Ionics

DOI 10.1007/s11581-015-1426-y

ORIGINAL PAPER

Conductivity of phosphoric acid: an in situ comparative study

of proton in phosphoric acid fuel cell
T. Paul 1,3 & D. Banerjee 1 & K. Kargupta 2

Received: 30 December 2014 / Revised: 26 February 2015 / Accepted: 17 March 2015

# Springer-Verlag Berlin Heidelberg 2015

Abstract Phosphoric acid fuel cell (PAFC) unit has been con- Introduction
structed to obtain the concentration-dependent proton conduc-
tivity at of aqueous phosphoric acid at different cell tempera- Concerns over the conventional fuels and techniques have
tures and fixed humidifier temperature. The proton conductiv- been stated in context of new energy technology. Batteries
ity of the acid has also been measured at different temperatures and fuel cells are more promising as clean and renewable
using an electrolysis cell. The experimental results obtained energy sources. Studies of electrolyte and cathode materials
from different cells have been compared with the conductivity for high energy density batteries and electrochemical devices
obtained from the theoretical simulation. The simulated pro- are very helpful [1, 2]. Being an electrochemical device, fuel
ton conductivity values at different cell temperatures matched cell converts chemical energy to electrical energy directly [3].
very well with that obtained from electrolysis cell. However, The studies on proton conductors as an electrolyte material for
the proton conductivities obtained from PAFC unit are found fuel cell has surged recently [4–6]. Phosphoric acid fuel cells
to be three orders less than those of the simulated values which (PAFCs) with proton conductors are advantageous compared
may be due to transport limitations. Geometry optimization to conventional electric power generators because of its high
and potential energy surface scans using ab initio calculations fuel conversion efficiency [7]. Also, a PAFC unit can maxi-
indicate that the proton transfer in the cell can be enhanced by mize its output power up to 80 % by considering the thermo-
the dilution of the phosphoric acid which significantly im- dynamic ideality within the normal operating conditions [8].
proves the generation of hydrogen bonds. The low vapour pressure of phosphoric acid [9] helps the
electrolytic behaviour of phosphoric acid and also evaluates
the evaporation of phosphoric acid and dissipation with the H2
Keywords Phosphoric acid fuel cell . Proton transfer . Proton gas. More specifically, the cell performance is very much de-
conductivity . Geometry optimization pendent on the electrolyte-retaining matrix layer [10] and ion
conduction through the matrix. The best matrix material for
PA F C i s s i l i c o n c a r b i d e ( S i C ) w i t h a b i n d e r
polytetrafluroethylene (PTFE) content of about 3–13 wt%,
as the PTFE binder prevents the movement of the SiC parti-
* T. Paul cles inside the PAFC [10]. The orthophosphoric acid can be
[email protected] taken as a binary mixture of H2O and P2O5 at 3:1 molar ratio.
The structure of phosphoric acid confirms the self-dissociation
1
Department of Physics, Indian Institute of Engineering Science and
to H4PO4+ and H2PO4− which enables to operate the cell up to
Technology, Shibpur Howrah, India 200 °C [8, 11]. A comparison on the proton conductivity of
2
Department of Chemical Engineering, Jadavpur University, Jadavpur
phosphoric acid is reported in the concentration range 62.14–
Kolkata, India 85.58 % with a temperature range from 25 to 60 °C [12].
3
Present address: Department of Solid State Physics, Indian
Additionally, the proton conductivity study of the aqueous
Association for the Cultivation of Science, Jadavpur Kolkata 700032, phosphoric acid in the temperature range 25–200 °C indicates
India that the conduction mechanism follows the Grotthuss
 Ionics

