Journal of The Electrochemical Society, 150 ͑6͒ D99-D107 ͑2003͒
D103
and cathode, which were oriented between the faces of the housings.
The system was gradually heated to an operating temperature of
315-330°C in an electrically-heated Lindberg furnace. When the
electrolyte became molten, weight was applied to the top of the cell
using a pneumatic piston. This provided a wet electrolyte seal
around the lips of the housings to prevent gas leakage.
Table II. Estimated mass transfer limiting current densities.
HCl mass
transfer step
iL
2
͑A/cm ͒
Bulk diffusion
Pore diffusion
Thin film diffusion
Membrane migration
8.9
3.6
1.6
1.7
Half-cell housings, which must be electrically conductive,
chemically resistant, and thermally stable, were fabricated by The
Electrosynthesis Co. Housings were machined from graphite and
impregnated with glassy carbon to minimize porosity and enhance
chemical and thermal stability. 3-D porous electrodes were chosen
to provide a high surface area for the electrode reactions involving
a
a
Achievable current density per applied volt.
2
three-phase contact. Pressed graphite felt (A ϭ 14.5 cm ; Elec-
Eq. 22, effective conductivity is defined as a function of the molten
2
trosynthesis Co.͒ and porous carbon aerogel paper (A ϭ 11.9 cm ;
salt electrolyte conductivity ͑͒, membrane porosity ( ) and tortu-
m
Marketech International͒ were chosen as the gas-diffusion electrodes
due to their physical, mechanical, and chemical stability.
osity ( ).
m
m
Data collection began when the electrolyte became molten and
the lips of the housings were sufficiently sealed to permit gas flow
through the system. A concentrated mixture of anhydrous HCl in
nitrogen, ranging from 25 to 50% HCl ͑Matheson͒, was fed to the
cathode through an alumina tube at flow rates of 100 to 300 mL/
min. Flowmeters equipped with Viton O-rings and tantalum and
sapphire beads ͑Brooks Instrument͒ were used to measure the flow
rates of the HCl-containing streams. As needed throughout the trial,
small amounts of electrolyte were fed to the membrane in situ
through a glass pipette to make up for losses due to incomplete
sealing at the cell edges.
eff
ϭ
͓26͔
m
The conductivity of the molten salt mixture at 573 K ͑0.767
S/cm͒ was estimated using data for the binary system LiCl-KCl
59-41 mol %͒. The system, consisting of a 65% porous membrane
with a thickness of 1.0 mm and a tortuosity of 3, is capable of
achieving up to 1.7 A/cm per applied volt.
Table II summarizes the predicted limiting current densities dic-
tated by mass transfer control in the experimental separation cell.
Based on these calculations, it is predicted that mass transfer will
2
4
͑
2
A Princeton Applied Research model 371 potentiostat supplied
direct current to the cathode. At each applied current, Simpson
model 460 multimeters monitored the applied current, anode-to-
cathode cross-cell voltage, and electrode potentials vs. a graphite rod
2
not limit HCl conversion to current densities over 1 A/cm .
Experimental
Hydrogen chloride removal was tested for the first time in a
single electrochemical membrane cell to experimentally verify the
concept. Experiments were initially based on bromine recovery tri-
als conducted in this laboratory, with the implementation of opti-
mized materials as established in the previous tests. Process per-
formance was characterized by HCl removal efficiency, chlorine
production and purity, polarization data, and cell longevity. Material
compatibility aspects were closely examined.
͑0.125 in. diam. POCO Graphite Inc͒ pseudoreference electrode.
Cell ohmic resistance was determined at each current by applying
the current interrupt technique and analyzing the output on a Tek-
tronix model 5111A storage oscilloscope. The cathode outlet stream,
containing hydrogen and unreacted HCl, was analyzed using Fourier
transform infrared spectroscopy ͑FTIR͒. Teflon tubing that directed
the outlet gas to the FTIR was wrapped in heating tape to maintain
the stream at 100°C. The absorbance spectrum generated by the HCl
dipole moment was easily identified using a Perkin Elmer FTIR
Paragon 500 with a 10 cm path glass cell and calcium fluoride
windows. To quantify HCl concentration, the absorbance of the
2
5
The single-cell design is shown in Fig. 5. Two electrically con-
ductive half-cell housings directed the flow of current and gas-phase
reactants and products to and from the electrodes. A woven mat of
͑
8% yttria-stabilized͒ zirconia with a thickness of 1 mm and an
Ϫ1
strongest peak at 2944 cm was recorded at 5-10 min intervals
estimated porosity of 65% was fabricated by Zircar Products and
used as a nonconductive, chemically, and thermally stable mem-
brane to absorb the molten salt electrolyte and separate the elec-
trodes. The electrolyte was a eutectic mixture of lithium chloride
57.5 mol %͒, potassium chloride ͑13.3%͒, and cesium chloride
29.2%͒ with a reported melting temperature of 265°C. Due to their
hygroscopic nature, the reagent grade salts ͑Sigma Chemical͒ were
mixed in a nitrogen environment inside a glove box. The membrane
was then presaturated with the electrolyte in a separate furnace by
heating it to 350-400°C under a nitrogen purge for a minimum of 4
h to remove residual moisture. The electrolyte/membrane was then
cooled and incorporated into the cell at 150°C between the anode
throughout each trial. FTIR calibration curves were constructed at
known HCl concentrations. A second order fit to the data provided
the concentration relationship.
Chlorine generated at the anode was removed with a continuous
͑
͑
high purity nitrogen purge ͑Air Products͒ and occasionally indicated
1
using starch-iodide paper, a common Cl detection technique. The
2
chlorine was condensed with a liquid nitrogen bath and collected
until the experimental run was terminated. Following the trial the
captured chlorine was purged through the FTIR unit to identify po-
tential contaminants. Of particular interest was the presence of HCl
in the anode outlet, possibly indicating a crack in the membrane.
Results and Discussion
Each experimental cell was operated periodically over the course
of several days, with the longest-lasting cell enduring 11 days of
intermittent operation; afterwards it was voluntarily terminated. The
2
cell was operated at current densities exceeding 400 mA/cm , based
on a superficial electrode area. Data were collected over a maximum
duration of 225 min.
Hydrogen chloride concentration as a function of applied current
is plotted in Fig. 6. Conversion values lie in close agreement with
those predicted by Faraday’s law, based on a process involving two
electrons per mole of chlorine produced. Current efficiencies ap-
proached 100% in every case. Using a feed of 25% HCl at 100
mL/min, HCl concentration dropped to less than 2% when current
was applied at 96% of the stoichiometric current.
2
5
Figure 5. Experimental single cell ͑Wauters ͒.