148
Ind. Eng. Chem. Prod. Res. Dev., Vol. 17, No. 2, 1978 a
R-CH2-
CH+CH2I SiMe
~
J
I
RCH, Me3Si0=CHExpected masses for fragment (c) are CH3(CH&OH CH3(CHz)loOH CH3(CH2)110H
-
mle 112 (CgH16) m/e 126 (CgHls)
m/e 140 (C10H20)
(iv) All the alcohols will show substantial hydrocarbon-type peaks a t mle 41/43,55/57,69,83,97, 111,and 129. I t was decided to confirm the nature of all the compounds by converting the alcohols to their trimethylsilyl (MesSi) derivatives. Each specimen was treated with "Trisil," a commercial silylating mixture, and subjected to GC/MS under the same condition applied to the alcohols, except that the column temperature was 100 "C. Each chromatogram (Figures 2B, 3B, and 4B) shows four peaks and the interpretation of the mass spectra (Table 111) of these peaks is given in Table V. The assignments in Table V are based on the following criteria. (i) All the silyl derivatives will give small M - 1 and larger M - 15 peaks. (ii) All, except terminal OH, compounds will give characteristic cleavage products as follows
R'
R'CH? -
CH2R
Me3SiO=CH-
+
CH,R
Expected fragments from these cleavages are given in Table VI. (iii) The terminal -0-MesSi derivative would be expected to give only M - 15+ ions of appreciable intensity but these ions are also shown by all other isomers. Therefore it appears that all possible alcohol derivatives are obtained from a straight-chain hydrocarbon. Gas chromatograph conditions used did not completely separate all isomers, but combined GCIMS makes the assignments fairly certain. Conditions could probably be worked out for complete separation of the isomeric alcohols. Literature Cited Am. SOC.Test. Mater., "ASTM Standard", Designation, D 2003 (1973). Bateman, L., Cunreen. J. I., Koch, H. P.. Nature, 164, 242 (1949). Boocock. J. R. B., Hickinbottorn, W. J., J. Cbem. SOC.,20, 1319 (1963). Cox, R. A., Swallow, A . J., J. Cbem. SOC.,4, 3727 (1958). Cram, D. J., Hammond, G. S.,"Organic Chemistry', 2nd ed,McGraw-Hill. p 524, New York, N.Y., 1964. Noller. C. R., "Chemistry of Organic Compounds", 2nd ed,p 798, W. B. Saunders Co., Philadelphia, Pa., 1957. Qvist, W., Chem. Abstr., 49, 201d(1955).
Received for r e u e u April 25, 1977 Accepted November 18, 1 9 7
Regeneration of Waste Acid from a New Ilmenite Treatment Process by Electrodialysis John J. Barney*' and James L. Hendrlx Chemical and Metallurgical Engineering Deparfment, Mackay School of Mines, University of Nevada, Reno, Nevada 89557
Recovery of waste sulfuric acid by electrodialysis using anion-exchange membranes was studied. The waste acid originates from the digestion of sodium titanate in a new process developed by the United States Bureau of Mines for the treating of ilmenite. A continuous multichambered operation was used to produce a concentrate of 50 wt YO H2S04and a depleted waste acid of 5-10 wt % H2SO4. Reagent grade waste acid solutions and actual titanate waste acids were treated at various current densities. Current efficiencies were highest at 30 A (136.4 mA/cm2), but power consumptions were about the same for 20 A and 30 A. Power requirements at 15 A were much higher than at either 20 A or 30 A. Power consumptions for the actual waste acids were slightly higher than for reagent grade acids. The minimum power requirement measured for processing actual waste acid was 9.48 kWh/kg of H2S04in the concentrate. Acid regenerated by electrodialysis has a lower concentration of contaminants than acid regenerated by evaporation.
Introduction In the production of titanium pigments from titanium rich ores, the chloride or the sulfate process is usually employed. Address correspondence t o this author a t t h e U n i t e d States Department of the Interior, Bureau of Mines, P.O. B o x 1660, T w i n Cities, Minn. 55111.