mechanism [13, 14]. An ab initio density functional study H2 was passed through a humidifier to moisturize the gas
regarding proton dissociation and transfer inside proton ex- and then connected to the anode through the graphite grooved
change membrane is also reported [15]. Results indicate that plates. On the other hand, pure O2 was directly connected to
first proton dissociation is possible even in low level of hy- the inlet of cathode through the graphite grooved plates. The
dration, whereas second proton dissociation requires more outlets of these graphite grooved plates were connected to
water molecules with wide distribution of charge and hydro- adsorbers to adsorb the gained moisture. The cross section
gen bonding between them. Electrochemical impedance spec- of the graphite grooved plate for maximizing the gas flow is
troscopy (EIS) technique has been used as a diagnostic marker also shown as an extended figure in Fig. 1a. Two stainless
to probe the cell performance [16]. Recently, our group has steel plates were placed at two ends as shown in Fig. 1a and
evaluated the state of health in terms of drying, normal and used as current collectors. A heating plate was inserted after
dilution of a unit PAFC at different cell temperatures and insulating electrically on the lower current collector to main-
humidifier temperatures using EIS [17]. A unit PAFC indi- tain the cell temperature. The total arrangement was kept com-
cates a single PAFC operating in its normal operating condi- pact by two pusher plates. Throughout all the experiments, the
tions. In this paper, we have reported in situ, concentration- inflows of H2 and O2 were maintained at 100 and 50 cc/min,
dependent proton conductivity of aqueous phosphoric acid by respectively, as monitored by rotameter. The terminals of cath-
constructing a unit PAFC and an electrolysis cell. The varia- ode and anode were connected to impedance analyzer
tion of conductivity at different PAFC cell temperatures has (Metrohm Autolab, Model PGSTAT 302 N). Measurements
been studied also. The experimental results obtained from were performed at cell temperature from 85 to 130 °C and at
different cells have been compared and analyzed by theoreti- fixed humidifier temperature 70 °C with a fluctuation of
cal simulation and ab initio calculation using density function- ±0.1 °C and after 30 min of temperature stabilization. Ten-
al theory of phosphoric acid which has not yet been reported. millivoltz amplitude of sinusoidal excitation was applied to
obtain data in the frequency range from 1 mHz to 10 kHz.
The initial concentration of aqueous phosphoric acid was
88 % as obtained from Merck, India. The concentration of
Experimental procedure phosphoric acid was lowered by adding appropriate amount
of deionized water. Following the conductivity measurements,
Proton conductivity measurement using unit PAFC the acid concentration was determined using a pH meter to
detect the first equivalent point.
The schematic representation of unit PAFC experiment is
shown in Fig. 1a. The unit PAFC consists of anode, cathode
and electrolyte. Two electrodes were composed of thin layer Proton conductivity measurement using phosphoric acid
of platinum (20 % Pt/C) deposited on to carbon plate. A glass electrolysis cell
mat soaked in aqueous phosphoric acid was used as an elec-
trolyte. This electrolyte electrode assembly was placed be- To measure the proton conductivity directly, we have chosen
tween two graphite grooved plates as seen in Fig. 1a. Pure an electrolysis cell made of glass with two holes on the upper

Fig. 1 a Unit PAFC experimental

set-up with electrolyte electrode
assembly. A cross-sectional view
of graphite grooved plates is also
shown as an extension. b The
aqueous phosphoric acid
electrolysis cell (Colour figure
online)
Ionics

