You are on page 1of 13

J Therm Anal Calorim

DOI 10.1007/s10973-016-5474-y

On the low- to high proton-conducting transformation


of a CsHSO4CsH2PO4 solid solution and its parents
Physical or chemical nature?

E. Ortiz1 I. Pineres1 C. Leon2

Received: 21 January 2016 / Accepted: 9 April 2016


 Akademiai Kiado, Budapest, Hungary 2016

Abstract The first-order transition from low to superpro- Keywords KDP-type crystals  Superprotonic phase
tonic conducting phase (at 119 \ Tsp \ 145 C) of a solid transition  Decomposition reactions  Thermal analysis
solution of CsHSO4 and CsH2PO4 [Cs3(HSO4)2(H2PO4)]
was carefully examined by using modulated and conven-
tional differential scanning calorimetric, simultaneous Introduction
thermogravimetric and differential scanning calorimetric,
simultaneous thermogravimetric and mass spectroscopy, Polymer electrolyte membrane fuel cells constitute one of
impedance spectroscopy, and temperature evolution of the most promissory clean energy technological strategies
X-ray diffraction and scanning electron microscopy. Our for transportation, mobile, and stationary applications [1].
results show evidence that at this temperature, the Nevertheless, their efficiency is reduced because the
endothermic anomaly associated with the physical trans- proton-conducting membrane is permeable to hydrogen
formation is, instead, the response of a chemical-surface and due to the low operation temperature below 100 C,
thermal decomposition. One of its products is water: part of which is imposed by the requirement of liquid water for
it is evaporated and part is strongly bonded to the other proton conduction [2]. This is because this system does
decomposition products. Given that these are of polymeric not correspond to a truly solid phase; it contains liquid-
nature, they constitute a host matrix that contains liquid like regions between the polymer molecules and, more-
water regions. Therefore, as part of liquid water dissolves a over, does not conduct protons exclusively [3]. In con-
superficial portion of salt (providing protons), this system trast, a truly solid compound, such as acid salts CsHSO4
behaves in similar manner to a polymer electrolyte mem- (CHS) and CsH2PO4 (CDP), has been proposed by
brane (located over the salt surface) where the proton Hailes research group [2, 4] as an electrolyte for fuel
transport mechanism might include the vehicle type, using cells. When these salts are heated through 140 and
H3O? as a charge carrier. A discussion that favors the 230 C, respectively, an anomalous phase transition from
non-existence of the superprotonic conducting phase of low- to high proton-conducting phase takes place [5, 6],
CsH2PO4 is also presented. with conductivity values in the 10-2 S cm-1 order of
magnitude. This offers three fuel cell technical advan-
tages: its electrolyte is a solid material, it uses an anhy-
drous proton transport mechanism, and has intermediate
& E. Ortiz operating temperature range (200300 C) [7]. However,
everortiz@mail.uniatlantico.edu.co like polymer electrolyte membrane fuel cells, they are
1
also not free of operating problems: from the electromo-
GFM, Department of Physics, Faculty of Basic Sciences,
tive force measurements generated across a CHS sample
Universidad del Atlantico, Km 7 Antigua va Puerto
Colombia, Barranquilla, Colombia placed in an oxygen concentration cell at 145 C, a very
2 long time was required for the voltage to equilibrate
GFMC, Department of Applied Physics III, Faculty of
Physics, Universidad Complutense de Madrid, 28040 Madrid, (about 5 h) [8]; furthermore, hydration of the feeding
Spain gases by partial water pressure must be used to avoid

123
E. Ortiz et al.

possible decomposition, e.g., a CDP fuel cell requires a nCsH2 PO4 s ! Csn H2 Pn O3n1 s n  1H2 Ov 1
value of 0.4 atm [7].
Although the existence of the superprotonic conducting Here, n is the number of molecules participating in thermal
phase of CDP above 230 C is now widely accepted, it decomposition; s or v in parentheses denote that the cor-
was not always so. For example, Haile et al. [9] proposed responding compound is in the solid or vapor state,
mixing, in different proportions, the superprotonic CHS respectively. This statement is partially supported on Thi-
conductor with the non-superprotonic CDP to examine los [24] earlier review on polymerization of phosphates
whether the phosphorous atom destroys its superprotonic and arsenates. According to Taninouchi et al. [25], the
conducting state. While pursuing this goal, the new stable phase just above the dehydration temperature is the
superprotonic conductor, Cs3(HSO4)2(H2PO4), was dis- full dehydrated product of solid CsPO3, but partially
covered. This system, with a superprotonic conducting dehydrated products, such as solid Cs2H2P2O7 and an
phase transition temperature, Tsp, of 119 C [9] is an unknown phase, appeared as transient phases in the course
important aspect of the present research work. The to complete dehydration. Park [26] described the mass loss
assumption that CDP was not a superprotonic conductor at the 230245 C temperature range as difficult to explain,
[9] was supported by the fact that while the sole work by given that polymerization processes depend on temperature
Baranov et al. [10] suggests the presence of such a and on an unstable intermediate product. However, the
transition at 230 C into a cubic phase, numerous other author concludes that around 230 C, an onset of thermal
studies reveal the onset of decomposition at this tem- decomposition of CDP in CsnH2PnO3n?1 takes place at the
perature [1113]. Nevertheless, since the discovery of this crystal surface, but probably with n  1, [(CsPO3)n]. A
researcher, many solid acid superprotonic conductors similar interpretation was proposed for CHS by Ortiz et al.
have been reported; probably, the most recent is [18], who found that when it is heated through 141 C the
Cs2(SO4)0.9(HAsO4)0.1Te(OH)6 [14]. chemical reaction
Even though much research supports the existence of a 2CsHSO4 ! Cs2 S2 O7 H2 O 2
superprotonic phase transition in CDP [7, 1517], it is still a
takes place at the crystal surface. A parallel explanation for
controversial issue [7, 16, 17]; if when CDP is heated above
the case of CsHSeO4 has also been presented by the same
230 C, it presents a physical or chemical transformation. In
research group [27].
other words, it is the high conductivity observed above
The aim of the present work was to study a CHS and
230 C due to an intrinsically bulk solid physical protonic
CDP solid solution [Cs3(HSO4)2(H2PO4)] to examine
transport mechanism, or is it due to a (surface) chemical
whether it could be object of the same controversial subject
decomposition? CHS is not free of the same debate [18].
of its parents (CDP and CHS); if this is so, we hope to
Some authors believe that a physical transformation
contribute to shed light on the dilemma regarding the
takes place when CDP (CHS) is heated through 230 C
nature (physical or chemical) of the transformation of all
(141 C), consisting of the monoclinic structural phase
these salts.
transition to the cubic (tetragonal) phase. This transition to
a higher structural symmetry would largely enhance proton
transport, governed by the Grotthuss mechanism, which
includes two stages: proton transfer and reorientation of the Experimental
tetrahedron [1922]. The first consists of a proton dis-
placement along the hydrogen bond closer to one of the Cesium carbonate (Cs2CO3, 99.9 %) and sulfuric acid
PO4 (SO4) groups participating in the bond; the second (H2SO4, 9597 %) were obtained from Sigma-Aldrich Co.,
corresponds to a breaking of the longest half of the USA, and phosphoric acid (H3PO4, 85 %) from Merck,
hydrogen bond followed by a reorientation of the HPO4 Germany. Cs3(HSO4)2(H2PO4) crystals and crystalline
(HSO4) group to a new position. The reorientation con- powder samples were grown by slow evaporation of the
tinues until a second PO4 (SO4) group is found in an aqueous solution containing stoichiometric amounts of
appropriate orientation to form a new hydrogen bond. Cs2CO3, H2SO4, and H3PO4. The sample was vacuum-
Then, the process is repeated, and, as a result, long-range dried at room temperature and transferred into a test tube
proton transport through the lattice is possible. On the other that was placed into a hermetic glass bottle with moisture
hand, following the hypothesis by Lee [23], other authors indicator silica. After keeping the samples during more
believe that a chemical transformation occurs when CDP is than 2 months in this condition, they were pre-treated
heated above 230 C, giving rise to partial thermal during 1 hour at 65 C (under 99.999 pure helium gas),
decomposition (polymerization) at reaction sites on the before starting any experiment.
surface of the crystal, according to the following chemical Conventional (DSC) and modulated differential scan-
reaction: ning calorimetric measurements (MDSC) were conducted

