Environ. Sci. Technol. 2007, 41, 984-989
Regenerating Fuel-Gas Desulfurizing Agents by Using Bipolar Membrane Electrodialysis (BMED): Effect of Molecular Structure of Alkanolamines on the Regeneration Performance CHUANHUI HUANG, TONGWEN XU,* AND XIAOFEI YANG Laboratory of Functional Membranes, School of Chemistry and Material Science, University of Science and Technology of China, Hefei 230026, P. R. China
Alkanolamine sulfates are the heat-stable salts formed in the fuel-gas desulfurization by using alkanolamines, and they can cause the deterioration of absorption performance and loss of absorbents. In this paper, a method was reported to regenerate three alkanolamines (monoethanolamine, MEA; diethanolamine, DEA; and N,N′dimethylethanolamine, DMEA) by using BMED. The effects of operation parameters (electrolyte concentration, alkanolamine sulfate concentration, and current density) on regeneration were analyzed on the basis of ion dimensions and intrinsic transport velocities, ion concentration, Donnan dialysis, ion orientation, and the interaction between alkanolamines and membranes. The process cost is estimated to be 0.48, 0.32, and 0.30 $/kg for MEA, DEA, and DMEA, respectively. BMED is not only feasible for alkanolamine regeneration but also environmental-friendly and economically attractive, especially as the bipolar membrane cost decreases and pollution control is strengthened.
Introduction Both flue-gas desulfurization and fuel-gas desulfurization are important measures taken against SO2 pollution. In comparison, the later is forethoughtful since low-valent sulfurs (H2S, CS2, COS, etc.) are removed before forming sulfur oxides (SO2 and SO3) during combustion. Alkanolamines are widely used to desulfurize and decarbonize such fuel gases as natural gas and synthesis gases from the gasification of coal and heavy oils (1). However, these fuel gases contain SO3 (2), and it reacts with alkanolamine and forms heat-stable saltssalkanolamine sulfates ((RH)2SO4), which cannot be regenerated by simply heating. If untreated, it will accumulate and cause a loss of amine absorbent and decrease in the efficiencies of desulfurization and decarbonization. In the end, the absorption has to be stopped due to excessively bubbling, great viscosity, and column flooding. Currently, there are two treating methods: metathesis and ion exchange. In the former case, NaOH is added to regenerate alkanolamines, but the formed salt (Na2SO4) will reduce the gas solubility due to the salting-out effect and cause an extra salt pollution in the absorbing sys* Corresponding author phone: +86-0551-3601587; fax: +860551-3601592; e-mail:
[email protected]. 984
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 41, NO. 3, 2007
tem. In the latter case, ion exchange resins are used to extract sulfates and regenerate alkanolamines, but a great amount of acid, base, and water are needed to rehabilitate the ion exchange resins. Furthermore, salt formation cannot be avoided because the acid’s anion and the base’s cation have no other outlets. In general, both of the two methods cannot achieve the regeneration of alkanolamines without a second pollution. The aforementioned shortcoming can be overcome by applying bipolar membranes electrodialysis (BMED), which can realize direct salt conversion without second salt pollution or provide H+ and OH-/alkoxide ions in situ without salt introduction (3). In a BMED stack, (RH)2SO4 can be separated from the solution and form the corresponding alkanolamine (R) and sulfuric acid by utilizing the H+ and OH- supplied by the water splitting in bipolar membranes. In this way, BMED can not only regenerate alkanolamines and keep the circulation steady and stable but also achieve the recycling of resources. In our previous paper (4), BMED has been approved feasible to regenerate piperazine, a desulfurizing agent. That research focused on the operation parameters but did not correlate operation performances with molecular structure and properties. Accordingly, in this paper, feasibility experiments were performed on BMED convention of the sulfates of three alkanolamines: monoethanolamine (MEA, a primary amine), diethanolamine (DEA, a secondary amine), and N,N′-dimethylethanolamine (DMEA, a tertiary amine). Moreover, the convention peculiarities were expected to be found by investigating operation parameters (electrolyte concentration, alkanolamine sulfate concentration, and current density) and the three amines’ properties. The findings will better the understanding of BMED and lay a foundation for treating industrial heat stable salts formed in the process of fuel-gas desulfurizing.