0019-7890/78/1217-0148$0.100/0
Rutile (TiO2) is the preferred feed for the chloride process because it has a high titanium dioxide content (95%)and a low iron content. Ilmenite (FeOTi02) can be used, but chlorination of the iron causes excessive chlorine consumption and the ferric chloride thus produced presents a waste disposal problem (Pings, 1972). Since domestic and worldwide resources of rutile are very limited (Stamper, 1970; Wessel,
0 1978 American Chemical Society
Ind. Eng. Chem. Prod. Res. Dev., Vol. 17, No. 2, 1978
1973), the exploitation of abundant ilmenite resources will become increasingly important to meet domestic needs in the near future. It has been projected that known reserves of rutile may become exhausted by the mid-1980’s (Stamper, 1970). Ilmenite is normally treated by the sulfate process, wherein the ore is digested with concentrated sulfuric acid to produce titanyl sulfate and iron sulfate. T h e sulfate process has been plagued with a solid waste disposal problem associated with the hydrated iron sulfate (copperas) by-product and a water pollution problem associated with the depleted sulfuric acid solution. The common practice of dumping raw sulfate process wastes a t sea may soon no longer be permitted (Wessel, 1973). A new ilmenite smelting process developed by the United States Bureau of Mines (USBM) produces a pig iron product and a titanium rich slag (Gomes et al., 1974). T h e disposal problem associated with the iron is obviated and a saleable product produced. The remaining waste sulfuric acid solution is a lingering problem and must be treated for recovery or disposal. Recovery of the waste acid is a desirable solution to the disposal problem since acid is also conserved. Recovery of the waste acid was the objective of the investigation reported here. T h e waste acid must be upgraded to about 50 wt % HzS04 to be reused for the digestion of sodium titanate in the USBM process (Gomes, 1975). Since electrodialysis had been applied to the treatment of pickling liquors, and since highly acid resistant ion-exchange membranes are commercially available, it was thought that electrodialysis had good potential for upgrading the titanate waste acid. Several laboratory studies have considered the application of electrodialysis for the recovery of spent pickling liquors but there have not yet been any industrial applications of electrodialysis for treating strong acid solutions. The first reported laboratory investigation was by Bramer and Coull (1955). They employed a two-chambered setup with a single anion permeable membrane separating the anolyte and the catholyte. Their experiments demonstrated the feasibility of recovering electrolytic iron and sulfuric acid. Also, they found current efficiency to be lowered by the leakage of acid produced a t the anode through the membrane to the cathode chamber. Bramer and Coull’s experiments were of too short duration (1 h) for meaningful determination of power requirements. Horner et al. (1955) have thoroughly evaluated the electrodialytic treatment of a spent pickle liquor by both batch and continuous operation. Continuous operation was found to be more efficient than batch operation and the threechambered arrangement more efficient than the two-chambered one. Multicompartment operation is, thus, suggested to reduce back-diffusion of sulfuric acid and enable stepwise depletion of the waste liquor. Though typical power requirements were 5.5-6 kWh/kg of sulfuric acid recovered, the process when credited for electrolytic iron compared favorably with lime neutralization. Lewis and Tye (1959) have made a generalized treatment for a range of pickle liquors. Two-chambered cells were used to obtain power requirements and establish the quantitative effects of cell configuration, solution flow arrangements, solution concentration and current density. A significant finding was that, when diffusional effects are unimportant, current efficiency is independent of the waste liquor concentration but dependent on the concentration of the recovered acid. I t was also shown that the effects of back diffusion of acid could be eliminated by increasing the current density. Other processes have been proposed for recovering acid and iron from hydrochloric acid pickle liquor and for incorporating electrodialysis into an evaporation-crystallization scheme for treating waste sulfuric acid pickle liquor (Farrell and Smith, 1962). More recently, Smith (1974) has demonstrated the
149
effectiveness of using a sequestrant to prevent fouling of anion-exchange membranes by iron deposits. Only one reference was located which discussed electrodialytic regeneration of waste acid having a concentration typical of that from titanium dioxide manufacture. In that study (Nishiba and Shoda, 1974), a 27 w t % H z S 0 4 solution was upgraded to 50 wt % using a three-chambered stack arrangement. Membranes used by Nishiba and Shoda were 1-3 mm thick. Ion-exchange membranes which are commonly used for electrodialysis are only 0.15 to 0.61 mm thick (Lacey, 1972). Though electrolysis with commercially available membranes has been successfully applied to the treatment of waste sulfuric acid solutions, recovered acid concentrations have been limited to about 6 N. The present study is concerned with acid concentrations up to 14 N HzS04. In addition, titanate waste