in Figure 1. In Figure 1a, the thermo-
for all other compounds were measured (only) in the presence gravimetric data for the material alone are presented,
of Pt, again, in mixtures containing 35 wt % catalyst. Ad-
whereas in Figure 1b the weight loss curve at 1 °C/min
ditional catalysts examined in the case of CsHSO4 were Ru,
Pd, Au, Ir, Ni/Al, and WC. The characteristics of the catalyst for samples with and without Pt catalyst are presented.
powders are presented in Table 1. For experiments with Ru, In addition, the inset in Figure 1a shows the tempera-
the metallic powder was obtained by reduction of RuO2 under ture at which 0.05% of the initial weight has been lost
flowing H2 165 °C for 13 h. In some cases, the reaction product as a function of (heating rate)1/2. Plotting the data in
(of the solid acid with H2) was examined by X-ray powder this manner allows one to extrapolate to zero heating
diffraction. For this, a Scintag Pad-V diffractometer, equipped rate and determine at which temperature this magni-
with a sealed tube Cu KR X-ray source, was utilized. tude of weight loss will be encountered even under
Performance of CsHSO4-based fuel cells over a prolonged isothermal conditions.11 The results here demonstrate
period of time under typical operating conditions has also been that CsHSO4 is generally stable against dehydration to
measured. A single cell membrane electrode assembly (MEA) ∼175 °C. The erratic behavior of the data in Figure 1a,
was fabricated as reported earlier.1 Briefly, a layer of the solid however, indicates that the initial dehydration steps are
acid was sandwiched between two electrocatalysis layers
rather sample dependent. Moreover, for ground powder
comprised of CsHSO4, Pt black, carbon black, and the volatile
organic naphthalene, in a mass ratio of 6:10:1:1 (approximate samples, a slight, but measurable, weight loss under
volume ratio of 4:1:2:2). These layers were, in turn, placed inert atmosphere was occasionally observed under
between two sheets of porous graphite current collectors. The isothermal conditions for temperatures as low as 145
entire assembly was uniaxially pressed at 490 MPa, to yield °C. The diffraction pattern of such samples was identical
a dense electrolyte membrane (1.3 mm in thickness) with good to that of standard CsHSO4. Thus, this small weight
mechanical contact to the electrocatalyst layers. The MEA was change is likely attributable to loss of H2O from the
placed in a standard graphite test station and fuel cell surface region of CsHSO4 particles. In the case of the
polarization curves obtained at 154 ( 2 °C. Upon heating to slowest heating rate, 0.5 °C/min, a plateau in the TGA
the measurement temperature, the organic phase in the
electrode evaporated, leaving behind a porous electrode struc- (11) Gray, A. P. Analytical Calorimetry; Plenum Press: New York,
ture. The Pt loading was ∼15 mg/cm2. 1968; pp 209-218.
212 Energy & Fuels, Vol. 17, No. 1, 2003 Merle et al.

Figure 2. Weight loss isotherm of CsHSO4 under flowing H2


(20 cm3/min) and at a temperature of 160 °C with and without
35 wt % Pt catalyst. To allow for direct comparison, the weight
of Pt has been accounted for in the curve for CsHSO4 + Pt.

