Recycling kraft pulping chemicals with molten salt electrolysis

Article abstract of DOI:10.1149/1.1495907

Molten salt mixtures containing sodium carbonate, sodium sulfide, and sodium sulfate have been electrolyzed to generate sodium oxide and sodium sulfide while removing carbon from the system in the form of gas and maintaining a sulfur balance in the melt. This investigation leads toward the development of an electrolysis-based recycle process for pulping chemicals. The molten salts are presently found in the chemical recovery process of kraft pulping. Electrolysis was performed in a divided melt/undivided atmosphere and divided melt/divided atmosphere to avoid consumption of the oxide and sulfide products. The anodic reaction was carbonate oxidation to carbon dioxide and oxygen while sulfide oxidation to polysulfides occurred to a lesser extent at less positive potentials; sulfate oxidation was not observed to occur. The cathodic reaction was sulfate reduction to sulfide and oxide, the desired molten precursors for recycled pulping liquors.

Full text of DOI:10.1149/1.1495907

Journal of The Electrochemical Society, 149 ͑9͒ D125-D131 ͑2002͒
D125
0
013-4651/2002/149͑9͒/D125/7/$7.00 © The Electrochemical Society, Inc.
Recycling Kraft Pulping Chemicals with Molten Salt
Electrolysis
a, ,z
a,
b
Ryan Wartena, * Jack Winnick, ** and Peter H. Pfromm
a
School of Chemical Engineering, Georgia Institute of Technology, Atlanta, Georgia USA 30332
Institute of Paper Science and Technology, Atlanta, Georgia USA 30318
b
Molten salt mixtures containing sodium carbonate, sodium sulfide, and sodium sulfate have been electrolyzed to generate sodium
oxide and sodium sulfide while removing carbon from the system in the form of gas and maintaining a sulfur balance in the melt.
This investigation leads toward the development of an electrolysis-based recycle process for pulping chemicals. The molten salts
are presently found in the chemical recovery process of kraft pulping. Electrolysis was performed in a divided melt/undivided
atmosphere and divided melt/divided atmosphere to avoid consumption of the oxide and sulfide products. The anodic reaction was
carbonate oxidation to carbon dioxide and oxygen while sulfide oxidation to polysulfides occurred to a lesser extent at less positive
potentials; sulfate oxidation was not observed to occur. The cathodic reaction was sulfate reduction to sulfide and oxide, the
desired molten precursors for recycled pulping liquors.
©
2002 The Electrochemical Society. ͓DOI: 10.1149/1.1495907͔ All rights reserved.
Manuscript submitted October 12, 2001; revised manuscript received March 4, 2002. Available electronically July 16, 2002.
The Kraft Pulping Process and Chemical Recycle
͑Fig. 1A͒ after the molten salts ͑kraft melt͒ are dissolved in water.
This process proceeds with a chemical equilibrium-limited ion trans-
fer involving calcium hydroxide ͑from lime͒. In the causticizing
reaction the hydroxide replaces the carbonate and forms sodium
hydroxide and a precipitate of calcium carbonate
In the United States more than 50 million tons of kraft pulp are
1
produced every year. Regeneration of active pulping chemicals,
sodium hydroxide and sodium sulfide, from spent chemicals ͑mainly
sodium carbonate and sodium sulfate͒ is one of the largest inorganic
chemical processes in the world; in the U.S. alone 25 million tons of
the molten pulping salts, also known as kraft melt, are recovered
annually. The development of an electrochemical process is moti-
vated by the ability to supply incremental production capacity to
pulp mills ͑both through reduced deadload and incremental capacity
increases for chemical recovery͒, improved energy efficiency, re-
duced introduction of contaminants into the process, and simplified
operation ͑shortened holdup time, elimination of multiple
equilibrium-limited chemical reaction/separation steps, and elimina-
tion of the cumbersome lime cycle involving a kiln͒. The purpose of
this work has been to verify the electrochemical reactions occurring
in a simulated molten kraft melt ͑mixtures of Na CO , Na S,
Ca͑OH͒ ͑aq͒ ϩ Na CO ͑aq͒ → 2NaOH͑aq͒ ϩ CaCO ͑s͒
2
2
3
3
͓
1͔
The precipitated calcium carbonate is separated from the solution
and then re-burned to calcium oxide in a lime kiln
CaCO ϩ heat → CaO ϩ CO
͓2͔
3
2
The needed calcium hydroxide ͑Eq. 1͒ is formed by slaking, thereby
closing the lime cycle
2
3
2
Na SO ͒ to develop an alternative process to the traditional
equilibrium-limited wet chemical method for recovery of pulping
chemicals.
CaO ϩ H O → Ca͑OH͒
͓3͔
2
4
2
2
2
The requirements in kraft chemical recovery cycle for recausticizing
are then: (i) remove the carbon from the pulping chemicals, (ii)
produce sodium hydroxide ͑or a precursor to sodium hydroxide͒,
and (iii) maintain a sulfur balance on the melt, preferably in the
form of sulfide.
In an attempt to alleviate the bottleneck of many pulp mills in-
duced by the lime cycle and reduce the energy consumption by
circumventing the necessity of the lime kiln and associated equip-
ment, we have developed a concept for a process using molten salt
electrochemistry.
