Separation

IHPurification
Technology
Separation and Purification Technology 11( 1997) 159- 171

Enhancement of the current efficiency for sodium hydroxide

production from sodium sulphate in a two-compartment
bipolar membrane electrodialysis system
M. Paleologou *, A. Thibault, P-Y. Wong, R. Thompson, R.M. Berry
Paprican, 570 Boulevard St-Jean, Pointe-Claire, PQ, Canada H9R 3J9

Received 21 June 1996; accepted 28 February 1997

Abstract

Sodium sulphate, a by-product of chlorine dioxide generators and probably of most totally chlorine free (TCF)
pulp bleaching sequences, was split into sodium hydroxide and sulphuric acid by using a two-compartment bipolar
membrane electrodialysis (BME) system employing WSI4010/1030 bipolar membranes and Nafion417 cation-selective
membranes. The current efficiency was enhanced by using the buffering effect of divalent anions such as sulphate to
reduce the free hydrogen ion concentration in the feed compartment. In the optimized experiment, sodium hydroxide
(1.08 M at 46 ml min-‘) and sodium sesquisulphate (0.74 M at 68 ml min-‘) were produced at a current efficiency of
78% for the production of both acid and base. The power requirement was 1986 kWh t-’ of sodium hydroxide at a
current density of 142.9 mA cmm2 and with an effective membrane area of 728 cm’, Under the conditions of this
experiment, it was possible to extract 25% of the sodium content of sodium sulphate in the form of sodium hydroxide.
0 1997 Elsevier Science B.V.

Keywords: Bipolar membrane electrodialysis; Chlorine dioxide generators; Current efficiency; Product purity; Sodium
bisulphate; Sodium sesquisulphate; Sodium sulphate; Sulphuric acid

1. Introduction the inorganic load to the effluent treatment system

and avoid chemical imbalances in the recovery
The environmental pressures leading to further cycle [ 3-71.
regulation of a wide range of parameters and In most commercially available chlorine dioxide
priority pollutants [l] are increasing the interest generators, sodium chlorate and sulphuric acid are
in low effluent, closed-cycle pulp and paper mills. reacted to produce chloric acid and sodium sul-
The low-cost regeneration of presently sewered phate by-product. Chloric acid is reduced in the
chemicals can improve the economic attractiveness generator to chlorine dioxide using any one of
of the changes required [2]. In particular, the a number of reducing agents including meth-
regeneration of sewered sodium sulphate into anol (R8, SVP-LITE, Solvay), sulphur dioxide
sodium hydroxide and sulphuric acid can decrease (Mathieson), sodium chloride (R2, R3) and
hydrogen peroxide (SVP-HP, SVP HPA) [ 8,9].
* Corresponding author. Older generators are operated under atmospheric

1383-5866/97/$17.000 1997 Elsevier Science B.V. All rights reserved

PIZ S1383-5866(97)00018-X
160 M. Paleologou et al. / Separation and Pur$cation Technology 11 (I 997) 159-I 71

conditions (Solvay, Mathieson, SVP HPA) and inorganic by-product of most mills which employ
produce a highly acidic solution of sodium sul- totally chlorine free (TCF) bleaching sequences;
phate by-product, whereas modern generators sodium sulphate is, for example, a by-product of
operate under vacuum and produce a solid the generation of activated oxygen [21], and it will
by-product, sodium sesquisulphate, which is a salt also be formed from the neutralization of the
composed of sodium sulphate and sulphuric acid acidic effluent from ozone bleaching [ 221.
in a 3:l stoichiometric ratio (R8, SVP-LITE). In Bipolar membrane electrodialysis (BME) in a
general, the effluent from chlorine dioxide genera- three-compartment configuration was previously
tors is neutralized with sodium hydroxide and then proposed as one approach for extracting the
returned to the recovery cycle to maintain both sodium and sulphur values out of the sewered
the sodium and sulphur balances. With the tighten- portion of sodium sulphate from chlorine dioxide
ing of the recovery cycle with respect to sulphur generators [23,24]. Using the same BME configu-
losses, however, increasing amounts of sodium ration, the direct splitting of sodium chlorate into
sulphate have had to be sewered and their sodium sodium hydroxide and chloric acid [25] was also
value lost. In cases in which the chlorine dioxide reported. In order to reduce the capital and operat-
effluent is acidic, the excess sulphuric acid is usually ing costs associated with BME in the three-com-
neutralized with lime before being dumped. Since, partment configuration, a two-compartment BME
however, the solubility of calcium sulphate is very system was recently described for the production,
low, most of this salt ends up in the primary from sodium chlorate, of a pure sodium hydroxide
and/or secondary sludge thereby leading to solution and a solution comprising a mixture of
increased sludge disposal costs. Moreover, under chloric acid and residual sodium chlorate [26].
anaerobic conditions in the effluent treatment This configuration is useful in applications in
system, certain bacteria are known to reduce sul- which only one of the two components of a salt
phate to sulphite and/or sulphide [lo]; increased (i.e. acid or base) is required in pure form.
bacterial growth can lead, in turn, to the clogging A disadvantage of the two-compartment BME
of filters, while high sulphide levels can lead to configuration [2628] as well as other two-com-
corrosion problems [ 11,121 and/or an increase in partment electromembrane systems (e.g. electro-
the toxicity of the treated effluent [ 13-171. lytic systems [29-311) is the generation of acid
Sodium sulphate also concentrates in the electro- within the feed compartment. The concentration
static precipitator (ESP) catch of the chemical of acid in this compartment leads to a reduction
recovery boiler and is generally returned to the in current efficiencies for both the production of
recovery cycle. For coastal mills which use logs base and acid. For example, the steady-state cur-
floated in sea water, chloride concentrates with the rent efficiency for the production of sodium
sulphate in the ESP catch and part of this has to hydroxide from sodium chlorate was 40% for the
be sewered or leached out to limit chloride build-up two-compartment configuration [ 261 compared
in the recovery cycle [ 18,191. Build-up of chloride with 60% in the three-compartment configuration
and potassium in the black liquor can contribute under similar conditions [25]. This difference
to deposit formation and fire-side corrosion of the between the two configurations is caused by the
recovery boiler superheater tubes [20]. This sew- competing transport of Naf and H+ ions across
ered discharge containing sodium sulphate could the cation-selective membranes, an effect which is
become more common in the future, if chloride- magnified in the two-compartment configuration.
containing bleach-plant effluents are taken back to The free H+ ion concentration in the salt/acid
the recovery cycle. In those mills practising closed- loop is generally much higher than that in the salt
cycle operation while using chlorine-containing loop of the three-compartment configuration [26].
bleaching agents, chloride levels would increase The magnitude of the difference is determined by
throughout the liquor cycle and particularly in the the nature of the membrane, the [free H+]/[Na+]
ESP catch. ratio in the salt or the salt/acid compartments of
Sodium sulphate is also expected to be the major the two configurations, and the relative mobilities
 M. Paleologou et al. / Separation and Pur$eation Technology 11 (1997) 159-I 71 161