curve at ∼96 wt % corresponds to the loss of 1/2 mole of


H2O (per mole CsHSO4), and is consistent with the
formation of Cs2S2O7. Examination of the data in Figure
1b clearly reveals that the Pt in the system strongly
influences the dehydration process. The onset of CsH-
SO4 dehydration is significantly decreased while the
rate at which dehydration occurs is greatly increased
in the presence of the catalyst material.
In Figure 2 the weight loss isotherms (T ) 160 °C) of
CsHSO4, both alone and in the presence of platinum
catalyst material, under reducing atmospheres are
presented. In the absence of any catalyst material, the
solid acid exhibits a weight loss of 0.03% over a 10 h
period during the isothermal hold. In the presence of
Pt, however, the sample loses 4.4 wt %, or 6.8 wt % of
solid acid. These results demonstrate quite clearly that,
whatever the reaction between H2 and CsHSO4, it is
highly catalyzed by Pt.
The temperature-dependent measurement, Figure 3,
reveals that weight loss is steady but slight at temper-
atures between 120 and 140 °C (<0.01%/°C), and then
increases rapidly beyond ∼140 °C to a much more severe
rate of 0.07%/°C. The differential scanning calorimetry
(DSC) curve exhibits an exothermic peak at this tem-
perature, which closely corresponds to the superprotonic Figure 3. Combined thermal and evolved gas analysis of
transition, reported to occur at 141 °C.2 The rapid CsHSO4 mixed with 35 wt % Pt catalyst, under flowing 4%
weight loss beyond the transition suggest that the H2/96% Ar (40 cm3/min) and a heating rate of 1 °C/min; (a)
kinetics of the reaction are correlated to the structural thermal gravimetric and differential scanning calorimetry
state of the solid acid, perhaps being enhanced by the traces; (b) mass spectrometry signals for H2O and H2S; and
(c) ratio of the intensities of the mass spectra signals for these
dynamics of the SO4 group reorientation in the high- two species.
temperature, disordered phase. At higher temperatures,
an anomaly is evident in the DSC, TGA, and MS data as a consequence of the reaction of the hydrogen with
at ∼198 °C. The precise chemistry associated with this residual oxygen in the experimental apparatus. Defining
feature is unclear, but it is again apparent that the the background signal as the value at 120 °C, the ratio
decomposition of CsHSO4 under reducing conditions is of the water to hydrogen sulfide signals, after back-
much more severe than under inert conditions and in ground subtraction, is as high as ∼26:1 at the transition
the absence of a catalyst material. and decreases gradually with increasing temperature,
The mass spectrometry results show a dramatic Figure 3b. The observed ratio is significantly greater
increase in both the H2S and H2O content of the evolved than the 4:1 value expected for reaction 1, and only some
gas at precisely the position of the first DSC peak. This of this discrepancy can be explained by differences in
result is in general agreement with the reaction pro- sensitivity of the MS detector to these species. A likely
posed earlier1: explanation is that, in addition to catalyzing the reduc-
tion reaction, Pt also catalyzes the simple dehydration
2CsHSO4 + 4H2 f Cs2SO4 + 4H2O + H2S (1) reaction:
2CsHSO4 f Cs2SO4 + H2O + SO3 (2)
Quantitative analysis of the H2O and H2S contents as
determined from the mass spectrometry experiments is as suggested by the results in Figure 1b. It should be
difficult because the background H2O signal was high noted that at temperatures above ∼140 °C, SO and SO2
Sulfate and Selenate Solid Acids in Fuel Cell Environments Energy & Fuels, Vol. 17, No. 1, 2003 213

Figure 4. X-ray powder diffraction pattern of the reaction byproduct of H2 with CsHSO4 (in the presence of Pt catalyst).