Experimental Apparatus and Procedure Sample Preparation. (RH)2SO4 was prepared by adding sulfuric acid dropwise into an alkanolamine (MEA, DEA, or DMEA) solution until the equivalent pH was reached. The equivalent pH values were 5.59, 5.28, and 5.47 for the sulfates of MEA, DEA, and DMEA, respectively (25 °C, c(RH)2SO4) ) 1 mol/L). Apparatus. The laboratory-scale experimental equipment (4) was mainly comprised of four parts: (a) the direct current power supply (DF1731SD2A, Zhongce Electronics Co. Ltd., China), (b) beakers to store the feed, (c) immersible pumps (AP1000, Zhongshan Zhenghua Electronics Co. Ltd., China) to circulate the solutions at the maximal speed of 27 L/h, and (d) BMED stack. The BP-C-A configuration with one repeating unit was applied in these experiments to regenerate alkanolamines (Figure 1). Specifically, the stack has (a) a cathode and an anode, which are made of titanium coated with ruthenium; (b) Plexiglas spacers (thickness)9 mm) to separate the membranes with Viton gaskets as the seals; (c) an anion-selective membrane (FT-FAB, Table S1 [Supporting Information]), a cation-selective membrane (FT-FKB, Table S1) and a bipolar membranes (Neosepta BP-1, Table S1), all with an effective membrane area of 7.07 cm2; and (d) supporting electrolyte solutions and electrode rinsing solutions, which were both prepared from sodium sulfate (Na2SO4). For industrialization, the unit can be repeated between a pair of electrodes in order to treat more feed solution. In these multiunit stacks, such electrode factors as hydrolysis and electrode electrical resistance have less influence on the whole stack. 10.1021/es061918e CCC: $37.00
2007 American Chemical Society Published on Web 12/29/2006
FIGURE 1. Schematic of the BMED stack used for experiments: A, anion-selective membrane; C, cation-selective membrane; BP, bipolar membrane; R, alkanolamine; RH+, alkanolamine cation; (RH)2SO4, alkanolamine sulfate. All the experiments were undertaken at a certain constant current strength, and the voltage drop across the stack was measured with a digital multimeter (GDM8145, Good Will Instrument Co. Ltd., Taiwan). Before a current was applied, the independent solutions were circulated for 30 min, and all the visible gas bubbles in every compartment were eliminated. The alkanolamine concentration was determined by using UV-spectrophotometry. The pH was measured with a pH meter (PHS-3C, Shanghai Leici Instrument Co. Ltd., China). Calculation of Current Efficiency and Energy Consumption. The current efficiency η was calculated as eq 1
η)
(C0 - Ct)BF It
(1)
where C0 and Ct are the concentration of alkanolamines at time 0 and t, respectively. B is the circulated volume of solution in the base cycle, I is the current, and F is the Faraday constant (5). In this work, t was equal to 1 h, and the change of fluid volume in each cycle was negligible, i.e., B ) 0.5 L. The energy consumption E (kW‚h/kg) was calculated as eq 2
E)
∫CUIdt BM
(2)
t
where U is the voltage drop across the BMED stack (V), and M is the molar weight of alkanolamines. All the experimental data were determined through three independent measurements, and the uncertainty with those results was estimated to be approximately (5%.
