Environ. Sci. Technol. 2009, 43, 8315–8319
H2S(g) Removal Using a Modified, Low-pH Liquid Redox Sulfur Recovery (LRSR) Process with Electrochemical Regeneration of the Fe Catalyst Couple
unfriendly, the iron redox couple (Fe3+/Fe2+) is to date predominantly used within LRSR processes (4, 6). The two steps of the LRSR method with iron as the catalyst couple are shown in eqs 1 and 2.
YOURI GENDEL, NO’OMI LEVI, AND ORI LAHAV* Faculty of Civil and Environmental Engineering, Technion, Haifa, 32000, Israel
Due to the inherently slow kinetics of spontaneous Fe(II) oxidation with O2(aq) at low pH (7) and the low solubility of Fe(III) species at pH values higher than ∼pH 3, the use of chelating agents (EDTA, NTA, and others) has been suggested to attain high Fe(II) oxidation rates while at the same time preventing Fe(III) species from precipitating at the neutral to slightly basic pH of typical LRSR solutions (pH 7-9) (4). However, complex-based LRSR methods suffer from relative rapid degradation of the chelating ligands and eventual catalyst loss (4, 6). The common alternative to the use of chelating agents is LRSR processes that apply biological regeneration of Fe(III). This process has been performed at acidic conditions (pH 1.5-2.7) where the Fe(III) precipitation rate is presumably low (8, 9). In this process Fe(III) is achieved by biological oxidation of ferrous iron to ferric iron by specialized Fe(II) oxidizing bacteria such as Acidithiobacillus ferrooxidans (9-11). Apart from inherent instability associated with methods based on the performance of autotrophic bacteria (sensitivity to low temperatures, low bacterial yield, long recovery periods after failure, need for continuous feeding, etc.) the main reported drawback of biologically regenerated LRSR processes is the formation of Fe(III) precipitates, mainly from the jarosite group (10, 12). Formation of jarosite species can be reduced by lowering pH but at the pH range at which ironoxidizing bacteria operate (1.5 < pH < 2.5) jarosite precipitation will invariably occur and constitute a major problem for long-term operation (12, 13). At pH < ∼1.5 the activity of ferrous-oxidizing bacteria has been reported to significantly decrease (8, 10). The aim of the present study was thus to develop a stable, feasible, and cost-effective physicochemical Fe3+/Fe2+-LRSR process that can be operated for a long period of time with minimal iron loss due to precipitation of Fe(III) species, and with minimal chemicals addition. To this end an electrochemical cell was employed as a novel nonbiological alternative for oxidizing ferrous iron to its active ferric form at low pH (i.e., to regenerate the “catalyst”). An electrolytic approach for Fe(III) regeneration was proposed by ref 14, however this particular process was designed for highly concentrated H2S(g) streams (30%) and operated at extremely low pH (pH < -0.5) and extremely high Fe (56 g L-1) and Cl(200 g L-1) concentrations. Such conditions are neither environmentally/economically sustainable nor necessary for most H2S(g) containing streams. Contrary to LRSR processes which rely on unstable organic ligands, the electrolyte solution in the investigated alternative is potentially stable and its H2S/Fe(II) oxidizing ability does not change with time. To test the feasibility of this approach we investigated whether this process could be sustained for an extended period under conditions conducive to high H2S and Fe(II) oxidation rates, while avoiding the precipitation of Fe(III) species. As shown below, the practical pH at which precipitates do not form was found to be pH e 1.4. To operate with a safety margin, the process was tested at pH 1.0. Apart
Received June 1, 2009. Revised manuscript received September 11, 2009. Accepted September 11, 2009.