123
On the low- to high proton-conducting transformation of a CsHSO4CsH2PO4 solid solution and its

on this salt with a TA Instruments 2920 analyzer. The DSC 71.4 C


curves were collected from several samples by using dif- 1
106.6 C
ferent heating rates. Similarly, thermogravimetric analysis
H = 26.4 J g1 127.5 C
(TG), on three samples, was performed with a TA Instru- H = 44.0 J g1
ments 2050 analyzer at different heating rates. For one of

Heat flow/mW
1
these TG experiments (using a heating rate of 2 C min-1),
a Balzers Thermostar mass spectrometer was coupled to 103.1 C

Endothermic
the TG equipment to execute a simultaneous TGMS
experiment. Also, simultaneous thermogravimetric and 2
differential scanning calorimetric (TGDSC) measure-
ments were obtained with a TA Instruments SDT 2960 by
using a heating rate of 1 C min-1. The MDSC thermo- 3 135.2 C
grams were made on sealed (using a Loctite Superflex Red
10 30 50 70 90 110 130 150
High-Temp silicone adhesive sealant) and unsealed crys-
Temperature/C
tals with an average heating rate, amplitude, and modula-
tion period of 0.2 C min-1, 0.2 C, and 60 s, respectively. Fig. 1 DSC curve for a Cs3(HSO4)2(H2PO4) crystal under a 5150
Regarding the DSC, MDSC, and SDT experiments, the 5 C heatingcooling cycle. The heating rate was fixed at 1 C min-1
equipment was carefully calibrated for each heating rate,
including temperature, enthalpy, and onset slope (thermal measured an DH of 40 J g-1, this event should be associ-
resistance). Pure helium gas (99.999 %) with a ated with the first-order transition from low to superpro-
60 mL min-1 flux was used for all these experiments. tonic conducting phase, discovered by Haile et al. [9, 30].
Conducting silver paint on both faces of a flat crystal was This transition is a consequence of a simultaneous struc-
used as electrodes for the complex impedance measurements, tural phase transition from a monoclinic to a cubic (CsCl
which were performed by means of a Solartron 1260A ana- type) symmetry [9]. The latter is isostructural with the
lyzer, using an average heating rate of 0.2 C min-1. The superprotonic conducting phase of CDP [31, 32]. Checking
dc conductivity was calculated from the frequency depen- the DSC baseline, the small thermal response at the peak
dence of the real part of the complex conductivity following a temperature of 103.1 C could be interpreted as part of the
well-established procedure [28, 29]. earlier thermal behavior of the endotherm. A similar
The temperature evolution of X-ray powder diffraction observation was reported, but at 110 C [30]. On cooling,
patterns at different isotherms (from 25 to 170 C) was the curve shows a flat, broad exothermal peak with onset
taken by using a Bruker-AXS D8 Advance X-ray diffrac- and peak temperature values of 106.6 and 71.4 C,
tometer equipped with an Anton Paar HTK-1200 high- respectively. This sample thermal response could be related
temperature furnace. These data were obtained by using to the observation that the cubic superprotonic conducting
CuKa1 radiation (k = 1.5406 A) with a 2h scan step of phase remains down to 87 C on the cooling cycle [33].
0.02 every 5 s. A scanning electron microscope (FEI However, it should be noted that the DH value of the
XL30 ESEM) was used to obtain surface sample images at cooling exothermic peak does not correspond to that of the
different temperatures (from 25 to 143 C) under 5-mbar heating endothermic one; it is about 40 % less. So that,
pressure. according to this measurement, the reported first-order
phase transition is not of the reversible type. Additionally,
from literature data, a wide dispersion of the Tsp value (into
Results and discussion a 26 C temperature range) is found: 119 [9], 131 [34], 135
[31], 143 [9], and 145 C [30]. This behavior is not
A DSC heatingcooling cycle on a Cs3(HSO4)2(H2PO4) exclusive to this CHSCDP solid solution salt. Although in
crystal, using 1 C min-1 as heating rate, is plotted in lesser proportion, it is also present in its parents; while the
Fig. 1. The upper temperature limit was imposed by the Tsp value on CDP spreads from 126 [35] to 231 C [32];
reported value of 150 C, where the salt starts to decom- CHS does the same, but from 140 [36] to 147 C [37].
pose [30]. According to the fundamental thermodynamic first-order
On heating, an endothermic anomaly with 127.5 and phase transition theory, when a system is heated (under
135.2 C, as respective onset and peak temperature values, standard pressure conditions) through its respective tem-
was registered. The corresponding enthalpy change value, perature phase transition, it must correspond to a constant
DH, was 44.0 J g-1. Comparing this result with that by value independent of the heating rate used. However, it is
Suzuki and Hayashi [31], who reported an endothermic not the case for any of these salts. If instead, the DSC peak
DSC peak at 135 C, and that by Haile et al. [9] who enthalpy is related to a partial thermal decomposition

123
E. Ortiz et al.

(polymerization) at reaction sites on the surface of the 100.0


70 90 110 130 150 170 190
crystal, the wide literature dispersion on the reported Tsp
values of Cs3(HSO4)2(H2PO4) and its parents could be well (a) TG
explained. When a sample with a higher surface quality 0.25 %
99.8
(less concentration of surface structural defects) is heated 150.0 C

Mass/%
with a higher heating rate (less time for the thermal kinetic 150 C
reaction process), the onset decomposition temperature is 100
99.6
expected to be higher. Therefore, we propose that the DSC 5.2 %

Mass/%
95
peak of Cs3(HSO4)2(H2PO4) corresponds to a chemical 373 C 16.6 %
transformation instead of a physical one. As indicated in 90 650 C
99.4
the introduction, a similar hypothesis was suggested for its 85
parents, concluding that instead of a physical transforma- 75 150 225 300 375 450 525 600
Temperature/C
tion, a surface chemical decomposition takes place at
230 C for CDP [23, 26, 38, 39], and 141 C for CHS [18], 135.4 C
0.5
according to Eqs. (1) and (2), respectively. Given that (b) DSC 2
Cs3(HSO4)2(H2PO4) is a solid CHS and CDP solution, it is

Endothermic

DTG/(102)% C1
likely that its thermal decomposition products involve the

Heat flow/mW
1.0
same as its parents. If the proposed hypothesis was true, a 170.0 C
mass-loss step, related to the evolution of water vapor as a 113.5 C
1
product of decomposition, should be detected in the same 1.5
temperature range where the DSC peak is registered.
The temperature dependence of the mass loss for the DTG
wide temperature range from 35 to 650 C, using 2.0
135.2 C 0
1 C min-1 as heating rate, is shown in the insert in
Fig. 2a. At first glance, the TG curve might be described as 70 90 110 130 150 170 190