Results and Discussion Alkanolamine Properties and Regeneration Behaviors. Naturally, the difference between the regeneration behaviors of MEA, DEA, and DMEA depends on their intrinsic properties and their effects impacting on the conversing system. For comparison, some data were collected or calculated (see Table 1) (6-10). Based on the data, a qualitative prediction was made on the regeneration behaviors of the three alkanolamines before discussing the effects of operation parameters. (1) Ion dimensions and intrinsic transport velocities. Whether it comes to molecular weight, molecular volume, van der Waals volume, the alkanolamines present the same following order: DEA > DMEA > MEA. As concerns monovalent protonated amine ions in an infinitely diluted
aqueous solution, the larger the amine molecule, the slower the protonated ion moves. The protonated alkanolamine ions follow the same order as the above when compared in terms of diffusion coefficient and limit mobility. (2) Ion concentration. The three alkanolamine sulfates are all strong acid-weak base electrolytes, and they will form RH+, H+, R, SO42-, and HSO4- if dissolved in water. As far as the sulfate aqueous solutions (c((RH)2SO4) ) 0.1-0.3 mol/L, pH ) 5-6) used in this work, RH+ and SO42- account for more than 99.9% of the total cations and anions, respectively, according to the calculation based on dissociation constants. When it comes to the basic dissociation constant (pKb), the three alkanolamines have the following order: MEA < DMEA < DEA. Since H+ competes with RH+ when migrating, the RH+/H+ ratio was also considered, and the calculated ratio order is MEAH+/H+ > DMEAH+/H+ > DEAH+/H+. Although the intrinsic mobility of H+ (3.63 × 10-7 m2/(sV), 25 °C, aqueous) is greater than that of alkanolamine ions, its concentration is much less than that of each alkanolamine ion. Therefore, RH+ has the transporting predominance over H+ in spite that protons have the greater intrinsic competence. (3) Donnan dialysis. Another competing ion is Na+ (µ∞ ) 5.19 × 10-8 (m2/(sV)), which originates from Donnan dialysis. A Na2SO4 solution and a (RH)2SO4 solution were separated by a cation-selective membrane. Driven by the concentration gradient, RH+ and Na+ migrate into adjacent compartments. In order to keep neutrality, equivalent RH+ and Na+ exchange across the membrane. Regardless of the applied electrical field, this Donnan dialysis works once the two solutions are circulated beside the cation-selective membrane. The amount of Na+ exchanged increases with the mobility of RH+, contact time, Na2SO4 concentration, and (RH)2SO4 concentration. Given the same other conditions, the concentration of competing Na+ decreases with RH+ mobility and thus has the following order: cNa+(-MEAH+) > cNa+(-DMEAH+) > cNa+(-DEAH+). If all these Na+ ions are diluted with the water in the salt compartment, the resulted concentration (5.09 × 10-8 mol/L, suppose: DNa+ ) 1.33 × 10-9 m2/s, dc/dx) 5 mol/m4, time ) 90 min, 25 °C) will be much less than that of H+ (2-6 × 10-6 mol/L, c((RH)2SO4) ) 0.1-0.3 mol/L, 25 °C) and should have much less competence. However, once the current is switched on, many Na+ ions in the boundary layer will migrate in the opposite direction and compete with RH+. (4) Ion orientation. MEAH+, DMEAH+, and DEAH+ will be shaped like “s“, “NH, or >NH- and will be oriented toward water due to the hydrophilic interaction; (2) amino groups will head RH+ since the proton is oriented toward the cation-selective membrane. In comparison, DMEAH+ is the worst configuration for ion transport since it has the biggest volume and largest cross section due to the long “wing spans”. Without enough space or time for adjustment, DMEAH+ ions will jam in small pores of the membrane, and small ions (Na+ and H+) will carry more current. (5) Interaction between alkanolamines and membranes. This interaction can be approximately predicted by solubility parameters (δsp), but the prediction is very complicated. Since MEA, DEA, and DMEA have the lower δsp values than water (δsp(H2O) ) 23.4 cal0.5/cm1.5), they all can decrease the polarity 986
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 41, NO. 3, 2007
of water and thus make the membrane swell. Membranes will have the greatest swelling ratio if immersed in DMEA due to its lowest δsp. Naturally, when predicting membrane swelling, one has to consider at least 3 other factors. First, considering only the similarity between solubility parameters, MEA will make the membrane swell the most due to its greatest similarity. Second, from the viewpoint of hydrophobicity, DMEA will decrease the water uptake of the cationselective membrane after more DMEA enters the membrane. Third, since the volume of water is much more than that of the membrane, a little free RH can be distributed into the membrane. In summary, the membrane dimensions will not change much since those factors have reverse effects. Even if immersed in pure RH, the cation-selective membrane has no apparent swelling when compared with the membrane
