Dev. Chem. Eng. Mineral Process. 14(1/2), pp. 71-84, 2006. Production of Hydrogen by Electrolytic Purification of Water G. Mathieson”*,A. Langdon’ and G. Jamieson’ Departments of Chemistry and Materials and Process Engineering, University of Waikato, Private Bag 3105, Hamilton, New Zealand Works Filter Systems Limited, PO Box 20-271 Te Rapa, Hamilton, New Zealand ~ Electrolytic purification of water is not new but has never been widely applied because of failure to meet realistic standards of economy. A continuous process with low cost consumables can compete with standard water treatment processes, even without energy recovery in the form of hydrogen. In order to extend the lower limit of feasible conductivities a specialised cell conjguration was developed. This utilkes a novelform of corroding anode that produces a highly eflective flocculant, in close proximity to a cathode, separated by a simple porous membrane. The active surface of the corroding anode is continually replenished. Flocculant clearance is achieved by high water flow velocity in a flume. Hydrogen is produced separately from anodic products. Other functions like electrolytic flotation, redox and sterilisation are synergistic. Because electrical energy input dominates the operating cost there is a strong incentive to recover the hydrogen, which is produced at about 0.02 Nm’ per mJ of water treated and about 25% eficiency. Introduction Establishing economic means of producing hydrogen is by definition essential to use of hydrogen as an energy carrier in a hydrogen economy. While electrolysis of water is a well-known means of producing hydrogen, the required electrical energy has a * Author for correspondence (gam 1O@waikato.ac.nz). G.Mathieson, A . Langdon and G.Jamieson relatively high value and some is lost in the process. If the hydrogen that is produced is used either for heating by combustion, or to make electricity in another lossy step, then the entire process cannot be justified on the basis of efficiency. While using the hydrogen as a form of energy storage from low electrical tariff periods has been considered, there is another possibility. By achieving another process goal simultaneously, electrolysis may become economic. Electrolytically induced flotation, flocculation and sterilisation are mechanisms that can be used to purify water (Sixpence, 1977; Musquere, Ellingsen et al., 1983; Osasa, Nakakura, et al., 1993; Ordonez, 1997; Tao, 1999), which would otherwise have to be carried out by other means. The hydrogen and oxygen bubbles produced by electrolysis are small enough to attach to fine contaminants and float them out, in analogy to traditional dissolved air flotation systems. The corrosion of a trivalent anode produces metal ions that rapidly hydrolyse into amorphous floc particles, in analogy to familiar chemical flocculation. The production of oxidising agents at a non-corroding anode leads to disinfection of the treated water. The striking advantage of electrolytic treatment is that combined functions of flotation, flocculation and sterilisation can be integrated in a single structure. For example, if both bubbles and flocculant are released from an electrode, the bubbles will rapidly float out the floc, allowing for simple separation. Moreover, there appears to be a synergy between nucleation of bubbles and flocculation (da Rosa and Rubio, 2005). This is taken to the extreme by production of bubbles and flocculant in close proximity at an electrode surface. Theoretical analyses, field trials and laboratory experiments have been carried out to determine the most promising commercial applications. Both the conductivity and pH of the water are crucial to whether or not the processing is cost-effective. In certain conditions electrolytic water treatment is both more effective and potentially less costly than established methods (Rohland, 1915 ; Stuart, 1946; Bonilla, 1947; Khosla, Venkatachalam, et al., 199I. ; Robinson, 2000; Tetrault, 2003A). In all cases, maximising the production efficiency of hydrogen, i.e. lowering cell voltage by reducing cathodic overvoltage, was considered as a sub-goal to maximising the cost-effectiveness of the entire process. From t h s point of view, energy recovery in the form of hydrogen is seen as a bonus. Municipal 72 Production of Hydrogen by Electrolytic Purlfication of Water wastewater plants that generate electricity from the natural gas fermented from the bio-solids in wastewater can use the recovered hydrogen with little or no modification to existing equipment. Because electrolytic processing is also applicable to relatively small-scale operation