2460
Ind. Eng. Chem. Res. 1996,34, 2450-2454
Separation of Weak Acids and Bases by Neutralization Dialysis Hideo Tanabe, Hiroshi Okochi, and Manabu Igawa* Faculty of Engineering, Kanagawa University, Rokkakubashi, Kanagawa-ku, Yokohama 221, J a p a n
Neutralization dialysis is a deionization method based on a Donnan dialysis. Weak acids and bases, even urea and alcohols, were also removed by this method, and their transport rates increased with the decrease of their dissociation constants and with the increase of the concentrations of acid and base solutions in the neutralization system. Aqueous silica, which is a typical inorganic weak acid and abundant in surface water, was also effectively removed by this method. Therefore, tap water was effectively deionized to be pure water with a resistivity of 15 MQ cm and a silica concentration of less than 10 ppb as Si02 and the neutralization dialysis was proven to be a n effective deionization method.
Introduction
t
Neutralization dialysis (Igawa et al., 1986,1987) is a new deionization process based on the principle of Donnan dialysis (Donnan, 1924) using a cation- and an anion-exchange membrane simultaneously. The feed solution is separated from an acid and a base solution by a cation- and an anion-exchange membrane, respectively. The cations in the feed solution are exchanged with protons across a cation-exchange membrane, the anions in the solution are exchanged with hydroxide ion across an anion-exchangemembrane, and the deionization proceeds with the neutralization reaction in the feed solution. Deionization with ion-exchange resin requires intermittent regeneration, and some contamination occurs in the regeneration process, although it is an effective process of deionization. The neutralization dialysis is, however, a continuous process and does not require the regeneration process, so that little contamination occurs. Therefore, deionization with ion-exchange resins may be substituted by the neutralization dialysis in the future and the kinetic model has been investigated (Sato et al., 1993). Neutralization dialysis has already found an industrial application in the deionization of milk whey (Bleha and Tishchenko, 1992). It can also be used to maintain a constant pH value of the solution in the transport system facilitated by the countertransport of hydroxide or hydrogen ions (Igawa et al., 1990). In deionization with ion-exchange resins, weakly dissociated materials are difficult to be removed and colloidal particles, which are weakly dissociated in general, are irreversibly adsorbed to the resin, which drops the efficiency and makes trouble in the regeneration of the resin (Streat, 1988). Representative of colloidal particles are fulvic acids, colloidal silica, and metal hydroxide complexes. The colloidal particles are also irreversibly adsorbed to the ion-exchange membrane, and it makes the membrane resistance increase in the electrodialysis (Watabe et al., 1980). Therefore, we have studied the transport properties of weak acids and basis in neutralization dialysis and applied neutralization dialysis to deionize tap water. The results will be reported in this paper.
Experimental Section Apparatus. Neutralization dialysis was carried out in the cell shown in Figure 1. The feed solution, the acid solution, and the base solution were circulated from the respective reservoirs with pumps at a rate of 110 mumin. The compartments in the cell were labeled as
t
i
Feed solution
t
Acid solution
Figure 1. Neutralization dialysis cell: a, anion-exchange membrane; c, cation-exchange membrane; R, W o n rubber sheet; S, spacer; T, Teflon plate.