acids which are treated contain a variety of impurities whereas the pickling liquors which have been previously studied contained only iron as an impurity. The purpose of the present study was to determine the optimum conditions of operation for recovery of titanate waste acid and to compare the energy requirements for electrodialysis with those for evaporation. The specific effects of impurities on the operation of the electrodialytic process were not studied but their concentrations in the waste and regenerated acids were measured. Both one- and two-stage operations were investigated in this study. Multiple stage operation is suggested for applications where the degree of demineralization or concentration desired is beyond the capability of a single stack (Mintz, 196X) Therefore, to achieve the extremes of a 50-6096 H2S04 concentrate and a depleted waste of about 6% HLS04,a two-stage process was examined. A single-stage process was also evaluated since it offers the advantage of simpler operation. A study of an evaporation process for upgrading waste sulfuric acid from the digestion of ilmenite was made by the New Jersey Zinc Co. (Peterson and Kern, 1975). In that study it was noted t h a t the economics of the process were sensitive to the cost of fuel. Power requirements, considering only the fuel oil required for evaporation, were the equivalent of 4.8 kWh/kg of HzS04 for production of an 80% HzSO4 concentrate from a feed solution containing about 25% HISOI. Experimental Section Acid Solutions. The makeup acid solution which fed the concentrate chambers of the electrodialysis stack was, in all runs, made from reagent grade H2S04. Two actual waste acid solutions were treated. Though prepared in the laboratory, the actual waste acids had compositions typical of waste acids which would be generated in the digestion of sodium titanate. One was obtained from digestion of high sodium titanate (HST) material and the other from the digestion of low sodium titanate (LST) material. Both acid solutions contained several impurities. The impurity and acid concentrations of the H S T and LST waste acids are given in Table I. Equipment. Membranes. The major membrane w e d was produced by Ionac Chemical Co. and denoted MA-3475. It is a strongly ionized, “heterogeneous” anion-exchange membrane. The second membrane used was the A-101, produced by Asahi Chemical Industry. I t is a “homogeneous” membrane made chiefly of divinylbenzene-styrene copolymer with ion exchange groups attached. It is reinforced with a wide mesh backing and is supplied wet. Characteristics of both the A-101 and MA-3475 are listed in Table 11. Electrodes. A cathode was made of the acid resistant Hastelloy-C and the anode consisted of a thin platinum coating on titanium. The platinum which is heat reduced onto
150
Ind. Eng. Chem. Prod. Res. Dev., Vol. 17, No. 2, 1978
Table I. Composition of Titanate Waste Acids Which Were Treated
ComDound
i 3
Concn, g/L From high sodium From low sodium titanate digestion titanate digestion
STAINLESS STEEL BOLTS
7.0 0.6
2.9 0.4 3.0 4.3 31.0 304
1.6 2.2
3.5 370 MEMBRANE SPACERS
Table 11. Characteristics of Ion-Exchange Membranes Which Were Used Model 'Type Composition Ion exchange group
A-101
MA-3475
Anion-exchange Homogeneous Quaternary amine 1.5
Anion-exchange Heterogeneous Quaternary amine
Ion exchange capacity, mequiv/g Water content, % 24 Water permeability (mL/h/ft2/30 psi) ... (mL/h/ft2/10 psi) ... Electrical resistance, ohm- 1.9 (0.5 N NaC1) cmz Tensile strength, kg/mm2 3.2 Mullen burst strength, psi . . . Thickness, mm 0.21 Density, g/m2 ... % Permselectivity (0.25 N NaCVO.5 N NaC1) 98 (0.5 N NaC1/1 N NaC1) . . . Chemical stability 60t% HC1 ... NaOH ... Available size, m 1.3 X 1.3 Approximate cost, $/ft2 10.00
1.13 .
I
.
Less than 7 Less than 3 5.2 (1N NaC1)
... 200 0.37 360 ... 99 up to 125 "C 35t% Concentrated 5006 1x3
4.30-4.50
the titanium is employed mainly to lend desirable electrochemical properties to the electrode rather than for corrosion resistance (Bomberger and Kessler, 1969). Since the platinizing is porous, the titanium is readily attacked a t high sulfuric acid concentrations. Periodically, then, the anodes were returned to the manufacturer for replatinizing. All electrodes were 22.9 cm wide and 25.4 cm high. Flow Channels. Flow channels (membrane spacers) were of the tortuous path design. Three types of materials were used: polyethylene, polybutylene, and polypropylene. Polybutylene and especially polypropylene were used as alternatives to polyethylene because they have better resistance to degradation a t elevated temperatures. Heat Exchangers. Heat exchangers were constructed by laminating three 6.4 mm thick sheets of Plexiglas with a flow pattern cut out of the center piece. One such heat exchanger was located behind each electrode and cooled with tap water. Electrodialysis Stack. T h e t,ransfer area of the electrodialysis stack consisted of membranes separated by spacers between the cathode and the anode with a n effective area of 220 cm2. The electrodes, bounded by heat exchangers, polyvinyl chloride (PVC) end blocks, and steel end plates, were all held together by four stainless steel bolts. T h e stack is schematically depicted in Figure 1. Test Procedures. M e m b r a n e Suitability Tests. Tests were first conducted using only the MA-3475 membranes.
