Electrowinning Al from Al2 S3 in molten salt

Article abstract of DOI:10.1149/1.2728263

In order to investigate an alternative process for the production of primary aluminum via a sulfide intermediate, the electrochemical behavior of Al2 S3 in molten salt has been studied on a laboratory scale. The effects of electrolyte composition, temperature, and cell design on the cell performance have been investigated. Temperature and cryolite addition have positive effects on the current density. Increasing the anode-to-cathode surface area (closer to unity) and shortening the interelectrode distance lead to higher current density. It is concluded that the electrolytic process is governed by the ohmic drop, caused mainly by the sulfur bubbles at the anode.

Full text of DOI:10.1149/1.2728263

Evaluation Only. Created with Aspose.PDF. Copyright 2002-2021 Aspose Pty Ltd.
D334
Journal of The Electrochemical Society, 154 ͑6͒ D334-D338 ͑2007͒
0013-4651/2007/154͑6͒/D334/5/$20.00 © The Electrochemical Society
Electrowinning Al from Al2S3 in Molten Salt
,z
Y. Xiao,a, D. W. van der Plas,b J. Bohte,a S. C. Lans,a A. van Sandwijk,a and
*
M. A. Reutera,c
aDepartment of Materials Science and Engineering, Delft University of Technology, 2628 CD Delft,
The Netherlands
bCorus Research, Technology and Development, 1970 CA IJmuiden, The Netherlands
In order to investigate an alternative process for the production of primary aluminum via a sulfide intermediate, the electrochemi-
cal behavior of Al2S3 in molten salt has been studied on a laboratory scale. The effects of electrolyte composition, temperature,
and cell design on the cell performance have been investigated. Temperature and cryolite addition have positive effects on the
current density. Increasing the anode-to-cathode surface area ͑closer to unity͒ and shortening the interelectrode distance lead to
higher current density. It is concluded that the electrolytic process is governed by the ohmic drop, caused mainly by the sulfur
bubbles at the anode.
© 2007 The Electrochemical Society. ͓DOI: 10.1149/1.2728263͔ All rights reserved.
Manuscript submitted October 9, 2006; revised manuscript received February 9, 2007. Available electronically April 23, 2007.
In terms of primary energy consumption, the Hall–Héroult pro-
promising with regard to energy consumption,6 i.e., it has the lowest
decomposition potential. The alkali chloride electrolyte permits a
relatively low operating temperature of about 700°C, just above the
melting point of aluminum. Because of operation with nonconsum-
able anodes, the interelectrode gap can be reduced and a multipolar
cell operation is possible, which increases productivity and reduces
energy consumption and capital costs. As Al2S3 reacts with air and
water, operation under an inert atmosphere is necessary.
The eutectic of a MgCl2–NaCl–KCl mixture ͑50–30–20 mol %͒
has been selected as an appropriate electrolyte for the electrolysis of
Al2S3. Electrochemical studies of Al2S3 in a chloride melt showed
that the reduction of Al ions at a graphite electrode is a diffusion-
controlled process and proceeds via a reversible three-electron
charge transfer.7,8 Minh et al.8 reported a limiting current density of
0.3 A cm−2 at the saturation solubility of Al2S3 ͑ϳ3 wt %͒ and
0.2 A cm−2 in the MgCl2–NaCl–KCl eutectic containing 2 wt %
Al2S3. The current efficiency was about 80% at a current density of
0.2 A cm−2, a cell potential of about 1.5 V, and interelectrode gap
of 3 cm. By adding 10 wt % AlCl3 to increase the solubility of
Al2S3 to 5 wt % and to increase the concentration of Al ions, current
densities of up to 2 A cm−2 were obtained. In this situation, the
current density was limited by the anodic reaction.
cess requires about 140 MJ energy to produce 1 kg of aluminum
metal, which is about four times the theoretical thermodynamic
minimum ͑34 MJ/kg Al͒. Apart from substantial CO2 and SO2 emis-
sions from the production of electrical energy as well as the con-
sumption of carbon anodes, the Hall–Héroult process creates green-
house gas emissions such as CF4 and C2F6 stemming from the
electrolyte that consists mainly of liquid cryolite ͑Na3AlF6͒. These
and other apparent disadvantages of the Hall–Héroult process have
led to numerous research efforts to find an alternative route for
primary aluminum production throughout the 20th century.