of the H+ and Nat ions. Since the free H+ ions 2.1. Two-compartment conjiguration
are much more mobile than the Na+ ions, a large
drop in current efficiency occurs as the free H+ Fig. 1 shows a unit BME cell in a two-compart-
concentration in the salt or the salt/acid loop ment configuration. In this instance, only bipolar
increases. and cation-selective membranes were used. The
The current efficiency in a two-compartment unit cell comprised two compartments, namely
BME system can be increased by raising the con- salt/acid and base: the first of these compartments
centration of the sodium ion in the salt/acid loop was bounded by the cation-selective side of a
by either increasing the concentration of the bipolar membrane and the cation-selective mem-
sodium salt in the feed solution or by adding brane; the second compartment was bounded by
another compatible sodium salt. This procedure the cation-selective membrane and the anion-selec-
decreases the [free H ‘]/[Na+] ratio in the salt/acid tive side of a second bipolar membrane. As is the
compartment. The same effect can also be achieved case with the three-compartment configuration
by adding to the salt/acid compartment a weak [23], as many as 200 unit cells can be stacked in
base that does not interfere with the application between the two electrodes; this means that a
under examination. The base binds to the free single Faraday of electricity does a significantly
Hf ions thereby decreasing the [free H+]/[Na+] greater amount of useful work than in an equiva-
lent electrolysis system in travelling from the anode
ratio.
to the cathode and before being dissipated, in the
The objective of this paper is to evaluate the
production of gases, at the two electrodes. The
buffering effect of divalent anions such as sulphate
results of this study were obtained using a BME
in improving current efficiency in a two-compart-
cell stack comprising eight unit cells incorporating
ment BME system used for the splitting of sodium
WSI4010/1030 bipolar membranes and Nafion417
sulfate into sodium hydroxide and a mixture of
cation-selective membranes. The total effective
sulphuric acid and sodium sulphate. In particular,
membrane area was 728 cm’ while the electrode
this paper demonstrates how an experimental run
area was 91 cm2. The membranes were kept apart
in operating mode 1 (batch salt/acid compartment
by mesh-like Vexar-type spacers (thickness =
and feed-and-bleed base compartment) can be used 1.35 mm) made of Kraton and two polyethylene
to enhance the current efficiency in the industrially gaskets (0.15 x 2 mm). The distance between the
realistic case of operating mode 2 (feed-and-bleed membranes was, therefore, 1.65 mm. All flows
salt/acid and base compartments). The integration
of this technology into pulp mill operations is Na,SO,
subsequently discussed in terms of methods that + H,SO,
can further improve its economic viability. 4

2. Experimental

The experimental system used in this work is

described in detail elsewhere [23]. The reader is Na,SO,

also referred to that paper for a more detailed LEGEND:

- Bipolar mtmbnne
account of the general process guidelines, basic 8+ Cation-selectivemembrane
principles, terminology and technology of BME.
Fig. 1, A two-compartment bipolar membrane electrodialysis
The experimental conditions and membrane mate-
unit cell for generating a mixture of acid and salt (sulphuric
rials used for operating in the two-compartment acid and sodium sulphate) and base (sodium hydroxide) from
configuration are described below. salt (sodium sulphate).
162 M. Paleologou et al. / Separation and Purification Technology II (1997) 159-I 71