Valuable cellulose fibers are liberated from wood by the kraft
process, using an aqueous mixture of sodium hydroxide and sodium
sulfide at an elevated temperature and pressure. A liquid to be re-
cycled ͑called black liquor͒ is separated from the fibers. Black liquor
is an aqueous mixture of sodium carbonate (Na CO ), sodium sul-
2
3
fate (Na SO ), sodium sulfide (Na S), other inorganics at lower
2
4
2
concentrations ͑chloride, metals, transition metals͒ and the dissolved
organics from the wood. Water is partially evaporated from the
black liquor, and the remaining solution is fed into the kraft recovery
3
furnace where it acts as a fuel ͑combustion of the organics͒. A
molten salt mixture recovered from the bottom of the recovery
boiler contains mainly sodium carbonate, sodium sulfide, and some
sodium sulfate, the balance being mainly sodium/potassium chlo-
ride, metal and transition metal compounds, and carbon. This melt is
continuously recovered at 750-900°C by gravity drainage through
smelt spouts from the bottom of the recovery boiler where it is
immediately dissolved in water. The conversion of sodium sulfate to
sodium sulfide in the recovery boiler depends upon the extent of
reduction within the char bed at the bottom of the recovery boiler.
The molten salt composition of Na CO :Na S ϩ Na SO ranges
The Proposed Electrochemical Recycle Process
The method is a single-step electrolysis process, eliminating the
addition of inorganic chemicals ͑Fig. 1B͒. The proposed process
2
Ϫ
2Ϫ
2Ϫ
produces reduction products of O and S
while oxidizing CO3
to CO2 and O2 , thereby generating the molten precursors to white
liquor and removing carbon from the system in the form of carbon-
containing gas. This paper provides evidence of the desired electro-
chemical reactions occurring in the molten phase of recovered kraft
pulping chemicals. The off-gas from the electrochemical cell has
been analyzed for comparison with the coulometrically anticipated
generation of the hypothesized oxidation. Second, the postelectroly-
sis inorganics are dissolved into water after being cooled to room
temperature and analyzed for compositional change. The sodium
oxide produced during electrolysis reacts with water to form sodium
hydroxide
2
3
2
2
4
generally from 1:1 to 4:1 ͑mole basis͒.
The recausticizing process ͑meaning the process to re-make
caustic͒ has the goal of replacing carbonate ion with hydroxide ion
*
Electrochemical Society Student Member.
*
* Electrochemical Society Active Member.
E-mail: wartena@ccs.nrl.navy.mil
z
Na O ϩ H O → 2NaOH
͓4͔
2
2
Downloaded on 2015-03-14 to IP 131.211.208.19 address. Redistribution subject to ECS terms of use (see ecsdl.org/site/terms_use) unless CC License in place (see abstract).

D126
Journal of The Electrochemical Society, 149 ͑9͒ D125-D131 ͑2002͒
Figure 2. Cyclic voltammogram vs. carbonate reference electrode scanned
at 0.1 V/s of a pure sodium carbonate melt on gold electrodes at 860°C under
an argon atmosphere. The carbonate oxidizes at the positive limit of the
voltammogram while carbonate reduces at the negative limit. Intermediate
4
reactions of sodium reduction and oxide oxidation also occur.
6
Lorenz and Janz report that the reduction of carbonate to carbon
monoxide and oxide becomes thermodynamically favored around
8
70°C in a (Li/Na/K) CO melt. Our prior investigation of pure
2 3
sodium carbonate melts by cyclic voltammetry confirmed the occur-
rence of these reactions in addition to oxide oxidation.4
Cyclic voltammetry has also been conducted on mixtures similar
5
to the kraft melt. Figure 3 shows a cyclic voltammogram for a
mixture of sodium carbonate, sodium sulfide, sodium sulfate, and
sodium oxide. The oxidation is similar to the pure sodium carbonate
system where the discharge of carbonate occurs according to Eq. 5.
The reduction differs from the pure carbonate system ͑Fig. 2͒ as
determined by the smaller potential span between the limits of the
potential window. The span between the limits of the potential win-
dow indicates sulfate was the main reduction reactant ͑Fig. 3͒. The
reduction of sulfate occurs in many steps as suggested by Rapp and
Figure 1. ͑A͒ The conventional kraft chemical recovery cycle. ͑B͒ Kraft
chemical recovery via molten salt electrolysis ͑only the main chemicals and
reactions shown͒.
7
Goto
SO ϩ 2e → SO3 ϩ O2Ϫ
2
Ϫ
Ϫ
2Ϫ
E
0
ϭ Ϫ1.822 V
͓9͔
4
900°C
The results of the analysis are compared to the expected composi-
tion based upon the hypothesized electrochemistry and electrical
charge passed during the experiment.
Thermodynamic predictions and the literature both point to the
possibility of converting a molten mixture of sodium carbonate, so-
dium sulfide, and sodium sulfate into a mixture of sodium oxide and
sodium sulfide while simultaneously discharging carbon-containing
gases with electrochemistry. Prior investigations with cyclic volta-
mmetry have supported the hypothesis that used pulping chemicals,
in the molten phase, can be recycled ͑or recausticized͒ into molten
precursors of pulping solutions of mainly sodium hydroxide and
sodium sulfide.4 At the limit of the potential window for the posi-
tive branch of the cyclic voltammograms ͑Fig. 2 and 3͒, sodium
carbonate oxidizes to carbon dioxide and oxygen. In pure sodium
carbonate melts, sodium ions preferentially reduce to sodium metal
yet the limit of the potential window for the negative branch has
been determined to be carbonate reduction to carbon monoxide and
,5
4
oxide ͑Fig. 2͒
2
Ϫ
1
2
Ϫ
0
CO3 → CO ϩ
O ϩ 2e
E
ϭ 0 V
͓5͔
͓6͔
͓7͔
͓8͔
2
2
900°C
Figure 3. Cyclic voltammogram vs. Pt/PtS scanned at 0.1 V/s of a melt
containing sodium carbonate, sodium sulfide, sodium sulfate, and sodium
oxide on platinum electrodes at 820°C under an argon atmosphere. The car-
bonate oxidizes at the positive limit of the voltammogram while sulfate
ϩ
Ϫ
0
0
Na ϩ 1e → Na
E9
ϭ Ϫ2.19 V
00°C
CO 23 ϩ 4e → 3 O ϩ C
Ϫ
Ϫ
2Ϫ
E
0
ϭ Ϫ2.28 V
900°C
reduces at the negative limit. Intermediate reactions of sulfide and oxide
2
Ϫ
Ϫ
2Ϫ
0
CO3 ϩ 2e → 2 O ϩ CO
E
ϭ Ϫ2.625 V
oxidations also occur.5
900°C
Downloaded on 2015-03-14 to IP 131.211.208.19 address. Redistribution subject to ECS terms of use (see ecsdl.org/site/terms_use) unless CC License in place (see abstract).