follows: compartment D for the feed solution, compartment A for the acid solution, and compartment B for the base solution. Each membrane area was 30 cm2 (2 cm x 15 cm), the thickness of each compartment was 1.0 mm, and a net type spacer 1.0 mm thick was placed in each compartment. The cation-exchange process proceeded after the anion-exchange process in the cell because it was difficult to seal the solutions in a cell composed of three compartments, where cation- and anion-exchange processes proceed simultaneously,when the membrane area was large. In all of the experiments in this paper, two cells were used sequentially, the cation-exchange process and the anion-exchange process proceeded alternately, the total membrane area of the cation-exchangemembranes was 60 cm2,and that of the anion-exchange membranes was also 60 cm2. The flow system in contact with the feed solution was made of Teflon or Viton to diminish the contamination from the apparatus. A cation- and an anion-exchange membrane, Selemion CMV and AMV (Asahi Glass Co. Ltd.), were used for the separation of weak acids and bases. A perfluorocarbon cation-exchange membrane, Nafion 117 (Aldrich Chemical Co. Inc.), and a perfluorocarbon anion-exchange membrane, Tosflex IE-SF34 (Tosoh Co.), were used to deionize an aqueous silica solution and tap water because perfluorocarbon ion-exchange membranes are more stable in strong acid and base solutions than hydrocarbon membranes. The detector of a resistivity meter (Toa Denpa Ind. Co. Ltd.) was placed in the line just after the outlet of the deionization cell to measure the resistivity of the deionized water continuously. Purified nitrogen gas was bubbled into the feed solution in its reservoir to prevent the dissolu-
0888-5885/95/2634-2450$09.00/00 1995 American Chemical Society
Ind. Eng.Chem. Res., Vol. 34, No. 7,1995 2451 tion of carbon dioxide into the deionized water, because the carbon dioxide dissolved in the deionized water lowers the resistivity of the water.
Measurement of Selectivity Coefficient of IonExchange Membrane. The selectivity coefficient of the anion-exchange membrane was necessary to estimate the flux of a weak acid, and it was measured for Selemion AMV by the general procedure of Yeager and Steck (1979);the anion-exchange membrane was soaked in 1M NaOH solution for 1h to obtain OH- form of the membrane; the membrane was soaked in the mixed solution of an organic acid and sodium hydroxide, which was added to dissociate the organic acid, for 3 h; the organic acid concentration and the pH value in the solution equilibrated with the membrane were measured. The membrane phase concentration of the organic acid was calculated by the concentration change of the organic acid solution, the membrane thickness, and the membrane area. The selectivity coefficient,K e x , of Selemion AMV for the organic anion, A-, over hydroxide ion was defined as follows:
where the ion formula with a line over it indicates the ion in the membrane. The ion-exchange capacity of Selemion AMV is 2.4 mmoVcm3 ( h i , 1981), and the concentration of the hydroxide ion in the membrane is calculated as the difference between the capacity and the concentration of the organic anion in the membrane. The equilibrium concentration of each organic acid was measured for three conditions changing its concentration, the selectivity coefficient was calculated for each condition, and the mean value was determined as the coefficient. Reagents. Weak acids and bases used in this study have various p K a values: acetic acid, 4.8; phenol, 10.0; 2,2,2-trichloroethyl alcohol, 12.2; 2,2,2-trifluoroethyl alcohol, 12.4; 2-chloroethyl alcohol, 14.3; ethyl alcohol, 15.9; 2-propyl alcohol, 18.0; ammonium ion, 9.2; and urea (protonated cation), 13.9 (Wade, 1987; Handbook of Chemistry and Physics, 1985). The acids and bases were of reagent grade and used without further purification. The solution of aqueous silica was prepared from sodium silicate, and it was filtered through a 0.22 pm membrane filter (Millipore Co.) after adjusting to pH 7 with hydrochloric acid. Tap water used as a test solution in this study was that in our university after filtration through the 0.22 pm membrane filter. Analysis. The concentration of organic acids and urea was determined by a total organic carbon analyzer (Shimadzu TOC-5000),and the concentration of dissolved silica was determined by an inductively coupled plasma atomic emission spectrometer (Seiko Ins. Inc., SPS1500). The concentration of ammonia was determined by the indophenol method, using a spectrophotometer (Shimadzu W-265).