The best investigated alternatives are inert anodes,1 the use of
AlCl3 as an intermediate step,2 and the carbothermic reduction of
alumina.3 Inert anodes would enable a closed cell concept that is
more favorable in terms of environment, energy efficiency, and capi-
tal cost. The aluminum chloride process comprises the chlorination
of Al2O3 to AlCl3 as an intermediate, and subsequent reduction of
AlCl3 to aluminum in an electrolysis cell. Due to the lower cell
voltage required, the carbon anode remains inert and less electrical
energy is required compared to an inert electrode process based on
the reduction of alumina. Carbothermic reduction would eliminate
the electrolysis step altogether, but high temperatures in excess of
2000°C would be required. Despite the substantial amount of capital
spent on these and other processes, no proven and economically
feasible alternative to the Hall–Héroult process to date exists.
Another alternative is to use aluminum sulfide as an intermediate
step, named the Compact Aluminum Production Process
͑CAPP™͒.4,5 In this process, Al2S3 is produced by the reaction of
CS2 and alumina. Subsequently, the aluminum metal can be ex-
tracted by electrolysis, producing sulfur gas at the anode, which can
be recycled to the production of CS2. The simplified reactions ͑as-
suming no complex ions͒ of the electrolysis process can be de-
scribed as follows6
In the present work, the behavior of electrowinning Al from
Al2S3 under different conditions is investigated in order to validate
and improve the operating conditions described in the above.
Experimental
Raw materials
.— Al2S3 with 98% purity was used for the ex-
periment. The NaCl, KCl, Na3AlF6, and MgCl2 were of laboratory
grade with minor impurities. The chemicals were stored and handled
in a glove box in an argon atmosphere.
Cathode
Anode
2Al3+ + 6e− → 2Al
2AlS33− − 6e− → Al2S3 + 1.5S2͑g͒
͓1͔
͓2͔
͓3͔
Overall
Al2S3 → 2Al + 1.5S2͑g͒
Due to the lower cell voltage required, the carbon anode remains
inert and less electrical energy is required compared with the Hall–
Héroult process. As a result of the advantageous electrolysis pro-
cess, substantial benefits arise in terms of energy efficiency and
environment. Figure 1 shows that the electrolysis of Al2S3 is very
*
Electrochemical Society Active Member.
c Present address: Department of Civil and Environmental Engineering, University
of Melbourne, Melbourne, 3010 Victoria, Australia.
z E-mail: y.xiao@tudelft.nl
Figure 1. Decomposition potential of aluminum species for aluminum pro-
duction.
Downloaded on 2015-03-13 to IP 169.230.243.252 address. Redistribution subject to ECS terms of use (see ecsdl.org/site/terms_use) unless CC License in place (see abstract).

Evaluation Only. Created with Aspose.PDF. Copyright 2002-2021 Aspose Pty Ltd.
Journal of The Electrochemical Society, 154 ͑6͒ D334-D338 ͑2007͒
D335
Figure 3. ͑Color online͒ Electrowinning of Al from Al2S3 in
MgCl2–NaCl–KCl based electrolyte of 50–30–20 mol % with cryolite addi-
tion at 725°C.
amounts of added aluminum sulfide, temperature, anode-cathode
distance, and the ratio of the anode-to-cathode surface area.
Electrolysis of Al2S3 in MgCl2–NaCl–KCl eutectic-based
system.— In order to saturate the electrolyte with aluminum sulfide,
4 wt % of Al2S3 was added into the eutectic MgCl2–NaCl–KCl of
50–30–20 mol %. Figure 3 shows the results of the electrolysis of
Al2S3 with the ternary eutectic at an operating temperature of
725°C. A linear relationship between the cell potential and the cur-
rent density is observed. The kick off of the reaction was determined
by extrapolation of the j-E relationship to the intercept with the x
axis. The apparent decomposition potential is E = −1.1 V, which is
close to the theoretical value of −0.98 V at 725°C.
The results from the experiment with no cryolite addition indi-
cate a limiting current density at approximately 0.17 A/cm2, occur-
ring at −2.4 V. The effective electrode area of the cathode was
assumed to be the cross-sectional area of the cell. This is somewhat
less than the limiting current density of 0.3 A/cm2 as reported by
Minh et al.8 In both cases, the limiting current density in the
MgCl2–NaCl–KCl eutectic electrolyte is too low to compete with
that of 0.7–1.2 A/cm2 in the Hall–Héroult process.