Journal of The Electrochemical Society, 149 ͑9͒ D125-D131 ͑2002͒
D127
SO3 ϩ 2e → SO2 ϩ O2Ϫ
2
Ϫ
Ϫ
2Ϫ
͓10͔
2
Ϫ
Ϫ
1
2
0
2Ϫ
SO ϩ 2e
→
S ϩ 2 O
͓11͔
͓12͔
2
2
1
2
0
2
Ϫ
2Ϫ
0
S ϩ 2e → S
E
ϭ Ϫ0.701 V
900°C
The cyclic voltammogram in Fig. 3 has a span comparable to the
first step of the reduction series which is significantly less, by ca.
0.8V, than the span in the pure sodium carbonate melt. Since the
oxidations in the two systems are similar, the difference in span
supports sulfate as the limiting reduction over sodium or carbonate
reduction.
Intermediate reactions oxide and sulfide oxidation are also indi-
cated by the cyclic voltammogram in Fig. 35
O2
S2
Ϫ
Ϫ
→
→
1
2
O ϩ 2e
Ϫ
E
0
ϭ Ϫ0.888 V
ϭ Ϫ0.701 V
͓13͔
2
900°C
1
2
Ϫ
0
S2 ϩ 2e
E
͓14͔
͓15a͔
͓15b͔
͓16͔
900°C
o
2Ϫ
2Ϫ
xϩ1
xS ϩ S ↔ S
1
2
2
S ϩ Na S ↔ Na S K900°C ϭ 1.27 • 10
2
2
2
2
Figure 4. Configurations of electrochemical cells for divided melt-
undivided atmosphere ͑A͒ and divided melt-divided atmosphere ͑B͒.
2
2
Ϫ
Ϫ
2Ϫ
0
S
ϩ 2e � 2S
E
ϭ Ϫ0.988 V
900°C
In order to minimize these reactions, which act to decrease the cur-
rent efficiency of the desired recausticizing reactions ͑Eq. 9 through
12͒, physical separation of the cathodic products must be made from
the anode. Additionally, oxygen generated by Eq. 5 and 13 serve to
oxidize sulfide to sulfate, which is an undesirable process
gineering͒, and thermocouple. The fittings also provided electrical
isolation of the electrodes from the reactor vessel. The cell was
operated in a two- or three-electrode configuration. Two arrange-
ments of the electrochemical cell were used. The first provided di-
vision between the anolyte and catholyte, but did not separate their
atmospheres ͑Fig. 4A͒. This was accomplished by drilling a hole
2
8
Na S ϩ 2 O ↔ Na SO K900°C ϭ 1.29 • 10
͓17͔
2
2
2
4
Therefore, gas-phase separation of the individual electrode compart-
ments must also be investigated. It is also important to verify that
͑0.051 cm diam͒ on the sidewall near the bottom ͑Fig. 4A, a͒ of an
sulfate oxidation does not occur anodically as it does in pure sodium
sulfate melts8
alumina crucible ͑Fig. 4A, b; 3 cm outside diam, 4.1 cm outside
height, 0.2 cm wall thickness, CoorsTek, Golden, CO͒. This alumina
crucible served as the anolyte chamber and fits into another alumina
crucible of larger diameter ͑8.3 cm diam, 16 cm tall, 99.8% Al O ,
2
4
Ϫ
1
2
Ϫ
0
SO → SO ϩ
O ϩ 2e
2
E
ϭ ϩ1.203 V ͓18͔
3
900°C
2
3
which can be verified by off-gas analysis and by a sulfur balance of
the melt.
CoorsTek͒ serving as the other electrode chamber. The hole in the
side of the smaller crucible allowed for communication between the
two electrode compartments. The second configuration was devel-
oped to provide a division between both the melts and associated
atmospheres ͑Fig. 4B͒. A closed-end alumina tube ͑Fig. 4B, a; 61
cm in length and 2.54 cm diam͒ had a hole drilled in the end with a
diameter of 0.051 cm ͑Fig. 4B, b͒ and served as the anolyte cham-
ber. This tube allowed division not only of the molten salts, but also
of the gases evolved from each compartment. Individual argon
purges were passed over each compartment to facilitate removal of
gaseous products, as shown by the curved arrows in Fig. 4. The
electrode materials were all platinum wires ͑0.05 cm diam͒ ͑Alfa
Aesar, Ward Hill, MA͒ attached to the potentiostat through insulated
fittings in the reactor.