at the scale of individual factories, there is an opportunity to develop smaller scale hydrogen recovery and usage apparatus. Theory and Modelling For a simple direct current supply voltage VDc, the main sources of electrical energy loss in an electrochemical reaction are shown as either opposing voltages, or resistances, in Figure 1. RE VDC RC VC, Figure I . The DC electrical equivalent of an electrolytic cell. The equilibrium potentials for the anode and cathode are VAeand VC, respectively. These cannot necessarily be considered constant with respect to current because both the pH and temperature are subject to alteration by the passage of current. The resistance of the supply, Rs, including metallic leads and the electrolyte are considered to be ohmic (linear) functions of the current. This is well justified for Rs, but as shall be explained later, the electrolyte resistance is likely to depart from ohmic behaviour at high current densities. The anodic and cathodic overvoltages are qA and qc. These are defined by the Butler-Volmer relation in Equation (1) and its simplified 73 G. Mathieson, A. Langdon and G. Jamieson form, the current-centric Tafel equation in Equation (2) (Bockris and Reddy, 1970), accurate when In11 > 118 mV (Riley, 2003), where n is the number of electrons transferred by the reaction and 7 is the overpotential. These overpotentials may be considered as equivalent to the voltages across the non-linear resistors RA and RC respectively, and b s is reflected in Equation (3) where REis the electrolyte resistance, hence: ...(2) ...(3) Equation (2) shows that in order to get significant current to pass, an overvoltage must be applied, yet Equation (4) shows that an overvoltage contributes directly to energy losses. Hence all practical designs must apply some form of compromise, so that the area of the electrodes is not too large while the energy efficiency is not too low. The acceptable limits for either depend on the application. Standard industrial electrochemical processes use high concentration, high conductivity electrolytes, since this minimises the loss attributed to the term VRE in Equation (4). For situations where the electrolyte is dilute and cannot be altered for economic or environmental reasons, such as in wastewater, the loss due to electrolyte resistance can be significant, even dominant, because while the electrode overvoltages are functions ofthe logarithm of the current density, the voltage across the electrolyte is linear or supra-linear with current. The conductance of the electrolyte can be estimated from its conduct:ivity (a), the area of the electrode interface ( A ) , and the average gap between the electrodes ( I ) , as given by: s, =u-A 1 74 .. . ( 5 ) Production of Hydrogen by Electrolytic Purification of Water -Cathodic O\snaltage for Hydrogen edutlon (V) 100 10 E -j -Electrolyte Voltage Drop (V) for 10 mm gap - ’ Electrolyte Voltage Drop (V) for 1 mm gap 0.1 >” 001 -Electrolyte Voltage Drop (V) for 0.1 mm gap 0 001 . . .”. .. ElectrolyteVoltage Drop (V) 0.0001 0 20 40 60 80 Current Dendty A m“ 100 at 10 mm gap for 6.6 M KOH at 70 C a r d 100 Slrn Figure 2. How electrolyte voltage drop is lowered to the order of the cathodic overvoltage by reducing the gap; based on an iron cathode and fluid with 0.02 S m-’ conductivity like tap water (resultfor concentrated electrolyte shown for comparison). Whle increasing the surface area offers lower losses, it also leads to an increase in both size and capital cost. The next obvious target is to lower the distance between the electrodes. Figure 2 shows how the voltage drop across the electrolyte gap becomes dominant if the electroIyte conductivity is reduced, and how reducing the gap size can lower the losses. This result is based on iron cathode exchange current data (Pletcher and Walsh, 1990). Whle Equations (1) and (2) are accurate at low to moderate current densities, at high current densities the overvoltages can be raised by: inadequate diffusion of dissolved reactants to the electrode, inadequate diffision of dissolved or precipitated products from the electrode and poor clearance of gas bubbles from the electrodes. However, if the bubbles of around 10 pm diameter that are formed are released rapidly enough from the surface they stir the diffusion layers effectively, greatly improving mass transport (Wendt and Kreysa, 1999). Partial occlusion of the electrolyte gap by gas bubbles leads to an increase in effective electrolyte resistance. Two extreme forms are explored for illustration. (i) If a contiguous sheet of gas forms between two electrodes from bubbles that coalesce, ths will grossly increase the effective resistance of the electrolyte even if it does not fill the entire space, hence this is to be avoided if at all possible. 