Electrolysis of HBr using molten, alkali-bromide electrolytes

Article abstract of DOI:10.1016/j.electacta.2010.10.036

The electrolytic oxidation of HBr was investigated using molten alkali-bromide salts in a porous scaffold of yttria-stabilized zirconia (YSZ), with porous Pt electrodes. Despite a relatively thick (1.0 mm) and dense (35% porous) YSZ scaffold, a total cell impedance of less than 3 Ωcm ² was achieved with NaBr at 1033 K. The open-circuit potentials also agreed with theoretical potentials for the reaction H₂ + Br ₂ = 2HBr. Lower operating temperatures were made possible by using a eutectic mixture of alkali bromides, (Li_0.56K_0.19Cs _0.25)Br. The electrolyte losses for the (Li_0.56K _0.19Cs_0.25)Br electrolyte, determined from the ohmic component of the impedance spectra, were less than 1.5 Ωcm² at 673 and 773 K, similar to the value found for NaBr at 1033 K, indicating that the ionic conductivities of the alkali bromides are high so long as the salts are molten. However, the electrode losses were dependent on temperature, decreasing from 18.1 Ωcm² to 3.0 Ωcm² in going from 673 to 773 K. Implications of these results for recycling HBr in hydrocarbon bromination reactions are discussed.

Full text of DOI:10.1016/j.electacta.2010.10.036

Electrochimica Acta 56 (2011) 1581–1584
Contents lists available at ScienceDirect
Electrochimica Acta
journal homepage: www.elsevier.com/locate/electacta
Electrolysis of HBr using molten, alkali-bromide electrolytes
J.-S. Park∗, Chen Chen, N.L. Wieder, J.M. Vohs, R.J. Gorte
Department of Chemical and Biomolecular Engineering, University of Pennsylvania, Philadelphia, PA 19104, USA
a r t i c l e i n f o
a b s t r a c t
Article history:
The electrolytic oxidation of HBr was investigated using molten alkali-bromide salts in a porous scaf-
fold of yttria-stabilized zirconia (YSZ), with porous Pt electrodes. Despite a relatively thick (1.0 mm) and
dense (35% porous) YSZ scaffold, a total cell impedance of less than 3 ꢀcm2 was achieved with NaBr at
1033 K. The open-circuit potentials also agreed with theoretical potentials for the reaction H2 + Br2 = 2HBr.
Lower operating temperatures were made possible by using a eutectic mixture of alkali bromides,
(Li0.56K0.19Cs0.25)Br. The electrolyte losses for the (Li0.56K0.19Cs0.25)Br electrolyte, determined from the
ohmic component of the impedance spectra, were less than 1.5 ꢀcm2 at 673 and 773 K, similar to the
value found for NaBr at 1033 K, indicating that the ionic conductivities of the alkali bromides are high so
long as the salts are molten. However, the electrode losses were dependent on temperature, decreasing
from 18.1 ꢀcm2 to 3.0 ꢀcm2 in going from 673 to 773 K. Implications of these results for recycling HBr
in hydrocarbon bromination reactions are discussed.
Received 10 September 2010
Received in revised form 11 October 2010
Accepted 13 October 2010
Available online 20 October 2010
Keywords:
Alkali-bromide
HBr
Electrolysis
© 2010 Elsevier Ltd. All rights reserved.
1. Introduction
compounds is that Br2 is produced on the feed side of the mem-
brane, so that the desired product is diluted by whatever comes in
The electrochemical oxidation of HBr to Br2 could find applica-
tions in several important industrial processes. HBr is a by-product
in the production of some brominated organic compounds, which
are commonly used as flame retardants and as agricultural chemi-
cals [1]. It is also formed in large quantities in a recently developed
process for converting stranded natural gas to liquid fuels [2,3]. In
the conversion of natural gas, methane is reacted with Br2 to form
CH3Br and HBr and the CH3Br is subsequently reacted to form liq-
uid hydrocarbons and more HBr. Recycling the HBr back to Br2 is
essential in each case.
At present, the standard method for converting HBr back to Br2
involves chemical oxidation of the HBr with a second compound,
such as chlorine [4] or a metal oxide [2], which in turn must also be
recycled. Together with the fact that this brings additional materials
and complexity into the process, the reactions require several sep-
aration steps to produce the ultimate Br2 product. Electrochemical
oxidation of the HBr is obviously simpler conceptually and has been
achieved using proton-conducting electrolytes. For example, elec-
trolysis of HBr in aqueous solutions has been reported, although
subsequent steps are then required to separate the Br from the
solution [5]. Gas-phase electrolysis has also been reported, using
nafion-based membranes [6] and phosphoric-acid films [7].