4i \
\_STEEL END PLATE
i \\PVC
END BLOCK
\,NODE
MEMBRANES
Figure 1. Exploded view of electrodialysis stack.
They were run at 5,10,12, and 15 A and provided information concerning the ranges of flow rates and voltage requirements as well as the stability of MA-3475 membranes in H2SO4 solutions. The tests were of fairly short duration, lasting from only 1 to 6 h. Four Ionac membranes were used in all runs. The flow arrangement consisted of one depleting chamber and four concentrating chambers in series. Solution was pumped through the depleting chamber a t a rate of about 1.0 mL/min but allowed to flow by gravity through the concentrating chambers a t about 0.05 to 0.30 mL/min. T h e dialyzed solutions were continuously collected and analyzed for HzSO4 content. The stack voltage was also closely monitored. Large increases were interpreted as an indication of possible membrane failure. Batch tests were made using only the A-101 membranes. Two A-101 membranes were used in the first test to divide the stack into one concentrating chamber and two depleting chambers. T h e second arrangement consisted of three A-101 membranes giving two chambers for each solution. The tests were made at 10 A. Samples were taken periodically to monitor the HzS04 concentrations. When a sample was taken, the feed was replenished with an equal volume of fresh solution. At the end of each run, the concentrated and depleted solutions were collected and analyzed for HzSO4 t o obtain the total acid gained by the concentrate and that lost by the depleted solution. A third group of batch tests employed a combination of A-101 and MA-3475 membranes. T h e tests were designed so that comparisons could be made with the tests using only A-101 membranes. Some of the tests employed one A-101 membrane and two MA-3475 membranes giving two concentrating chambers and two depleting chambers. Other tests employed three MA3475 membranes and two A-101 membranes giving two depleting chambers and four concentrating chambers. The stack operation and sampling procedures for this group of tests were the same as for the tests using A-101 membranes. Current was passed a t a level of 10 A. Two-Stage Operation. A two-stage arrangement was constructed wherein the waste solution and the concentrate solution would flow countercurrently between two stacks arranged hydraulically in series. T h e arrangement with approximate sulfuric acid concentrations for the various streams is shown in Figure 2. In terms of the primary purpose of each component, the qecond stage would be considered the concentrating stage and the first stage, the depleting stage. Though the waste acid solution is shown to be depleted to a concentration of 60g/L,, a wide range of concentrations could be obtained by varying the waste acid flow rate. T o evaluate the feasibility of stage 11, a multicompartment arrangement employing three MA-3475 membranes and two A-101 membranes was used. At steady state, current ef-
Ind. Eng. Chem. Prod. Res. Dev., Vol. 17, No. 2, 1978 WASTE ACID
Table 111. Data for Stage I of the Two-Stage Process
i-300gm/~1
I OEPLE'ED WASTE (-6Oqm/I:
Current, Concentrations, N
INTEQMEDIbTES
I
STAGE I
I
(-200 Qrn/li
1
STAGE 2
MbKEUP
1-140 g m / i i
I
CONCENTRATED PRODUCT,
Figure 2. Schematic depiction of two stage process.
ficiencies were calculated and power consumption data were obtained. Runs were initiated to determine the range of current efficiencies and power requirements for stage I. A stack utilizing three MA-3475 membranes to give two concentrating and two depleting chambers was used. Current efficiency and power requirement data were obtained under steady state conditions a t 8,10, and 15 A. Next, two stacks were operated together as stages I and I1 of the two-stage process. The initial concentrate makeup and intermediate solution contained 142 and 435 g/L, respectively. T h e initial waste feed and intermediate solution contained 295 and 192 g/L, respectively. The concentrating solution feed rates were the same for both stages. The depleting solution feed rates were also the same for both stages. This arrangement was operated a t 8 and 10 A. Each stack employed the membrane and flow arrangement of stages I and I1 described above. One-Stage Operation. T o produce a regenerated acid of 50% it was decided a one-stage operation would be adequate. T h e remainder of the tests used a single-stage process. A membrane arrangement providing four chambers (two concentrated and two depleted) was used in all runs, and solution flows were continuous. One MA-3475 membrane was placed adjacent to the cathode, another in the center of the stack, and two A-101 membranes were placed together adjacent t o the anode so as to offer a greater resistance to back diffusion of sulfuric acid from the anode chamber toward the depleted solution. Reagent grade waste acid solutions used in these tests contained 292-298 g/L of HzSO4. The concentrations of waste acid solutions which contained impurities are given in Table I. Data were collected when steady state was achieved. Power consumptions were calculated and product rates were obtained by measuring the volume of product plus samples collected over a specific period of time. In all runs, the pressure between the concentrated and depleted solutions was balanced so as to prevent bulk solution flow across the membrane separating the two solutions. Results Membrane Suitability Tests. Tests with only MA-3475 membranes in the stack proved that the MA-3475 membranes, alone, were unsatisfactory to produce the 50 or 60% acid goal. Although acid of the proper strength was produced, no extended run could be conducted due to membrane failure. After producing desirable concentrates for approximately 1 h, the stack voltages would rise from approximately 4.0 V to voltages greater than 6.5 V. Disassembly of the stack revealed the membrane bordering the concentrate chamber to be blackened and rippled. Current efficiencies were very low in tests employing Asahi's A-101 membranes, exclusively. Efficiencies of less than 10% were common. Also, the maximum concentrate of 8.85 N was well short of the goal of approximately 15 N.