The gas and chemical analyses are evaluated for current effi-
ciency with respect to the reactions occurring at the limits of the
voltage window. Specifically, sulfate reduction, where the full reduc-
tion from Eq. 9 to 12 is assumed to occur, along with carbonate
oxidation
2
Ϫ
Ϫ
2Ϫ
ϩ 4 O2Ϫ
0
Cathode SO4 ϩ 8e → S
E
ϭ Ϫ1.663 V
900°C
͓19͔
The anode follows Eq. 5 and leads to the overall reaction
Na SO ϩ 4Na CO → Na S ϩ 4Na O ϩ 4CO
2
4
2
3
2
2
2
0
ϩ 2 O2
E
ϭ Ϫ1.663 V
900°C
Materials.—Sodium carbonate ͑anhydrous, granular, 99.5% as-
say, major impurity is sodium sulfate, VWR͒ and sodium sulfate
͓20͔
͑
anhydrous, granular, 99.5% assay, VWR͒ were dried for 24 h in an
These reactions need to be experimentally verified, as is now dis-
cussed.
oven at 115°C. Sodium sulfide ͑89.8% purity determined by titra-
tion͒ was obtained from Alfa Aesar where the impurities included
sodium sulfate, sulfur, and sodium polysulfides as indicated by the
slight yellow color of some of the reactants received. Sodium oxide
was also obtained from Alfa Aesar ͑manufacturer’s assay 85 wt %
Na O, 15 wt % Na O ͒. Industrial grade argon ͑Air Products͒
served as the purge gas. The molten salt mixture was contained in a
flat-bottomed cylindrical alumina crucible ͑99.8% Al O , 8.3 cm
diam, 16 cm tall, Coors Technical Ceramics, Golden, CO͒. A larger
diameter crucible ͑99.8% Al2O3 , 10.5 cm diam, 19.4 cm tall, Coors
Technical Ceramics͒ was used to contain the melt in the event of
crucible failure.
Experimental
Electrochemical cell.—A flanged Inconel reactor ͑20 cm inner
diameter, 76 cm height͒ was inserted into a top loading electric
furnace ͑model 56822, Lindberg, Watertown, WI͒. The temperature
of the bulk molten salts in an alumina crucible was controlled to
Ϯ5°C as monitored by a K-type thermocouple ͑Omega Engineer-
ing, Inc., Stamford, CT͒ in a 0.25 in. round-bottom alumina well
2
2
2
2
3
͑
Omega Engineering͒. Multiple ports on top of the reactor were
equipped with plastic compression fittings ͑Swagelok, Solon, OH͒
facilitating height adjustment of electrodes, purge port ͑Omega En-
Downloaded on 2015-03-14 to IP 131.211.208.19 address. Redistribution subject to ECS terms of use (see ecsdl.org/site/terms_use) unless CC License in place (see abstract).

D128
Journal of The Electrochemical Society, 149 ͑9͒ D125-D131 ͑2002͒
Reactor setup and heating.—Solid, granular sodium salts were
placed in the alumina crucible and tube. The electrodes were in-
serted into the top of the reactor and lowered above the melt. The
loaded alumina crucible was positioned in the bottom of the reactor,
the lower part of the reactor was raised into place, and the flange
was closed. The ceramic purge port was secured 20-30 cm above the
material in the crucible while the sheathed thermocouple was posi-
tioned 1 cm from the bottom of the melt crucible. The system
was purged with argon ͑0.5-1.2 LSTP /min for the outer system,
0
.05-0.15 LSTP /min for the tube͒ for 3 h before heating. The elec-
Ϫ1
trochemical cell was heated at approximately 100°C h to the op-
erating temperature (Ϯ5°C).
Electrolysis experiments.—During the electrolysis experiments
the following quantities were measured: time (Ϯ60 s), temperature
of the molten salt at a distance approximately 3 cm away from the
electrodes, the current and voltage applied to the cell, and the rel-
evant volume percentages of product gas. The number of coulombs
transferred during the experiment was determined by integrating the
current data (Ϯ0.003 A) with time. The IR-compensation of the
voltage was determined by current interrupt.
Figure 5. Gas evolution data as a function of time for an electrolysis experi-
ment of sodium carbonate, sodium sulfide, and sodium sulfate in a divided
melt and common atmosphere at 820°C.
tation, but suffered from sulfide oxidation by oxygen to produce
sulfate ͑Eq. 17͒ and less than 10% current efficiencies for oxide
production ͑Eq. 19͒. Cyclic voltammetry confirmed that the cathodic
products were oxidized at the anode. In the experiments described
below, deviations from stoichiometric current for the desired reac-
tions ͑Eq. 9 through 12͒ are explained by the participation of sulfide
and polysulfide in a current shuttle between the electrodes ͑Eq. 14
through 16͒, while oxidation of oxide resulted in oxygen gas ͑Eq.
Gas analysis.—Purging the electrochemical cell, either as a
whole or by compartment, assisted the removal of gases produced
during electrolysis. Both the inlet and the outlet flow rates were
measured; deviations between the two were significant due system
resistance. The outlet flow rate closest to the gas analyzers was used
to calculate the rate of gas being produced by the electrolysis.
Carbon monoxide (Ϯ0.005%) and carbon dioxide (Ϯ0.01%)
were measured in percent volume by an infrared analyzer ͑IR-702
gas analyzer, Infrared Industries Inc., Santa Barbara, CA͒. Oxygen
13͒ whose fate was to leave the system or oxidize sulfide to sulfate
͑Eq. 17͒. The IR-corrected cell potentials for the experiments were
(
Ϯ0.01%) was measured in percent volume by an electrochemical
typically 2.5 to 3 V.
cell analyzer ͑model 8000, Illinois Instruments, Johnsburg, IL͒. The
Sulfur species were not detected in the atmosphere by either gas
chromatography analysis of the vapor phase or ICP analysis of hy-
drogen peroxide solutions through which the effluent was bubbled.