were directed from the bottom to the top of the 2.2. Operating mode 1: batch salt/acid
BME cell through the mesh-like spacers by appro- compartment andfeed-and-bleed base compartment
priately spaced ports and manifolds. In each spacer
there are two ports diagonally opposed. The mesh- Sodium sulphate was split in a two-compartment
like pattern of the Vexar spacers has been shown BME configuration where the base compartment
to give a very good mixing action as compared was run in the feed-and-bleed mode and the
with other commercially available spacers. It is salt/acid compartment was run in the batch mode.
believed that the good mixing action in the case In this mode of operation, a constant concen-
of these spacers results from a semi-helical flow tration of sodium hydroxide was continuously
path of solution in which first the surface of one being produced while the concentration of acid
membrane and then the surface of the opposite was increasing and the concentration of sodium
membrane is swept by rapidly moving solution. sulphate was decreasing in the salt/acid loop.
The BME stack was hydraulically connected to Table 1 shows the conditions for the operation
two separate holding tanks (salt/acid and base) of the cell stack. The salt/acid loop was initially
each one being part of a different recirculation filled with 8.0 1 of approximately 1.4 M sodium
loop. The BME stack was also hydraulically con- sulphate and 0.25 M sodium bisulphate, and the
nected to a further two tanks, both part of the base loop with 0.91 M sodium hydroxide. The base
same loop, containing electrode rinse solution. Use was initially primed with the indicated solution to
of a basic electrode rinse (10% NaOH w/w) maintain high conductivity at the start of the run.
allowed the cell to operate with a nickel anode The circulation rates were chosen such that the
and a 316 stainless steel cathode. The current pressure drops in the two compartments were
was held constant at 13 A (current density = roughly the same at 63 kPa to avoid bursting the
142.9 mA cm-‘) and the applied voltage was membranes. The feed solution for the base loop
allowed to vary in order to maintain constant was water. The desired feed rate of water into the
current in spite of changes in the conductivity and base was delivered by a solenoid-driven dia-
solution composition during the experiments. This phragm-type metering pump operating on PID
current density was chosen based on the economic control with a conductivity set point adjusted at
optimum defined by the operating costs associated 174 mS cm-‘. PID control was achieved using
with membrane area replacement (they decline Specifix software (0 1988 Interllution, Inc.) [32].
with increasing current density) and power require- Conductivity was continuously monitored in the
ments (they increase with increasing current salt/acid and base compartments using a Foxboro
density). 87 1EC-SPO electrodeless conductivity sensor
The acid and base concentrations were deter- interfaced to a Foxboro 872 conductivity transmit-
mined by titration. The sodium and sulphate con- ter. For the measurement of pH, a combination
centrations in samples taken from the salt/acid pH electrode interfaced to a Signet Compak8700
and from the base were determined using atomic pH transmitter was used. The initial temperature
absorption spectroscopy and ion chromatography, was 24°C and was allowed to rise as a result of
respectively. In order to evaluate the buffering heat produced by the system; it reached 34°C after
effect of sodium sulphate on the current efficiency, 10.5 hours of operation. Over the course of the
the salt/acid compartment of the two-compartment experiment, the voltage dropped from 2.6 to
BME system was first run in the batch mode, to 1.8 V unit cell -I. The voltage due to the electrode
allow the accumulation of acid in this compart- compartments ranged from 5.1 to 5.5 V.
ment, while the base compartment was run in the
feed-and-bleed mode. Based on the first experi- 2.3. Operating mode 2: feed-and-bleed salt/acid and
ment, conditions were chosen for running both the base compartments
salt/acid and base compartments in the feed-and-
bleed mode while maintaining high current Sodium sulphate was split in a two-compartment
efficiencies. configuration using the feed-and-bleed mode for
 hf. Paleologou et al. /Separation and Purijcation Technology 11 (1997) 159-I 71 163

Table 1
Operating conditions for the splitting of sodium sulphate (batch salt/acid and feed-and-bleed base mode)

Salt/acid Base

(Na$O,) (NaHSO,) (NaOH)

Initial concentration in tank (M) 1.4 0.25 0.91

Initial volume in tank (1) 8.0 9.34
Circulation rate (I min ‘) 3.60 3.26

both the salt/acid and base compartments. Using steady state, the voltage was constant at 2.2 V unit
this mode of operation, a constant concentration cell -I. The voltage due to the electrode compart-
of sodium hydroxide and sodium sesquisulphate ments was 5 V.
(1: 1 sodium sulphate, sodium bisulphate) were
continuously produced by feeding water into the
base recirculation tank and a sodium sulphate 3. Results and discussion
solution into the salt/acid tank, respectively. The
feed rates into the base and salt/acid compartments 3.1. Operating mode 1: batch saltlacid
were chosen to be such that the desired concen- compartment andfeed-and-bleed base compartment
trations of sodium hydroxide and sodium
sesquisulphate would be bled off from the system Fig. 2 shows the progress of the experiment
as overflows from the recirculation tanks. under operating mode 1. During the run of 10.5 h,
Table 2 shows the conditions used for the opera- the concentration of titratable H+ in the salt/acid
tion of the cell stack. The experiment was started loop gradually increased from 0.25 to 2.6 M,
with the salt/acid loop containing 8.98 1 of 0.69 M whereas the concentration of base in the base loop
sodium sulphate and 0.73 M sodium bisulphate was maintained at 1.08 M using PID control of
and the base loop containing 1.0 M NaOH. The the water feed rate with conductivity
circulation rates were chosen so that the pressure (174 mS cm-‘) as the control variable.
drop in the two compartments was roughly 62 kPa. Fig. 3 shows the changes in conductivity and
The feed solution for the base loop was water, pH with increasing concentration of titratable
while 1.42 M sodium sulphate was fed into the H’ in the salt/acid compartment. The shape of
salt/acid loop. The feed rates into the base and these curves can be explained in terms of the
salt/acid compartments were 43 and 73 ml mini, various acid-base equilibria that occur in the
respectively. The initial temperature was 22°C and salt/acid compartment as affected by the increasing
was allowed to rise as a result of heat generated hydrogen ion and decreasing sodium ion concen-
by the system to a steady-state level of 35°C. At tration in this compartment. Before any current is

Table 2
Operating conditions for the splitting of sodium sulphate (feed-and-bleed mode)

Salt/acid Base

(Na,SO,) (NaHSO,) (NaOH)

Initial concentration in tank (M) 0.69 0.73 1.00

Initial volume in tank (I) 8.98 9.15
Circulation rate (1 min -‘) 3.60 3.26
Feed solution concentration (M) 1.42 Water
Feed rate (ml min ‘) 73.0 43.0
Actual feed rate (mmol min- ‘) 104
164 M. Paleologou et al. / Separation and Purijication Technology 11 (1997) 159-I 71