the sequence of two big mass-loss steps with subtotal Temperature/C

values of 5.2 and 16.6 % at around 373 and 650 C, Fig. 2 The insert shows the TG curve of a Cs3(HSO4)2(H2PO4)
respectively. However, careful examination allows seeing a crystal, under the wide temperature range from 35 to 650 C.
small additional earlier temperature mass-loss step below a amplified 70190 C temperature range marked with the dashed
150 C. In order to increase the graphic resolution, the ellipse, in the insert. b DSC curve of another dry Cs3(HSO4)2(H2PO4)
crystal overlapped with the derivative curve, DTG, of a. The heating
small area marked with a dashed ellipse has been amplified rate of these experiments was fixed at 1 C min-1
in Fig. 2a. Now, a clear mass-loss step of 0.25 % at 150 C
is evident. Also, the DSC curve of another crystal sample
using the same heating rate of the TG experiment temperature where the DSC and DTG signals leave their
(1 C min-1) is plotted in Fig. 2b. Because the temperature respective baseline, it can be noted that decomposition
scales for these data plots (Fig. 2a, b) are the same, it is begins at around 113 C. If it is assumed that the whole
clear that the DSC endothermic peak temperature at constitutional water is thermally removed from the salt, we
135.2 C fits almost exactly with the peak temperature propose the following chemical reaction:
(135.4 C) of the temperature derivative of the thermo-
Cs3 HSO4 2 H2 PO4 s ! Cs2 S2 O7 s CsPO3 s
gravimetric (DTG) curve, meaning that the DSC
2H2 Ov 3
endothermic peak is directly associated with the mass-loss
step. where s and v in parentheses have the same meaning as in
To confirm that the DSC and TG events occur simul- Eq. (1). The respective theoretical sample mass loss cor-
taneously, we measured these two signal responses from responds to 5.22 %. This value agrees with the first major
the same crystal by using simultaneous DSCTG equip- mass loss of 5.2 % at 373 C. This proposed decomposi-
ment, using the same heating rate used in Fig. 2. The exact tion is consistent with the fact than CDP decomposes to
coincidence of their respective DSC and DTG temperature CsPO3 at around 380 C [40] and CHS to Cs2S2O7 at
peak values of 135.0 and 135.1 C confirms the direct around 400 C, when it is heated using a low heating rate
relation between these signals. Given that the results [41], as the one used here for Cs3(HSO4)2(H2PO4). If the
exhibit very similar data as that shown in Fig. 2, we sample is heated to higher temperature values (above
decided not to present it. These results show that decom- 447 C), the cesium pyrosulfate decomposes forming
position starts well below the temperature reported by cesium sulfate (Cs2SO4) and sulfur trioxide (SO3) [41]. The
Haile et al. [30] (150 C). By comparing the starting whole reaction requires a total mass loss of 16.84 %. This

123
On the low- to high proton-conducting transformation of a CsHSO4CsH2PO4 solid solution and its

value also agrees with our measured value of 16.6 % at Note that this temperature is higher than the respective value
650 C and with that reported by Haile et al. [9] (16.5 %). for the case when lower heating rate is used, see Fig. 2b.
However, because our interest is related to the small mass- In order to check the effect of the heating rate and the
loss step, we focused on the low temperature range, samples morphology, we compared the DSC and TG ther-
70190 C. mal response of crystalline powder samples (as grown)
To confirm whether water corresponds to a decomposition heated at 0.2 C min-1 (Fig. 4) with those crystal samples
product, a simultaneous TGMS experiment was performed heated at 2 C min-1 of Fig. 3. To assist this task, Figs. 34
on a crystal by using a heating rate of 2 C min-1 (Fig. 3). have been edited by using the same temperature and mass
Overlapped with these data, Fig. 3b also shows a DSC scales. For comparison, a dc conductivity measurement of a
experiment of a crystal heated under the same experimental flat crystal, but with the same (average) heating rate of 0.2 C
conditions. The results in Fig. 3 are quite similar to those min-1 used on the experiments shown in Fig. 4 has been
shown in Fig. 2: a DTG peak (Fig. 3a) appears simultane- overlapped (Fig. 4b). While Fig. 4a presents a mass loss of
ously with the DSC anomaly, it can be noticed that both 0.80 % at 150 C, Fig. 3a exhibits 0.23 % mass loss at this
temperature peaks are very close (137.8 and 137.3 C, same temperature value. Furthermore, while Fig. 4b pre-
respectively). Hence, these measurements are interpreted in sents a 99.5 J g-1 broad endothermic anomaly, essentially
the same way as those shown in Fig. 2. Moreover, the composed of three peaks (covering the whole temperature
m/z = 18? amu MS ion current signal presents an event with range associated with the mass-loss step), Fig. 3b shows a
a temperature peak value (138.0 C) that also coincides with 44.5 J g-1 single-endothermic peak (covering around half of
the DSC and DTG temperature peaks. This confirms that the corresponding mass-loss step temperature range). The
water is one of the decomposition products obtained from a higher mass-loss step of 0.80 % (Fig. 4a) compared to the
chemical transformation whose maximal rate (at this heating 0.23 % mass loss in Fig. 3a is a consequence of the sum of
rate) is reached at the average temperature value of 137.7 C. two effects: the first sample (shown in Fig. 4) had a higher
specific surface area and more time to yield more chemical
product (during the kinetics of the reaction) than the second
70 90 110 130 150 170 190
100.0 sample (shown in Fig. 3). The same reason also explains why
TG 1.5
137.8 C 0.23 % the DH measured value is higher for the first sample: given
99.8 (a) 150.0 C that it yields larger amounts of chemical products, it requires
more energy to break up chemical bonds. Considering that
DTG/(102)% C1

99.6
1.0 the first two peaks of the DSC curve in Fig. 4b are within the
Mass/%

99.4 reported temperature range where the superprotonic phase


99.2
transition takes place, one of these could be assigned to this
0.5 physical transformation. Nevertheless, both have their
99.0 respective DTG peak at quite the same DSC peak tempera-
DTG
ture values. In addition, the last DSC peak at 171.3 C does
98.8
0.0
not have a homologous peak on the DTG curve, probably at
138.0 C 188.5 C. However, it is 17.2 C above. Because thermal
0.5
128.4 C
DSC
decomposition phenomena usually tend to occur on surfaces
H = 4.6
1.0 where heat contact is direct and chemical binding is weaker,
44.5 J g1
Heat flow/mW

rather than on the bulk crystal [26], we believe that at the very
Ion current/nA

MS
1.5 earlier decomposition stage (at the onset of the first DSC
4.5
177.5 C endodermic anomaly), a big proportion of the water pro-
2.0 +
m/z = 18 aum duced from the most external surface easily transforms to gas
and evaporates. However, when the chemical reaction front
2.5 4.4 migrates below the most external surface, part of the water
(b) 137.3 C diffuses to the surface to evaporate, but another significant
3.0
part of it, as a liquid phase, is strongly bonded to the other
70 90 110 130 150 170 190
decomposed products or even to the salt. Liquid water
Temperature/C
might dissolve parts of the salt providing protons to the
Fig. 3 Simultaneous TGMS and DSC measurements of two Thermal Surface Decomposition Products System (TSDPS)
Cs3(HSO4)2(H2PO4) crystals, respectively, in the 70190 C temper- of Cs3(HSO4)2(H2PO4). Taking into account that except for
ature range. a TG and its corresponding DTG curve. b DSC overlapped
the liquid water, its composition is not exactly known, we
with the m/z = 18? amu MS ion current signal. The heating rate was
fixed at 2 C min-1. For comparison purposes, the mass and temper- prefer to use this definition. Probably, in addition to the
ature scales of this figure are the same as those of Fig. 4 liquid, this system contains Cs2S2O7 and decomposition

123
E. Ortiz et al.

100.0
70 90 110 130 150 170 190 range, where the transition is found to occur, is also not
188.5 C consistent with the thermodynamic first-order phase transi-
TG
99.8
125.2 C 4
tion theory where the low- and high-temperature phases can
(a) coexist only at the phase transition temperature. Considering

DTG/(102)% C1
99.6 134.7 C 0.80 % that except the liquid water, all the products constituting the
Mass/%

150 C
99.4 TSDPS of Cs3(HSO4)2(H2PO4) are of polymeric nature; it
2
DTG could behave similar to a polymer electrolyte membrane
99.2 111.6 C
(located over the salt surface), where the polymers constitute
99.0 a host matrix that contain liquid water regions. Therefore, as
0 part of liquid water dissolves a superficial portion of salt
98.8 (providing protons) [18, 27], the proton transport mechanism
1
0.0 H = 99.5 J g1 of TSDPS of Cs3(HSO4)2(H2PO4) might include the vehicle
DSC
type, using H3O? as a charge carrier. If the sample is heated
(b) 2
further (Fig. 4), a second-stage chemical-surface decompo-
0.2 126.0 C
sition takes place but the DTG peak temperature presents a