This supports the proposition that sulfate oxidation ͑Eq. 18͒ was not
an electrochemical reaction occurring at the anodic potential limit.
Electrolysis of molten paper pulping chemicals ͑850-900°C͒ was
performed in melts where (i) the anolyte and catholyte molten salts
were separated but their atmospheres were not and (ii) both the melt
and the atmospheres were separated. Sodium carbonate, sodium sul-
fide, and sodium sulfate comprised the initial mixture for the system
without the divided atmosphere ͑Fig. 4A͒. To verify sulfate reduc-
tion to sulfide and oxide, the second experimental configuration was
utilized to divide the melt and the atmosphere while the cell was
loaded with carbonate and sulfate of sodium ͑Fig. 4B͒.
oxygen analyzer was prone to deviations at flow rates below 0.5 L
min . The analyzers were calibrated with gases of known concen-
Ϫ1
trations ͑Holox, Atlanta, GA͒ of the range being measured. The
outlet flow rate was measured with a bubble flow meter (Ϯ1 mL)
and a stopwatch (Ϯ0.01 s).
The stoichiometric rate of carbon dioxide and oxygen evolution
at any current was calculated according to Eq. 5 and Faraday’s law.
Both actual and stoichiometric rates of gas evolution are presented
on the same graph ͑Fig. 3 and 5͒, allowing a comparison of the
actual gas evolution to the theoretical rates based upon coulometry.
Postelectrolysis chemical analysis.—After the electrolysis ex-
periments the reactor was cooled and the inorganic chemicals were
dissolved in a known mass of deionized water. Standard analytical
techniques used in the pulp and paper industry were applied to de-
termine the chemical composition. Analysis was performed as soon
as possible to minimize atmospheric oxidizing and carbonating of
Divided melt and undivided atmosphere.—Figure 5 shows the
time variation of the measured carbon dioxide, carbon monoxide,
and oxygen for the duration of an electrolysis experiment at 820
Ϯ 10°C of a mixture containing sodium carbonate, sodium sulfide,
and sodium sulfate ͑1:0.25:0.25 mol ratio, column 1, Table I͒. A
total of 26,766 Ϯ 918 C were transferred, equating to a 15.4
Ϯ 0.5% ͑extent of conversion͒ conversion of carbonate loaded into
the cell based upon carbonate oxidation.
the solutions. A three-step acid titration based upon the standard
TAPPI method T-5869 ͑ABC titration,1 Mettler DL70ES titrator,
Schwerzenbach, Switzerland͒ determined the concentrations of so-
dium hydroxide, sodium sulfide, and sodium sulfate. Sodium hy-
droxide was formed from the reaction of sodium oxide with water
͑
Eq. 1͒; this assumption was used when reporting sodium oxide
In Fig. 5, gas evolution increased proportionally to the applied
current, which is also calculated as the expected rate of production
through the stoichiometry of Eq. 5. The gas production dropped off
after 23000 s. when the current was turned off. It began to rise again
at 25000 s. after the current was turned back on. This verifies that
gas production is due to the applied current. Although neither the
carbon dioxide nor oxygen reached the stoichiometric amount for
carbonate oxidation, the oxygen approaches stoichiometry closer
than carbon dioxide. This is interpreted as additional oxygen gen-
eration from oxidation of oxide. Although oxide was not initially
loaded into the melt, it appears in the anode chamber by thermal
decomposition of sodium carbonate
values from aqueous analysis. Capillary ion electrophoresis ͑CIE,
Waters capillary ion analyzer, Milford, MA͒10 determined the con-
centration of carbonate and sulfate in solution. Inductively coupled
plasma analysis ͑ICP, Perkin Elmer Optima 3000DV, Wellesley,
MA͒ determined the metal contents of the solution and provided
verification for the sulfur balance in the melt, as well as a check on
the ABC titration ͑ABC͒ and CIE. The analysis is reported in moles
for comparison to the initial moles loaded into the cell. All analyses
have an uncertainty of р10%.
1
0
Results and Discussion
Cell arrangement and design is important in the electrolysis ex-
periments. Initial investigations operated in an undivided cell orien-
Ϫ8
Na CO ↔ Na O ϩ CO K900°C ϭ 7.23 • 10
͓21͔
2
3
2
2
Downloaded on 2015-03-14 to IP 131.211.208.19 address. Redistribution subject to ECS terms of use (see ecsdl.org/site/terms_use) unless CC License in place (see abstract).

Journal of The Electrochemical Society, 149 ͑9͒ D125-D131 ͑2002͒
D129
Table I. Material balance for divided melt-undivided atmosphere experiment with initial components of Na CO , Na S, and Na SO .