immersed in water. This was confirmed by the pure-solvent swelling experiment. The little swelling ratios also confirm the good dimension stability of the membrane FT-FKB. It was also true of the other two membranes. Based on the above analysis, some regeneration performances can be predicted. The regeneration of MEA cannot have the highest current efficiency at high electrolyte or (RH)2SO4 concentrations due to the greatest Na+ competition. In the case of DEA, its regeneration will have the greatest voltage drop due to the least dissociation degree of DEA and the lowest mobility of DEAH+. High current density will not favor its regeneration since the jam deteriorates. As for DMEA, most of its properties are in the middle of those of MEA and DEA, and its regeneration performances depend on the balance between positive and negative effects. Those predictions can be confirmed by the following experiments, and more convention peculiarities will be discussed. Effect of Electrolyte Concentration on Alkanolamine Regeneration. Figure 3a demonstrates the change of alkanolamine yield as electrolyte concentration increases. The yield was not affected significantly by electrolyte concentration except for an increase with time elapse in each case. In comparison, the yield has the following order: DMEA > MEA > DEA. This order can be explained by referring to the above theoretical analysis. MEAH+ has the greatest mobility and RH+/H+ ratio, the best ion configuration, but the greatest Na+ competition. DEAH+ has the least Na+ competition, but the least mobility and RH+/H+ ratio, and the worst ion configuration. In result, DMEA has the biggest yield though it has moderate aforementioned properties. Figure 3b shows the effect of electrolyte concentration on the voltage drop across the stack. In all cases, the higher the electrolyte concentration, the lower the voltage drop. The sharp voltage increase shortly after a current is applied is due to a sharp increase in the bipolar membrane resistance and a decrease in the apparent conductivity of the electrode rinsing solutions (4). In the case of DEA regeneration, it has the greatest voltage drop since (DEAH)2SO4 has the least dissociation degree and cation mobility. In the cases of MEA and DMEA, they have similar voltage drops due to the similar dissociate degree and cation mobility. Figure 3c shows the effect of electrolyte concentration on energy consumption and current efficiency. In all cases, as the electrolyte concentration increases, the energy consumption decreases since electrolyte can reduce the electrical resistance of the stack, especially that of the bipolar membrane. As concerns the current efficiency curves, MEA, DEA, and DMEA have distinctive regeneration behaviors. DEAH+ has the least Na+ competition, and an increase in Na2SO4 concentration will cause a greater RH+ flux before the current is applied due to the Donnan dialysis, which resulted in an increase in current efficiency. However, a higher Na2SO4 concentration will lead to a greater Na+ competition, which decreases the current efficiency. For MEA and DMEA, the negative effect of Na+ competition on current efficiency is more obvious, and this is especially true of MEA, which has the greatest amount of competing Na+. Naturally, another reason for the decline of current efficiency is that high concentration of electrolyte solutions also leads to a decrease in the water supply into the bipolar membrane and an increase in transition time (4). Effect of Alkanolamine Sulfate Concentration on Alkanolamine Regeneration. Figure 4 shows the effects of (RH)2SO4 concentration on alkanolamine yield, the voltage drop across the stack, energy consumption, and current efficiency. Figures 3 and 4 have many similarities because both Na2SO4 and (RH)2SO4 are electrolytes, but there are some differences. Increasing the concentration of (RH)2SO4 in the salt ycle results in a greater RH+ flux and RH+/H+ ratio but also a greater Na+ competition. In the cases of MEA and
FIGURE 3. Effect of electrolyte concentration on alkanolamine regeneration: (a) Effect of electrolyte concentration on alkanolamine yield; (b) effect of electrolyte concentration on the voltage drop across the stack; and (c) effect of electrolyte concentration on energy consumption and current efficiency. The other operation conditions were current I ) 0.35 A (current density I ) 50 mA/cm2), the concentration of (RH)2SO4 c ) 0.2 mol/L, fluid flow speed v ) 27 L/h. DEA, current efficiency increased and then decreased with (RH)2SO4 concentration, which can be explained by the above reason. In the case of DMEA, current efficiency kept decreasing. DMEA has a weaker hydrophilicity (δsp ) 11.3 cal0.5/cm1.5) than the cation-selective layer (δsp ) 11.7-12.2 cal0.5/cm1.5) of the bipolar membrane, and the regenerated DMEA will decrease the layer’s hydrophilicity and cause a less water supply for the water splitting, which is the source of OH-. Effect of Current Density on Alkanolamine Regeneration. Figure 5 indicates that, as current density increases, alkanolamine yield, the voltage drop across the stack, and energy consumption increase. As for current efficiency, it increased with current density in the cases of MEA and DMEA since the operation at a high current density will have (a) less VOL. 41, NO. 3, 2007 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
987
FIGURE 4. Effect of alkanolamine sulfate concentration on alkanolamine regeneration: (a) effect of alkanolamine sulfate concentration on alkanolamine yield; (b) effect of alkanolamine sulfate concentration on the voltage drop across the stack; and (c) effect of alkanolamine sulfate concentration on energy consumption and current efficiency. The other operation conditions were current I ) 0.35 A (current density I ) 50 mA/cm2), the concentration of Na2SO4 c ) 0.4 mol/L, fluid flow speed v ) 27 L/h.