151
A 8 8 8 10 10 10 10 15 15 15
Volts
C"c
cod
cc
Cd
4.05 4.20 3.90 4.80 4.20 4.80 4.80 5.40 5.30 5.20
3.20 3.20 3.20 3.14 3.14 3.20 3.20 3.00 3.10 3.10
3.90 3.84 3.88 3.69 3.78 3.84 3.84 3.92 3.88 3.92
7.75 8.99 10.20 8.80 8.90 9.79 12.23 9.18 10.80 11.20
1.53 1.55 1.82 1.23 1.70 1.27 1.23 1.25 1.04 1.62
Power consumption, kWh/kg of HzSO4 5.07 5.78 6.12 6.13 5.78 6.00 c r n
1.31
6.75 7.22 7.81
All tests employing a combination of A-101 and MA-3475 membranes gave higher efficiencies than those using A- 101 membranes alone. During these runs the concentrate continuously increased in concentration while the waste solution continuously decreased, reaching a maximum of 11.7 N and a minimum of 2.37 N a t the end of one run. Two-Stage Operation. Tests were conducted to demonstrate the operating performance of stage I1 of a two-stage process. Of significant note are the very low flow rates (0.1 to 0.2 mL/min) which were required t o produce a concentrate of 5 5 6 0 % (16.2-18.3 N ) H2S04. Changes of approximately 0.05 mL/min in the flow rates caused relatively large changes in the product concentration. One run was terminated before steady state was reached because the concentrate concentration (C,) had risen to 62.5% (19.4 N) H2SO4 and was continuing to rise as a result of the very low concentrate flow rate, 0.13 mL/min. Some runs did demonstrate the feasibility of stage 11. Concentrates of 55-60% H2SO4 were produced while 17.3 and 19.8% of the waste acid was recovered in the runs. In one run steady-state production of 60% (18.3 N) H 2 S 0 4was maintained for 33 h. The indicated power requirements for stage I1 were 4.6-5.5 kWh/kg of H2S04. The use of Plexiglas heat exchangers in tests representing stage I of the two-stage process permitted operation a t currents above 10 A. The heat exchangers therefore were used in all subsequent tests. Results of the stage I tests are given in Table 111. Current efficiencies for these tests were much higher than for stage I1 but power consumptions were also high. A t 10 A, for example, production of a concentrate of 12.2 N and a depleted waste of 1.2 N would require about 7.6 kWh/kg of H2SO4. Though data are limited, they do show a trend of increasing power consumption with increasing concentrate concentration for a given current density. There is also a trend of increasing power consumption with increasing current density. Though the current efficiencies were somewhat higher a t 15 A than a t 8 or 10 A, a higher voltage caused the power requirements to be higher as well. By combining the results of stage I and stage I1 a power consumption for the two-stage process of about 12.1 kWh/kg of HzS04 for production of a 60% sulfuric acid concentrate a t 10 A can be estimated. The two stacks were also run together. Down time due to corrosion of an electrode and long operating times required to reach steady state inhibited the gathering of power consumption data for this setup. The datum point obtained in one run, however, agreed quite well with the individual operation of stages I and 11, discussed above. The power requirement for production of a 56.9% (17.0 N)H2SO4 concentrate and a 2.2 N H2S04 depleted waste was 12.5 kWh/kg of H2SO4. One-Stage Operation. Power consumptions and current efficiencies were obtained for concentrates containing 525-791 g/L of HzSO4 while producing a depleted waste solution of (usually) 50-125 g/L of HzS04. Similar data were obtained for
152
Ind. Eng. Chem. Prod. Res. Dev., Vol. 17, No. 2, 1978
A-
- --- --
-A,
-----. *--A
IO
I2
I4
C,
16
INORMALITYI
Figure 5. Current efficiencies as a function of concentrate concentration for treatment of a reagent grade waste acid: V, 15 A (68.2 mA/cm2);0 , 20 A (90.9 mA/cm2);0 ,30 A (136.4 mA/cm2).