A modified pH 1.0 liquid redox sulfur recovery (LRSR) process, based on reactive absorption of H2S(g) in an acidic (pH 1.0) iron solution ([Fe(III)] ) 9-8 g L-1, [Fe(II)] ) 1-2 g L-1) and electrochemical regeneration of the Fe(III)/Fe(II) catalyst couple, is introduced. Fe(II) was oxidized in a flow-through electrolytic cell by Cl2(aq) formed on a Ti/RuO2 anode. pH 1.0 was applied to retard the potential precipitation of predominantly jarosite group Fe(III) species. At pH 1.0, the presence of chloride ions at [Cl-] ) 30 g L-1 allows for both efficient (indirect) electrochemical oxidation of Fe(II) and efficient H2S(g) reactive absorption. The latter observation was hypothesized to be associated with higher concentrations of Fe(III)-Cl complexes that are more highly reactive toward H2S(aq) than are free Fe(III) ions and Fe-SO4 complexes that otherwise dominate pH 1.0 Fe(III) solutions in the absence of a significant Clconcentration. At the described operational conditions the rate of Fe(II) oxidation in the experimental system was 0.793 kg Fe h-1 per m2 anode surface area, at a current efficiency of 58%. Electricity cost within the electrochemical step was approximated at $0.9 per kg H2S(g) removed.
Introduction Development of effective, stable, and feasible methods for hydrogen sulfide removal from gaseous streams has been the focus of extensive research. The approaches for H2S(g) removal can be roughly divided into biological (1, 2) and physicochemical methods (3). Liquid Redox Sulfur Recovery processes (LRSR) (4) are among the most promising and thus intensively studied techniques for H2S(g) removal. LRSR techniques comprise two simultaneous reactions: (1) H2S(g) is absorbed into a solution that contains an oxidizing agent capable of rapidly oxidizing H2S(aq) to solid elemental sulfur which is thereafter removed by gravity settling and (2) the oxidizing agent is oxidized back to its active form, typically using atmospheric oxygen. The first step has been termed “reactive absorption” of H2S(g) (5). The following couples: As5+/As3+, V5+/V3+, Co2+/Co3+, and Fe3+/Fe2+, have been used in various LRSR processes (4, 6). Due to high and specific oxidation potential toward H2S(aq) and the fact that cation species of As, V, and Co are toxic and environmentally * Corresponding author e-mail:
[email protected]; tel: 972 4 8292191; fax: 972 4 8228898. 10.1021/es901594j CCC: $40.75
Published on Web 09/23/2009
2009 American Chemical Society
H2S + 2Fe3+ f S0 + 2Fe2+ + 2H+ 2Fe2+ + 2H+ +
(1)
1 O f 2Fe3+ + H2O 2 2
VOL. 43, NO. 21, 2009 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
(2)
9
8315
from pH, the other pertinent operational parameters that were investigated were the absorption/oxidation solution composition (this same solution also served as the electrolyte in the electrochemical reaction) and the current density applied in the electrochemical cell. Similarly to classic LRSR operation, the proposed modified process consists of two steps: the first step (H2S(g) absorption and oxidation to S0 by a concentrated Fe(III) solution) was carried out using a bubble column reactor at pH 1.0. The effect of this significantly lower pH (relative to biological LRSR systems) was tested to determine its affect on both H2S(g) reactive absorption efficiency and Fe(III) regeneration. To maintain close to steady-state conditions in terms of the Fe(II):Fe(III) ratio, the absorbing/oxidizing Fe(III)/Fe(II) solution is recycled between the bubble column reactor and a much smaller electrolytic cell, where the second step, i.e., Fe(II) oxidation to Fe(III), takes place. The total dissolved iron concentration in the absorption/oxidation solution in the proposed process is 10 g Fe L-1 with the Fe(II) concentration ranging between 1 and 2 g Fe L-1, while the Fe(III) concentration ranges between 9 and 8 g Fe L-1. When Fe(II) reaches a concentration of 2 g L-1 the electrochemical oxidation reactor is switched on until the Fe(II) concentration drops back to 1 g L-1. Maintaining a relatively high and constant Fe(III) concentration is imperative for efficient H2S(aq) reactive absorption (5, 15, 16) while avoiding low Fe(II) concentrations (90%) at an acceptable gas flow rate the Fe(III) concentration has to be at least 9 g Fe(III)/L. The results conform to the conclusions of previous works (5, 15, 16). Figure 1 shows the results of H2S(g) reactive absorption experiments III, IV, V, and VI (Table S1), operated at pH 1.0 with various Cl- concentrations. It can be seen that at this very low pH, in the absence of Cl- ions, the H2S(g) removal efficiency was very low (not higher than 20%). However, in the presence of Cl- ions (starting from a minimum concentration of 10 g Cl L-1) the efficiency improved dramatically to around 70% and went up to 80% at Cl- concentrations of 30 and 50 g L-1 (no significant difference was found between