One problem with using proton-conducting electrolytes to oxi-
dize HBr for processes involving production of brominated organic
with the feed. The use of a membrane that is a bromide-ion con-
ductor would produce concentrated Br2 that could be fed directly
back into the process. While there has been some effort to pro-
duce solid, bromide-ion conductors [8], we are aware of only one
report of gas-phase electrolysis of HBr using a membrane that
conducts bromide ions [9]. That study used porous zirconia mem-
branes saturated with a molten eutectic mixture of alkali bromides,
(Li0.575K0.133Cs0.292)Br. The gas-diffusion electrodes, composed of
either vitreous carbon or graphite felt, were then pressed onto the
electrode. Although the authors demonstrated removal of 95% of
the HBr from a process stream at 573 K [9], a cell potential of greater
than 3 V was required to achieve a current density of 100 mA/cm2.
Since the standard Nernst Potential for electrolysis of HBr is approx-
imately 0.6 V under these conditions, significant inefficiencies must
have existed in these cells.
In the present study, we set out to determine whether the
performance of an electrolyzer based on molten salts could be
improved substantially by using Pt electrodes, higher tempera-
tures, and different salt compositions. We will demonstrate that
reasonably good performance can be achieved with Pt electrodes
and at higher temperatures. The salt compositions that we studied
did not seem to matter, so long as the salts were molten.
2. Experimental procedures
A diagram of the cell design used in this study is shown in
Fig. 1. The support structures for these cells were porous wafers of
yttria-stabilized zirconia (YSZ), 1.0 cm in diameter. These were pre-
∗
Corresponding author. Tel.: +1 215 898 4439; fax: +1 215 573 2093.
E-mail addresses: jspark.phd@gmail.com, jongsung@seas.upenn.edu (J.-S. Park).
0013-4686/$ – see front matter © 2010 Elsevier Ltd. All rights reserved.
doi:10.1016/j.electacta.2010.10.036

1582
J.-S. Park et al. / Electrochimica Acta 56 (2011) 1581–1584
formers [10], the YSZ has a sponge-like structure, with pores rang-
ing in size from 1 to 5 ␮m. The porosity of the Pt layer appeared to
be significantly lower than that of the YSZ, with fewer but larger
pores.
After attaching the YSZ scaffold with Pt electrodes to the end
of a 1-cm diameter, alumina tube using a ceramic adhesive (522,
Aremco, USA), molten salts were added to the YSZ scaffold by
simply exposing its edges to the salts in a furnace at just above
the melting temperatures of the salts. Capillary forces were found
to be effective in drawing the salts into the pores. Two alkali-
bromide salts were used in these measurements: (1) NaBr and (2)
(Li0.56K0.19Cs0.25)Br. The raw powders used for making the elec-
trolytes were LiBr (99%, Alfa Aesar, USA), NaBr (99%, Alfa Aesar,
USA), KBr (101%, Fisher Science, USA) and CsBr (99.9%, Alfa Aesar,
USA). The (Li0.56K0.19Cs0.25)Br mixture was prepared by simply mix-
ing the appropriate amounts of LiBr, KBr and CsBr, melting the
mixture in an alumina crucible, and then quenching the mixture.
In order to conduct the fuel-cell and electrolysis experiments
under well defined partial pressures, the alumina tube with the cell
at one end was inserted into a second tube, so that the gas-phase
composition could be controlled on both sides. This apparatus is the
same as that used in an earlier study of Natural Gas Assisted Elec-
trolysis, where it is described in more detail [11]. We introduced
Br2 vapor to one side of the cell (the Br2 electrode) and a mixture
of H2 and HBr to the other side (the HBr electrode). The Br2 was
fed to the cell by flowing 100 ml/min of N2 through a bubbler of
liquid Br2 (99.8%, Alfa Aesar). The bubbler was held at 298 K, so
that the partial pressure of Br2 was 0.3 atm. At the HBr electrode,
HBr was produced by reacting excess H2 with a second N2–Br2
stream. The H2–N2–Br2 mixture was allowed to flow through a
quartz tube at temperatures above 673 K before being exposed to
the HBr electrode. To check that the Br2 was completely converted
to HBr by reaction with H2, the outlet gas from the reactor was bub-
bled through water to look for color changes characteristic of Br2
but none was ever observed. The partial pressures of HBr and H2
could be controlled by changing the relative flow rates of the H2
and Br2–N2 streams.