\
4
9-
D
\
+e
13--
,
\
\
\
7°C
02
\
\\
Figure 3. Concentrate concentration as a function of concentrate flow rate: V , 15 A (68.2 mA/cm2);0 , 20 A (90.9 mA/cm2);0 , 30 A (136.4 mA/cm2).
, -
IO--
9
I2
---------
4
, 15
, 10
21
'\. -v
0
3
,
, 24
27
30
33
36
C: (NGRMALITYI
Figure 6. Power consumption as a function of depleted solution concentration for treatment of a reagent grade waste acid and production of a 50 w t O/O H2S04 concentrate: V , 15 A (68.2 mA/cm2); 0 , 20 A (90.9 mA/cmZ);0 ,30 A (136.4 mA/cm2).
34r
/-----7)-
;/'
w 24
-A-
, 14
,z
15
; : /V /
ia
2I
__---V ,
,
24
27
3o
33
33
C INORMALITY)
Figure 4. Current efficiencies as a function of depleted solution concentration for treatment of a reagent grade waste acid and production of a 50 wt O h HzS04 concentrate: V , 15 A (68.2 mA/cm2);0, 20 A (90.9 mA/cm2); 0 ,30 A (136.4 mA/cm2). a range of depleted solution concentrations of 40-135 g/L of H2SO4 while producing a concentrate of 50 wt % (14.3 N) HzSO4. The makeup solution contained 139-141 g/L of HzS04 in all runs. The strong influence of product flow rate on the concentration achieved is shown in Figure 3. When the concentrate concentration is above 13 N, flow rate changes of only 0.020.03 mL/min cause concentration changes of, at least, 1 N at all three current densities. Figure 4 shows current efficiency as a function of the depleted solution concentration (Cd) at 15, 20, and 30 A. The data correspond to the treatment of a reagent grade waste acid and production of a 50% concentrate product. It is clearly seen in Figure 4 that increasing current density increases current efficiency. Similar data as a function of the concentrate concentration are given in Figure 5. T h e power consumption for production of a 50% sulfuric acid concentrate as a function of a depleted waste of varying concentration is shown in Figure 6. T h e power requirements at 15 A were much higher than those for operation a t 20 or 30 A. T h e power requirements a t 20 A and those at 30 a were nearly the same until the waste acid was depleted to about 1.5 N HzS04. Power consumption and current efficiency data were also
obtained for the processing of actual titanate waste acid solutions at 15 and 20 A. Power consumptions for the HST waste acid as a function of Cd at 15 A are compared to those for reagent grade acid in Figure 7. There is no clear trend in the comparison due to the limited number of data points. T h e range of power consumptions for both is fairly close, however, except for the HST waste depleted to 2.63 N H2S04. T h e corresponding concentrate produced in obtaining that point had a concentration of 15.1 N H 2 S 0 4while concentrates of only 14.1-14.2 N HzSO4 were produced in obtaining the other three points. The higher concentrate concentration, then, probably caused the unexpectedly high power consumption for the H S T waste acid depleted to 2.63 N H2SO4. Figure 8 shows power consumptions for a range of concentrate concentrations while treating HST and reagent grade waste acids. Though both acids are depleted to comparable levels, power consumptions for treating the HST waste acid are consistently higher. T h e requirements for the HST acid are about 1-2 kWh/kg higher than for the reagent acid throughout the entire range of concentrations investigated. Power consumption data for treatment of H S T and L S T acids a t 20 A are shown in Figure 9. These data are for production of a 50% (14.3 N ) sulfuric acid concentrate. T h e data for the HST waste acid compare quite closely with the reagent grade sulfuric acid but three of four L S T points show considerably higher power requirements. Inspection reveals that the three high points correspond t o concentrate products of 15.0-15.15 N H2S04which is a somewhat higher concentration than most of the data points for the H S T and reagent acids. A similar occurrence for treatment of HST acid a t 15 A was mentioned above. Figure 10 compares power consumptions for the HST, LST, and reagent waste acids a t 20 A as a function of C,. Again, the data for HST acid agree quite closely with that for reagent grade sulfuric acid but power con-
Ind. Eng. Chem. Prod. Res. Dev., Vol. 17, No. 2, 1978
15
25
20 C,
153
30
(NORMALITY)
Figure 7. Comparison of power requirements for treatment of HST waste acid and a reagent grade waste acid as a function of c d : V ,15 A, reagent waste acid; ~ , l A,5 HST waste acid.