= 3Qr , 50 membrane within the salt/acid compartment, and

water feed rate these replace the sodium ions migrating to the
- 40 f
base compartment. A small portion of these
hydrogen ions are also used to neutralize the
2 hydroxide ions that back-diffuse into the salt/acid
- 30 d compartment from the base compartment. One
!!
OH-in Base might expect a steady increase in the conductivity
- 20 $
L
and a steady decrease in the pH of the salt/acid
compartment as sodium ions, which have an equiv-
- 10 3Em alent ionic conductivity of 50.11 cm’ R-’ equiv-’
at 25°C are replaced by the more highly mobile
i; - hydrogen ions which have an equivalent ionic
c 0 100 200 300 400 500 600 700
k conductivity of 349.82 cm2 R-i equiv-’ [33]. In
Time, min
the initial stages of the run, however, and despite
Fig. 2. Under operating mode 1, the base compartment was run the increase in the concentration of titratable H+
in the feed-and-bleed mode, whereas the salt/acid compartment (0 to 0.7 M), the conductivity increases only grad-
was run in the batch mode. Over the course of the 10.5 h run,
ually from 125 to 145 mS cm-‘, while the pH
the concentration of titratable acid in the salt/acid loop grad-
ually increased. The concentration of sodium hydroxide in the declines slightly from 1.6 to 1.5; beyond this point
base loop was maintained steady using PID control with con- the conductivity and pH change quite sharply to
ductivity used as the control variable. Over the course of the reach values of 230 mS cm- ’ and 1.O, respectively,
experiment the temperature rose from 24 to 34°C. at a titratable Hf concentration of 1.4 M. The
reason for the slow change in conductivity and pH
in the initial part of the run is that a portion of
the hydrogen ions produced by the bipolar mem-
brane in the salt/acid compartment combine with
sulphate ions to form bisulphate [3436]:
H+ + SOi-%HSO; K= 76.9 1mole-‘, (2)
where
K=[HSO,]/[H+][SO;-1.
This equilibrium reduces the concentration of free
hydrogen ions in solution by partially converting
Titratable H’ concentration, M
sulphate to bisulphate to the extent determined by
Fig. 3. Over the course of the run, the conductivity in the K. As a result of this equilibrium, a lower rise in
salt/acid compartment remains initially steady due to the buff- conductivity is observed at the lower acid concen-
ering effect of sodium sulphate. Once the buffering capacity of trations shown in Fig. 3. The curves of Fig. 3, can
sodium sulphate is reached, the conductivity increases with
be regarded as representing a conductometric and
increasing concentration of titratable acid and declining concen-
tration of sodium sulphate. a potentiometric titration in which hydrogen ions
are added as the titrant to a solution of sodium
sulphate while sodium ions are taken out of solu-
applied, the conductivity and pH of the solution tion:
are due to the complete dissociation of sodium
sulphate in solution: Na,SO, + H+ (added)*NaHSO,
Na,SO,+Na+ + SO:-. (1) + Na+ (removed). (3)
Once the current is applied across the membrane The equivalence points occur when all the sodium
stack, hydrogen ions are produced by the bipolar sulphate has been converted to sodium bisulphate
 M. Paleologou et al. / Separation and PuriJication Technology 11 (1997) 159-l 71 165

(when the titratable H+ concentration is about of the run from 43.3 ml min-’ at 90min to
1.4 mole 1-i in Fig. 3). These equivalence points 15.6 ml min-’ at 390 min after the start of the run
are not very sharp because the equilibrium of (see column 3). This is due to the decreasing
Eq. (2) is gradually shifted to the left as a result current efficiency of sodium ions crossing over to
of, initially, the increasing ratio of bisulphate to the base compartment as a result of sodium deple-
sulphate in the salt/acid compartment and, subse- tion in the salt/acid compartment (see column 4)
quently, as a result of the movement of hydrogen and acid accumulation. Some sulphate ions do
ions to the base compartment, in place of sodium accumulate in the base loop (1.97 g 1-i by the end
ions, in response to the electric field gradient. of the run). The presence of these ions in this loop
At the midpoint between the beginning of the is due to the limited selectivity of the cation-
titration and the equivalence point, the concen- selective component of the bipolar membrane
tration of sodium bisulphate in the solution is [23,26]. The levels, however, are so low that they
equal to that of sodium sulphate. If all the water are only of concern in cases in which high purity
were to be evaporated from this solution, the salt base is required. The last two columns of Table 3
obtained would be sodium sesquisulphate: indicate that the total sodium content in the two
NaHSO, + Na,SO,$Na,H (SO& compartments remains constant throughout the
run (values are within ~3% of each other); this is
(or Na,SO,.NaHSO,) (4) also the case with the sulphate content in the two
Fig. 3 shows that as long as the solution main- loops. A mass balance was thus maintained
tains a stoichiometric excess of sodium sulphate throughout the run.
over sodium bisulphate (approximately between 0 All the changes happening in the salt/acid com-
and 0.5 mole 1-l titratable H+), the hydrolysis of partment can be represented as shown in Fig. 4.
sodium bisulphate is significantly suppressed and In this figure, the concentrations of sodium sul-
the changes in conductivity and pH in the solution phate, sodium bisulphate and sulphuric acid were
of the salt/acid compartment are small. calculated based on the analytical data appearing
Table 3 presents the concentrations of Na+ and in Table 3 for sodium and sulphate as well as the
SO:-, as determined by atomic absorption and acid concentration in the feed compartment at
ion chromatography, respectively, in the salt/acid various times during the experiment. For simplic-
and base loops. The concentration of sodium ions ity, given the large value of the equilibrium con-
in the base loop remains constant since this loop stant for Eq. (2), it was assumed that the
was operated under PID control. It is interesting equilibrium is shifted completely to the right. For
to see, however, the gradual decline in the bleed the first 3.5 h of the run, the concentration of
rate from the base compartment over the course sodium bisulphate in the salt/acid loop gradually

Table 3
Ion concentration and mass balance for BME of sodium sulphate (batch salt/acid and feed-and-bleed base mode)

Time Loop volume Bleed rate Na+ concentration (g 1-l) SO:- concentration (g l-‘) Total (g)
(min) (1) (mlmin-‘) ~ ~ ~

Salt/acid Base Salt/acid Base loop Base bleed Salt/acid Base loop Base bleed Na’ so:-

