Desulfurization of Hot Syngas Containing ... - ACS Publications

either in a combined gas-turbine, a steam-turbine cycle (IGCC), or in a fuel cell (MCFC) are ... 750 °C when low-steam operating conditions were pres...
0 downloads 0 Views 165KB Size
Download PDF
610

Ind. Eng. Chem. Res. 2000, 39, 610-619

Desulfurization of Hot Syngas Containing Hydrogen Chloride Vapors Using Zinc Titanate Sorbents Raghubir P. Gupta* Center for Engineering and Environmental Technology, Research Triangle Institute, P.O. Box 12194, Research Triangle Park, North Carolina 27709

William S. O’Brien† Mechanical Engineering & Energy Processes Department, Southern Illinois University at Carbondale, Carbondale, Illinois 62901

There is a primary need to increase the use of Illinois coal resources by developing new methods of converting the coal into electricity with highly efficient and environmentally acceptable systems. New coal gasification processes that can generate electricity with high thermal efficiency either in a combined gas-turbine, a steam-turbine cycle (IGCC), or in a fuel cell (MCFC) are being developed. Both of these new coal-to-electricity pathways require that the coal-derived fuel gas be at a high temperature and free of potential pollutants, such as sulfur compounds. Unfortunately, some high-sulfur Illinois coals also contain a significant amount of chlorine, which converts into hydrogen chloride (HCl) in the coal gas. In this study, two simulated gasifierproduct streams were contacted with the zinc titanate desulfurization sorbent in a bench-scale atmospheric fluidized-bed reactor at temperatures ranging from 538 to 750 °C (1000-1382 °F). The first set of experiments involved treating a medium-Btu fuel gas (simulating that of a “Texaco” oxygen-blown, entrained-bed gasifier) containing 1.4% H2S and HCl concentrations of 0, 200, and 1500 ppmv. The second set of experiments evaluated the hot-gas desulfurization of a low-Btu fuel gas (simulating the product of the “U-Gas” air-blown gasifier), which had a 0.5% H2S content and with HCl concentrations of 0, 200, and 800 ppmv. These operating conditions were typical of the gas-treatment requirements of gasifier systems fueled by Illinois basin coals containing up to 0.6% chlorine. The results of the experiments at 538 and 650 °C at all the HCl concentrations revealed no deleterious effects on the capability of the sorbent to remove H2S from the fuel-gas mixtures. In most cases, the presence of the HCl significantly enhanced the desulfurization reaction rate. Some zinc loss, however, was encountered in certain situations at 750 °C when low-steam operating conditions were present. Also of interest, a portion of the incoming HCl was removed from the gas stream and was retained permanently by the sorbent. Introduction It is vital to the industry of Illinois that new approaches soon be developed to increase the use of the abundant bituminous coal resources in the Illinois coal basin by processes that are thermally efficient and environmentally acceptable. Two of the most attractive techniques now being actively developed to cleanly convert coal into electricity are the integrated gasification combined cycle (IGCC) and the molten carbonate fuel cell (MCFC) power generation systems. In both of these innovative systems, the coal is gasified to form a fuel gas that must then be purified of particulate matter and undesirable gas species, mainly sulfur compounds, before being oxidized in the IGCC combustion turbine or in the MCFC fuel cell vessel. To realize the highest thermal efficiency for the overall conversion of coal to electricity, both of these processes require that the temperature of the coalderived fuel gas flowing from the gasifier be as high as * To whom correspondence should be addressed. Fax: (919) 541-8000. E-mail: [email protected]. † Current address: 5313 Glow Drive, Cross Lanes, WV 25313.