75 G.Mathieson, A. Langdon and G. famieson (ii) Evenly dispersed fine bubbles will reduce the electrolyte conductivity from KO to K slightly more than predicted by the linear gas void to total volume ratio, 4, (Wendt and Kreysa, 1999). Theoretical estimates of the conductivity ratio are in a narrow range defined by Equation (6), i.e. if the gas bubbles occupy half the space then the resistance will be a little more than doubled: 3 (I-&,)* - .. K <-<I-&, KC! For the simple case of parallel electrodes with plug flow of fluid that dominates the motion of the bubbles, the terminal part of the flow will have the highest gas loading. The build-up can be modelled as spatially linear so that the average level is the same as that at the halfway point. Hence the term triangular gas void ratio, EA shown in Equation (7) below, equal to half the value of the voiding at the exit point, which can be used in place of cg in Equation (6).Given that V,,, is molar volume in m3 per mole and D, is the required faradic volumetric rate of hydrogen and oxygen bubbles dose per fluid flow for an example of flotation treatment by simple electrolysis of water, a low value of DIof 60000 A m*3s (1 N(Vmin) ) results in a of about 1%. Note that this does not depend on the gap size, though for very narrow gaps coalescence may cause the sheeting problem. Furthermore the utility in reducing the gap size is limited by increases in fluid pressure drops and greater difficulty in preventing contact of the electrode pairs. For example, the pressure drop in laminar flow between parallel plates is proportional to the cube of the reciproc.al of the gap (Wendt and Kreysa, 1999). &, =--3 IVm 4 F Dl (m3of hydrogen and oxygen gas per m3 of liquid) .. Experimental Details Consider an example of field work carried out at a tannery. Tannery effluent is particularly noxious, because the chemicals used in the tanning process are preservatives, including chromium compounds, and the pH tends to be high. 76 Production of Hydrogen by Electrolytic Purification of Water Electrolytic processing is feasible because the high salt content of the effluent gives a high electrical conductivity of around 1.5 S m-'. While research on the subject dates back to early in the 20" century, (Rohland, 1915; Fassina, 1938) commercialisation has not occurred, perhaps due to excessive power consumption. Other researchers have produced promising results with rendering plant effluent (Tetrault, 2003A), but with cell voltages of about 50 V (Tetrault, 2003B). Toward the end of 2004, Works Filter Systems Limited was invited to trial an electrolytic effluent processor at a site in the Eastern Waikato region of New Zealand. The design of the field-trialled system is protected by NZ Provisional Patent 537700. This utilises a novel form of 6060 alloy aluminium as a corroding anode that produces a highly effective flocculant, in close proximity to a steel cathode, but separated by a simple porous nylon mesh membrane. The active surface of the corroding anode is continually replenished. Flocculant clearance into the water is achieved by high flow velocity in a flume. Hydrogen is produced outside this flow and is therefore not mixed with anodic products. The effective cathode area was about 0.4 m2, driven to current densities of up to about 100 A m-2.A smaller version was made for laboratory work. The composition of 6060 Aluminium alloy is shown in Table 1 and indicates that for levels of up to 100 mglL of A1 added by corrosion, the total level of additional metals is likely to be less than 1 mg/L with most elements well below 1 mg/L. In particular, Cr is added at the rate of 0.0005 mg per mg of Al, which is insignificant compared to the background level in Tannery Effluent at any reasonable level of A1 addition. Some of the more noble impurities such as Cu may not be anodically oxidised, but released to the flow in metallic form as the bulk alloy disintegrates. Because of the nylon membrane, the diffusion of metals from the main flow stream to the cathode is limited and the rate of plating-out is low. Power supplies: 1. D3800 DC power supply from Dick Smith Pty Ltd with < 1% voltage ripple. 