30 7.92 27.3 55.0 23.5 23.5 139 1.11 1.11 657 1120
90 7.59 43.3 47.7 24.8 24.7 136 1.17 1.17 668 1060
150 7.22 44.3 41.8 23.8 23.1 141 1.18 1.18 659 1060
210 6.98 31.3 36.4 24.7 24.5 145 1.33 1.34 668 1060
270 6.69 22.3 32.0 24.1 23.9 151 1.51 1.52 662 1070
330 6.52 18.6 27.7 25.3 25.0 156 1.72 1.73 673 1090
390 6.23 15.6 25.1 25.5 25.2 151 1.97 1.99 675 1030
166 M. Paleologou et al. / Separation and Purijication Technology II (1997) 159-I 71

membrane in

. .
” 01, I ,

0 100 200 300 400 500 600 7 0 1 2

Time, minutes Titratable H+ concentration, M
Fig. 4. The changes happening in the salt/acid compartment are Fig. 5. The current efficiency for the production of sodium
shown in this figure. For the first 3.5 h of the run, the concen- hydroxide remains largely unaffected up to a concentration of
tration of sodium bisulphate gradually increased while the con- titratable acid of 0.7 N; it then declines as a result of an
centration of sodium sulphate gradually decreased. In the last increased flux of free hydrogen ions competing with the sodium
7 h of the experiment, the concentration of sodium bisulphate ions in crossing over to the base compartment.
decreased while the concentration of sulphuric acid gradually
increased. The concentration of sodium hydroxide in the base
ions competing with the sodium ions in crossing
loop was maintained at 1.04 M.
over to the base compartment because of an
insufficient concentration of sulphate ions in solu-
increased from about 0.3 to 1.5 M while the con- tion to buffer the hydrogen ions produced by the
centration of sodium sulphate in this compartment bipolar membrane through the formation of bisul-
gradually decreased from approximately 1.2 M to phate. At the end of the experiment and at a
0. In the last 7 h of the experiment, the concen- concentration of titratable H+ of 2.6 M, the cur-
tration of sodium bisulphate in the salt/acid loop rent efficiency for sodium hydroxide production
decreased from 1.5 to 0.7 M, while the concen- declined to approximately 10%.
tration of sulphuric acid in this loop gradually In addition to the buffering effect of sulphate
increased from 0 to 1 M. The concentration of on current efficiency, however, there are some
sodium hydroxide in the base loop was maintained other aspects to be considered to explain the shape
at 1.04 M. The feed rate of water into the base of the curve of Fig. 5. The shape of this curve can
compartment was about 41 f4 ml min-’ for the be explained well by a model previously proposed
first 3 h of the run, but afterwards the rate dropped by Jorissen and Simmrock [29] for electrolytic
gradually to finally reach 7 ml min - ’ after 630 min systems. According to this model, at high sodium
of operation (see Fig. 2). hydroxide concentrations in the base compartment
As shown in Fig. 5, the equilibria of Eqs. (l)-(4) and low concentrations of acid in the feed compart-
have a direct impact on the current efficiency for ment (salt/acid in this case), losses in current
the production of sodium hydroxide. Current effi- efficiency are caused by the back-diffusion of
ciency is defined as the proportion of the current OH- ions into the salt/acid compartment. In this
that is being consumed in useful work (to make case, the cation-selective membrane is in the alka-
the desired products, e.g. acid and base). line state through its entire thickness and an alka-
Operationally, it is defined to be the ratio of the line boundary layer is developed on the salt/acid
actual over the theoretical production rate (based side of the cation-selective membrane [see
on Faraday’s laws) [23]. The current efficiency for Fig. 6(a)]. Under these circumstances, the acid
the production of sodium hydroxide remains concentration in the salt/acid compartment has no
largely unaffected (78%) up to a concentration of influence on the current efficiency (initial section
titratable H+ of 0.7 M. The efficiency then declines of the curve of Fig. 5) because the H+ ions have
as a result of an increased flux of free hydrogen already been neutralized in the alkaline boundary
 AL Paleologou et al. / Separation and Purification Technology 11 (1997) 159-l 71 161

alkaline state acidic state the acid boundary layer on the base side and do
not affect the behaviour of the membrane itself
[see Fig. 6(b)].
According to this model, there exists a definite
acid concentration in the salt/acid compartment,
for each sodium hydroxide concentration in the
base compartment, at which the cation-selective
membrane shifts from the alkaline state into an
acid state or vice versa (about 0.7 M titratable
HC for this experiment). This model explains the
obvious change in gradient of the current efficiency
Cation-Exchange function with increasing titratable H+ concen-
(a) tration as a change from the alkaline state to the
Fig. 6. Model of the alkaline (a) and acid (b) states of the acid state of the cation-selective membrane (see
cation-selective membrane in a two-compartment BME system. break point in Fig. 5).
When the membrane is in the alkaline state, the current effi- For as long as the membrane is in the alkaline
ciency is dependent on the caustic concentration in the base
compartment and independent of the acid concentration in the
state, the current efficiency for sodium ion trans-
salt/acid compartment. When the membrane is in the acid state, port through the Nafion417 cation-selective mem-
the current efficiency is dependent on the [free H+]/[Na+] in brane is independent of the concentration ratio of
the salt/acid compartment, but independent of the sodium [free H+]/[Na+] in the salt/acid compartment;
hydroxide concentration in the base compartment [29]. once, however, the membrane shifts into the acid
state, the current efficiency steadily declines with
layer and cannot reach the cation-selective mem- increasing [free H+]/[Na+] ratio. At all times the
brane itself. According to this model, if membrane current efficiency is independent of the individual
type, current density and temperature are fixed, as concentrations of sulphuric acid and sodium sul-
long as the cation-selective membrane is in the phate. When the cation-selective membrane is in
alkaline state, the current efficiency depends on the acid state, and assuming that the membrane
the sodium hydroxide concentration in the base has no special selectivity for Na+ over H+, the
compartment and is not at all affected by the acid current efficiency for sodium ion transport to the
concentration in the salt/acid compartment. base compartment is expected to be proportional
At the concentration of sodium hydroxide main- to the transport number of sodium ion through
tained in the base compartment (about 1.04 M) of the cation-selective membrane. The transport
this experiment, it appears that the back-diffusion number is defined as the portion of the total
of hydroxide ions into the salt/acid compartment electric current through a medium that is carried
leads to a constant loss in current efficiency of by a particular ion [ 371. The reader is referred to
about 20%. This situation changes only when, as Ref. [23] for a more comprehensive account of the
a result of an increase in the acid concentration in impact of sodium hydroxide concentration in the
the salt/acid compartment, more Hf ions migrate base compartment and acid concentration in the
from the bipolar membrane to the boundary layer feed compartment on current efficiency.
of the cation-selective membrane than OH- ions
arrive through the cation-selective membrane from 3.2. Operating mode 2: jked-and-bleed salt/acid and
the base compartment. Under these circumstances, base compartments
the alkaline boundary layer disappears. In this
case the H+ ions migrate through the cation- Table 4 and Fig. 7 show the results from an
selective membrane to the base compartment and experiment run under operating mode 2. As shown
form an acid boundary layer. In this case, the base in Fig. 7, during the last 3.5 h of the 8 h run,
concentration has no influence on the current equilibrium was maintained and the desired steady-
efficiency because the OH- ions are neutralized in state concentrations of sodium hydroxide and
168 M. Paleologou et al. / Separation and Purl$cation Technology I1 (1997) 159-l 71