Results and Discussion Separation of Weak Acids and Bases by Neutralization Dialysis. The conjugate base of a weak acid is exchanged with hydroxide ion in the membrane even when the acid was essentially in the undissociated form in the feed solution and is transported very rapidly against its concentration gradient in neutralization dialysis (Igawa et al., 1993). Figure 2 shows the
0
20
40 Time (min)
60
Figure 2. Concentration change of weak acids and bases in neutralization dialysis: 0, acetic acid; 0 , ammonia; A, phenol; A, urea; 0,ethyl alcohol; - - -, potassium chloride; compartment A, 10 mM HCl(100 mL); compartment B, 10 mM NaOH (1OOmL); compartment D, 10 mM weak acid or base (100 mL).
separation of weak acids and bases in neutralization dialysis, and acetate ion and ammonium ion were transported almost as rapidly as potassium chloride (dashed line). Other acids and bases with large pKa or pKb values were also removed, although the transport rate decreased with the increase of the values regardless of acid or base. Therefore, the transport characteristics of weak acids and bases in neutralization dialysis were investigated in further detail using organic acids with various pKavalues. In the experiments, concentration polarization might occur at the surface of the dialysis membrane. However, the effect was minimized in the cell by a large linear velocity, which was calculated to be 9.2 cm s-l from the pumping rate and the cross sectional area of each compartment, and spacers were inserted t o form a turbulent flow in the flow system. Concentration of Weak Organic Acids in the Membrane Phase. The transport rate was primarily controlled by the concentration difference in the membrane phase. Weak acid, HA, in the feed phase solution is partly dissociated, and the dissociation equilibrium is described as follows:
The organic anion is readily exchanged with hydroxide ion in the membrane at the membrane surface as
A-
+
- OH- = A-
-
+ OH-,
Kex=
[A-l[OH-l
-
(3)
[A-I[OH-l After hydroxide ion is transferred to the feed solution by the ion-exchange reaction, a neutralization reaction occurs between a proton dissociated from the organic acid and the transferred hydroxide ion as in the following equation. OH-
+ H+ = H,O,
K, = [H'l[OH-l
(4)
where K, is the ion product of water. The total ion concentration in the membrane is equal to the ionexchange capacity, Ce,, as
Cex= [A-I
-
+ [OH-]
(5)
2452 Ind. Eng. Chem. Res., Vol. 34, No. 7, 1995
N
. -E, E
1
v
E 2
pH 13 ...........................
K
.
---------
-pH
14
pH 12
(B)
n
15
10
~
20
PKa
5
15
10
20
PKa Figure 3. (A) Theoretical values of the membrane phase concentration and (B) the concentration difference in the membrane phase: the solid line is the concentration of the surface adjacent to the feed solution of pH 7, and the other lines are those of the surface adjacent to the base solution with various pH values (upper figure).
The concentration of the organic anion a t the membrane surface is calculated from eqs 3 and 5 as follows:
(6) The concentration of the total organic acid in the solution is as follows:
C , = [HA]
+ [A-I
(7)
The concentration of the organic anion at the membrane surface is obtained from eqs 2, 4, 6 , and 7 as follows.
In the calculation of the concentration of the anion in the membrane phase, the ion-exchange selectivity coefficient is required. The measured values were as follows: acetic acid, 1.01 & 0.04; phenol, 11.1 f 0.6; 2,2,24richloroethyl alcohol, 3.07 f 0.15; and 2,2,2trifluoroethyl alcohol, 1.21 f 0.03. The values were the average of the experimental values in the range of 90% confidence level. The value of phenol is much larger than the others because of the affinity of the aromatic rings between phenol and the base of the ion-exchange membrane, styrene-butadiene copolymer. It was very difficult to measure the value of K,, for the organic acid with a PKa value larger than 14, because its concentration change caused by adsorption to the membrane was less than 0.5 ppm of TOC, which was smaller than the detection limit of the TOC analyzer. Then, in the following calculation,the values of Kexfor the very weak organic acids were assumed to be 2, the intermediate value of the coefficients measured except for phenol. Figure 3A shows the membrane phase concentrations at the surface adjacent to the feed solution (solid line) and the receiving phase solution (other lines) of the base solution, which were calculated from eq 8. In the calculation, the concentrations of the organic acid were
Figure 4. Flux of organic acid by neutralization dialysis: 0, 10 mM NaOH in compartment B for the initial condition (measured); 0 , 1 mM NaOH in compartment B for the initial condition (measured); -, 10 mM NaOH in compartment B (calculated), - - -, 0.5 mM NaOH in compartment B (calculated). In the calculation, the solution pH of the feed solution was assumed to cm2 be 7, the diffusion coeficient was assumed to be 1.1 x min-', which was the coefficient of acetic acid, and the selectivity coefficient was assumed to be 2.