2. Continuously adjustable full-wave rectifier composed of an earthleakage protected 230 V single phase supply, a 4 A variac kindly supplied by Physics Department, Waikato University, a step-down transformer, and four 50 A bridgerectifier modules in parallel on a heatsink. The output was fed to the elecrolyser via 77 G. Mathieson, A. Langdon and G. Jamieson Table 1. Impurities in A1 Alloy EN A W-6060 (MA TTER Project 2001)). Element I Mg Si Fe Total other Zn Ti Other elements Cr I A1 remainder 1 Weight % 0.35 - 0.6 0.30 - 0.6 0.10 - 0.30 < = 0.15 < = 0.15 < = 0.10 < = 0.10 < = 0.10 < = 0.05 < = 0.05 > 98.35 80 A cable, and protected from short-circuit by a 100 A high rupture current (HRC) fise. This power supply could operate continuously at 0 V to 30 V fill wave rectified rms at 0 A to 40 A rms,and supply up to 60 A for a few minutes at up to 10 V, limited by rectifier heat sinking. The peak voltage of this supply is approximately 1.5 times the rms value, and the 100 Hz harmonic component is significant. The continuous variability enabled fine control of the current delivery during experimental work. Electrical measurements were made in the field with a Fluke 337 clamp-meter. pH measurements were made in reference to regular 2 or 3 point calibration. Results and Discussion The most surprising result in the field trials was the effectiveness of the treatment at high pH, including very low residual aluminium and chromium levels for the pH of the outflow (Pontius, 1990). The processing can lower the pH when the initial pH is above 8. This would be expected where the anolyte and catholyte are separated, but a pH drop was observed even in the whole outflow. This suggests that the normal protective layer is continuously disrupted by electrolysis, and oxygen sourced from outside the electrolytic system is able to corrode additional aluminium. ‘These effects are still under investigation in the laboratory, supported by past studies (IA, Qu, et al., 1999; Zeppenfeld, 2003). 78 Production of Hydrogen by Electrolytic Pur fication of Water The current voltage characteristic was largely dependent on the conductivity and state of the anode surface (Chen, Chen, et al., 2002). To a great extent the characteristic of initial current with voltage was linear, suggesting that the electrolyte and membrane dominate the total resistance. Most interestingly, in situations where the treatment was effective, the initial current with a new anode was low, but the current increased steadily over an hour or two to a level that could exceed double the initial value. In cases of low salinity the current would usually decline monotonically from the initial value and treatment was unlikely to be effective for very long, if at all. Cell voltages from 3.5 V to 10 V were effective in treating tannery effluent. Current doses (current per volumetric flow rate, or charge per volume) of between 1 and 5 A/(litre/min) were necessary to cause good flocculation. The low end was sufficient for saline solutions, while the high end was necessary for heavily loaded tannery effluent. Note that 1 A/(litre/min) corresponds faradically to about 5.59 ppm of A13+ or 10.56 ppm A1203equivalent. Figures 3, 4 and 5 show the results from a trial on 12 May 2005. Turbidity, measured in nephleometric turbidity units (NTU), is a good (though not linear) indicator of the amount of suspended solids in the water. Because the quality of the tannery effluent changes so quickly, the samples were taken over just a half hour period, with inflows collected at either end of the time period. The figures show that generally, the higher the dose of aluminium, the greater the reductions in outflow turbidity, pH and chromium, though the change of the inflow also has an impact. The cost-effectiveness of the electrolytic processing, including cost of electricity and aluminium, is favourable compared to dosing with polyaluminium chloride (a commercial, partly hydrolysed, flocculant). The specialised anode has a clear advantage over other corroding anodes. Trackmg of the water flow due to the normally inevitable uneven corrosion is eliminated. The clearance of flocculant from the specialised anode is high, as long as the flow speed is at least about 0.1 m s-’. This is in stark contrast to earlier models where the floc would rapidly clog the anode and block both the flow and the current. Hydrogen production was estimated faradically around 6.96 x lo” m3 of gas (at STP) per m3 of fluid treated per charge dose. Moderate treatment at about 79 G. Mathieson, A. Langdon and G. famieson 3 N(litre/min), equal to 1.8 x lo5 C m-3, gives about 20 litres of hydrogen per cubic metre of water treated. The efficiency of hydrogen production and the fraction of energy that can be recovered from the electrical input are identical, and may be estimated by the ratio of the standard potential for electrolysis of water (1.23 V) to the cell voltage. For a cell voltage of 5 V, the efficiency