Figure 2. Experimental setup.
Experimental equipment and procedure.— A schematic view of
the experimental cell is shown in Fig. 2. The electrochemical cell is
assembled in a corundum alumina tube under inert Ar atmosphere in
a glove box. The cathode is a pool of molten aluminum with effec-
tive area of 8.1 cm2, which is polarized by a graphite block con-
nected by a rod of stainless steel shielded by a quartz tube. The
anode is constructed of a graphite block of 1 ϫ 1 cm and 5 cm
high, which is immersed 2 cm into the electrolyte and is connected
by a stainless steel rod. The surface area ratio of anode to cathode is
0.12. The interelectrode gap is 2 cm. The anode acts also as the
reference electrode; thus, the cell potential is measured during the
electrolysis.
In the experiment, the cell is externally heated by a 2100 W
cylindrical furnace equipped with Kanthal Super heating elements,
with maximum operating temperature of 1400°C. The temperature
is measured with Pt–PtRh10 thermocouples. In the off-gas line, two
water bottles were installed to wash the off-gas and to remove the
sulfur before gas release to the atmosphere. A buffer bottle was
installed before the water bottles to prevent a backflow in the sys-
tem.
During the experiment, a continuous flow of argon ͑0.5 L/min͒
was led through the cell to protect it from contact with air. The
furnace was first heated and remained at 200°C for 45 min to re-
move the moisture in the system. Then the furnace was further
heated to the experimental temperature and was equilibrated at that
temperature for 1 h before the measurement. To investigate the elec-
trochemical behavior of the cell, linear sweep voltammetry and am-
perometry experiments were performed. At the end of the measure-
ment the electrolyte was taken out and the aluminum metal was
collected to determine the current efficiency.
In order to increase the current density, cryolite was added to the
electrolyte to increase the solubility of Al2S3 and to increase the
activity of aluminum ion in the melt. Linear extrapolation of the j-E
curve gives a decomposition potential of −0.98 V, which equals the
theoretical value. The amount of cryolite added was varied between
5 and 30 wt %. The highest current density was achieved with
10 wt % of added cryolite. In this case the current density increased
by more than a factor of three. This is most probably due to both the
increased Al2S3 solubility and the increased conductivity. From
solubility experiments it was observed that the solubility of Al2S3 in
the electrolyte increases by a factor of four when adding 10 wt % of
cryolite to the electrolyte.9 In addition, the specific conductivity of
cryolite ͑2.79 ⍀−1 cm−1 at 1000°C10͒ is higher than that of the ter-
6
nary eutectic ͑estimated to be about 1.6 ⍀−1 cm−1 ͒. However,
when further increasing cryolite from 10 to 30%, the current density
is lowered. This is most probably due to the increased viscosity of
the electrolyte. In general, it is demonstrated that the addition of
fluoride contributes to a positive effect on the electrowinning pro-
cess to a certain extent.
Effect of temperature.— The operating temperature in an elec-
trolytic cell used for the winning of aluminum is an important pa-
rameter that influences many of the properties of the electrolyte, for
example density, viscosity, conductivity, and vapor pressure. All
these properties have more or less influence on the electrochemical
behavior in the cell and further on the current efficiency. However,
none of these properties are known for the MgCl2–NaCl–KCl eu-
tectic with cryolite and Al2S3 additions. Voltammograms and linear
sweeps were performed at the temperatures of 725, 775, 785, and
825°C. The results are shown in Fig. 4. The temperature has a
slightly positive effect on the cathodic current density.
Results
The effects of various experimental parameters on the electrow-
inning behavior have been investigated, including cryolite addition,
Downloaded on 2015-03-13 to IP 169.230.243.252 address. Redistribution subject to ECS terms of use (see ecsdl.org/site/terms_use) unless CC License in place (see abstract).

Evaluation Only. Created with Aspose.PDF. Copyright 2002-2021 Aspose Pty Ltd.
D336
Journal of The Electrochemical Society, 154 ͑6͒ D334-D338 ͑2007͒
Figure 6. ͑Color online͒ Effect of interpolar distance ͑gap͒ and the surface
area ratio of anode to cathode on the electrowinning of Al from 4 wt %
Al2S3 in MgCl2–NaCl–KCl-based electrolyte with 10 wt % cryolite addition
at 725°C.