10 and 1 mM in the feed and the receiving phase solution, respectively,and the solution pH in the receiving phase solution was altered from pH 10 to 14. The solution pH of the feed solution was assumed to be 7, because the acid solution in the feed solution becomes neutral in the deionization process of neutralization dialysis. The value of the organic acid concentration in the receiving phase was a typical value, where the receiving phase concentration changed linearly with time and the flux was obtained. Figure 3B shows the difference between both sides of the membrane corresponding to the values shown in Figure 3A. Since the solute flux is proportionalto the concentration difference in the membrane phase, we can predict the dependence of the fluxes on pKa values from Figure 3B. The concentration difference for the organic acid with a pKa value larger than 15 is very small. The concentration differencefor the organic acid with a pK, value of about 12 becomes the maximum when the solution pH of the receiving phase is lower than 11. The PKa value where the concentration difference is the maximum changes with not only the base solution pH but also the organic acid concentration of the feed solution as expected from eq 8. Transport Characteristics of Weak Organic Acids. The flux of the organic anion across the anionexchange membrane can be calculated as follows:
The flux of undissociated organic acid, the second term in eq 9, was estimated by the measurement of the acid flux across the chloride ion-type anion-exchange membrane from the feed solution to pure water. The flux of the organic anion can be calculated as follows: Js,A-
[ A - l , - [A-lb)
Q,*- -
=y
(10)
where the subscripts f and b refer to the membrane phase adjacent to the feed solution and that to the base solution, respectively, and each membrane phase concentration can be calculated from eq 8. Figure 4 shows the measured fluxes of the organic acids when the solution pH of the receiving phase was adjusted to be 11 and 12. The fraction of the flux of the undissociated organic acid in the flux of the neutralization dialysis was very small, less than 4%, for the
Ind. Eng.Chem. Res., Vol. 34, No. 7, 1995 2463 lS
7
iij 0"
lo%
I\
3
0.5
'
0
2
3 Time (hour)
4
1
5
Figure 5. Concentration change of ethyl alcohol in neutralization dialysis: 0, compartment B; 0, compartment D; compartment A, 10 mM HCl 1 mM Ca&OH (100mL); compartment B, 10 mM NaOH 1mM CaH50H (100 mL); compartment D, 1mM C a b OH (100mL).
+
0
1
1
I
+
organic acids with pKa values less than 13, and the flux of the undissociated organic acid was approximately equal to the flux of the neutralization dialysis for the organic acids with pKa values over 15. Therefore, the flux in the neutralization dialysis can be estimated approximately by eq 10. The solid line and the dashed line in Figure 4 are the calculated values for the cases in which the pH of the base solution was 10.7 and 12, using the diffusion coefficient of acetic acid, 1.1 x cm2min-l, which was obtained experimentally from ita flux and the membrane phase concentration, an ionexchange selectivity coefficient of 2, and an organic acid concentration in the base solution of 1mM. The reason why the pH of the base solution was assumed not to be 11 but to be 10.7 for the case in which the initial experimental condition was pH 11 was because the solution pH was lowered by the ion-exchange reaction when the flux was measured experimentally. In the case of pH 12, the pH was maintained practically constant. Corresponding to the calculated values, the measured fluxes of the organic acids with pKa values larger than 15 were very small and there was a maximum flux for 2,2,2-trifluoroethylalcohol 12.4) when the pH of the base solution was 11. The difference between the calculated values and the experimental data is brought about from the differences of the true values and the assumed values for the selectivity and the diffusion coefficients. With the increase of the selectivity coefficient,the concentration difference in the membrane phase decreases for the concentration increase at the base solution side membrane surface and, then, the flux decreases. The decrease of the diffusion coefficient also decreases the flux. Phenol has a small diffusion coefficient, 8.0 x cm2 min-l, and a large selectivity coefficient, and thus the flux of phenol deviated from the calculated value.