Log/S cm1
Heat flow/mW

Endothermic

3
delay of 6.5 min (1.3 C) with respect to the respective DSC
171.3 C
0.4 peak temperature. A similar, but more pronounced, effect
4
133.4 C corresponds to the third-stage chemical decomposition,
111.0 C
where the corresponding DTG peak temperature is delayed
0.6 5 86 min (17.2 C) with respect to the respective DSC peak.
IS This suggests that the total additional water produced in this
125.1 C
6 event is strongly bonded to the hygroscopic products of
70 90 110 130 150 170 190
decomposition. This explains the fact that the corresponding
Temperature/C
DTG peak temperature around 188.5 C appears at a higher
Fig. 4 TG and DSC measurements of two Cs3(HSO4)2(H2PO4) temperature value where no DSC peak appears. This DTG
crystalline powder samples (as grown), respectively, in the peak could also involve the release of part of the liquid water
70190 C temperature range. a TG and its corresponding DTG formed at the first two DSC peaks. This phenomenon is also
curve. b DSC overlapped with a dc conductivity measurement of a
crystal sample. Heating rate of all these experiments was fixed at
evident in the results shown in Figs. 23 where after the DTG
0.2 C min-1 peak (associated with the endothermic DSC peak), a second
DTG peak starts to appear at 170.0 and 177.5 C, respec-
tively, without any associated endothermic DSC peak. Note,
products of CDP as oligomer transient decomposition also, that while the DSC baseline stays flat, a second MS
products to CsPO3 [23, 25] or even long chains of CsPO3 water vapor increase starts simultaneously with its respective
monomers [CsnH2PnO3n?1 polymer with n  1, (CsPO3)n], DTG signal at 177.5 C (Fig. 3b). This evidence confirms
as proposed by Park [26]. From conductivity data (Fig. 4b), it the existence of liquid water strongly bonded to the TSDPS
can be seen that conductivity increases by 300 times. How- of Cs3(HSO4)2(H2PO4). Otherwise, our proposed proton
ever, it does not correspond to an abrupt conductivity jump, conduction mechanism cannot be possible.
as that observed through the well-known first-order transi- The fact that liquid water is present at the temperature
tion from low to superionic conducting phase of AgI salt at range where the existence of the superprotonic conduction
147 C [42, 43]. Instead, through a wide temperature range phase of Cs3(HSO4)2(H2PO4) has been reported is evident
of 15 C, a gradual monotonically conductivity increase from the Suzuki and Hayashi [31] 1H solid-state NMR
from 111 to 126 C is observed. This measurement agrees studies. On heating the salt, at 127 C, two narrow signals
with that reported by Haile et al. [9], who indicated that the coexist, where one of these (with a chemical shift value at
temperature range (14 C) over which this transition occurs around 6.4 ppm) was ascribed to physically absorbed
corresponds to 111125 C. Certainly, by comparing the water. At 137 C, the two signals collapse into one sharp
temperature value where the DTG and DSC signals leave signal. From Fig. 3, it can be seen that this temperature
their respective baselines (Fig. 4a, b, respectively), both corresponds to the same value where the higher decom-
coincide with the onset temperature value where conduc- position rate of the surface of the crystal, heated to
tivity starts to jump. Similarly, the respective DSC and DTG 2 C min-1, is reached. On the other hand, a room tem-
temperature peaks of 125.1 and 125.2 C coincide with the perature 1H NMR study [35] of a wet powdered CDP
offset temperature where the conductivity stops jumping. At sample, than was pretreated isothermally at 200 C for a
this temperature, the highest first-stage chemical-surface day, shows three peaks with the main chemical values of
decomposition rate takes place. This wide temperature 14.5, 10.5, and 6.6 ppm, respectively. The first two peaks

123
On the low- to high proton-conducting transformation of a CsHSO4CsH2PO4 solid solution and its

correspond to two crystallographically distinct hydrogen 135 47


80
atoms of the salt and the last one to chemically surface- S
absorbed water, which (as noted) was particularly difficult
45
to remove. In addition, from the respective 1H NMR data

S/J mol1 K1
76

H/J g1
by Yamada et al. [33], when CDP is heated through the

Tonset/C
130
superprotonic phase transition, a single peak that appears at 43

the average chemical shift value of the first two peaks, 72


reported by Boysen et al. [35], decreased to 6.5 ppm. 41
Tonset
Therefore, we believe this peak corresponds to the 1H 125 H
NMR response of protons from the liquid water produced 68
39
as a consequence of thermal decomposition. Certainly, this
statement is consistent with Lees [44] optical observations 0 2 4 6 8 10 12 14 16
of non-equilibrium processes at the CDP sample surface, Heating rate/C min1
where a liquid-like behavior, such as water flow, takes
Fig. 5 Heating rate dependence of the onset temperature, Tonset,
place when the sample is heated above Tsp. Therefore, we enthalpy change, DH, and entropy change, DS, of the endothermic
trust that when CDP is heated through the reported peak associated with the first-order superprotonic conducting phase of
superprotonic phase transition, instead of this physical Cs3(HSO4)2(H2PO4). Five crystals were heated to 0.2, 2, 5, 10, and
transformation a chemical one takes place at its surface, 15 C min-1, respectively
following Lees [23] reaction (1). However, we must
emphasize that not all the H2O evaporates because part of
the liquid water is strongly bonded to the other polymeric plotted, a discontinuous change, DS, takes place at the
product(s) and to the salt. Hence, we propose the following respective transition phase temperature. This is equal to the
chemical reaction: corresponding DH or latent heat, Q, divided by this tem-
perature. Thereby, independent of the heating rate used,
nCsH2 PO4 s ! Csn Hn Pn O3n1 s 1  xn
DS and its respective DH should be constant. Certainly,
 1H2 Ol xn  1H2 Ov 4
because the theoretical DS calculated by Haile et al. [9]
Here, n and x are the number of molecules participating in (51.5 J mol-1 K-1) corresponds to the difference between
thermal decomposition and the fraction of the water that the configurational entropy associated with the protonic
evaporates, respectively. The letters s, l, or v in parentheses configuration at the cubic superprotonic conducting phase
denote solid, liquid, or vapor. Herein, it is important to (57.3 J mol-1 K-1) and that associated with the room
point out that the strongly bonded liquid water may affect temperature monoclinic phase (5.8 J mol-1 K-1), it does
the dry structure of the polymeric product(s). The high not depend on any particular heating rate value.
conductivity values reported above 230 C could be In this sense, an additional kinetic DSC experiment is
interpreted as proposed here for Cs3(HSO4)2(H2PO4). presented in Fig. 5. Here, the onset temperature, Tonset,
Returning to the comparison from the measurements of enthalpy, DH, and entropy changes, DS, of the thermal
Figs. 3 and 4, where the endothermic DSC anomaly on the anomaly are measured from five crystals heated to 0.2, 2, 5,
latter corresponds to an overlapping of three peaks while 10, and 15 C min-1, respectively. It can be seen that none
on the former it corresponds to only one, it is explained by of these quantities remains constant. Comparing the result
the fact that at a lower heating rate (when the reaction has associated with the extreme heating rate values of 0.2 and
10 times more time to develop), the sequence of decom- 15 C min-1, it is observed that while Tonset shifts 11.3 C
position events is better resolved. Also, the well-matched to higher temperatures from 123.5 to 134.8 C, DH de-
(125.1 and 125.2 C, respectively) first DSC and DTG creases 6.2 J g-1 (13.5 %) from 46.0 to 39.8 J g-1, and
peak temperatures of the samples heated at low heating rate DS decreases 12.7 J mol-1 K-1 (15.9 %) from 80.0 to
(0.2 C min-1) shift together to a higher temperature value 67.3 J mol-1 K-1. This experiment proves that the DSC
(12.2 C above) when the crystals are heated at a higher anomaly is governed by a chemical instead of a physical
heating rate (2 C min-1). This is the usual behavior of the transformation. As indicated above, higher heating rates
kinetic nature of a chemical transformation. Moreover, it mean less time to start the chemical bond breaking process
demonstrates that the whole enthalpy of the DSC peak (higher Tonset) and less broken chemical bonds (lower DH).
must be associated with the chemical decomposition This also explains the wide spread of the values reported in
reaction. the literature for the superprotonic conducting phase tran-
Given that the low to superprotonic conducting phase sition temperature.
transition is of the first-order type [9], when the tempera- Chisholm [45] reported an extended research of several
ture dependence on entropy thermodynamic function, S, is mixed CHSCDP compounds, and the X-ray powder