Figure 4. ͑Color online͒ Effect of temperature on the electrowinning of Al
from 4 wt % Al2S3 in MgCl2–NaCl–KCl eutectic based electrolyte with
20 wt % cryolite addition.
to unity has positive effect on the cathodic current density. The
cathodic graphite block is 2.2 cm in diameter and 2 cm in height. If
the 3.80 cm2 top surface of the graphite block is considered as the
effective surface area of the cathode, the anode-to-cathode surface
area ratio increases to 0.26. This is more than twice the original
value of 0.12 based on the surface area of the aluminum cathode
pool, which equals the cross section of the cell, 8.04 cm2. In both
cases, the same anode was used with an assumed effective surface
area of 1 cm2.
To investigate the effects of an increased distance on the current
density, experiments were carried out with the graphite block cath-
ode, using a vertical interpolar distance of 2.0, 2.5, and 3.0 cm,
respectively. The shorter the distance, the higher the cathodic current
density.
Effect of cell design.— In the electrolysis process, both interpo-
lar distance and surface area ratio of anode to cathode are important
for the cell performance. The left side of Fig. 5 illustrates the origi-
nal layout of the cathode and anode in the cell. The anode is not
symmetrically located above the middle of the cathode. In addition,
the anode surface opposing the cathode is much smaller than the
cathode surface. These geometrical constraints reduce the effective
cathode surface area to a value much smaller than the cross section
of the alsint cell tube. As this cross section was used in calculating
the limiting current density, the real limiting current density for a
more favorable geometry would be substantially higher. This was
investigated by changing the cell layout and using the graphite block
as the cathode, as illustrated on the right side of Fig. 5.
The electrolysis results under different arrangements are depicted
in Fig. 6. The experiment with surface area ratio of 0.12 was carried
out with an aluminum pool as cathode. The other experiments were
performed with a graphite cathode without Al pool. When using the
graphite block as the cathode, the cathodic current density increased
significantly, in comparison to that with the Al pool as the cathode.
Increasing the anode-to-cathode surface area ratio to a value closer
Current efficiency.— The current efficiency ͑CE͒ of aluminum
electrolysis is an important factor in both technical and economical
aspects. CE is defined as the percentage of the metal amount actu-
ally produced divided by the theoretical metal output. The theoreti-
cal metal output is calculated as below
Q
theoretical
Al
W
=
MAl
͓4͔
nF
where Q is the electric charge that passes through the cell, n is the
valence of aluminum ion in the electrolyte, F is Faraday’s constant,
and MAl is the mole mass of aluminum metal.
The current efficiency is influenced by a number of variables
including temperature, composition of the electrolyte, and cell de-
sign. In the present study, in order to determine the current effi-
ciency, a fixed potential of 2.2 V was supplied to the cell with
graphite cathode at 725°C to prevent codecomposition of other spe-
cies like MgCl2. The electrolyte contains MgCl2–NaCl–KCl eutectic
with 10 wt % cryolite and 4 wt % Al2S3. The result is shown in Fig.
7. From the linear scan it was indicated that after 1300 s the con-
centration of aluminum ion in the electrolyte is very low. The total
amount of current passed through the electrolyte is 7000 C at this
stage. This equals 0.66 g of theoretical aluminum output. As no
aluminum pool was used in this experiment, the metal produced was
dispersed into the electrolyte as fine droplets. After the experiment,
the electrolyte was dissolved in the distilled water. The aluminum
metal was collected and further dissolved in HCl. The actually pro-
duced amount of aluminum from electrolysis was determined by
analyzing the HCl solution with atomic absorption spectrophotom-
etry ͑AAS͒. The total amount of actually produced aluminum is
0.39 g, which is equivalent to the consumption of 1.08 g of Al2S3.
The Al2S3 added is 2.6 g; thus, about 42% of Al2S3 has been con-
verted into aluminum metal during the electrowinning process. The
current efficiency for this experiment was determined to be about
Figure 5. ͑Color online͒ Cell layout: ͑left͒ Al pool as cathode and ͑right͒
graphite block as the cathode.
Downloaded on 2015-03-13 to IP 169.230.243.252 address. Redistribution subject to ECS terms of use (see ecsdl.org/site/terms_use) unless CC License in place (see abstract).

Evaluation Only. Created with Aspose.PDF. Copyright 2002-2021 Aspose Pty Ltd.