Results and discussion

Figure 2 illustrates voltage and power density variation with

different current densities for the as-fabricated unit PAFC.
Here, the cell temperature and humidifier temperature are cho-
sen as the cell performance is satisfactory [17]. The open-
circuit voltage is obtained as 0.85 V inferring the healthy
response of the glass mat. The sharp fall of the voltage at high
current density indicates the higher discharge rate of the cell.
Figure 3a shows the Nyquist plot obtained from unit PAFC.
The data in Fig. 3a have been fitted using a modified Randels-
Ershler model [25], and the fitted values are also shown in
Table 1. The pictorial representation is also given in the inset
Fig. 2 Voltage and power density variation with current density for a unit
of Fig. 3a. Here, RΩ, Rct and Ws represent the ohmic resistance,
PAFC at cell temperature 150 °C and humidifier temperature 70 °C charge transfer resistance and finite-length Warburg imped-
ance, respectively. The conventional double-layer capacitance
side as seen in Fig. 1b. Sufficient amount of aqueous phos- is replaced by a constant phase element (CPE) because the
phoric acid was taken in the electrolysis cell so that the two Pt capacitance caused by the double-layer charging is distributed
disc electrodes were dipped into the acid through the holes of along the length of the pores inside the porous electrode [26].
the electrolysis cell and the Pt disc electrodes were connected Ws signifies the diffusion of protons through the pores of the
to the two probes of the impedance analyzer using Pt wires. glass mat. In Fig. 3b, the Nyquist plot is shown as obtained
The set-up was placed on the heating plate as seen in Fig. 1b. from the electrolysis cell. The data in Fig. 3b are also fitted
The conductivity measurement was performed after 15 min of using the Randles model [25], and also, the fitted values are
temperature stabilization. written in Table 1. The pictorial representation of the Randles
model is shown in the inset of Fig. 3b. Here, only Rct and CPE
are used to fit the data. The circuit components also exhibit the
Computational method simple electrochemical reaction in the electrolysis cell. The
single semicircle for both cases represents the occurrence of
Geometry optimizations were performed using Jaguar 8.5 im- reactant reduction reaction at the electrode-electrolyte inter-
plemented in Schrödinger material suite [18]. Full optimiza- face. As Pt supported carbon substrate is used as
tions over all degrees of freedom were carried out by conjugate electrocatalyst, so the overpotential for the oxidation of pure
gradient method [19] without symmetry constraints initially hydrogen is negligible due to very facile hydrogen oxidation
invoking in Hartree-Fock theory with the 6-31G** split valence kinetics on Pt. But, the strong kinetic inhibition at cathodic
basis set [20]. The resulting structures were refined thereafter oxygen reduction reaction leads to the high overpotential as
with density functional theory employing Becke’s three- represented by these semicircles [27].
parameter hybrid functional (B3LYP) [21, 22] and the same In Fig. 4a, b, we have presented the temperature depen-
basis set. Final optimizations were performed at the B3LYP/6- dence of the proton conductivity for two different concentra-
311G** level of theory [23]. Water molecule was added indi- tions of phosphoric acid and also by comparing both experi-
vidually to the optimized H3PO4 molecule at a variety of initial mental results. Water evaporation and condensation of phos-
positions, and full optimizations were performed using the pro- phoric acid above 100 °C may be a problem in proton con-
cess described earlier. Partial charges and electrostatic potentials ductivity measurement. In our experiments, the phosphoric
were obtained according to the CHelpG scheme at the highest acid is used in its concentrated form (88 and 80 %). Thus,
level of theory [24]. Potential energy surface (PES) scans relat- the use of concentrated acid minimizes the water vapour pres-
ed to proton transfer were performed at the B3LYP/6-31G** sure, and hence, water management in the cell is not difficult
level for all the optimized structures. to achieve [8, 9]. Additionally, if there is a chance for water

Table 1 Fitted parameters using

modified Randels-Ershler model Type of experiment RΩ (Ω) (±0.01) Rct (Ω) (±1) Diffusion time (ms) Qg (Fsn−1)×10−9 n
and Randles model
Unit PAFC 0.4517 961 14.95 4.83 0.88
Electrolysis Cell – 18 – 4.53 0.78

The errors are indicated in parenthesis

 Ionics

Fig. 3 Nyquist plots obtained for

a unit PAFC experiment at cell
temperature 150 °C and
humidifier temperature 70 °C and
b phosphoric acid electrolysis cell
at 70 °C. The equivalent circuits
are also shown in the insets