Fig. 1. Schematic diagram of the cell used for electrolysis of HBr.
pared by first ball milling equal weights of YSZ (Tosoh Corp., 8 mol%
Y2O3-doped ZrO2, 0.2 ␮m) and graphite (Aldrich, USA) for 24 h in
ethanol. After drying, the mixed powder was pressed into 2-mm
thick wafers and fired to 1773 K for 4 h. Finally, the calcined wafers
were polished to a thickness of 1.2 mm. The porosity of these wafers
was determined to be 35%, by comparing the apparent density to
the bulk density of YSZ and by measuring its weight after saturation
with water. Platinum electrodes were added to both sides of the YSZ
wafer using Pt paste (6926, BASF, USA). The paste was applied as
0.64-cm circles on both sides of the wafer using a template, after
which the entire structure was fired to 1473 K for 30 min to cure
the Pt.
Fig. 2 shows an SEM micrograph of the interfacial region of the
Pt-YSZ structure. The Pt electrodes were approximately 300 ␮m
thick and adhesion to the porous YSZ was very good. As in previous
studies with porous YSZ made by tape casting with graphite pore
The performance of the cells, both V–i polarization curves and
impedance spectra, was monitored using a Gamry Instruments
potentiostat. Impedance spectra were measured at various bias
Fig. 2. SEM micrograph of the interfacial region of the Pt-YSZ structure. The Pt layer is at the top.

J.-S. Park et al. / Electrochimica Acta 56 (2011) 1581–1584
1583
Fig. 4. (a) V–i polarization curves for the cell using the (Li0.56K0.19Cs0.25)Br elec-
trolyte, measured at 673 (ꢀ) and 773 K (ꢀ), with partial pressures for H2 and HBr
of 0.1 and 0.4 atm on the HBr electrode and a Br2 pressure of 0.3 atm on the Br2
electrode. (b) Impedance data measured at open circuit at 673 (ꢀ) and 773 K (ꢀ).
Fig. 3. (a) The V–i polarization curve for the cell with the NaBr electrolyte, measured
at 1033 K using a P(Br2) of 0.3 atm at the Br2 electrode and P(H2) and P(HBr) of 0.5
and 0.2 atm, respectively, at the HBr electrode. (b) Impedance spectra measured at
0.56 (ꢀ) and 0.76 V (ꢀ).
loss in the cell, determined from the high-frequency intercept with
the real axis in the Cole–Cole Plot, is approximately 1.8 ꢀcm2, while
the non-ohmic loss, determined from the lengths of the arcs in the
impedance data, was 1.1 ꢀcm2. The ohmic losses could certainly be
reduced by making the YSZ scaffold thinner and by increasing its
porosity. The non-ohmic losses are associated with the electrodes
but the factors influencing these losses are more difficult to iden-
tify. The fact that these losses were independent of current density
implies that diffusion was not important in this measurement.
A problem associated with working at high temperatures is that
the vapor pressure of NaBr at 1033 K is approximately 10−3 atm,
and loss of the electrolyte was observed after a day of operation. In
order to work at lower temperatures, we next examined a cell with
(Li0.56K0.19Cs0.25)Br as the electrolyte. This electrolyte has a melt-
ing temperature of approximately 523 K [12], so that experiments
with molten electrolytes could be performed at much lower tem-
peratures. Fig. 4 shows V–i polarization and impedance data for this
cell, measured at 673 and 773 K, with partial pressures for H2 and
HBr of 0.1 and 0.4 atm on the H2 side and a Br2 pressure of 0.3 atm
on the Br2 side.
As shown in Fig. 4(a), changing the temperature had a significant
impact on the polarization curves. First, the measured OCV was
0.45 V at both 673 and 773 K. Although the Nernst Potential does
not change significantly over this temperature range, the value of
0.45 V is slightly lower than the calculated potential for these partial
pressures, 0.54 V. Since the OCV is a function of the H2:HBr ratio at
the H2 electrode, it seems likely the actual ratio was slightly lower
than we had targeted. More significant for this discussion, the slope
of the V–i curve at 673 K was much steeper than at 773 K. The V–i
voltages in the frequency range from 100 kHz to 0.1 Hz, with a
20 mV AC perturbation.