FIGURE 5. Effect of current density on alkanolamine regeneration: (a) effect of current density on alkanolamine yield; (b) effect of current density on the voltage drop across the stack; and (c) effect of current density on energy consumption and current efficiency. The other operation conditions were the concentration of Na2SO4 c ) 0.4 mol/L, the concentration of (RH)2SO4 c ) 0.2 mol/L, fluid flow speed v ) 27 L/h.
relative Na2SO4 diffusion and (b) a shorter transition time and less alkanolamine migrating out of the base compartment due to the concentration gradient. MEA regeneration has the lowest current efficiency due to the greatest Na+ competition. When it comes to DEA, the current efficiency decreased with current density due to the worst ion configuration. The more current density, the less time and space there is for DEAH+ to adjust and the more DEAH+ jams the pores. Process Economics. The cost estimation is made on the basis of the laboratory-scale experimental equipment (Table 2), and the corresponding calculation was conducted using the method reported in literature (11). The BMED stack has
the best performance for DMEA regeneration in view of its lowest cost per molar. The process costs for alkanolamine regeneration are 0.48, 0.32, and 0.30 $/kg R for MEA, DEA, and DMEA, respectively, which are lower than their market prices (ca. 1.1, 1.0, and 1.2 $/kg R, respectively, in China). When compared to piperazine regeneration (4), the cost of alkanolamine regeneration is less than that for regenerating piperazine (0.96 $/kg of piperazine, BMED of BP-C-C configuration). Regardless of the difference in experimental conditions, double demand of H+ (2 mol H+/1 mol piperazine) mainly accounts for the higher process cost when regenerating piperazine. Notably, as the manufacture cost of bipolar membranes reduces (the cost of a bipolar membrane is assumed, in this
988
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 41, NO. 3, 2007
TABLE 2. Estimation of the Process Cost alkanolamine MEA
DEA
Operation Conditions current density, mA/cm2 50 50 effective membrane 7.07 7.07 area, cm2 the concentration of 0.4 0.4 Na2SO4, mol/L the concentration of 0.2 0.2 amine sulfate, mol/L stack configuration BP-C-A BP-C-A repeating unit number 1 1 the energy consumption, 1.49 1.25 kW‚h/kg R the energy consumption, 0.091 0.132 kW‚h/mol R the process capacity, 6.24 10.81 kg R/year Energy Cost electricity charge, $/kW‚h 0.1 0.149 the energy cost for the regeneration, $/kg R the energy cost for the 0.007 peripheral equipment, $/kg R the total energy cost, 0.16 $/kg R
DMEA 50 7.07 0.4 0.2 BP-C-A 1 0.97
Acknowledgments This research was supported in part by the National Science Foundation of China (nos. 20636050 and 20376079), Key Foundation of Educational Committee of Anhui Province (KJ2007A016). and National Basic Research Program of China (no. 2003CB615700).
0.087
Supporting Information Available
9.98
Table for membrane properties. This material is available free of charge via the Internet at http://pubs.acs.org.