(but no SO3) were also detected, as might be expected


from an overall decomposition reaction that includes
both reactions 1 and 2.
While these mass spectrometry experiments were
repeated several times to ensure reproducibility, it
should be noted that, on occasion, the mass spectrom-
etry data showed a sharp peak rather than a simple
increase in the H2O and H2S signals at 140 °C and that
the H2O:H2S signal ratio was extremely high (∼200:1).
Moreover, beyond 140 °C, the intensity of the H2S signal
gradually increased, whereas that of the H2O signal
gradually decreased, and, simultaneously, the DSC
signal was measurably endothermic at these high
temperatures. These observations are consistent with
a reduction process that generates H2S that is heavily
adsorbed onto the surface of the Pt particles, and then Figure 5. Weight loss isotherms of solid acid and salt
compounds (as indicated) under flowing hydrogen (20 cm3/min)
gradually desorbs with increasing temperature. In such in the presence of Pt catalyst at 160 °C.
cases, the SO and SO2 signals were much higher than
in the “standard” case. It could not be established to reaction 2. However, at this low reaction temperature,
whether the differences in these experimental results dehydration alone, even in the presence of the catalyst,
were due to residual oxygen in the system, or true Figure 1b, would not be expected to proceed to comple-
sample-to-sample variations. tion.
The diffraction pattern of the byproduct of the reac- The reduction reaction 1 can be expressed generally
tion between hydrogen and cesium hydrogen sulfate is for MHXO4 and M3H(XO4)2 compounds (where M )
shown in Figure 4. The sample was examined after ∼15 alkali metal or ammonium ion and X ) S or Se):
h of reaction at 160 °C in the presence of Pt catalyst
(again, 35 wt %). Prior to exposure to H2, the diffraction 2MHXO4 + 4H2 f M2XO4 + 4H2O + H2X (3a)
pattern of the as-synthesized material corresponded
exactly to that of CsHSO4 phase III (not shown), 2M3H(XO4)2 + 4H2 f 3M2XO4 + 4H2O + H2X (3b)
whereas after reduction, Figure 4, the diffraction peaks
correspond to those of Pt and Cs2SO4. This result is in A comparison of the stabilities of various sulfate and
agreement with the reaction 1. Complete thermal selenate compounds is presented in Figure 5 (atm )
decomposition of CsHSO4 under inert atmosphere would flowing H2, T ) 160 °C, catalyst ) 35 wt % Pt). Each of
also yield Cs2SO4 as the final reaction product according the acid compounds undergoes a superprotonic transi-
214 Energy & Fuels, Vol. 17, No. 1, 2003 Merle et al.

Figure 6. Weight loss isotherms of CsHSO4 under flowing


hydrogen (20 cm3/min) in the presence of various (potential)
fuel cell catalysts (as indicated) at 160 °C.

tion at a temperature below the measurement temper-


ature. It is apparent that the simple salt, Cs2SO4, is
stable under reducing conditions, despite the 6+ oxida-
tion state of sulfur and thus it is likely the final reaction
product. Disappointingly, from an applications perspec-
tive, none of the sulfate or selenate solid acids is stable
under these conditions. The rate of weight loss follows
the sequence (NH4)3H(SO4)2 > CsHSO4 > Rb3H(SeO4)2
> (NH4)3H(SeO4)2 > Cs2(HSO4)(H2PO4). Given the
fractional weight change per formula unit reacted, eq
3, one would expect weight loss to occur more quickly
for MHXO4 than M3H(XO4)2 compounds. The rapid
degradation of (NH4)3H(SO4)2 reveals that this material
is particularly unsuitable for fuel cell applications.
Figure 7. (a) Potential and (b) power density for a CsHSO4-
Indeed, attempts to demonstrate fuel cell performance based single cell fuel cell. Initial curves (circles) obtained
using (NH4)3H(SO4)2 were uniformly unsuccessful.12 The immediately upon reaching the measurement temperature of
lower reactivity of Cs2(HSO4)(H2PO4) may be a result 154 °C, post-degradation curves (triangles) obtained after 5 h
of the lesser tendency of phosphorus to form H3P than of operation, and post-recovery curves (squares) obtained after
sulfur to form H2S. an additional 3 h of exposure to stagnant air.
The effect of alternative catalyst materials on the
stability of CsHSO4 is evident in Figure 6. Here again, circuit of 17 mA/cm2. After flushing both the anode and
the results are undesirable for fuel cell applications. cathode with compressed air and then holding the
Even relatively inactive catalysts such as Au lead to a system in stagnant air for 5 h, the fuel cell performance
significant increase of the rate of reaction 1 compared recovered to a level close to that measured initially.
to its rate in the absence of catalysts. In general, the Peak power was 10 mW/cm2 and current density at
reaction rate increases according to the sequence Au < short circuit 41 mA/cm2. These results indicate that the
Ni/Al < RuO2 < Pt < WC < Pd < Ir. Given the wide loss in performance cannot be associated with degrada-
variation in specific surface areas of the catalyst materi- tion of the membrane directly, which would be irrevers-
als employed, Table 1, it is not possible to establish ible, but some other reversible process. Adsorption of
whether this sequence is a direct consequence of catalyst small quantities of H2S onto the surface of the Pt
chemistry. Nevertheless, it is surprising that WC, with catalyst, as generated according to reaction 1, and its
a relatively low specific surface area, serves as a very desorption upon exposure of the anode to air, is consis-
effective catalyst for sulfur reduction. There is no reason tent with the observed fuel cell behavior, although other
to expect these catalyst materials would have a lesser possibilities cannot be unequivocally ruled out. It has
impact on alternative superprotonic solid acids of sul- been established, in the case of phosphoric acid fuel
fates or selenates that might be considered for fuel cell cells, that as little as 50 ppm H2S in the incoming fuel
applications. stream significantly degrades anode performance, and
The performance of a typical CsHSO4-based single- that this degradation is similarly reversible.13
cell fuel cell is shown in Figure 7. Upon initial examina-
tion, the peak power was 12 mW/cm2, and the current Concluding Remarks
density at short circuit 45 mA/cm2. Damage to the fuel
cell under operating conditions was rapid. After 5 h of In the experiments carried out here the catalyst
continuous operation, the performance fell to a peak content (35 wt %) is much greater than is used in a
power of 6 mW/cm2, and a current density at short typical fuel cell. Thus, the loss of electrolyte in a solid