2
3
2
2
4
Postelectrolysis analysis
Total
%
͓(Cat ϩ An)/
Expected] ϫ 100
Ϯ13.4%
Initial
Expecteda
Ϯ3.4%
Cat ϩ An
Ϯ10%b
Moles
Catholyte
Ϯ10%
Anolyte
Ϯ10%
ϮϽ1%
Method
Na CO
0.900
0.761
0.903
0.814
0.781
0.686
0.015
0.409
0.058
0.053
0.089
0.083
0.072
0.001
0.058
0.005
0.004
118.7
113.5
99.6
6.0
238.3
45.3
ABC
ABC
CIE
ABC
CIE
2
3
0
0
.864
.758
Na2S
0.232
0.231
-
0.267
0.196
0.139
0.016
0.467
0.063
Na SO
2
4
Na O
ABC
2
0.057
41.0
Na
S
2.726
2.726
2.962
108.7
93.3
107.3
103.0
109.5
90.3
ICP
ICP
ABC and CIE
ABC and CIE
ICP
ICP
ABC and CIE
ICP
2.543
2.924
2.808
c
0.463
0.463
0.507
0
0
.418
.482
104.1
Al
CO2
-
-
-
0.041
0.047
Ϯ0.004e
0.030
0.011
d
0.139
33.8
Ϯ11.9%
Trapezoidal
O2
-
-
0.069
-
0.030
Ϯ0.005
0.008
43.5
Ϯ20.1%
Trapezoidal
Trapezoidal
CO
-
Ϯ0.002
a
Based upon stoichiometry of carbonate oxidation and sulfate reduction and coulombs passed during experiment ͑26,766 Ϯ 918 C or Ϯ3.4%͒.
Unless otherwise noted, acid titration ͑ABC͒, capillary ion electrophoresis ͑CIE͒, and inductively coupled plasma ͑ICP͒ were reported to have an error
b
less than Ϯ10%.
c
d
e
Moles of metal are determined by considering NaOH, Na S, and Na CO from ABC acid titrations and Na SO from CIE.
Trapezoidal rule was used to integrate the rate of evolution (mol min ) with time.
Gas analysis error was propagated from the flow rate and smallest increment of the analyzer.
2
2
3
2
4
Ϫ1
or by inefficient separation of oxide produced by reduction ͑Eq. 19͒
in the cathode chamber. Carbon monoxide production is also coin-
cident with the passing of charge. This is attributed to the chemical
reduction of carbonate by sulfoxylate4
͑Eq. 19͒. Aluminum was found by ICP analysis and is attributed to
either sodium carbonate or sodium oxide fluxing with the alumina
oxide cell components
1
Na CO ϩ Al O ↔ 2NaAlO ϩ CO
K900°C ϭ 2.13 • 10
SO2 ϩ 2CO3 → SO4 ϩ 2 O2Ϫ ϩ 2CO
2
Ϫ
2Ϫ
2Ϫ
2
3
2
3
2
2
͓22͔
͓
23͔
which is an intermediate in the sulfate reduction sequence.
The postelectrolysis chemical analysis and integrated gas are
summarized in Table I; the method of analysis is indicated in the last
column. The second to last column of Table I provides the percent-
age of combined anolyte and catholyte moles present to the expected
moles based upon stoichiometry of Eq. 20, charge passed and as-
suming a mass balance of sodium and sulfur in the melt. Since
chemical separation between the anolyte and catholyte is not ex-
pected to be completely efficient, the combined moles found in each
compartment are compared. The postelectrolysis sodium carbonate
determined by ABC titration and CIE range from indicating no oxi-
dation of carbonate to stoichiometric conversion. From the carbon
dioxide gas analysis data that evolves proportionally to the applied
current, it is inferred carbonate oxidation ͑Eq. 5͒ does occur since
carbonate is the only source of carbon in the system. Therefore, the
carbon dioxide evolved determines the current efficiency for carbon-
ate oxidation, which is 33.8%. When sulfoxylate reduction of car-
bonate ͑Eq. 22͒ is considered by accounting for the carbon monox-
ide, the carbon removal efficiency is 39.6%. The remaining
oxidative current is attributed to a combination of oxide oxidation
8
Na O ϩ Al O ↔ 2NaAlO
K900°C ϭ 8.04 • 10
͓24͔
2
2
3
2
Three times as much aluminum was found in the catholyte as in the
anolyte. With the consideration that 75% of the aluminum found
was due to sodium oxide fluxing ͑Eq. 24͒, the current efficiency for
sodium oxide production is 54 Ϯ 2%. Both sodium and sulfur are
maintained in the melt, as shown by ICP analysis and confirmed by
combined consideration of the ABC titrations ͑for sulfide͒ and CIE
analysis ͑for sulfate͒. This supports the fact that sulfur-containing
gases are not produced during electrolysis.
Figure 6 summarizes the electrochemical reactions, and subse-
quent chemical reactions, occurring in a melt-separated cell of mix-
tures containing Na CO , Na SO , Na S. When the atmospheres of
the electrode chambers are not separated, oxygen generated from the
anode oxidizes the sulfide to sulfate ions in both electrolytes. Addi-
tionally, the sulfide ions undergo electrochemical oxidation to form
sulfur that chemically reacts with more sulfide to produce polysul-
fide ions.
2
3
2
4
2
͑
Eq. 13͒ and sulfide/polysulfide shuttling between the anode and
cathode ͑Eq. 14 through 16͒. A majority of the sulfide was oxidized
to sulfate ͑Eq. 17͒ during the experiment by oxygen generated at the
anode ͑Eq. 5 and 13͒ and due to atmospheric oxidation during dis-
solving the chemicals and before analysis. Sodium oxide was pro-
duced at 41 to 45.3% current efficiency for sodium sulfate reduction
Divided melt and atmosphere.—With a cell configuration where
the atmospheres are not separated it was not possible to decouple the
individual electrode contributions to the evolved gas. Additionally,
sulfide reduction products will be oxidized by anodic oxygen ͑Eq.
17͒. Therefore we separate both the melt and the atmosphere of
Downloaded on 2015-03-14 to IP 131.211.208.19 address. Redistribution subject to ECS terms of use (see ecsdl.org/site/terms_use) unless CC License in place (see abstract).