C, ( N O R M A L I T Y )
Figure 10. Comparison of power requirements for treatment of HST, LST, and reagent grade waste acids as a function of C, at 20 A: 0 , reagent waste acid; 0 , HST waste acid; + , LST waste acid. Table IV. Impurity and Acid Concentrations of Treated HST Waste Acid and Concentrate Product"
Constituent Ti02
Fe A1 74 9
10
ll
12
13
14
15
16
Mg
17
C, (NORMALITY)
Figure 8. Comparison of power requirements for treatment of HST waste acid and a reagent waste acid as a function of C,: V , 15 A, reagent waste acid; V , 15 A, HST waste acid.
l4
1
IC
5
20
2 5
30
Cd I N O R M A - T Y I
Figure 9. Comparison of power requirements for treatment of HST, LST, and reagent grade waste acids as a function of C d at 20 A: 0 , reagent waste acid; 0 , HST waste acid; +, LST waste acid. sumptions are slightly higher throughout the range of concentrate concentrations examined. An analysis of concentrate and of depleted solution produced from HST waste acid is given in Table IV. T h e concentrations of all constituents except, of course, sulfate and sulfuric acid were higher in the depleted H S T than they were in the feed. Conductivities of HST and L S T waste acids both before and after treatment are given in Table V. For comparison, conductivities of reagent grade HzS04 solutions of the same concentration are also given. In all cases, the specific conductance of the waste acid was significantly lower than the reagent grade acid of the same concentration. T h e differences in conductivity ranged from 11.7 to 23.3%. Discussion Membrane Suitability Tests. T h e first rims clearly showed that the Ionac membranes would not be suitable for
Depleted HST 3.17 0.47 3.30 5.40 33.0 210
Concn. gIL Concentrate product (composite)
Na Total SO4 HyS04 122 a HST = from high sodium titanate.
0.47 0.08 0.25
0.38 2.20 669
680
contact with the anolyte. Though membrane properties were not determined, the darkened and burnt appearance of membranes were obvious indications of degradation. Results of tests using A-101 membranes alone were, also, not surprising. Permeabilities of the Asahi membranes were expected to be high because the membranes had been used in dialysis experiments (Senoo e t al., 1973). Also, the thinness (0.21 mm) of the A-101 is a desirable characteristic for dialysis membranes t o facilitate diffusion. T h e low efficiencies obtained with A-101 membranes can most likely be attributed to back-diffusion of sulfuric acid. Minimum concentrations exhibited by the depleted solutions indicated that there was considerable diffusion of sulfuric acid from the concentration t o the depleted solution. Since neither the A-101 nor the MA-3475 membranes was acceptable alone, a combination was attempted. Two-Stage Operation. Results of two-stage continuous tests did show considerable improvement over batch tests, but quantitiative results were difficult to obtain. Fluctuations of only 0.03 mL/min in flow rates caused noticeable changes in product concentrations. It seems contradictory that both current efficiencies and power requirements for the concentrating state (11)are lower than those for the depleting stage (I) but this result can be readily explained. Membrane permselectivity decreases with increasing concentration in solution, as seen in
where h ' is the ~ Donnan ~ ~ potential, ~ 4 is electric potential, z, is the electrochemical valence of species i, F is the Faraday constant, R is the gas constant, T is temperature, a, is the thermodynamic activity of species i, and L', is the partial molal volume of species i. Since all feed and product concentrations are higher in stage I1 than in stage I, a decrease in membrane
154
Ind. Eng. Chem. Prod. Res. Dev., Vol. 17, No. 2, 1978
Table V. Specific Conductance of Waste Acid Solutions Compared t o Reagent Acid Solutions Concn, Acid
g/L of H2S04
O a t
Specific conductance, mho/cm
HST 304 0.63 Treated HST" 104 0.30 LST 370 0.65 Treated LST 121 0.33 Reagent 304 0.72 Reagent 104 0.385 Reagent 370 0.736 Reagent 121 0.43 HST = from high sodium titanate. LST = from low sodium titanate. WT %
selectivity is to be expected. Also, a combination of membranes was used in the concentrating stage while the depleting stage used only the less permeable Ionac membranes. Both selectivity and permeability accounted for the relative magnitude of the efficiencies for both stages. Because higher efficiencies were achieved in stage I, it was expected that power requirements would be lower than for stage 11. Yet, higher power requirements actually were obtained. One might expect that result to be caused by a lower conductivity of the less concentrated solutions in stage I. Figure 11,however, shows that conductivities for concentrated solutions decrease just as steeply as for depleted solutions. Also, the voltage for operation a t 10 A was about the same for both stages. The difference, therefore, is artificial, not real. Though power inputs to each stage are about the same, the higher concentration of the concentrate product in stage I1 lowered the consumption per kilogram of sulfuric acid. The indicated trend of increasing power consumption with increasing current density in stage I, though not conclusive, is probably real. This same trend was observed by Horner et al. (1955) for continuous treatment of pickling liquors. When current is increased, power consumed by overvoltages and by solution and membrane resistance is increased. One hopes, then, for a lowered relative effect of diffusion on the total flux to increase current efficiency. Such a result might be inferred from either the general equation Jz