possible. However, the coal-derived fuel gas contains reduced sulfur species which must be removed at this high temperature for the overall process to be efficient. Solid regenerable sorbents of mixed-metal oxides that efficiently remove hydrogen sulfide (H2S) and carbonyl sulfide (COS) at temperatures in the 500-700 °C range have recently been developed. Zinc-based sorbents are currently the leading sorbent candidates. In three of the five IGCC demonstration plants being constructed in the U.S. Department of Energy’s (DOE’s) Clean Coal Technology Program (DOE, 1993), zinc titanate is described as the sorbent to use in a hot-gas desulfurization unit. Although most of the mixed-metal sorbents have been developed for fixed-bed operation, fixed-bed systems suffer many limitations, including poor temperature control and unsteady-state operation. The recent development of an attritionresistant, fluidizable zinc titanate sorbent at the Research Triangle Institute (RTI) offers excellent H2S and COS removal efficiency, good sulfur absorption capacity, and excellent regenerability (Gupta et al., 1993). In addition, this sorbent has high attrition resistance, which offers the opportunity to conduct the hot-gas desulfurization in a bubbling or circulating fluidized-

10.1021/ie990533k CCC: $19.00 © 2000 American Chemical Society Published on Web 02/18/2000

Ind. Eng. Chem. Res., Vol. 39, No. 3, 2000 611

bed reactor system, leading to much better gas-solid contact efficiency, better temperature control, and a much greater flexibility in designing alternative subsystems for the continuous regeneration of the sorbent. Unfortunately, some Illinois high-sulfur coals contain a significant amount of chlorine, which evolves as hydrogen chloride (HCl) in the coal-gas product stream from the gasifier. This HCl reacts with certain inorganic compounds, usually with undesirable results. In studying the fly ash deposited during the combustion of coal in boilers, investigators found that chlorides in the gas stream tended to react with the metallic oxides (zinc and ferric) in the slag deposits on the boiler tubes, forming metal chlorides that had lower melting temperatures than the oxides and were more chemically reactive. On the basis of mass-balance calculations, the gasification of the higher chlorine Illinois coals (containing 0.1-0.6% chlorine, by weight) in the typical presentday coal-gasification reactor designs (“Texaco,” “UGas”, “Lurgi”, etc.) would produce from 100 to 1400 ppmv HCl (ppm by volume) in the gasifier effluent stream. Very little is known about how the HCl will interact chemically with the zinc titanate desulfurization sorbents at high temperatures and in the presence of the various fuel-gas compositions that could be produced in the different present-day gasifier designs, except for very limited work reported by Jain et al. (1991) on the effect of chloride on zinc ferrite and zinc titanate sorbents. They found that the presence of HCl in syngas had little deleterious short-term effect on the desulfurization performance of zinc ferrite and zinc titanate sorbents. No long-term exposure tests were conducted to study this effect. Almost all the gaseous chlorine exiting from the gasifier has been analyzed (or assumed) to be in the form of HCl, this form predominating because of the high hydrogen content in the gasifier effluent gas. A portion of the HCl may be scavenged by the alkaline species in the mineral slags which, after evolving from the coal matrix, would cling to the walls of the productgas main or to any heat-transfer surfaces placed in the path of the gasifier product-gas stream. The chlorinegas species, however, would still remain as HCl as long as the gas contains substantial hydrogen and is reducing in its reactive behavior. In this study, the effects of HCl, in concentrations typical of a gasifier fed by high-chlorine Illinois coals, on the fluidizable zinc titanate sorbents were examined. These sorbents are being developed by RTI, under the sponsorship of the U.S. DOE/National Energy Technology Laboratory (NETL), for H2S and COS removal from hot coal gas in the IGCC processes. The major objectives of this project are to identify any deleterious changes in the sorbent caused by the HCl, both in the absorptive operation and in the regeneration cycle, and to determine the fate of the chloride chemical species. Background Since the beginnings of the coal-conversion industry, the presence of chlorine and other trace elements in certain coals has been reported as having significant deleterious effects on the use of these coals (Chou, 1991). Coals that were high in alkalis, sulfur and chlorine were reported to cause major problems, causing the efficiencyrobbing fouling and slagging on the heat-transfer sur-