Table 4
Throughput rates and mass balance for BME of sodium sulphate (feed-and-bleed mode)

Salt/acid Base

(NazSOJ ( NaHSO,) (NaOH) (NaGSO,)

Bleed solution concentration (M) 0.71 0.74 1.09 0.02

Bleed rate (ml rain-‘) 68 68 46 46
Actual bleed rate (mm01 min-‘) 48.3 50.3 50.1 0.92
Net production rate (mm01 mini) -55.4 + 50.3 $50.1 +0.92
Expected production rate (mm01 min-‘) -39.1 +64.6 +64.6 0.0

1986 kWh t - ’ of sodium hydroxide as compared

with 3457 kWh t-l of sodium hydroxide in the
case of an equivalent three-compartment BME
system [23]. This difference in energy consumption
translates to a significant decrease in the payback
‘g 0.9 - period for a two-compartment BME system over
5 NaHSO, in Salt/Acid
co 0.8 - 4 an equivalent three-compartment BME system ( 1.9
versus 3.1 years) [24].
6 0.7 - A mass balance was maintained throughout the
0.6 b ’ ’ ’ ’ ’ ’ ’ ’ ’ run. The mass balance for Na+ and SOi- were
100 200 300 400 500 - 2.6 and -2.O%, respectively, without taking into
Time, minutes consideration the amounts in the rinse loop. These
Fig. 7. In the feed-and-bleed mode of operation, a constant results show that in the feed-and-bleed mode,
concentration of sodium hydroxide and sodium sesquisulphate current efficiencies of 78% for sodium hydroxide
was produced once steady-state operation was reached. By and sulphuric acid production can be maintained
operating the salt/acid compartment with a sodium sesquisul-
at a ratio of 1:1 NaHSO,:Na,SO, (concentration
phate composition, high current efficiencies were obtained for
the production of base. As a result of heat generated by the of 0.75 M sodium sesquisulphate) in the salt/acid
system, the steady-state temperature of operation was 35°C. compartment.

sodium sesquisulphate were produced. At steady 3.3. Integration into pulp mill operations
state, a solution of 1.09 M NaOH was bled from
the base tank at 46 ml min-’ and a solution A problem of operating in the two-compartment
approximating the composition of sodium BME configuration with a sodium sesquisulphate
sesquisulphate (0.71 M in sodium sulphate and steady-state composition in the salt/acid compart-
0.74 M in sodium bisulphate) was bled from the ment is that only a quarter of the saltcake can be
salt/acid tank at 68 ml min-‘. The contamination converted to sodium hydroxide at high current
of the base bleed with sodium sulphate was typi- efficiencies. This problem can be resolved by oper-
cally 2.2 mole% Na,SO,/NaOH. The presence of ating the two-compartment BME system in associ-
sodium sulphate in the base compartment was due ation with a technology that continuously removes
once again to the limited selectivity of the compo- the acid component of the sodium sulphate after
nent parts of the bipolar membranes. The current it has gone through the salt/acid compartment of
efficiencies for the production of sodium hydroxide the BME system. One such technology is the
and acid under these conditions were 78% in both generator acid purification (GAB) system [ 381. On
cases, which showed that high current efficiencies a net basis, the two-compartment BME-acid
can be obtained by operating at a nearly 1:1 ratio removal system combination can produce sulphu-
of Na,SO,:NaHSO,. The energy requirement was ric acid and sodium hydroxide from sodium sul-
 M. Paleologou et al. 1 Separation and Purification Technology 11 (1997) 159-l 71 169