is about 25%. The purity of the hydrogen obtained depends both on the quality of the water and the use of separating membranes. The high oxygen demand of many wastewaters will consume at least some of the oxygen produced, but the traces of CH,,CO, C02 and sulphur compounds that are present in mast organic wastewaters will inevitably contaminate the hydrogen if the fluid flows along the cathode surface. This is of little or no consequence if the fuel gas (BactogasQ - a registered trademark of Works Filter Systems Limited) is simply combusted. Most of these contaminants are either harmful or fatal to known fuel cell catalysts (Vennekens, 2000). For premium uses of pure Figure 3. Comparison of electrolytic treatment on tannery efluent with polyaluminium chloride dosing (inflow at 2:30pm: pH 8.9,850 NTU, I7250 uS/cm at I9”C; at 3 pm: pH 8.6, 900 NTU, 16560 uS/cm at 19°C). 80 Production of Hydrogen by Electrolytic Purification of Water Figure 4. Residual metals comparisonfor electrolytic treatment of tannery effluent. Figure 5. pH data comparisonfor electrolytic treatment of tamely efluent.. 81 G. Mathieson, A . Langdon and G.Jamieson hydrogen, this suggests that a gas impervious, but ion-permeable membrane, be employed and that uncontaminated liquid be used as the catholyte. While this will restrict the treatment to that offered by the anode compartment, the only known hnction that would be impaired is electroplating of impurities onto the cathode. In some cases there is overall merit in having a relatively high cell voltage for a corroding anode system, because it both enhances the pH changes in the vicinity of the electrodes and the space-time yield, but at very high anodic overvoltage the current fraction of anodic oxygen production increases at the expense of flocculant production and quality. The ultimate limit for anodic overvoltage is when partly corroded fragments of the anode are released by the disintegration of' the anode surface. These fragments are not effective as flocculants and they greatly increase the amount of metal added to the water. Applications The ideal treatment system depends on the water to be treated and the required quality of the outflow. Some examples of the various field applications that have been tried are listed by water type and suitable treatment combination in Table 2. Beside iron water (Coup and Campbell, 1964), tannery effluent (Fassina, 1938) and laundry water recycling, other trialled field applications include: landfill leachate, municipal wastewater terminal effluent, and sludge returned from a clarifier. The most promising of these, despite its relative difficulty, is landfill leachate (Ihara, Shimada, et al., 2003) because of the value in making an inflow that contains numerous pervasive and persistent toxins into an outflow that is acceptable for discharge into stormwater drains. A novel specialised processor has been designed for this application. A longer term strategy is to target point sources of, for example, heavy metals and treat these streams before they enter municipal wastewater conduits. If the treatments are successful, large amounts of impurities will be separable from the bulk of the water. The means of separation, besides electroflotation, include settling and filtration. In some ways, the challenge of disposal will remain, from a much more concentrated form. 82 Production of Hydrogen by Electrolytic Purification of Water Further laboratory work is continuing in order to elucidate some of the mechanisms of treatment and establish measurable figures of merit, such as pH change, for use in control and evaluation in applications. Table 2. Examples of electrolytic water treatment options. Influent water type Flocculation Flotation Redox Sterilisation Ideal anode Bore water with ferrous iron at about 20ppm. Conductivity of about 2000 pS cm.' by NaCI. Achieved by rapid oxidation of the iron to Fe 111 using a noncorroding anode - no tlocculant needed. Cathodic hydrogen gas will float out the flocc. Direct anodic oxidation of Fe II to Fe Ill may contribute directly. Chloride IS converted to chlorine at the anode. That remaining atter Fe oxidation is free chlorine. Noncorroding. Must not release toxic substances. I 0000 ps cm-' Tannery Effluent with Cr, suspended solids and high but variable pH. Corrode an aluminium or iron anode to produce amorphous hydroxide. Cathode gas will float out the ilocc. Numerous but not easily controllable. pH falls in most cases. Processing improves the health of oxidation ponds downstream. Near pure Aluminium or Iron. Laundry water recycling, water that IS high In pH and salinity with a