evaporation, the inlets of the experimental set-up (see Fig. 1b) bonds ~2.60 Å [14]. Here, we have extended this idea by
are made in such a fashion that the tendency gets reduced. We measuring the temperature-dependent proton conductivity.
have also checked the concentration of H3PO4 after the proton Figure 4a, b suggests that the proton conductivity is indepen-
conductivity measurement at temperature 130 °C, and we dent of thermal energy but depends on the concentration of
have found no significant variation in concentration. On the phosphoric acid. So, the formation of contact ion pair and the
other hand, the proton conductivity measurement has per- build-up of Grotthuss chain both are dependent on the acid
formed readily as described in the BExperimental procedure^ concentration. It is observed in Fig. 4a, b that with decreasing
section. Consequently, the linear response of the proton con- phosphoric acid concentration, the proton conductivity in-
ductivity indicates that the protons follow the Grotthuss mech- creases. Recently, it has been reported that the self-
anism. The Grotthuss mechanism is a common parlance for dissociation of phosphoric acid is responsible for high con-
describing anomalous transport processes in aqueous acidic centration of protons [14] and this self dissociation is high in
and basic solutions [14]. But, in a pure Grotthuss transport its high aqueous solution. So, the proton conductivity in-
process, a concerted proton hopping along chains of creases with decreasing concentration of phosphoric acid.
hydrogen-bonded polar molecules is not supported for the Theoretical calculations to support this context are given
bulk solutions [14]. As a result, protons are displaced on short afterwards.
distances with its accompanying solvent. Here, the proton The proton conductivities at different concentrations of
conduction is possible due to formation of contact pair be- phosphoric acid and temperatures are given in literature [13].
tween H4PO4+ and H2PO4− and migration of the ions and, The variation of proton conductivity with concentration of
subsequently, neutralization of these ions in water solution phosphoric acid is fitted with fourth-order polynomial at a
of the phosphoric acid [14]. The main driving force behind particular temperature to find a relation between them. We
these steps is the transient response of very short hydrogen have picked 70, 90, 100, 120, 140 and 160 °C to obtain a
Ionics

Fig. 4 Temperature dependence of proton conductivity of aqueous

phosphoric acid: a 88 % and b 80 % using different measurements
(Colour figure online)

Fig. 5 Optimized (B3LYP/6-311G**) geometries of H3PO4 with

relation between the proton conductivity and concentration of different hydration levels at top panel and potential energy surfaces
phosphoric acid from Table 1 in ref. [13]. The fourth-order respective of each figure at the bottom panel: a H3PO4 + H2O, b
polynomial is used to fit the data, as at this order, the goodness H3PO4 + H2O, c H3PO4 + 2H2O and d H3PO4 + 2H2O with their
of fit increases. Later, Lagrange interpolation technique is ap- different orientations (Colour figure online)
plied to find proton conductivity at any intermediate temper-
ature. The detail of the simulation is given elsewhere [28]. The concentrations of phosphoric acid. It can be seen that the pro-
simulation data are also shown in Fig. 4a, b for both ton conductivity measured in phosphoric acid electrolysis cell

Table 2 Energetics of optimized

geometries at the B3LYP/6- n H2O Reference figure Total internal energya Zero-point energy (ZPE)b Interacting protons
311G** level
0 – −644.094486 30.349 0
1 Fig. 5a −720.487127 46.253 2
1 Fig. 5b −720.527323 46.720 2
2 Fig. 5c −796.924352 62.328 3
2 Fig. 5d −796.942402 62.580 3
3 Fig. 6a −873.364115 79.063 4
3 Fig. 6b −873.356301 79.204 4
3 Fig. 6c −873.384168 78.857 5
3 Fig. 6d −873.346523 78.752 5

Here, 0 stands for H3PO4 only

a
Total internal energy = (SCFE + ZPE + U) in hartrees
b
Zero-point energy in kilocalorie per mole
 Ionics

Table 3 Assignment of infrared active bands at different frequencies

for optimized geometries at the B3LYP/6-311G** level

n H2O Reference figure νs (cm−1) νb (cm−1) νs + νb (cm−1)