phate. Compared to the three-compartment con- sodium sulphate-acid-water system. Fig. 8 [40]
figuration which can achieve the same outcome, presents the solubility of sodium sulphate at vari-
the two-compartment BME-acid removal system ous concentrations of sulphuric acid at temper-
combination produces acid and base with signifi- atures ranging from 0 to 82.5”C. It appears that
cantly lower capital and operating costs [24]. at 0°C the sodium sulphate solubility increases in
A further problem of operating the two-com- the O-2.5 M acid concentration range, while it
partment BME system without an acid removal declines quite sharply in the 2.5-4.0 M range; it
system is the limited uses of sodium sesquisulphate levels off in the 4.0-7.0 M range. Increasing the
at the mill site. Having an acid removal system temperature appears to expand the range of acid
operate in association with the BME system pro- concentrations for which a solubility increase is
vides the opportunity of producing relatively pure observed. To prevent possible damage to the mem-
sulphuric acid, which could be used in a number branes, it is recommended that the temperature of
of mill applications such as the dilution of concen- operation does not exceed 60°C [23]. According
trated sulphuric acid prior to addition in the to Fig. 8, at this temperature, a solution as high
chlorine dioxide generator, the acidulation of tall as 2.2 M in sodium sulphate can be prepared and
oils, the regeneration of ion-exchange columns assuming this solution is acidified to the sodium
used in the treatment of boiler feed water, the sesquisulphate composition, the solubility can be
production of magnesium sulphate from magne- increased to about 2.3 M (also 0.76 M in sulphuric
sium oxide, and the adjustment of pH prior to acid). On the other hand, however, the equilibrium
ECF or TCF bleaching or for pitch control. constant K for the reaction represented in Eq. (2)
A key parameter which has a direct impact on decreases with temperature (e.g. K=41 at 55°C).
the economics of BME systems is membrane life; This is expected to have a negative impact on the
as we have previously shown, membrane life of at buffering capacity of sodium sulphate.
least 10 months is required for the two-compart- An increase in temperature will also bring about
ment configuration to be economically viable (pay- an increase in current efficiency because at high
back period of three years) [24]. Membrane temperatures the ratio of equivalent ionic conduc-
durability for the splitting of sodium sulphate was tivities of [free H+]/[Na+] decreases. For example,
tested in a laboratory two-compartment BME cell at 25°C this ratio is 350/50.9 (=6.88) whereas at
under extreme conditions and was found to 50°C this ratio becomes 465/82 (=5.67) [41].
exceed 1 year. Higher temperatures will have additional benefits
One disadvantage of operating at the high cur- in terms of lowering the ohmic resistance of the
rent efficiency region is that the cation-selective
membrane is in the alkaline state. In this state,
extremely high purity of the feed is required as ‘0 3.5
was previously shown from chloralkali electrolysis .z 3.0
[ 391. These requirements can be achieved only by
means of expensive treatments with specific ion ; 2.5
0
exchange resins. However, by operating with the 8 2.0
cation-selective membrane in the acid state, with Y
just a little loss in current efficiency (see Fig. 5), +- 1.5

the expensive treatments can be avoided and the ZN 1.0

membrane life prolonged. The technical feasibility
of this idea was previously demonstrated in experi- z 0.5
ments with a sodium chloride/hydrochloric acid
feed solution to the anode compartment of a two- 1 2 3 4 5 6
mole HaSO, I 1000 g solution
compartment electrolysis system [29].
The buffering capacity of sodium sulphate is Fig. 8. The solubility in the Na,SOpH,SO,--H,O system at
expected to be limited by its solubility in a given various temperatures [40].
170 M. Paleologou et al. / Separation and Pkjication Technology II (1997) 159-I 71

salt/acid and base compartments as well as enhanc- Pulp and Paper Industry: Current Technology and
ing K, (water-dissociation constant) which has a Research Needs, Pulp and Paper Research Institute of
Canada, MR297, 1994.
direct impact on the efficient operation of the [3] D. Pryke, G. Brown, Chlorine Dioxide Substitution and
bipolar membrane in the interface of which water pulp Mill Chemical Balances, CPPA Pacific and Western
is split into hydrogen and hydroxide ions. For Branches Jt Conf. Preprints, Whistler, BC, 1989.
example, whereas at 25°C K,= 1 x 10ei4, at 50°C [4] L. Lewko, A. van Donkelaar, The New Reality - Softwood
the value of K, increases to 5.5 x lo-l4 [41], which Bleaching: Chemical Balances, Costs and Environmental
Impacts, CPPA Pacific and Western Branches Jt Conf.
represents more than a two-fold increase in the
Preprints, Jasper, AB, 1990.
concentration of hydrogen and hydroxide ions in [ 5] P. Ulmgren, R. Lindstrom, The manufacture of sulfur-free
the interface of the bipolar membrane. alkali in a kraft pulp mill. Part 1. A new process for balanc-
ing sulfur and sodium in a closed cycle mill, Nordic Pulp
Pap. Res. J. 4 (1988) 191.
[6] G. Williams, G. Berndt, Maintaining the Caustic/Chlorine
4. Concluding remarks Balance in a Bleached Kraft Pulp Mill, CPPA Pacific and
Western Branch Conference Preprints, Jasper, AB, 1990.
In this paper a two-compartment BME system [7] K. Lohrberg, G. Klamp, Chlorine dioxide can be generated
was used to produce pure sodium hydroxide and without sulfur balance problems, Pulp Pap. 56( 11)
(1982) 114.
a mixture of sodium sulphate and sodium bisul-
[S] M. Fredette, Chlorine Dioxide Generators for Chemical
phate from sodium sulfate. The free acid concen- Balance, Tech ‘91, CPPA Kraft Pulp Manufacturing
tration in the salt/acid compartment was found to Course, Whistler, BC, 1991.
have a negative effect on sodium hydroxide current [9] P. Stockburger, What you need to know before buying
efficiency as a result of the competing transport your next chlorine dioxide plant, Tappi J. 76 (3) (1992) 99.
of Na+ and H+ ions across the cation-selective [lo] K. Hanel, Biological Treatment of Sewage by the Activated
Sludge Process, Ellis Horwood, Chichester, UK, 1988,
membranes. Nevertheless, current efficiencies for pp. 209-212.
the production of sodium hydroxide and acid of [ 111T.M. Walski, T.S. Barnhart, J.M. Driscoll, R.M. Yencha,
78% can be obtained by maintaining in the Hydraulics of corrosive gas pockets in force mains, Water
salt/acid compartment a 1:l ratio of sodium sul- Envir. Res. 66 (6) (1994) 772.
phate to sodium bisulphate. This technology will [12] R.L. Islander, Microbial ecology of crown corrosion in
sewers, J. Environ. Engng 117 (6) (1991) 751.
be of use to mills which employ either elemental
1131J.R. Chevalier, Toxicity of sodium sulphide to common
chlorine free (ECF) or totally chlorine free (TCF) shiners - dynamic bioassay, Tappi J. 56 (5) (1973) 135.
bleaching sequences, as they move towards system [ 141 D.Y. Prasad, Effect of Kraft pulp mill waste components
closure, by reducing the inorganic waste load to on some aquatic organisms, Indian J. Envir. Health 23 (4)
their effluent treatment plants and addressing the (1980) 340.
[ 151 T.K. Ghoth, S.K. Konar, Toxicity of chemicals and waste
reuse of sodium/sulphur.
waters of paper and pulp mills to worms, plankton, and
molluscs, Indian J. Envir. Health 22 (4) (1980) 278.
[ 161 D.Y. Prased, Toxicity of Kraft pulp mill waste constituents
Acknowledgment on fresh water Gambusia afinis, Indian Pulp Paper 35 (2)
(1980) 9.
[17] D.J.S. Robinson, E.J. Perkins, Toxicity of some wood pulp
The authors would like to thank J-N. Cloutier effluent constituents, Cumbria Sea Fisheries Committee
of Paprican for useful discussions and careful Carlisle Scientific Report 74 (1) (1977) 22.
review of this manuscript. [ 181 W.A. Moy, P. Joyce, G.E. Styan, Removal of sodium chlo-
ride from kraft recovery systems, Pulp Pap. Can. 75 (4)
(1974) 88.
[19] C. Forget, V.C. Uloth, D.L. Singbeil, Potassium and
References Chloride Removal from the kraft Pulping Process, Pulp
and Paper Research Institute of Canada, PPR1020, 1993.
[l] W.F. Sinclair, Controlling Pollution from Canadian Pulp [20] R. Backman, M. Hupa, Fouling and corrosion mechanisms
and Paper Manufacturers: a Federal Perspective, in the recovery boiler superheater area, Tappi J. 70
Environment Canada, Ottawa, 1990. (1991) 185.
[2] M.T. Towers, J.T. Wearing, Closed-cycle Design in the [21] C.-L. Lee, K. Hunt, R.W. Murray, Activated oxygen, a
 icf. Paleologou et al. 1 Separation and Purzjication Technology I1 (1997) 159-171 171