D130
Journal of The Electrochemical Society, 149 ͑9͒ D125-D131 ͑2002͒
Figure 7. Gas evolution data as a function of time for an electrolysis experi-
ment of sodium carbonate and sodium sulfate in a divided melt and atmo-
sphere at 840°C.
Figure 6. Electrochemical reactions occurring at the limits of the potential
window in mixtures containing sodium carbonate, sodium sulfide, and so-
dium sulfate at 840 Ϯ 20°C in a cell where the melt of the anolyte and
catholyte are separated, but the atmospheres are not.
This experiment differs from the experiment described in the previ-
ous section by division not only of the melt, but also of the atmo-
sphere.
In Fig. 6 carbon dioxide generation in the anolyte was again
proportional to the current, verifying electrochemical carbonate oxi-
dation ͑Eq. 5͒. As the current density was increased, the rate of
carbon dioxide evolution approached the stoichiometric value. Lo-
the electrodes in an electrolysis experiment of sodium carbonate and
sodium sulfate to determine individual gas contributions and to
minimize sulfide oxidation.
Figure 7 shows the time variation of the measured carbon diox-
ide and carbon monoxide of the anolyte and carbon monoxide of
the catholyte for the duration of an electrolysis experiment at 840
Ϯ 10°C of a mixture containing sodium carbonate and sodium sul-
fate ͑1:0.33 mol ratio, column 1, Table II͒. Carbon dioxide was not
detected in the catholyte off-gas. The total charge transferred was
6
renz and Janz have reported this effect at 560°C in pure carbonate
melts; as oxide oxidation is the major current consumer at lower
current densities while carbonate oxidation dominates at higher cur-
rent densities. The fluctuation in carbon dioxide is attributed to sul-
fide shuttling and polysulfide passivation ͑Eq. 14 through 16͒ of the
electrode surface as well as oxide oxidation ͑Eq. 13͒, which con-
sumes current. If carbonate were the electrochemically reduced spe-
cies at the cathode, carbon monoxide would be expected to evolve at
the rate of carbon dioxide generated by oxidation. More carbon
monoxide was detected in the catholyte than the anolyte; this is
44,015 Ϯ 1,564 C, equating to a 22.8 Ϯ 0.8% conversion of car-
bonate loaded into the cell, based upon carbonate oxidation ͑Eq. 5͒.
Table II. Material balance for divided melt-divided atmosphere experiment with initial components of Na CO and Na SO .
2
3
2
4
Postelectrolysis analysis
Total
%
͓(Cat ϩ An)/
Expected] ϫ 100
Ϯ13.6%
Initial
Expecteda
Ϯ3.6%
Cat ϩ An
Ϯ10%b
Moles
Catholyte
Ϯ10%
Anolyte
Ϯ10%
ϮϽ1%
Method
Na CO
0.998
0.770
0.763
0.707
0.681
0.013
0.056
0.055
0.022
99.1
95.6
61.4
112.5
39.5
93.8
ABC
ABC
ABC
Difference
ABC
2
3
0.736
Na2S
-
0.057
0.272
0.228
2.654
0.035
0.306
0.090
2.489
c
Na SO
0.329
-
2.654
2
4
Na O
0.090
2.272
0
2
Na
0.217
ICP
c,d
2
.388
90.0
103.6
ABC and CIE
ICP
S
Al
CO2
0.329
-
-
0.329
-
0.228
0.341
0.072
0.150
0.313
0.058
0.003
0.028
0.014
0.147
ICP
e
65.8
Trapezoidal
Ϯ0.014f
NA
0.026
Ϯ0.001
Ϯ0.013
Ϯ12.9%
O2
CO
-
-
0.114
-
0.019
0.007
Trapezoidal
Ϯ0.007
Ϯ0.005
Ϯ0.002
a
Based upon stoichiometry of carbonate oxidation and sulfate reduction and coulombs passed during experiment ͑44,015 Ϯ 1,564 C or Ϯ3.6%͒.
Unless otherwise noted, acid titration ͑ABC͒, capillary ion electrophoresis ͑CIE͒, and inductively coupled plasma ͑ICP͒ were reported to have an error
b
less than Ϯ10%.
c
Sodium sulfate by difference between the sulfur determined by ICP and sodium sulfide determined by ABC titration.
Moles of metal are determined by considering NaOH, Na S, and Na CO from ABC acid titrations and Na SO from CIE.
Trapezoidal rule was used to integrate the rate of evolution (mol min ) with time.
d
2
2
3
2
4
e
f
Ϫ1
Gas analysis error was propagated from the flow rate and smallest increment of the analyzer.
Downloaded on 2015-03-14 to IP 131.211.208.19 address. Redistribution subject to ECS terms of use (see ecsdl.org/site/terms_use) unless CC License in place (see abstract).

Journal of The Electrochemical Society, 149 ͑9͒ D125-D131 ͑2002͒
D131
fluxing ͑Eq. 24͒, the current efficiency for sodium oxide production
becomes 52%. Both sodium and sulfur are maintained in the melt as
indicated by ICP analysis and confirmed by combined consideration
of the ABC titrations and CIE analysis. Again, this supports the fact
that sulfur-containing gases are not produced during electrolysis.
Figure 8 summarizes the significant electrochemical reactions
occurring in a fully separated cell of mixtures containing Na CO3
2
and Na SO , as discussed in this section. Carbonate ions are oxi-
2
4
dized to carbon monoxide and oxygen while sulfate ions are reduced
to sulfide and oxide ions. With a cell setup where both the melt and
the atmosphere are separated, neither oxide nor sulfide will be elec-
trochemically oxidized nor will cathodically produced sulfide react
with anodic oxygen.