1 Fig. 5a 140.108 213.892 -

321.919
352.302
1 Fig. 5b 192.898 320.949 -
245.631 359.014
2 Fig. 5c 169.019 247.888 -
266.151 260.684
293.325
305.013
333.854
361.561
2 Fig. 5d 115.474 321.581 201.207
138.437 224.416
269.736 245.140
329.060
335.630
3 Fig. 6a 45.800 276.220 378.054
74.560 291.856
97.932 301.726
186.516 323.338
217.268
243.461
249.964
3 Fig. 6b 29.209 334.096 238.883
37.255 348.214 251.482
42.970 362.259 274.502
134.059 380.550 294.752
161.821
194.120
211.729
3 Fig. 6c 38.465 28.893 251.267
84.346 317.164 281.391
107.162 340.668
Fig. 6 Optimized (B3LYP/6-311G**) geometries of H3PO4 with 139.463 353.419
different hydration levels at top panel and potential energy surfaces 186.801 362.034
respective of each figure at the bottom panel: a H3PO4 + 3H2O, b 212.201 385.787
H3PO4 + 3H2O, c H3PO4 + 3H2O and d H3PO4 + 3H2O with their 230.211
different orientations (Colour figure online) 3 Fig. 6d 185.301 32.115 280.972
202.672 47.828 300.487
231.854 99.561 331.364
and that obtained from literature value [13] match well (same 265.171 360.419 339.840
order of magnitude), whereas there is a third-order drop in 401.799 352.501
proton conductivity which has been obtained from PAFC unit
experiment. The drop in proton conductivity may be due to
transport limitations. of charges. The hydrogen bonds are denoted by dotted lines,
In this section, we have tried to explain the fact that with and the bond lengths are also given in unit of angstrom. These
decreasing acid concentration, the proton conductivity in- charge delocalization effects are found to play critical role in
creases. As it is known that hydrogen bonding is an important first proton dissociation at low level of hydration (see bottom
feature of proton transport as well as proton conductivity, the panels of Fig. 5). The term hydration level indicates the num-
interaction of hydrogen with oxygen through bonding is the ber of water molecules added to a H3PO4 molecule to find the
main focus in these theoretical studies. In Figs. 5a–d and 6a–d, interaction in computation. The total internal energy and the
the top panel of each figure shows the optimized geometry zero-point energy of all the studied systems are given in
structures with their electrostatic potential in the bottom panel. Table 2. It has been obtained from Table 2 that with increasing
For each bottom panel, the blue regions are electron deficient hydration level (say two and three molecules of water), total
and the red regions are electron-rich. Delocalization of charge energy gets favourable but for a particular type of hydration
is visible in the bottom panels. The flexibility of molecules for level (say two or three molecules of water), the value of total
hydrogen bonding between them results in the delocalization energies is nearly same. The same energy value for a particular
Ionics

frequencies are responsible for hydrogen bonds [29]. The in-

creasing number of infrared active bands with increasing hy-
dration level in Table 3 indicates the possibility of more hy-
drogen bonds. As a result, the proton conductivity of H3PO4
increases.

Conclusions

The temperature variations of the proton conductivity of aque-

ous phosphoric acid obtained from different cells are in line
with the simulated data. However, the conductivity values
obtained from PAFC are three orders lower than those obtain-
ed from electrolysis cell, which may be attributed due to trans-
port limitations at different cell temperatures and fixed humid-
ifier temperature. The Grotthuss mechanism is found to be
suitable to explain the temperature- and concentration-
dependent proton conductivity obtained from PAFC with
highest power density of 0.05 W cm−2. The ab initio density
functional theory calculations of phosphoric acid in its differ-
ent hydration levels indicate the higher possibility of hydrogen
bond formation with increasing water molecule which leads to
greater proton transfer. The infrared active bands obtained in
lower wavelengths suggest the same conjecture on hydrogen
bond formation.