= (Ji)diff
+ (Ji)el + (Ji)con
(2)
where J is the flux, J,,, is the convective flux, J&ff is the diffusional flux, and J,, is the electrical transference, or the sulfate ion flux equation
where C represents concentration, D is the diffusion coefficient, and f is the molar activity coefficient. If the effect of diffusion on the total flux is already low, however, no appreciable increase in current efficiency will be achieved by increasing current density. Likewise, if the relative effect of back-diffusion is large but not decreased by increases in current, the current efficiencies will not improve with increasing current density. The latter case could occur if diffusion is increased by a higher temperature a t a higher current density. The use of heat exchangers eliminated the possibility of such a temperature effect. One-Stage Operation. Data similar to those displayed in Figure 4 were obtained by Lewis and Tye (1959) for treatment of pickling liquor. They found that a t 100 mA/cm2, the catholyte concentration had no effect on the current efficimcy (with an anolyte of 2.5 N H z S O ~ )but , a t 40 mA/cm2, the cur-
H,SO,
Figure 11. Specific conductance of sulfuric acid solutions at 18 "C (Lange, 1967).
rent efficiency gradually decreased as the catholyte concentration was reduced below about 2 N. The effect of catholyte concentration on current efficiency is, essentially, the effect of diffusion for a constant anolyte concentration. The high anolyte concentrations required in the present study created large concentration gradients between the concentrated and the waste acid solutions. As a result, 136.4 mA/cm2 was not a high enough current density to completely eliminate the effect of diffusion on current efficiency. As predicted by theory and discussed earlier for two-stage operations, however, the relative effect of diffusion is obviously reduced by increased current density. The independent effect of concentrate concentration on current efficiencies (Figure 5) shows more than a diffusional influence on current efficiency. If diffusion were the only factor reducing efficiency, then the maximum current efficiencies in Figure 5 would be the same as those in Figure 4. The maximum current efficiencies in Figure 5, however, are higher for all three current densities. A likely explanation for this result is that the membrane permselectivities increased as the concentrate concentrations decreased. Comparing Figures 4 and 6, it can be seen that increases in current efficiency are beneficial for reducing power consumptions. A t 30 A, however, the point is reached where further increases in current will cause increases in power consumed by hydrogen and oxygen overvoltages and resistance heating that will not be compensated by additional increases in current efficiency. The minimum power consumption for treatment of reagent grade acid and production of a 50% (14.2 N) H2S04 concentrate was 9.2 kWh/kg of HzS04 a t 30 A. For the two-stage process a t 10 A, the comparable power requirement was 10.9 kWh/kg of H2S04. If current efficiency could be increased, power requirements could be greatly reduced. For a 100%efficient electrodialysis cell, power requirements can be readily calculated once voltages are estimated. Using a cell voltage of 4.2 V the 15 A power requirement is 2.0 kWh/kg of H2S04. Likewise, for 4.6 V a t 20 A, the power requirement is 2.2 kWh/kg and for 5.5 V at 30 A, the power requirement is 2.6 kWh/kg of HzS04. For a constant current efficiency, the power consumption always increases with increasing current density due to power losses to solution and membrane resistances. Actual Waste Acids. All actual waste acids had a slightly yellowish color when fed to the stack but the depleted waste solutions leaving the stack were deep violet. Since there was no color change when treating reagent grade acid, the color must have been produced by an impurity in the waste acid. The color change from yellow to violet can most likely be at-
Ind. Eng. Chem. Prod. Res. Dev., Vol. 17, No. 2, 1978
Table VI. Concentrations of Contaminants in Pregnant and Barren Solutions and Ti02 Precipitate Using Acid Regenerated by Electrodialysis Concentration, ppm Element Pregnant solution Barren solution Ti02 precipitate Cr
v
Mn Nh cu co Fe Zr
10 40 500 20
... ... ...
...
10 50 100 10
...
... ... ...