FIGURE 2. Distribution of soluble Fe(III) species as a function of [Cl-] at pH 1.0 [Fe(III)] ) 9 g L-1; [Fe(II)] ) 1 g L-1; [SO42-] ) 25 g L-1. concentrations of 30 and 50 g Cl L-1). Irrespective of the specific H2S removal efficiencies achieved in these tests (which were not optimized for H2S(g) removal and can be significantly improved) the main new information emanating from Figure 1 is that the presence of a threshold concentration of chloride ions in the absorption/oxidation solution significantly improves the H2S(g) removal efficiency of pH 1.0 Fe(III) solutions. Figure 2 shows the distribution of Cl-, OH-, and SO42ferric complexes at pH 1, as a function of the Cl- concentration (MINEQL+ simulation). In these calculations we assumed [Fe(III)T] ) 9 g L-1, [Fe(II)T] ) 1 g L-1, and [SO42-] ) 25 g L-1. It can be seen that the Fe-Cl complexes constitute a significant percentage of the total soluble Fe(III) concentration, even at relatively low total Cl- concentration of 0.2 M (7.1 g L-1) and that as the [Cl-] increases the concentration of the species Fe(OH)2+, previously reported as the main oxidizing species for H2S (16), significantly decreases. Based on these results we hypothesize that the significantly enhanced H2S(g) removal efficiency observed at pH 1.0 in the presence of a high Cl- concentration results from the formation of one or more Fe(III)-Cl complexes that are apparently much more reactive toward H2S(aq) than the free Fe(III) ion, Fe(OH)2+, and Fe-SO4 complexes that dominate pH 1 solutions in the absence of chlorides. Determining the precise H2S oxidation kinetics equation under these conditions is beyond the scope of this paper. VOL. 43, NO. 21, 2009 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
8317
FIGURE 3. H2S(g) reactive-absorption efficiency in 9 g L-1 Fe(III), 1 g L-1 Fe(II), and 30 g L-1 chlorides solution as a function of pH (Tests V, VII, and VIII in Table S1, [SO42-] ) 25-29 g L-1).
FIGURE 4. Change in Fe(II) concentration with time in indirect electrolysis experiments operated with varying current densities. Operational conditions: [Fe(II)]initial ) [Fe(III)]initial ) 5 g L-1, [Cl-] ) 30 g L-1, pHinitial ) 1.0 ( 0.2, pHfinal ) 1.2 ( 0.2. Linear regression approximation of the rate of Fe(II) oxidation in the interval 2 to 1 g L-1 is shown only for the 0.43 kA m-2 run. Figure 3 shows the pH dependency of the H2S(g) absorption efficiency when Cl- ions were present in the absorbing solution at a concentration of 30 g L-1 (Tests V, VII, and VIII in Table S1). Figure 3 shows that increasing pH from 1.0 to 1.7 in the presence of a high Cl- concentration improves the H2S(g) removal efficiency, but not dramatically so (from ∼80% at pH 1.0 to ∼88% at pH 1.7). Consequently, operation of the LRSR process at pH 1 seems advantageous as no ferric iron precipitation occurs at this pH and H2S(g) removal efficiency is still reasonably good. Electrolysis Experiments. The results of electrolysis experiments carried out with various current efficiencies are shown in Figure 4. With respect to the proposed LRSR process, the Fe(II) oxidation rates of interest are the ones obtained within the interval between 2 and 1 g Fe(II) L-1. Within this concentration range, Fe(II) oxidation rates (dFe/dt, presented in units of g Fe h-1) calculated from the slope of the linear regression of the measured Fe(II) values, are shown in Figure 5 along with the associated current efficiency value, both as a function of the current density (current efficiency is defined as dFe/dt · F · V/I · 100%, where F ) Faraday constant ) 96485.3399 (C (mole e-)-1; V ) volume of electrolyte (L); and I ) applied electrical current (A) and current density ) I/A where A is the anode surface area (m2)). Figure 5 shows that an increase in the current density resulted in higher Fe(II) oxidation rates. However, the highest current efficiency (58%) attained under the experimental 8318
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 43, NO. 21, 2009
FIGURE 5. Fe(II) oxidation rates and current efficiencies as a function of the applied current densities in the Fe(II) concentration interval 2 to 1 gFe L-1.