123
E. Ortiz et al.

Figure 6 indicates that the enthalpy change associated with


H = 45.9 J g1 the thermal event at around 123.5 C, as measured from
0.6 Total heat flow
the kinetic component, corresponds to 99.1 % (45.5 J g-1)
of the total DH value measured from the total heat flow
signal (45.9 J g-1). Considering that decomposition reac-
Heat flow/mW

tions are time-dependent phenomena, the results suggest


0.0
Non kinetic component H = 0.4 J g1
that 99.1 % of the total DH is associated with chemical
123.5 C
H = 45.5 J g1 processes. The remaining 0.9 % measured value for the
Kinetic component
heat capacity component (0.4 J g-1) of the very small
MDSC peak could be associated with the physical trans-
0.6 formation in the crystal from low to superprotonic con-
ducting phase. However, using our Tonset value (123.5 C),
126.2 C 135.3 C
the enthalpy change associated with this transformation is
75 90 105 120 135 150
much lower than that required by the theoretical config-
Temperature/C
urational DS calculated by Haile et al. [9],
Fig. 6 Total heat flow and its time-dependent (kinetic component) DH = 29.6 J g-1. Consequently, from our MDSC mea-
and time-independent (heat capacity component) components of a surements, the non-kinetic component of the enthalpy
sealed Cs3(HSO4)2(H2PO4) crystal under a heating cycle from 75 to change value (0.4 J g-1) cannot be related to a superprotonic
150 C. The heating rate was fixed at 0.2 C min-1
conducting phase transition of the Cs3(HSO4)2(H2PO4)
crystal. Further, Fig. 7a shows the MDSC cooling cycle of
1.2 the sealed crystal shown in Fig. 6, presented together with
0.9 H = 45.2 J g1 a cooling cycle from another unsealed crystal (Fig. 7b) that
Total heat flow
0.6
was previously heated to the same maximum temperature
Non kinetic component
H = 0.2 J g1
used in Fig. 6 (150 C). On the cooling cycle of the sealed
Heat flow/mW

0.3 (a) and unsealed samples, practically 100 % of the total heat
0.0 flux corresponds to the kinetic component. By comparing
Kinetic component H = 45.0 J g1
0.3 the DH of the total heat flux endothermal anomaly for the
heating (45.9 J g-1, Fig. 6) cycle with those cooling cycles
0.6
0.4 of the sealed (45.2 J g-1, Fig. 7a) and unsealed samples
0.3 (b) (30.1 J g-1, Fig. 7b), we may conclude that when the
Total heat flow
0.2 H = 30.1 J g1 sample is sealed it seems that the transformation is rever-
0.1
15 30 45 60 75 90 105 120 135 sible, but if it is unsealed it is not reversible (similar to the
Temperature/C results presented in Fig. 1). Given that the elastic cover
cannot alter the configurational entropy change, DS (which
Fig. 7 a Total heat flow and its time-dependent (kinetic component) only depends on the intrinsic structural properties of the
and time-independent (heat capacity component) components of a
sealed Cs3(HSO4)2(H2PO4) crystal under the cooling cycle from 150
low- and high-temperature phases), it constitutes a physical
to 5 C, immediately after the heating cycle shown in Fig. 6. b Total unrealistic phenomenon interpretation. For the sealed
heat flow of an unsealed Cs3(HSO4)2(H2PO4) crystal under a cooling sample, the total water produced on heating (as a conse-
cycle, after having been heated to 150 C. The cooling rate of these quence of the chemical reaction) is kept with the sample,
experiments was fixed at 0.2 C min-1
and when cooled, the reverse chemical reaction can take
place; thus, the DH value is the same as when the sample
diffraction patterns measured in the temperature transition was heated. But if the crystal is not sealed, a fraction of the
regions [around 30 C for Cs3(HSO4)2(H2PO4)] showed a water is evaporated, and in the cooling cycle, the liquid
reproducible mixture of the room and high-temperature water is not enough to react with the total amount of the
phases. Therefore, he concluded that the transition appears other products. Therefore, the measured DH value is lower
to be a thermodynamic, rather than kinetic, phenomenon. than when the sealed sample is cooled. Moreover, the
To examine this last statement, the thermal results for a unsealed endotherm anomaly has a different shape and
heatingcooling cycle of a sealed Cs3(HSO4)2(H2PO4) appears at lower temperatures (covering a wider tempera-
crystal, using the MDSC technique, are shown in Figs. 6 ture range) than the sealed one. When the water produced
7a, respectively. This technique allows separation of the is not enough in the unsealed sample, under a dry atmo-
time-dependent (kinetic) component from the time-inde- sphere, the inverse reaction is more sluggish than when
pendent (heat capacity) component of the thermal there is abundant liquid in the sealed one. In the unsealed
response of the sample to a modulated heat flow [46]. sample, it is, then, possible that other types of reactions

123
On the low- to high proton-conducting transformation of a CsHSO4CsH2PO4 solid solution and its

take place, which could explain why new peaks become 9

(110)
visible on cooling to 98 C, giving rise to an extremely 6

(200)
(100)

(210)
(111)
3
complicated pattern [9] on further cooling to 70 C. 170 C
Before proceeding with our XRD results, it is important to 9
6
examine previous results by Chisholm [45]. According to
3
this author, the X-ray powder diffraction patterns, some 160 C
9
Celsius degrees, above Tsp (taken at around 140 C) of
6
low-phosphate-content mixed CHSCDP compounds [like 3
150 C
Cs3(HSO4)2.5(H2PO4)0.5 or Cs3(HSO4)2.25(H2PO4)0.75]
9
consist of a combination of two phases: a tetragonal and a 6

Intensity/a.u.
cubic one. The first was associated with the well-known 3
140 C
tetragonal phase of the superprotonic high-temperature 9
phase of CHS. It appears as a consequence of a partial 6
thermal decomposition of the mixed CHSCDP system, 3 130 C
where sulfate and phosphate tetrahedral migrate into dif- 9
ferent phase domains. From our point of view, given that Tsp 6
of CHS is 141 C, this assumption is reasonable. However, it 3 120 C
is not the case for the other decomposed product: CDP. At 9
140 C, its thermodynamic stable phase is the low-temper- 6
ature monoclinic one, but no XRD peaks for this phase 3
110 C
appear. Therefore, the cubic phase observed of this CHS 9

(220) (402)

(023)(331)

(214)(114)
(330)(313)
(221)

(511)
(600)

(421)
6

(204)
(112)