3. Results and discussion
In order to determine the feasibility of using molten, alkali
bromides as electrolytes, we first examined a cell with a NaBr elec-
trolyte, using P(Br2) of 0.3 atm at the Br2 electrode and P(H2) and
P(HBr) of 0.5 and 0.2 atm, respectively, at the HBr electrode. Because
the melting temperature of NaBr is 1020 K, electrochemical mea-
surements were carried out at 1013 and 1033 K so as to work just
above or just below the melting temperature of the electrolyte. The
cell performance at 1013 K was so poor that it was unmeasure-
able, demonstrating the necessity of having the electrolyte salt be
molten. By contrast, the V–i polarization curve at 1033 K, shown
in Fig. 3(a), was reasonably good. The open-circuit voltage (OCV)
was 0.66 V, which is identical to the Nernst Potential that would be
calculated for this temperature at these partial pressures. The V–i
relationship is also a straight line for both HBr production (negative
currents) and HBr electrolysis (positive currents). The current den-
sity at 1.2 V was 200 mA/cm2, which corresponds to the production
of ∼84 cm3 Br2 cm−2 h−1 (STP).
The nature of the losses in the cell at 1033 K was investigated
by impedance spectroscopy, with the spectra measured at 0.56 and
0.76 V shown in Fig. 3(b). These two potentials were chosen so as to
measure the impedance while producing and consuming HBr. The
impedance plots are essentially identical and show that the ohmic

1584
J.-S. Park et al. / Electrochimica Acta 56 (2011) 1581–1584
Replacing Pt with other metals might be difficult due to corro-
sion issues. While Pt forms a bromide salt, PtBr2 is unstable above
523 K. However, there are conductive ceramics, such as dope SrTiO3
[13,14], that deliver high conductivities and are reasonably stable in
harsh environments. Although ceramic materials are less catalyti-
cally active than Pt, it may be possible to enhance their performance
by doping with small amounts of catalytically active metals [15,16].
Alternatively, the results presented here suggest that one could
work at much higher temperatures where the electrode, catalytic
requirements might be less stringent. Vaporization of the elec-
trolyte salt should not be a limiting factor, since the salt could
be easily replaced by a salt reservoir at the edges of the scaffold
medium.
In summary, the results in this paper suggest that reasonable
performance can be achieved for HBr electrolysis using molten,
alkali-bromide electrolytes. Further investigation into electrode
materials will be needed; however, the inherent advantages of pro-
ducing a concentrated Br2 stream suggest this work is justified.
Fig. 5. V–i polarization curves of the cell using the (Li0.56K0.19Cs0.25)Br electrolyte,
measured at 773 K using the following conditions: (P(HBr), P(H2)) = (0.05 atm,
0.3 atm) (ꢀ) and (0.4 atm, 0.1 atm) (ꢀ) on the HBr electrode. The P(Br2) was fixed at
0.3 atm on the Br2 electrode.
4. Conclusion
relationships were also not linear, with the “S” shape of the V–i
curve being especially noticeable at 773 K. The fact that there is no
break in the slopes of the V–i curves at open circuit implies that
the electrode reactions are probably not activated by polarization.
The changing slope at higher current densities, especially on the
HBr production side, is likely associated with the low H2 partial
pressures.
The Cole–Cole Plots of the impedance spectra, measured at
0.55 V and the conditions of Fig. 4(a), are shown in Fig. 4(b). Interest-
ingly, the ohmic losses at these two temperatures were similar and
relatively small, 1 ꢀcm2 at 773 K and 1.5 ꢀcm2. This would sug-
gest that the ionic conductivity of the alkali-bromide electrolytes
is weakly dependent on temperature, so long as the salts exist
in their molten state. The differences in the ohmic losses for the
molten NaBr and the molten eutectic mixture are also not sig-
nificant. The main difference in the cell performances at the two
temperatures in Fig. 4 was that the non-ohmic losses decreased
with increasing temperature, from 18 ꢀcm2 at 673 K to 6 ꢀcm2
at 773 K. Since the non-ohmic losses are associated with the elec-
trodes, this result implies that the performance is limited by the
electrodes. The strong temperature effect is likely due to increases
in the chemical reaction rates, since changes in gas-phase diffusiv-
ity with temperature would be inconsequential.
Electrolytic oxidation of HBr to H2 and Br2 using a mem-
brane that is a bromide-ion conductor has significant advantages
over membranes that are proton conducting because Br2 can be
produced as a concentrated stream. The results in this paper
demonstrate that molten alkali-bromide salts exhibit good ionic
conductivities and could be effective as electrolytes for this appli-
cation. Although Pt is an effective material for both electrodes in
this application, the possibility of using less expensive materials is
feasible due to the possibility of using high temperatures in this
application.