FIGURE 6. Fe(II) oxidation rate and current efficiency as a function of initial chloride concentration. Operational conditions: [Fe(II)] ) 2-1 g L-1, [Fe(III)] ) 8-9 g L-1, current density ) 0.64 kA m-2. conditions was recorded with a current density of 0.64 kA m-2. In this specific experiment the Fe(II) oxidation rate was 3.7 g Fe h-1. In normalized terms this rate equals 793 g Fe h-1 m-2 anode surface area. Although this is ∼50% less than the highest Fe(II) oxidation rate recorded in this set of electrolysis experiments (7.13 g Fe h-1), the latter was attained at a significantly lower current efficiency (42.8%). The volumetric Fe(II) oxidation rates attained in these experiments are approximately 1 order of magnitude higher than the highest Fe(III) regeneration rates reported in the context of biological LRSR systems (10). Figure 6 shows the results of a second set of experiments in which the current density was maintained constant at 0.64 kA m-2 and Cl- concentration was varied. As shown in Figure 6, the chloride concentration had a significant effect on the Fe(II) oxidation rate up to [Cl-] of about 30 g L-1 (∼0.82 M). Increasing the concentration above 30 gCl L-1 had little effect on the Fe(II) oxidation rate. However, it is noted that when the Cl- concentration was higher than 30 g L-1, lower voltage (corresponding to a lower energy demand) was required to achieve a current density value of 0.64 kA m-2. However, due to potential Cl- loss during inevitable blow-downs expected during operation of the LRSR process, the increase of Cl- concentration beyond 30 g L-1 is probably not justified. To sum up, [Cl-] ) 30 g L-1 was found in the current study to be the most advantageous chloride ion concentration, both from the electrochemical oxidation perspective and also from the H2S(g) reactive absorption standpoint. As mentioned in the introduction, at pH 1, Cl2(aq) (in practical terms) does not hydrolyze to HOCl, and is thus susceptible to volatilization. The potential Cl2(g) loss due to
the electrolysis step was quantified in all the electrolysis experiments. In 6 out of 8 experiments no Cl2 at all was found to escape the system. In the other two experiments (current densities of 0.64 and 1.71 kA m-2 and [Cl-] ) 30 g L-1), the total mass of Cl2(g) that escaped the system throughout the experiment was 7 and 15 mg Cl respectively, i.e., less than 0.06% of total Cl- mass in the experimental system. Despite these low values, minimization of Cl2(g) loss should be taken into account in the design of industrial systems. Summary of Process Considerations and Estimation of the Electricity Cost in the Electrolysis Step. To avoid the formation of Fe(III) precipitates during long-term LRSR process operation the pH of the absorbing Fe(III) iron solution should be maintained as low as pH 1.0. At such pH the efficiency of the H2S(g) reactive absorption by Fe(III) solutions is very low. The presence of chloride ions in solution at [Cl-] > 10 g L-1 was shown to significantly improve the H2S(g) reactive absorption efficiency. Concurrently, the presence of a high Cl- concentration makes the option of indirect electro-oxidation of Fe(II) logical because Fe(II) is readily oxidized by Cl2(aq) formed on the anode. Fe(II) oxidation rate as high as 793 g h-1 m-2 was achieved using a current density 0.64 kA m-2 at a current efficiency of 58%. Very little Cl2(g) mass