(621)
CDP solid solution was associated with an isostructural

(202)
3
phase with the superprotonic cubic one of CDP [45]. If this 25 C
interpretation is correct, why does not the XRD pattern from 18 21 24 27 30 33 36 39 42
the decomposed phosphate product(s) appear? Taking into 2 /
account our interpretation that the 141 C XRD pattern of
Fig. 8 XRD patterns of a Cs3(HSO4)2(H2PO4) powdered sample at
CHS corresponds to the XRD response of the TSDPS of CHS
25, 110, 120, 130, 140, 150, 160, and 170 C, respectively. Miller
[18], the 140 C XRD pattern form the low-phosphate- indices of (intensity-relevant) reflection peaks of the room temper-
content mixed CHSCDP compounds may correspond to ature phase (25 C) has been indicated, as well as those of the 170 C
an overlapped XRD pattern from the TSDPS of CHS and pattern, associated with the isostructural superprotonic conducting
phase of CDP. The peaks for the 25 C pattern that remain on
the TSDPS of CDP. The mass-loss step registered at
subsequent high-temperature patterns and those new ones are marked
around Tsp on Cs3(HSO4)2.5(H2PO4)0.5 supports this with solid and dotted arrows, respectively
interpretation [47]. The superprotonic conducting state of
a high-phosphorous content mixed CHSCDP compound,
like Cs3(HSO4)2(H2PO4), is purely cubic [45]. From ear- increase their intensity, probably by the order increase of
lier studies on pre-heated (during 1030 min at 210 C) the atomic planes induced by the thermal atomic binding
highly SiO2 content composites of (1 x)CsHSO4 - xSiO4, relaxation. From 110 to 170 C, a general behavior is
it was concluded that an amorphous phase of CHS is formed observed. The monoclinic peaks become weaker until
[48]. We trust that the amorphous phase might correspond 170 C when they completely disappear. Certainly, the
to an amorphous phase, but of TSDPS of CHS. Then, the 
peaks at 2h = 24.4 [(220) (402)] and 27.8 (600) persist
fact that only the cubic phase of a high-phosphorous content even until 160 C. New peaks appear (all those of the
mixed CHSCDP compound is present could be explained 170 C pattern, which have been associated with the
by assuming that TSDPS of CDP plays a similar role as isostructural phase of the superprotonic cubic phase of
SiO2 does on TSDPS of CHS. CDP [45]) with a general temperature intensity increase.
Figure 8 shows high-temperature XRD patterns of The 2h = 31.5 (111) peak is present from the early tem-
powdered crystalline Cs3(HSO4)2(H2PO4) at several iso- perature of 120 C. The most evident pattern that shows the
therms from 25 to 170 C, under stagnant ambient atmo- phase coexistence of the monoclinic room temperature and
sphere. The 25 C diffraction pattern corresponds to that that responsible for the new peaks phase(s) is that of
reported (ICSD 079,789) for a room temperature mono- 150 C. Even if we hypothetically assume that the low- to
clinic symmetry with space group P21/n. No other phase high proton (superprotonic)-conducting phase transition is
was detected. According to this reported data, Miller time-dependent, the very low average heating rate used in
indices of (intensity-relevant) reflection peaks of this phase this experiment (0.17 C min-1) does not offer the
were assigned. By heating from 25 to 110 C, some peaks opportunity for a phenomenon, like superheating. It should

123
E. Ortiz et al.

be noted that the higher reported Tsp (145 C, [30]) is the cubic cell constant, a, for CDP from 230 to 290 C.
15 C below 160 C; however, at this last high tempera- According to these data, the thermal expansion coefficient
ture, the sample still contains a room temperature mono- of the cubic phase of CDP [Da/ao = a(T - To) with ao =
clinic phase. By heating at such a low heating rate, how is 4.9724(2) and To = 260 C] was a = 3.92 9 10-5 K-1.
it possible to have a low-temperature phase 15 C above Although we know the superprotonic cubic phase of CDP
the higher reported first-order phase transition tempera- is thermodynamically not stable below the Tsp, using these
ture? This constitutes one of the most robust evidence that data, the ao(T) linear relation has been extrapolated to room
Cs3(HSO4)2(H2PO4) undergoes a chemical instead of a temperature. Comparing, in their order, the CDP a values
physical transformation. If we compare this result with the reported by Ikeda et al. [49, 52], Bronowska [32], and
TG curve of Fig. 4a, which shows the mass loss for a Yamada [33] (marked with a rectangle) at around the same
powder crystalline sample heated at similar heating rate temperature of 240 C, a progressive reduction of the
(0.2 C min-1) used in this experiment, at 130, 140, 150, a value is clearly noted. Because all powdered samples are
160, and 170 C, the sample contains 7.6, 13.4, 15.3, 17.8, pure CDP and are exposed to quite the same atmospheric
and 19.7 % of thermal decomposition products. Here, we pressure (the XRD chamber used by Ikeda et al. [49, 52] is
have assumed a total water mass loss of 5.2 %. However, if equipped with a check valve to limit the system pressure to
we consider that after thermal decomposition the sample 2 atm) the a value should be the same, the bar error of this
maintains 60 % of the water produced, a rough estimation data is smaller than the symbol size. The unique different
of the amount of decomposed product implies 2.5 times parameter between these experiments was the water partial
these values. Therefore, at 170 C, where the respective pressure. While Ikeda et al. [49, 52] exposed the sample to
pattern is composed entirely by the new XRD peaks, about a higher water partial pressure around 1 atm, Bronowska
50 % of the sample corresponds to decomposed products [32] exposed it to H2O-saturated atmosphere and Yamada
from which a third part belongs to TSDPS of CDP and the [33] made the experiment under the usual non-humidified
other two-thirds to the amorphous TSDPS of CHS. The atmosphere. From these results, it seems that the cubic cell
remaining 50 % of the sample belongs to the salt, which is constant, a (and therefore the volume cell value), rises with
not detected because all the X-ray energy is absorbed by increase in water partial pressure. This behavior is more
the decomposed product located at the sample surface. akin to that of a crystalline phase of a polymeric material
According to Ikeda [49], the Tsp of CDP, measured form than of an ionic solid compound.
the DSC onset temperature, occurs at 227 C. However, Considering that cell constant values for the CHSCDP
from his temperature evolution of XDR patterns, the strong mix compounds (Fig. 9) were measured under a similar
low monoclinic peaks kept their intensity until phase atmosphere as used by Yamada [33] for CDP, we used the
transition started at 232 C. If the sample if heated further, same ao(T) linear relation (the same slope) obtained by
these peaks decrease, but even at 238 C, they do not Ikeda et al. [49, 52] in their CDP experiment under high
disappear. Otomo et al. [50] have observed a similar but
more pronounced behavior. From their XRD data, they
conclude that the monoclinic phase of CDP was still pre- 4.98 Ikeda et al. [44, 47]
(water partial pressure ~ 1 atm)
sent at 243 C. This means that the low-temperature phase 237 C
4.97
is present even 12 C above the higher reported Tsp Bronowska [28]
(saturated water atmosphere)
CsH2PO4

(231 C [32]) for the superprotonic first-order phase tran- 4.96 140 C
Cell constant a/

Cs5(HSO4)3(H2PO4)2 242 C
sition of CDP. As for the Cs3(HSO4)2(H2PO4) case, this 4.95 130 C Yamada [29]
Cs3(HSO4)2.50(H2PO4)0.50
also constitutes one of the most overwhelming evidence 170 C
4.94 This work
that CDP undergoes a chemical transformation instead of a Cs3(HSO4)2(H2PO4)
physical transformation that leads to high proton conduc- 4.93

tivity. A similar behavior has also been registered on CHS 4.92 140 C
[18] and CHSe [27], where their respective (012) and (021) 119 C Cs3(HSO4)2(H2PO4)
4.91 50 C Cs2(HSO4)(H2PO4)
peaks persist 13 and 7 C above their corresponding higher Cs2(HSO4)(H2PO4)
reported Tsp of 147 C [37] and 128 C [51]. 4.90
0 50 100 150 200 250 300
Ikeda et al. [49, 52] built a high-temperature sealed
Temperature/C
XRD chamber with 150 C tempered walls to prevent
water condensation. Seeking to avoid the thermal decom- Fig. 9 Temperature and humidity dependence of the cell constant
position of a powdered CDP sample, a drop of water was value, a, associated with the cubic superprotonic conducting phase of
CDP [28, 29, 43, 46] and its comparison with the, respective, a values
placed inside the chamber, so, under a heating cycle above
of the high-temperature cubic phase of some CHSCDP solid
100 C, the water evaporates providing humidity. Figure 9 solutions [40]. The a value calculated from our 170 C XRD pattern,
shows the Ikeda et al. [49, 52] temperature dependence of of a Cs3(HSO4)2(H2PO4) powdered sample, is also indicated