Literature Cited
0.1 0.125
0.1 0.097
0.006
0.005
0.13
0.10
Investment Cost membrane lifetime and 3 3 the amortisation of the peripheral equipment, years membrane prices 135 135 (monopolar membrane), $/m2 membrane prices 1350 1350 (bipolar membrane), $/m2 membrane cost, $ 1.05 1.05 stack cost, $ 1.57 1.57 peripheral equipment 2.36 2.36 cost, $ the cost of the base tank 0.00 0.00 and its pump, $ total Investment cost, $ 3.94 3.94 amortisation, $/year 1.31 1.31 interest, $/year 0.31 0.31 maintenance, $/year 0.39 0.39 the total fixed cost, 2.02 2.02 $/year the total fixed cost, 0.32 0.19 $/kg R the total process cost, 0.48 0.32 $/kg R the total process cost, 0.029 0.033 $/mol R
general, BMED is not only scientifically novel but also economically feasible and attractive, in addition to its environmental benefits. When BMED is applied to treat the actual discharge from desulfurization, extra cost has to cover the prevention of scale formation in order to achieve process efficiency and stability. Naturally, more research needs to be conducted before industrialization.
3
135 1350 1.05 1.57 2.36 0.00 3.94 1.31 0.31 0.39 2.02 0.20 0.30 0.027
paper, ten times that of a monopolar membrane), BMED will obtain a more distinctive economical competence. In
(1) Jenab, M. H.; Abdi, M. A.; Najibi, S. H.; Vahidi, M.; Matin, N. S. Solubility of carbon dioxide in aqueous mixtures of N-methyldiethanolamine + piperazine + sulfolane. J. Chem. Eng. Data 2005, 50, 583-586. (2) Anoufrikov, Y.; Kamps, AÄ . P. S.; Rumpf, B.; Smirnova, N. A.; Maurer, G. Solubility of H2S in H2O + N-methyldiethanolamine + (H2SO4 or Na2SO4). Ind. Eng. Chem. Res. 2002, 41, 25712578. (3) Huang, C. H.; Xu, T. W. Electrodialysis with bipolar membranes for sustainable development. Environ. Sci. Technol. 2006, 40, 5233-5243. (4) Huang, C. H.; Xu, T. W.; Jacobs, M. L. Regenerating flue-gas desulfurizing agents by bipolar membrane electrodialysis. AIChE J. 2006, 52, 393-401. (5) Xu, T. W.; Yang, W. H. Effect of cell configurations on the performance of critic acid production by a bipolar membrane electrodialysis. J. Membr. Sci. 2002, 203, 145-153. (6) Van Krevelen D. W. Properties of polymers: their estimation and correlation with chemical structure; Elsevier Scientific Publishing Co.: Amsterdam-Oxford-New York, 1976. (7) La-Scalea, M. A.; Menezes, C. M. S.; Ferreira, E. I. Molecular volume calculation using AM1 semi-empirical method toward diffusion coefficients and electrophoretic mobility estimates in aqueous solution. J. Mol. Struct.: Theochem. 2005, 730, 111120. (8) Zhao, Y. H.; Abraham, M. H.; Zissimos, A. M. Fast calculation of van der Waals volume as a sum of atomic and bond contributions and its application to drug compounds. J. Org. Chem. 2003, 68, 7368-7373. (9) Aboudheir, A.; Tontiwachwuthikul, P.; Chakma, A.; Idem, R. Kinetics of the reactive absorption of carbon dioxide in high CO2-loaded, concentrated aqueous monoethanolamine solutions. Chem. Eng. Sci. 2003, 58, 5195-5210. (10) Dean J. A. Lange’s Handbook of Chemistry; McGraw-Hill Book Co.: New York, 1985. (11) Strathmann, H.; Koops, G. H. In Handbook on bipolar membrane technology; Kemperman, A. J. B., Ed.; Twente University Press: Enschede, The Netherlands, 2000; pp 191-220.
Received for review August 10, 2006. Revised manuscript received September 29, 2006. Accepted November 6, 2006. ES061918E
VOL. 41, NO. 3, 2007 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
989