loss was recorded during the electrolysis step. Considering a current efficiency value of 58% the cost of the electricity within the electrolytic step (the major operational cost) was approximated at 0.9 $ kg-1 H2S(g) removed (for the investigated laboratory electrolytic cell operated at 3.0 A and 3.0 V and using an electricity cost of 0.12 $ kWh-1). Note that the hydrogen gas formed on the cathode as a byproduct in the electrolysis reaction is a valuable product and, if collected, may significantly decrease the operational costs. The elemental sulfur produced in LRSR processes may be 99.98% pure (1), making it a high quality product, which may help in reducing the overall process cost. Based on the results the suggested electrochemically regenerated LRSR process appears to be stable, feasible, and effective. Further research is required, for example, to determine appropriate electrode material and optimal current density for attaining the highest possible energy efficiency at larger scales. For example, the current efficiency reported in this study included a certain cathodic back reduction of Fe(III) which can be minimized, for example by applying a high anode to cathode surface area ratio as suggested by ref 26. Another option to increase the current efficiency of Fe(II) oxidation is to prevent contact between Fe(III) species and the cathode using a membrane-separated electrolytic cell (14). Such approach however may lead to operational difficulties associated with membrane clogging.
Acknowledgments This research was supported by research grant IS-3522-04 from BARD, The United StatessIsrael Binational Agricultural Research and Development Fund, and The Grand Water Research Institute. We thank Dr. Barak Morgan from the University of Cape Town, South Africa, for his valuable help during the writing of the manuscript.
Supporting Information Available Three figures and 1 table: a detailed description of the proposed process and experimental setup (Figures S1 and S2), and further data on experimental conditions (Table 1) and H2S(g) removal results (Figure S3).This material is available free of charge via the Internet at http://pubs.acs.org.
Literature Cited (1) Jensen, A. B.; Webb, C. Treatment of H2S-containing gases: A review of microbiological alternatives. Enzyme Microb. Technol. 1995, 17, 2–10.
(2) Iranpour, R.; Cox, H. H. J.; Deshusses, M. A.; Schroeder, E. D. Literature review of air pollution control biofilters and biotrickling filters for odor and volatile organic compound removal. Environ. Prog. 2005, 24 (3), 254–267. (3) Eow, J. S. Recovery of sulfur from sour acid gas: A review of the technology. Environ. Prog. 2002, 21 (3), 143–162. (4) Hua, G. X.; McManus, D.; Woollins, J. D. The Evolution, Chemistry and Applications of Homogeneous Liquid Redox Sulfur Recovery Techniques. Comments Inorg. Chem. 2001, 22 (5), 327–351. (5) Ebrahimi, S.; Kleerebezem, R.; van Loosdrecht, M. C. M.; Heijnen, J. J. Kinetics of the reactive absorption of hydrogen sulfide into aqueous ferric sulfate solutions. Chem. Eng. Sci. 2003, 58 (2), 417–427. (6) DeBerry, D. Chemical evolution of liquid redox processes. Environ. Prog. 1997, 16 (3), 193–199. (7) Morgan, B.; Lahav, O. The effect of pH on the kinetics of spontaneous Fe(II) oxidation by O2 in aqueous solution - basic principles and a simple heuristic description. Chemosphere 2007, 68, 