123
On the low- to high proton-conducting transformation of a CsHSO4CsH2PO4 solid solution and its

Fig. 10 Temperature evolution


of SEM Cs3(HSO4)2(H2PO4)
surface images from 25 to
143 C, under 5-mbar pressure

humidified atmosphere, but subtracting the required dif- matrix with very high water content. At 143 C, this system
ference to match the Yamada [33] a value. As noted, the is in the course of drying.
indicated cell constant values of the mixed CHSCDP solid The thermal decomposition process observed from our
solutions agree with the new ao(T) linear relation, even SEM result is analogous with that described for KH2PO4,
with that calculated by our XRD pattern at 170 C for KDP (and CDP) [28]. When this salt is heated stepwise, to
Cs3(HSO4)2(H2PO4). Supported on this observation and the higher temperatures, a solid crust of high-molecular
extreme high coincidence between all the high-temperature polyphosphate forms around the crystallites, which breaks
XRD patterns (above their reported Tsp values) of the dif- down if the internal pressure of water produced becomes
ferent mix CHSCDP solid solutions (Fig. 4.3 of Ref. sufficiently high. Moreover, our microscopy images are
[45]), we propose that all the cubic high-temperature also consistent with Lees [44] optical observation (at
phases from all these samples correspond to the same atmospheric pressure) of the crystal surface of several solid
system: the TSDPS of CDP. acids, including CDP. By heating through their respective
Finally, Fig. 10 shows the temperature evolution of SEM transformation temperatures, the scene that comprises
Cs3(HSO4)2(H2PO4) surface images from 25 to 143 C. various kinds of patterns, such as cracks, pores, sheaves,
Even though the images were gotten from an under-atmo- spherulites, and flow of liquid (water), was compared to the
spheric pressure, it helps to understand the thermal evolution corresponding geological phenomena that involve earth-
behavior of the sample surface. At room temperature, the quakes and volcanos in eruption with lava flow production.
image shows a uniform surface containing some powder
particles. At 101 C, this seems to decompose and it is
observed that small white bubbles are formed, probably Summary and conclusions
from polymeric material. At 107 C, the image shows
intense surface chemical decomposition where entities like Under a heating cycle of the CHSCDP solid solution
volcanos seem to eject a liquid. A less amplified image, at Cs3(HSO4)2(H2PO4), a clear mass-loss step was simultane-
120 C, shows a smooth surface rounded by some volcano ously observed with the DSC endothermic peak, which has
craters. The imagen at 126 C shows a bigger portion of the been associated with the low to superprotonic conducting
smutted area. The image at 143 C shows a not so smooth phase transition. When heating rate increases, the DSC and
surface, but looks like a membrane covering the salt. DTG peaks move together to higher temperature values.
According to our research presented here, the liquid ejected Another DSC kinetic experiment shows that when heating
from volcanos at 107 C corresponds to the water. Proba- rate increases, the Tonset (associated with the superprotonic
bly, the smoothed area at 120 and 126 C is a polymeric conducting phase transition temperature, Tsp) increases but

123
E. Ortiz et al.

entropy and enthalpy changes decrease. These experiments Therefore, the high-temperature cubic phase observed
explain the broad literature Tsp value dispersion when CDP is heated above 231 C does not correspond to a
(119145 C). The MDSC experiment verified that practi- superprotonic conducting phase of CDP.
cally, the total enthalpy of the endothermic anomaly is of
kinetic nature. The simultaneous TGMS signals evidenced Acknowledgements We are grateful to Professor Bengt-Erik Mel-
lander (Department of Applied Physics, Chalmers University of
that water is a chemical product of the surface thermal Technology, Gothenburg-Sweden) for the useful discussions about the
chemical decomposition of the salt and (compared to DSC design of the project that generated, as a product, this contribution.
curves) that part of it is strongly bonded to other polymeric
products. The temperature evolution of SEM images con-
firms the chemical reaction at the surface of the sample,
including the production of water and a polymeric material.
All these experiments suggest that the observed high proton References
conduction, above Tsp, is a consequence of the chemical
transformation of Cs3(HSO4)2(H2PO4). Probably, the 1. Wang Y, Chen KS, Mishler J, Cho SC, Adroher XC. A review of
polymer electrolyte membrane fuel cells: technology, applica-
TSDPS of Cs3(HSO4)2(H2PO4) behaves similar to a poly- tions, and needs on fundamental research. Appl Energy.
mer electrolyte membrane (located over the salt surface), 2011;88:9811007.
where the polymers constitute a host matrix that contains 2. Haile SM, Boysen DA, Chisholm CRI, Merle RB. Solid acids as
liquid water regions. Therefore, as part of liquid water fuel cell electrolytes. Nature. 2001;410:9103.
3. Norby T. The promise of protonics. Nature. 2001;410:8778.
dissolves a superficial portion of salt (providing protons) 4. Boysen DA, Uda T, Chisholm CRI, Haile SM. High-performance
[18, 27], the proton transport mechanism of this system solid acid fuel cells through humidity stabilization. Science.
might include the vehicle type, using H3O? as a charge 2004;303:6870.
carrier. 5. Baranov AI, Fedoshuk RM, Schagina NM, Shuvalov LA.
Structural phase transitions to the state with anomalously high-
The highest reported Tsp value, where the monoclinic ionic conductivity in some ferroelectric and ferroelastic crystals
room temperature phase of Cs3(HSO4)2(H2PO4) transforms of the bisulfate group. Ferroelectr Lett. 1984;2(1):258.
to the superprotonic cubic phase, is 145 C [30]. Because 6. Baranov AI, Khiznichenko VP, Sandler VA, Shuvalov LA. Fre-
this phase transition is of the first-order type, the only quency dielectric dispersion in the ferroelectric and superionic
phases of CsH2PO4. Ferroelectrics. 1988;81:1836.
temperature where the low- and high-temperature phases 7. Haile SM, Chisholm CRI, Sasaki K, Boysen DA, Uda T. Solid
may coexist is Tsp. According to our temperature evolution acid proton conductors: from laboratory curiosities to fuel cell
of the XRD patterns (using a very low heating rate average electrolytes. Faraday Discuss. 2007;134:1739.
of 0.17 C min-1), the sample still contains room tem- 8. Uda T, Boysen DA, Haile SM. Thermodynamic, thermome-
chanical, and electrochemical evaluation of CsHSO4. Solid State
perature monoclinic phase 15 C above the highest repor- Ion. 2005;176:12733.
ted superprotonic temperature phase transition. From the 9. Haile SM, Lentz G, Kreuer KD, Maier J. Superprotonic con-
thermodynamic point of view, this probably constitutes the ductivity in Cs3(HSO4)2(H2PO4). Solid State Ion. 1995;77:128
most robust evidence that Cs3(HSO4)2(H2PO4) undergoes a 34.
10. Baranov AI, Merinov BV, Tregubchenko AV, Khiznichenko VP,
chemical transformation that leads to high proton conduc- Shuvalov LA, Schagina NM. Fast proton transport in crystals
tion instead of being due to a superprotonic physical with a dynamically disordered hydrogen bond network. Solid
transformation (first-order structural phase transition). State Ion. 1989;36:27982.
By using this same behavior, interpretation for the 11. Nirsha B, Gudinitsa EN, Fakeev AA, Efremov VA, Zhanadov
BV, Olikova VA. Study of the thermal dehydration of cesium
Cs3(HSO4)2(H2PO4), the fact that the low-temperature dihydrogen phosphate. Russ J Inorg Chem. 1982;27:13669.
monoclinic phase of CDP was still 12 C above the highest 12. Nelmes RJ, Choudhary RNP. Structural studies of the monoclinic
reported Tsp for the superprotonic first-order phase transi- dihydrogen phosphates: a neutron-diffraction study of paraelec-
tion of CDP [50], permits us to propose this as the most tric CsH2PO4. Solid State Commun. 1978;26:8236.
13. Osterheld RK, Markowitz MM. Polymerization and depolymer-
overwhelming evidence that CDP undergoes a chemical ization phenomena in phosphatemetaphosphate systems at
instead of a physical transformation. Certainly, taking into higher temperatures. IV. Condensation reactions of alkali metal
account our early results [38, 39] that CDP presents a clear hydrogen phosphates. J Phys Chem. 1956;60:8637.
mass-loss step when heated above 230 C, and the fact 14. Litaiem H, Garcia-Granda S, Ktari L, Dammak M. The structural
behaviour before the ionicprotonic superconduction phase
that the high-temperature cubic phase of our XRD mea- transition and thermal properties in the caesium sulphate arsenate
surement of Cs3(HSO4)2(H2PO4) at 170 C is, essentially, tellurate compound. J Therm Anal Calorim. 2016;123:391400.
identical to the high-temperature cubic phase of CDP (by 15. Hosseini S, Mohamad AB, Kadhum AH, Wan Daud WR. Ther-
Yamada [33]) at 242 C (except for the thermal expansion mal analysis of CsH2PO4 nanoparticles using surfactants CTAB
and F-68. J Therm Anal Calorim. 2010;99:197202.
effect), we conclude that the XRD signal responses of CDP 16. Goni-Urtiaga A, Presvytes D, Scott K. Solid acids as electrolyte
and Cs3(HSO4)2(H2PO4) at 242 and 170 C, respectively, materials for proton exchange membrane (PEM) electrolysis:
correspond to the same system: the TSDPS of CDP. review. Int J Hydrog Energy. 2012;37:335872.