Acknowledgements
This work was funded by the U.S. Department of Energy’s Hydro-
gen Fuel Initiative (grant DE-FG02-05ER15721). We would like to
thank Prof. Eric McFarland and Dr. Anna Ivanovskaya for helpful
suggestions.
Dr. J.-S. Park also acknowledges support from the National
Research Foundation of Korea Grant funded by the Korean Gov-
ernment (NRF-2009-352-D00134).
References
Finally, the effect of changing the partial pressures of H2 and
HBr was investigated on the (Li0.56K0.19Cs0.25)Br cell at 773 K, while
holding the Br2 pressure fixed at 0.3 atm on the Br2 electrode. Fig. 5
shows V–i polarization curves with a relatively low HBr partial pres-
sure (P(HBr) of 0.05 atm and P(H2) of 0.3 atm) and a relatively low
H2 partial pressure (P(HBr) of 0.4 atm and P(H2) of 0.1 atm). The OCV
measured with P(HBr) and P(H2) of 0.05 and 0.3 atm, respectively,
was 0.72 V, the same as the theoretical value. The OCV decreased
to 0.45 V at the higher P(HBr). Under HBr electrolysis conditions,
the slope of V–i polarization curve decreased dramatically with
increasing P(HBr).
[1] Bromine, organic compounds, in: David Ioffe, Arieh Kampf (Eds.), Kirk-Othmer
Encyclopedia of Chemical Technology, 2002, p. 340.
[2] A. Breed, M.F. Doherty, S. Gadewar, P. Grosso, I.M. Lorkovic, E.W. McFarland,
M.J. Weiss, Catal. Today 106 (2005) 301.
[3] X.-P. Zhou, A. Yilmaz, G.A. Yilmaz, I.M. Lorkovic, L.E. Laverman, M. Weiss, J.H.
Sherman, E.W. McFarland, G.D. Stucky, P.C. Ford, Chem. Commun. 18 (2003)
2294.
[4] J.E. Kogel, N.C. Trivedi, J.M. Barker, S.T. Krukowski (Eds.), Industrial Minerals
& Rocks, 7th ed., Society of Mining, Metallurgy, and Exploration, Littleton, CO,
2006, p. 289.
[5] J. Hanrahan, H.L. Heaton, R.Z. Parker, US Patent no. 5,833,834 (1998).
[6] P. Sivasubramanian, R.P. Ramasamy, F.J. Freire, C.E. Holland, J.W. Weidner, Int.
J. Hydrogen Energy 32 (2007) 463.
The results from this study demonstrate that the performance of
cells with molten alkali bromides as electrolytes can be reasonably
good. Even with the relatively thick, low-porosity scaffolds used
to hold the molten electrolytes, the impedance spectra demon-
strate that the electrolyte losses in these cells can be relatively low.
The electrode losses are the more significant issue. The Pt elec-
trodes used here appear to be much more effective than carbon
electrodes used in the previous study by Wauters and Winnick [9],
even without any attempt to increase their porosity. However, it
would obviously be better if less costly materials than Pt could be
used.
[7] W. Kondo, S. Mizuta, Y. Oosaw, T. Kumagai, K. Fujii, Bull. Chem. Soc. Jpn. 56
(1983) 2504.
[8] N. Imanaka, Y. Kato, J. Mater. Sci. 40 (2005) 6495.
[9] C.N. Wauters, J. Winnick, AIChE J. 44 (1998) 2144.
[10] S. Park, R.J. Gorte, J.M. Vohs, J. Electrochem. Soc. 148 (2001) A443.
[11] W. Wang, J.M. Vohs, R.J. Gorte, Top. Catal. 46 (2007) 380.
[12] G.G. Diogenov, V.I. Ermachkov, Russ. J. Inorg. Chem. 12 (1967) 436 (Engl.
Transl.).
[13] S.Q. Hui, A. Petric, J. Electrochem. Soc. 149 (2002) J1.
[14] J. Canales-Vazquez, S.W. Tao, J.T.S. Irvine, Solid State Ionics 159 (2003) 159.
[15] S. Lee, G. Kim, J.M. Vohs, R.J. Gorte, J. Electrochem. Soc. 155 (2008) B1179.
[16] G. Kim, S. Lee, J.Y. Shin, G. Corre, J.T.S. Irvine, J.M. Vohs, R.J. Gorte, Electrochem.
Solid State Lett. 12 (2009) B48.