(12) Boysen, D. A.; Merle, R. B.; Chisholm, C. R. I.; Haile, S. M. (13) Chin, D.-T.; Howard, P. D. J. Electrochem. Soc. 1986, 133,
Unpublished results. 2447-2450.
Sulfate and Selenate Solid Acids in Fuel Cell Environments Energy & Fuels, Vol. 17, No. 1, 2003 215

acid based fuel cell can be expected to be significantly dissolution of the electrolyte. The most ideal solution
less than shown in Figures 4 and 5. Accordingly, the to the stability problem is to develop completely alter-
decrease in performance of CsHSO4-based fuel cells with native superprotonic solid acids based on different
time is not a direct consequence of electrolyte damage, oxyanion (XO4n-) groups, in particular PO43- groups.
but rather, a consequence of the generation of H2S Efforts along these lines are underway, however, it must
which acts as a highly effective catalyst poison. Thermal be noted that the high-temperature behavior of CsH2-
desorption of hydrogen disulfide from the platinum PO4, the most likely candidate, simply based on its
catalyst surface is likely responsible for the recovery of chemical and structural similarity to CsHSO4, is not
fuel cell performance after exposure of the electrodes well-understood.15 Thus, a broad-based exploratory
to (humid) air. Identification of a fuel cell anode catalyst synthetic effort may be necessary to develop solid acids
that is not susceptible to H2S poisoning thus emerges that combine high conductivity with good stability.
as one strategy for enhancing the lifetime of solid acid
sulfate or selenate fuel cells. An alternative strategy is Acknowledgment. Dr. Tregubchenko kindly pro-
to increase the water partial pressure at the anode so vided the selenate crystals examined here. This work
as to drive reactions 2 and 3 to the left. Indeed, the high was supported by the California Energy Commission
water concentrations needed to operate direct methanol through the PIER grant program, by the California
fuel cells14 may suffice to prevent H2S formation. In such Institute of Technology through the Grubstake program
a case, however, the anode liquid (methanol plus water) and by the Office of Naval Research.
must be fed to the fuel cell in vapor form so as to prevent
EF0201174
(14) Wasmus, S.; Kuver, A. J. Electroanal. Chem. 1999, 461 (1-2),
14-31. (15) Lee, K.-S. J. Phys. Chem. Solids 1996, 57 (3), 333-342.

You might also like