123
On the low- to high proton-conducting transformation of a CsHSO4CsH2PO4 solid solution and its

17. Paschos O, Kunze J, Stimming U, Maglia F. A review on phos- 37. Lim AR, Chang JH, Kim HJ, Park HM. Phase transition and
phate based, solid state, protonic conductors for intermediate ferroelastic property studied by using the 133Cs nuclear magnetic
temperature fuel cells. J Phys-Condens Matter. 2011;23:234110. resonance in a CsHSO4 single crystal. Solid State Commun.
18. Ortiz E, Vargas RA, Mellander BE. Phase behaviour of the solid 2004;129:1237.
proton conductor CsHSO4. J Phys-Condens Matter. 2006;18:9561. 38. Ortiz E, Vargas RA, Mellander BE. On the high-temperature
19. Baranov AI, Shuvalov LA, Schagina NM. Superion conductivity phase transitions of CsH2PO4: a polymorphic transition? A
and phase transitions in CsHSO4 and CsHSeO4. JETP Lett. transition to a superprotonic conducting phase? J Chem Phys.
1982;36:45962. 1999;110:484753.
20. Pham-Thi M, Colomban P, Novak A, Blinc R. Phase transitions 39. Ortiz E, Vargas R, Mellander BE. On the high-temperature phase
in superionic protonic conductors CsHSO4 and CsHSeO4. Solid transitions of some KDP-family compounds: A structural phase
State Commun. 1985;55:26570. transition? A transition to a bulk-high proton conducting phase?
21. Belushkin AV, Adams MA, Hull S, Shuvalov LA. P-T phase Solid State Ion. 1999;125:17785.
diagram of CsHSO4. Neutron scattering study of structure and 40. Merle RB, Chisholm CR, Boysen DA, Haile SM. Instability of
dynamics. Solid State Ion. 1995;77:916. sulfate and selenate solid acids in fuel cell environments. Energy
22. Ke X, Tanaka I. Atomistic mechanism of proton conduction in Fuels. 2003;17:2105.
solid CsHSO4 by a first-principles study. Phys Rev B. 41. Fukami T, Tahara S, Nakasone K. Thermal properties and
2004;69(16):165114. structures of CsHSO4 and CsDSO4 crystals. Int Res J Pure Appl
23. Lee KS. Hidden nature of the high-temperature phase transitions Chem. 2014;4(6):62137.
in crystals of KH2PO4-Type: Is it a physical change? J Phys 42. Funke K. Solid state ionics: from Michael Faraday to green
Chem Solids. 1996;57:33342. energythe European dimension. Sci Technol Adv Mater.
24. Thilo E. Condensed phosphates and arsenates. In: Emeleus HJ, 2013;14:043502.
Sharpe AG, editors. Advances in inorganic chemistry and radio- 43. Hull S. Superionics: crystal structures and conduction processes.
chemistry, vol. 4. New York: Academic Press; 1962. p. 175. Rep Prog Phys. 2004;67:1233.
25. Taninouchi YK, Hatada N, Uda T, Awakura Y. Phase relation- 44. Lee KS. Surface transformation of hydrogen-bonded crystals at
ship of CsH2PO4CsPO3 system and electrical properties of high-temperatures and topochemical nature. Ferroelectrics.
CsPO3. J Electrochem Soc. 2009;156:B5729. 2002;268:36971.
26. Park JH. Possible origin of the proton conduction mechanism of 45. Chisholm CRI. Superprotonic phase transitions in solid acids:
CsH2PO4 crystals at high temperatures. Phys Rev B. parameters affecting the presence and stability of superprotonic
2004;69(5):054104. transitions in the MHnXO4 family of compounds (X = S, Se, P,
27. Ortiz E, Trochez JC, Vargas RA. Phase behaviour of the solid proton As; M = Li, Na, K, NH4, Rb, Cs). PhD thesis California Institute
conductor CsHSeO4. J Phys-Condens Matter. 2008;20:365218. of Technology. 2002. http://resolver.caltech.edu/CaltechETD:
28. Leon C, Luca ML, Santamara J, Sanchez-Quesada F. Universal etd-01292003-150309. Accessed 2 Jan 2016.
scaling of the conductivity relaxation in crystalline ionic con- 46. Reading M, Luget A, Wilson R. Modulated differential scanning
ductors. Phys Rev B. 1988;57(1):414. calorimetry. Thermochim Acta. 1994;238:295307.
29. Daz-Guillen MR, Moreno KJ, Daz-Guillen JA, Fuentes AF, 47. Ortiz E, Vargas RA, Trochez JC, Bornacelli J, Nunez H. On the
Ngai KL, Garcia-Barriocanal J, Santamaria J, Leon C. Cation size novel superprotonic conductor material b -Cs3(HSO4)2[-
effects in oxygen ion dynamics of highly disordered pyrochlore- H2-x(P1-x, Sx)O4)] (x * 0.5): Does it behave as a solid phase?
type ionic conductors. Phys Rev B. 2008;78(10):104304. Phys Status Solidi C. 2007;4:40704.
30. Haile SM, Kreuer KD, Maier J. Structure of Cs3(HSO4)2(- 48. Ponomareva VG, Uvarov NF, Lavrova GV, Hairetdinov EF.
H2PO4)a new compound with a superprotonic phase transition. Composite protonic solid electrolytes in the CsHSO4-SiO2 sys-
Acta Crystallogr B. 1995;51:6807. tem. Solid State Ion. 1996;90:1616.
31. Suzuki KI, Hayashi S. Proton dynamics in Cs3(HSO4)2(HPO4) 49. Ikeda A. Superprotonic solid acids: thermochemistry, structure,
studied by 1H NMR. Solid State Ion. 2006;177:287380. and conductivity. Ph.D. thesis California Institute of Technology.
32. Bronowska W. Comment on Does the structural superionic 2013. http://resolver.caltech.edu/CaltechTHESIS:09142012-
phase transition at 231 C in CsH2PO4 really not exist?. Chem 115522353. Accessed 2 Jan 2016.
Phys. 2001;114(1):6112. 50. Otomo J, Minagawa N, Wen CJ, Eguchi K, Takahashi H. Pro-
33. Yamada K, Sagara T, Yamane Y, Ohki H, Okuda T. Superpro- tonic conduction of CsH2PO4 and its composite with silica in dry
tonic conductor CsH2PO4 studied by 1H, 31P NMR and X-ray and humid atmospheres. Solid State Ion. 2003;156:35769.
diffraction. Solid State Ion. 2004;175:55762. 51. Yokota S. Ferrroelastic phase transition of CsHSeO4. J Phys Soc
34. Yamane Y, Yamada K, Inoue K. Superprotonic solid solutions Jpn. 1982;51:188491.
between CsHSO4 and CsH2PO4. Solid State Ion. 2008;179:4838. 52. Ikeda A, Kitchaev DA, Haile SM. Phase behavior and super-
35. Boysen DA, Haile SM, Liu H, Secco RA. High-temperature protonic conductivity in the Cs1-xRbxH2PO4 and Cs1-xKxH2PO4
behavior of CsH2PO4 under both ambient and high pressure systems. J. Mater Chem A. 2014;2(1):20414.
conditions. Chem Mater. 2003;15:72736.
36. Sinitsyn VV, Ponyatovski EG, Baranov AI, Tregubchenko AV,
Shuvalov LA. Proton-conductivity anisotropy in CsHSO4, and
CsDSO4 crystals and its response to hydrostatic pressure. Sov
Phys JETP. 1991;73:38693.

123

You might also like