Conclusions
The electrolysis experiments in a divided melt at 820-840°C con-
firm the electrochemical reactions suggested by cyclic voltammo-
grams. Sulfate reduction to sulfide and oxide ͑Eq. 9-12͒ occurred
with a current efficiency of 50 to 60% and carbonate oxidation ͑Eq.
Figure 8. Electrochemical reactions occurring at the limits of the potential
window in mixtures containing sodium carbonate and sodium sulfate at
5͒ occurred with a current efficiency of 66%. When sulfoxylate re-
duction of carbonate ͑Eq. 22͒ is considered in the catholyte, the
840 Ϯ 20°C in a cell where both the melt and the atmospheres of the anolyte
carbon removal efficiency is 74%. Nearly stoichiometric carbon di-
and catholyte are separated.
Ϫ2
oxide was generated at current densities at and above 700 mA cm
.
Separating the melt creates an increase in current efficiency for ox-
ide production by a factor of five. Separating the atmosphere of the
anolyte and catholyte had the effect of reducing the amount of sul-
fide oxidation to sulfate by anodic oxygen ͑Eq. 17͒. Analysis of the
reactor off-gas did not indicate the presence of sulfur species, and
the melt was determined to maintain a balance of both sodium and
sulfur. These experiments show promise for the electrolytic recy-
cling of kraft pulping chemicals in the molten phase.
attributed to sulfoxylate chemical reduction of carbonate ͑Eq. 22͒ as
opposed to carbonate reduction by direct electrolysis. From the gas
analysis it can then be concluded that carbonate was oxidized and
sulfate was the major reduction reactant.
The postelectrolysis analysis of the melt and gas evolution is
summarized in Table II; it is of the same format as Table I. The
carbonate loss was again nearly stoichiometric, yet the carbon diox-
ide generated is considered to give a more reliable value for the
current efficiency of carbonate oxidation ͑Eq. 5͒ as 66.2%. This is
nearly double the efficiency described in the divided melt/undivided
atmosphere experiment and is attributed to operating at higher cur-
rent densities, where carbonate is oxidized preferentially to oxide.
When sulfoxylate reduction of carbonate ͑Eq. 22͒ is considered by
accounting for the moles of carbon monoxide from the catholyte, the
carbon removal efficiency is 74.1%. Again, the remaining oxidative
current is attributed to a combination of oxide oxidation ͑Eq. 13͒
and sulfide/polysulfide ͑Eq. 14 through 16͒ shuttling between the
anode and cathode. No sodium sulfide was added to the initial melt;
therefore the sodium sulfide determined by analysis resulted from
sulfate reduction ͑Eq. 19͒ although its distribution was divided be-
tween the catholyte and the anolyte. Considering the combined
anolyte and catholyte for sodium sulfide, the current efficiency for
sulfate reduction ͑Eq. 19͒ is 61.4%. This does not take into account
sulfide that was oxidized to sulfate ͑Eq. 17͒ which is indicated by
the approximately proportional increase in sulfate over the expected
value.
Acknowledgments
The authors thank the United States Department of Energy ͑DE-
FC07-97ID13547͒ and the member companies of the Institute of
Paper Science and Technology for their financial support.
Institute of Paper Science and Technology assisted in meeting the publi-
cation costs of this article.
References
1
. Statistics of Paper, Paperboard and Wood Pulp, Technical Report, American Forest
and Paper Association, Washington, DC ͑1994͒.
2. T. M. Grace, B. Leopold, and E. W. Malcolm, in Pulp and Paper Manufacture, Vol.
, M. J. Kocurek and F. Stevens, Editors, p. 637, Joint Textbook Committee of the
5
Paper Industry of the United States and Canada, Atlanta, GA ͑1991͒.
. T. N. Adams, W. J. Frederick, T. M. Grace, M. Hupa, K. Iisa, A. K. Jones, and H.
Tran, Kraft Recovery Boilers, p. 381, TAPPI Press, Atlanta, GA ͑1997͒.
3
4. R. Wartena, J. Winnick, and P. H. Pfromm, J. Appl. Electrochem., Accepted.
5
6
7
. R. Wartena, J. Winnick, and P. H. Pfromm, J. Appl. Electrochem., Accepted.
. P. K. Lorenz and G. J. Janz, Electrochim. Acta, 15, 1025 ͑1970͒.
. R. A. Rapp and K. S. Goto, in Proceedings of the 2nd International Symposium on
Molten Salts, J. Braunstein and J. R. Selman, Editors, pp. 159-177, MRS, Pitts-
burgh, PA ͑1978͒.
8
9
. C. A. C. Sequeira and M. G. Hocking, Electrochim. Acta, 23, 381 ͑1978͒.
. TAPPI Test Methods, Technical Association of the Pulp and Paper Industry, Atlanta,
GA ͑1987͒.
Sodium oxide was produced at 39.5% current efficiency accord-
ing to sodium sulfate reduction ͑Eq. 19͒. Aluminum was again found
in higher quantities in the catholyte than the anolyte. With the con-
sideration that 80% of the aluminum found was due to sodium oxide
1
0. D. A. Skoog and J. J. Leary, Principles of Instrumental Analysis, 4th ed., Harcourt
Brace College Publishing, Fort Worth, TX ͑1992͒.
Downloaded on 2015-03-14 to IP 131.211.208.19 address. Redistribution subject to ECS terms of use (see ecsdl.org/site/terms_use) unless CC License in place (see abstract).