(sa:
Enrichment of Alcohols in Neutralization Dialysis, Even the concentrations of the organic acids with large pKavalues, such as ethyl alcohol (pK: 15.9), were lowered by the neutralization dialysis as shown in Figure 2. Then, it is possible t o enrich ethyl alcohol against the concentration gradient in the neutralization dialysis. The up-hill transport of ethyl alcohol from compartment D to compartment B is shown in Figure 5, where the concentration of each compartment was equal at the initial stage of the experiment. In contrast, the concentration of ethyl alcohol was maintained at a constant value in compartment A.
2
3
5
4
Time (hour)
Figure 6. Concentration change of dissolved silica in neutralization dialysis: concentrations of HC1 and NaOH in compartments A and B (1 L)are 0.01 M (O), 0.1M (01,and 1M (A);compartment D, 10 ppm as Si02 of sodium silicate solution (100mL).
The enrichment factor, the ratio of the equilibrium concentration of the receiving phase solution to the feed solution, can be estimated as follows. In an equilibrium, both sides of the membrane phase have the same concentration.
In eq 8, CHAand [OH-] are different on both sides of the membrane and the concentration ratio of both sides can be obtained as follows using eqs 8 and 11.
The denominators of both sides of eq 12 are approximately equal to K, because of the large pKa value. Then, the concentration ratio becomes a function of the ion-exchange capacity and the selectivity coefficient as follows.
(13) In Figure 5, the equilibrium concentration ratio was 1.31. The selectivity coefficient is a little different in the feed solution and the base solution depending on the composition in the equilibrated solution (Myers and Boyed, 1956). The ion-exchange capacity is also somewhat dependent on the solution equilibrated to the membrane because of the swelling of the membrane. These effects must be controlling the enrichment factor. Deionization of Dissolved Silica by Neutralization Dialysis, Figure 6 shows the concentration change of dissolved silica in the neutralization dialysis. The concentration of aqueous silica was 10 ppm as Si02 in this experiment, and the silica in the solution was not colloidal but dissolved (Stumm and Morgan, 1981). The form of the dissolved silica changed with the increase of pH as follows. Si(OH),
-
SiO(OH),-
-
-
Si406(OH),2SiO.,(OH):-
(14)
2454 Ind. Eng. Chem. Res., Vol. 34, No. 7, 1995
When the concentration of hydrochloric acid and sodium hydroxide in compartment A and B was 10 d, dissolved silica in the feed solution was removed only 55% after 5 h. However, the dissolved silica was effectively removed when the concentration of the acid and base solution became high. It must be caused by the ionization effect of the concentrated base solution on the silica in the membrane phase. Aqueous silica was accumulated and became polymerized in the membrane phase when the base concentration was not high. When the base concentration was high, the aqueous silica existed as silicic acid and was transported readily across the membrane. Silica can be dissolved to be a high concentration only in the high pH range. The equilibrium concentration in the feed solution can be estimated from eq 12. The effect of the values of K,, and C,, is not the dominant factor for both sides of the membrane phase when the pKa was less than 12, and the ratio can be obtained by the following equation.
Table 1. The Quality of Tap Water Deionized by Neutralization Dialysis wateln resitivity (MR-cm) [Si021(ppb) TOC (ppmC) tap water 7.6 10-3 2.1 104 4.2 after UF 2.1 x 10-2 2.1 x 104 2.4 after step I 6.8 x 99 1.3 after step I1 2.9 x 10-1