Journal of The Electrochemical Society, 154 ͑6͒ D334-D338 ͑2007͒
D337
Figure 7. ͑Color online͒ Plot of cathodic current density at 2.2 V in the
electrolyte of MgCl2–NaCl–KCl eutectic with 10 wt % cryolite and 4 wt %
Al2S3 at 725°C.
60%. The maximum current efficiency achieved under the present
conditions was 62% when potential of 2 V was applied for 5700 s.
Discussion
Current efficiency.— The current efficiency of 60% obtained
during the experiment is rather low compared to the average values
of 90-95% in the modern Hall–Hérout process for the Al production
process. This may be explained by various factors to be discussed
below.
Considering the low solubility of Al2S3 ͑ϳ0.32 wt %͒ in the
MgCl2–NaCl–KCl eutectic with 10 wt % cryolite at 850°C,9 ini-
tially only a small fraction of the total 4 wt % of Al2S3 is dissolved
in the melt. As the initially dissolved Al2S3 is reduced, more Al2S3
may dissolve. The dissolution of Al2S3 in the electrolyte may be as
fast as the reduction of Al2S3. However, the total time of the experi-
ment is about 90 min, which is substantially lower than the equilib-
rium time of 5 h derived from solubility experiments.9 Therefore it
is possible that the reduction process is faster than the dissolution
process, and the melt becomes gradually depleted of Al2S3. This
may explain the gradual reduction of the current density in Fig. 7. If
this happens, the conditions gradually become less favorable as the
experiment proceeds, and Al2S3 may have not dissolved into the
electrolyte at the end of the experiment, as indicated by the low
Al2S3 conversion of 42% in the experiment.
In addition, due to the use of a graphite cathode, which is not
wetted by aluminum metal, the formed aluminum droplets are dis-
persed in the electrolyte. This combined with the sulfur gas evolu-
tion at the anode can result in a convection flow inside the electro-
lyte whereby the aluminum droplets come into contact with sulfur
and form aluminum sulfide again. Besides, the possible codeposition
of sodium and potassium and dissolution of reduced metal in the
electrolyte may also lower the current efficiency. Using a liquid
aluminum pool as a cathode can avoid this problem.
Figure 8. ͑Color online͒ Schematic illustration of mass and current flow in
the electrowinning process.
increased, by at least a factor of four. Secondly, the current density
at a given cell potential increases by about a factor of three. These
effects may be attributed to different reasons.
The increase in limiting current density can be caused mainly by
the increased solubility of Al2S3. It can also be argued that the
amount of fluoride contributes to the positive effect on the electrow-
inning process, resulting from a higher activity of AlFmn − than AlS33−
species. When complexing of Al with F is more preferred, the con-
centration of S ions would be higher, thus favoring the anodic reac-
tion.
As long as the j-E relationship is almost linear, its slope is de-
termined by an ohmic drop, rather than a depletion of ionic species.
Thus, the increase of the slope of the j-E curve should be ascribed
merely to a reduction of the resistivity between anode and cathode.
The slope increases by almost a factor of three on the addition of
cryolite to the MgCl2–NaCl–KCl eutectic, which is much more than
expected only from the increased melt conductivity. A hypothetical
explanation is that upon the addition of cryolite, a complex ion is
formed, changing interfacial tension at the anode. This may result in
different characteristics of the sulfur bubble layer at the anode and
reduce the ohmic drop. Further research is needed to understand
these phenomena completely.
It must be kept in mind that the cell geometry differs substan-
tially from the industrial design and should be regarded as less fa-
vorable. Using an aluminum pool as a cathode would have counter-
acted the formation of fine aluminum droplets in the molten salt.
Furthermore, the presence of high electrical current densities affects
the flow pattern in an industrial cell, which in turn influences the
behavior of gas bubbles and the stability of the metal/salt interface.
As a consequence, current efficiencies derived from small-scale ex-
periments should be treated with some caution.
Current density.— The addition of cryolite in the electrolyte has
two distinct positive effects on the electrochemical behavior of the
cell, as seen in Fig. 3. First, the limiting current density is markedly
In general, the cell potential consists of thermodynamic, kinetic
͑activation potential and mass-transfer limitations͒, and ohmic con-
tributions. Figure 8 shows the schematic diagram of the experimen-
Downloaded on 2015-03-13 to IP 169.230.243.252 address. Redistribution subject to ECS terms of use (see ecsdl.org/site/terms_use) unless CC License in place (see abstract).