selective bleaching agent for chemical pulps. Part I. [31] M. Niksa, Acid/Base Recovery from Sodium Sulphate, 5th
Laboratory bleaching with isolated and in-situ generated International Forum on Electrolysis, Ft Lauderdale,
activated oxygen, J. Pulp Pap. Sci. 20 (5) (1994) Florida, 1991.
5125-5130. [32] Specifix Control Package, Copyright 01988, lnter-
[22] C.A. Rolf, E.O. Sture, A Method in Ozone Bleaching of llution@, Inc., Northwood, MA02062.
Pulp, PCT, International Patent Application Number WO [33] D.A. Macinnes, The Principles of Electrochemistry, Dover
93106297, 1993. Publications, New York, 1961, p. 342.
[23] M. Paleologou, P-Y. Wong, R.M. Berry, A solution to [34] C.F. Baes, Jr, The estimation of bisulphate ion dissociation
caustic/chlorine imbalance bipolar membrane electrodia- in sulfuric acid-sodium sulphate solutions, J. Am. Chem.
lysis, J. Pulp Pap. Sci. 18 (4) (1992) 5138-5145. Sot. 79 (1957) 5611.
[24] R. Thompson, M. Paleologou, R.M. Berry, Caustic soda [35] K.S. Pitzer, R.N. Roy, L.F. Silvester, Thermodynamics of
and sulphuric acid production from sodium sulphate electrolytes. 7. Sulfuric acid, J. Am. Chem. Sot. 99
by-product of chlorine dioxide generation by bipolar mem- (1977) 4930.
brane electrodialysis (BME) - economics, Tappi J. 78 (6) [36] R.E. Lindstrom, H.E. Wirth, Estimation of the bisulfate
(1995) 1277134. ion dissociation in solutions of sulfuric acid and sodium
[25] M. Paleologou, R.M. Berry, B.I. Fleming, Sodium chlorate bisulfate, J. Phys. Chem. 73 (1) (1969) 218.
splitting: a new solution to the problems of chemical imbal- [37] A.J. Bard, L.R. Faulkner, Electrochemical Methods:
ance in Kraft mills, J. Pulp Pap. Sci. 20 (2) (1994) 539-544. Fundamentals and Applications, Wiley, New York, 1980,
[26] M. Paleologou, P.-Y. Wong, R. Thompson, R.M. Berry, pp. 64-68.
Bipolar membrane electrodialysis for the production of [38] M. Paleologou, R. Thompson, R.M. Berry, C. Brown,
sodium hydroxide from sodium chlorate: a comparison of M. Sheedy, T. Gilliss, The generator acid purification
the two- and three-compartment configurations, J. Pulp (GAP) system reduces caustic make-up requirements at
Pap. Sci. 22 (1) (1996) 51-57. Kraft mills, 83rd CPPA Annual Meeting (Technical
[27] K.N. Mani, F.P. Chlanda, Electrodialytic Water-splitting Section), Montreal, PQ, 1997, pp. 28-31.
Process and Apparatus for Conversion of Alkali Metal [39] D. Bergner, M. Hartmann, R. Staab, Fortschritte auf dem
Sulphate Values Derived from Spent Rayon Spin Baths, Gebiet der Alkalichlorid-Electrolyse, Chem. lng. Tech. 59
US Patent no. 4 504 373 (March 1985). (4) (1987) 271.
[28] K.N. Mani, F.P. Chlanda, C.H. Byszewski, Aquatech 1401W.F. Linke, Solubilities of Inorganic and Metal-organic
membrane technology for recovery of acid/base values Compounds, 4th ed., vol 2, American Chemical Society,
from salt streams, Desalination 68 (1988) 149. Washington, DC, 1965, p. 982.
[29] J. Jorissen, K.H. Simmrock, The behaviour of ion [41] R.C. Weast, M.J. Astle, CRC Handbook of Chemistry and
exchange membranes in electrolysis and electrodialysis of Physics, CRC Press, Boca Raton. FL, 1981, pp.
sodium sulphate, J. Appl. Electrochem. 21 (1991) 869. D-145--D-146.
[ 301 G. Kreysa, Electrochemische umwelttechnik, Chem. Ing.
Technol. 62 (1990) 357.