Acknowledgments T. Paul wishes to thank Mr. Raghu Rangaswamy

for the license of Schrödinger suite.
Fig. 7 Infrared spectra for all the optimized structures of a H3PO4 + H2O
associated with Fig. 5a, b H3PO4 + H2O associated with Fig. 5b, c H3PO4
+ 2H2O associated with Fig. 5c, d H3PO4 + 2H2O associated with Fig. 5d, References
e H3PO4 + 3H2O associated with Fig. 6a, f H3PO4 + 3H2O associated
with Fig. 6b, g H3PO4 + 3H2O associated with Fig. 6c and h H3PO4 +
3H2O associated with Fig. 6d 1. Steele B-C-H, Heinzel A (2001) Nature 414:345
2. Jacquens J, Farrusseng D, Georges S, Viricelle J-O, Gaudillère C,
Corbel G, Lacorre P (2010) Fuel Cells 10:433–439
hydration level indicates that the addition of water molecule is 3. Roy Choudhury S, Deshmukh M-B, Rengaswamy R (2002) J
position independent to H3PO4. The higher value of total en- Power Sources 112:137
ergy indicates the increasing possibility of proton transfer. As 4. Giddey S, Ciacchi F-T, Badwal S-P-S (2005) Ionics 11:1
5. Wanyu L, Shanwen T, Dingkun P, Guanyao M, Bin Z (1999) Ionics
a result, the proton conductivity increases with decreasing acid 5:70
concentration. Both strong- (1.58 Å) and moderate (1.68 Å)- 6. Mellander B-E, Albinsson I (1998) Ionics 4:415
type hydrogen bonds have been formed in all these types of 7. Sammes N, Bove R, Stahl K (2004) Curr Opin Solid State Mater
geometries. As stated earlier, each hydrogen bond represents a Sci 8:372–378
8. Basu S (2007) Recent trends in fuel cell science and technology.
proton transfer and the number of interacting protons is listed
Springer, New York
in Table 2. The increasing value of interacting proton further 9. Handles A-E, Nixon A-C (1956) Ind Eng Chem 48:1960–1962
confirms the increasing proton conductivity with decreasing 10. Song R-H, Dheenadayalan S, Shin D-R (2002) J Power Sources
acid concentration. 106:167–172
In this section, we have studied the infrared active bands 11. Munson R-A (1964) J Phys Chem 68:3374–3377
12. Wydeven T (1966) J Chem Eng Data 11:174–176
obtained after optimization of the structures. The infrared 13. Chin D-T, Chang H-H (1989) J Appl Electrochem 19:95–99
spectra for all the optimized geometries are also shown in 14. Vilĉiauskas L, Tuckerman M-E, Bester G, Paddison S-J, Kreuer K-
Fig. 7. The vibrational bands are categorized into three such D (2012) Nat Chem 4:461
as νs (stretching vibrational band consisting both symmetric 15. Clark J-K II, Paddison S-J (2013) Electrochim Acta 101:279
16. Yuan X, Wang H, Sun J-C, Zhang J (2007) Int J Hydrog Energy 32:
and asymmetric), νb (bending vibrational band) and νs + νb
4365–4380
(both stretching and bending vibrational band). It is clearly 17. Paul T, Seal M, Banerjee D, Ganguly S, Kargupta K, Sandilya P
observed that the peaks of the infrared active bands in lower (2014) J Fuel Cell Sci Technol 11:041001–041006
 Ionics

18. Jaguar, version 8.5 2014 Schrödinger, LLC, New York 25. Xie Z, Holdcroft S (2004) J Electroanal Chem 568:247
19. Schlegel H-B (1982) J Comput Chem 3:214 26. Schmidt T-J, Paulus U-A, Gasteiger H-A, Behm R-J (2001) J
20. Harihara P-C, Pople J-A (1973) Theor Chim Acta 28:213 Electroanal Chem 508:41
21. Becke A-D (1993) J Chem Phys 98:5648 27. Ciureanu M, Roberge R (2001) J Phys Chem B 105:3531
22. Becke A-D (1993) J Chem Phys 98:1372 28. Paul T 2010 M. Sc Thesis, Bengal Engineering and Science
23. McLean A-D, Chandler G-S (1980) J Chem Phys 72:5639 University
24. Breneman C-M, Wiberg K-B (1990) J Comput Chem 11:361 29. Joesten M-D (1982) J Chem Educ 59:362