lot of fibrous matter. Flocculate after the gross solid is removed. Float out the gross solid first. Incidental. The flocculation may reduce PH. Sterilize after flocculation so that the chlorine demand is low. Noncorroding for floating/ sterilisation . Corroding for making fioeculant. Conclusions Electrolysis is demonstrably an economic way to purify some particularly noxious types of water, with the added benefit of energy recovery in the form of hydrogen. By using a specialised corroding anode (New Zealand Provisional Patent number 537700) and other proprietary water treatment apparatus including the Rutherford Award winning Porous Ceramic Dual Media Filtration (PCDM) system, Works Filter Systems Limited (New Zealand), and Works Filter Systems Pty (Australia) intend to supply commercial electrolytic water treatment products. 83 G. Mathieson, A. Langdon and G. Jarnieson Acknowledgments The PhD project work on which this paper is based is supported by Technology New Zealand and Works Filter Systems Limited. Special thanks to the Hamilton City Council for the use of their wastewater processing sites for field trials. References I. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16. 17. 18. 19. 20. 21. 22. 23. 24. 25. 26. 84 Bockris, J.O.M. and A.K.N. Reddy, 1970. Modern Electrochemistry. Plenum Press, New York. Bonilla, C.F.. 1947. Possibilities of the electronic coagulator for water treatment. Inter nndSewnge, 85,21,22,44,45. Chen, X . , G.Chen, et al., 2002. Investigation on the electrolysis voltage of electrocoagulation. Chem. Eng. Sci'.,57(13), 2449-2455. Coup, M.R. and A.G. Campbell, 1964. The Effect of Excessive Iron Intake upon the Health and Production of Dairy Cows. NZ Journnl ofAgriculturnl Resenrch, 7(4), 624-638. da Rosa, J.J. and 1. Rubio. 2005. The FF (flocculation-flotation) process. Miner. Eng.. 18(7), 701-707. Fassina, L., 1938. Purification of tannery effluents. J. Am. Lenther Chem. Assoc., 33, 380. Ihara, L,E. Shimada, et at., 2003. Treatment of landfill leachate. Absfmcts offnpers, 225th ACS Nntionnl Meefing, New Orleans, LA, United States, March 23-27,2003, 1EC-265. Khosla, N.K., S. Venkatachalam, et al., 1991. Pulsed electrogeneration of bubbles for i:lectroflotation. J. Appl. Electrochem., 21(1 I), 986-90. Lu, G.,J. Qu,et at., 1999. The electrochemical production of highly effective polyalurmnium chloride. Wnter Research, 33(3), 807-813. MATTER Project, The University of Liverpool, 2001. Alloy Composition Details. Musquere, P., F. Ellingsen, et al., 1983. Electrotechnics in drinking and wastewater. Wnrer Supply, 8(Special Subject (2-3)), SS 8-1 to SS 8-25. Ordonez, G.A., 1997. In situ electrolytic disinfection. Proceedings - water quality technology conference, Costa Rica. Osasa, K., H. Nakakura, et al., 1993. Treatment of colloidal waste material by electroflotation using sacrificial electrodes. Kngnku Kogaku Ronbunshu, 19(2), 317-324. Osipenko, V.D. and P.I. Pogorelyi, 1977. Electrocoagulation neutralization of Chromium Containing effluent. Mernllurgist (English translnfion),21(9-10),44-45. Pletcher, D. and F.C. Walsh, 1990. lndustriol Elecfrochemisfry,University Press, Cambridge. Pontius, F.W., 1990. Wnkr Qualiw nnd Trentment - A hnndbook ofcommunity wntersupplies, McGraw-Hill. Riley, J . , 2003. Dynnmic Elecfrochemisfry.University of Bristol, Bristol, UK. Robinson, V., 2000. A new technique for treatment of wastewater. Enviro 2000 Ozwn,rerOzwnste Conference. Darling Harbour, Sydney, Australia. Rohland, P., 1915. The action of electrolytes on eMuents containing organic waste. KislloidZeitschrifr, 16, 58-60. Stuart, F.E., 1946. Electronic water purification; Progress report on the electronic coagulator - a new device which gives promise of unusually speedy and effective results. Inter and Sewnge, 84(May), 24-26. Tao, Y., X.Zhang, et al., 1999. Electropurification of sulfur-containing sewage by electro-oxidizing desulfurization. Elecfrochemicnl nnd Solid-State Letters, 2(3), 133-134. Tetrault, A., 2003A. Electrocoagulation Treatment of Low Temperature Rendering P1,ant Wastewater at an Australian Abattoir. EnvironzO3, NZWWA, Auckland, New Zealand. Tetrault, A., 20038. Personal Communication. Environz03. NZWWA . Auckland, New Zealand: Cell Voltage of EC Pacific Unit is about 50 V. Vennekens, G . , 2000. Fuel Cell Catalyst Poisons - personal communication. G . Mathieson. Wendt, H. and G. Kreysa, 1999. Elecfrochemicnl Engineering: Science nnd Technology in Chemicnl nnd Other lndusrries, Springer. Zeppenfeld, K . , 2003. Electrochemical production of aluminum hydroxide from alkaline aluminate liquors. Aluminium (Isernhngen. Germnny), 79(9), 754-759.