2080–2084. (8) Smith, J. R.; Luthy, R. G.; Middleton, A. C. Microbial ferrous iron oxidation in acidic solution. J. Water Pollut. Control Fed. 1988, 60 (4), 518–530. (9) Pagella, C.; De Faveri, D. M. H2S gas treatment by iron bioprocess. Chem. Eng. Sci. 2000, 55, 2185–2194. (10) Nemati, M.; Harrison, S. T. L.; Hansford, G. S.; Webb, C. Biological oxidation of ferrous sulphate by Thiobacillus ferrooxidans: a review on the kinetics aspects. Biochem. Eng. J. 1998, 1, 171– 190. (11) Molchanov, S.; Gendel, Y.; Ioslvich, I.; Lahav, O. An improved experimental and computational methodology for determining the kinetic equation and extant kinetic constants of Fe(II) oxidation by Acidithiobacillus ferrooxidans. Appl. Environ. Microbiol. 2007, 73 (6), 1742–1752. (12) Grishin, S.; Bigham, J.; Tuovinen, O. Characterization of jarosite formed upon bacterial oxidation of ferrous sulfate in a packedbed reactor. Appl. Environ. Microbiol. 1988, 54, 3101–3104. (13) Grishin, S. I.; Tuovinen, O. H. Fast kinetics of Fe2+ oxidation in packed-bed reactors. Appl. Environ. Microbiol. 1988, 54 (12), 3092–3106. (14) Mizuta, S.; Kondo, W.; Fujii, K.; Iida, H.; Isshiki, S.; Noguchi, H.; Kikuchi, T.; Sue, H.; Sakai, K. Hydrogen production from hydrogen sulfide by the Fe-Cl hybrid process. Ind. Eng. Chem. Res. 1991, 30, 1601–1608. (15) Asai, S.; Konishi, Y.; Yabu, T. Kinetics of absorption of hydrogen sulfide into aqueous ferric sulfate solution. AIChE J. 1989, 35, 1271–1281. (16) Asai, S.; Nakamura, H.; Aikawa, H. Absorption of hydrogen sulfide into aqueous ferric chloride solutions. J. Chem. Eng. Jpn. 1997, 30 (3), 500–506. (17) Spalding, C. W. Reaction kinetics in the absorption of chlorine into aqueous media. AIChE J. 1962, 8 (5), 685–689. (18) Herrera, L.; Ruiz, P.; Aguillon, J. C.; Fehrmann, A. A new spectrophotometric method for the determination of ferrous iron in the presence of ferric iron. J. Chem. Technol. Biotechnol. 1989, 44, 171–181. (19) Zoltov, Y. A. Fundamentals of Analytical Chemistry. Practical Guide; Visshaya Shkola: Moscow, 2001; 461 pp; in Russian. (20) Yoshinaga, T.; Ohta, K. Spectrophotometric determination of chloride-ion using mercury thiocyanate and iron alum. Anal. Sci. 1990, 6 (1), 57–60. (21) Khelifa, A.; Moulay, S.; Hannane, F.; Benslimene, S.; Hecini, M. Application of an experimental design method to study the performance of electrochlorination cells. Desalination 2004, 160, 91–98. (22) Videm, K.; Lamolle, S.; Monjo, M.; Ellingsen, J. E.; Lyngstadaas, S. P.; Haugen, H. J. Hydride formation on titanium surfaces by cathodic polarization. Appl. Surf. Sci. 2008, 255, 3011–3015. (23) Heal, G. R.; Kuhn, A. T.; Lartey, R. B. A parametric study and computer-based simulation of an undivided sodium hypochlorite electrolyzer. J. Electrochem. Soc. 1977, 124 (11), 1690–1697. (24) Rudolf, M.; Rousar, I.; Krysa, J. Cathodic reduction of hypochlorite during reduction of dilute sodium chloride solution. J. Appl. Electrochem. 1995, 25, 155–165. (25) Kelsall, G. H. Hypochlorite electro-generation. I. A parametric study of a parallel plate electrode cell. J. Appl. Electrochem. 1984, 14, 177–186. (26) Bisang, J. M. Electrochemical treatment of waste solutions containing ferrous sulfate by anodic oxidation using an undivided reactor. J. Appl. Electrochem. 2000, 30, 399–404.
ES901594J
VOL. 43, NO. 21, 2009 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
8319