faces of modern high-pressure, high-temperature boilers (Gluskoter and Rees, 1964). A very through survey of the distribution of chlorine in Illinois Basin coals was presented by Gluskoter and Rees (1964), reporting that the chlorine content ranges from less than 0.1% to nearly 0.6%, with the highest chlorine coals found in the south-center portion of Illinois in coal seams located greater than 1000 ft below the earth’s surface. O’Brien and Gupta (1992) summarized the referenced research work describing where within the coal structure the chlorine is located and how it might evolve during the various coal-conversion processes. Order-of-magnitude calculations on HCl evolution were made using data from several pieces of literature that reported mass balances of low-Btu gas production (Westinghouse, 1973), medium-Btu syngas (the Texaco “Cool-Water” Process) production (Rib, 1989), a Winkler fluidized-bed gasifier in Germany (Engelhard and Adhoch, 1989), and an early theoretical study of low-Btu gas production (Wen, 1973; Wen et al., 1974). If Illinois Basin coals with chlorine contents ranging from 0.1% to 0.6% were gasified and if all the chlorine were converted to HCl gas, the fuel-gas product streams exiting the gasifier would contain as much as 1500 ppmv HCl, depending on the type and composition of the fuelgas product. An oxygen-fed “Texaco-type” medium-Btu syngas could contain as high as 1500 ppmv, while the HCl content in a low-Btu fuel gas such as that produced in an air-blown “U-Gas” gasifier may reach 800-1000 ppmv. In a study examining the hot-gas cleaning of a sulfurand chlorine-containing simulated coal gas, DOE/NETL researchers (DOE, 1986) reported that a concentration of 191-229 ppmv HCl was in the incoming hot-gas stream. In performing an exploratory study of the effect of HCl on the sulfidation kinetics of a zinc ferrite sorbent, Gangwal et al. (1988) included 1050 ppmv HCl in the incoming hot-gas stream, which simulated the fuel-gas exiting a coal gasifier. Chlorides in a sulfur-laden hot fuel-gas could very well interfere with the chemical behavior or the sulfurremoving sorbent in the fuel-gas cleanup processes. Gangwal et al. (1988) noted that the presence of HCl in the feed gas significantly decreased the H2S-absorbing capacity of the zinc ferrite sorbent after the sorbent’s second regeneration cycle. In summarizing the results of studies on the sulfur absorption capacity of the Battelle solid-supported molten salt (SSMS) system involving lithium aluminate ceramic pellets, the personnel at the DOE/NETL (DOE, 1986) stated that “Removal of up to 1 ppm or less was demonstrated for H2S, COS and HCl through multiple cycles...”. However, they further stated that “The chloride absorption capacity is high, but the reaction may be too irreversible for regeneration by simple gas purging.” On the basis of the above, an experimental program was initiated to investigate the effect of HCl vapors present in coal-derived syngas on the H2S absorption behavior of zinc titanate fuel-gas desulfurization sorbents. Bench-Scale Reactor System The cyclic (sulfidation/regeneration) desulfurization tests were performed on various zinc titanate sorbent materials in a semibatch bench-scale reactor designed

612

Ind. Eng. Chem. Res., Vol. 39, No. 3, 2000

Figure 1. Bench-scale experiment test setup.

for operations up to 850 °C. The flow scheme of this bench-scale experimental reactor system is shown in Figure 1. Because hydrogen chloride is a highly corrosive gas which severely attacks stainless steel in the presence of steam, the bench-scale test facility at RTI was extensively modified. In place of the previous stainless steel reactor, a 1-in. (26 mm) i.d. quartz reactor system was designed and constructed. Housed in a three-zone furnace, the new reactor had an inside diameter of 26 mm, a 2-mm wall thickness, and a total length of 44 in. (112 cm). Teflon parts replaced those components exposed to the HCl at the lower (less than 200 °C) temperatures. As shown in Figure 1, the dry coal gas (without steam) was introduced from a sidearm in the reactor. The bottom section of the quartz reactor was tightly packed with quartz wool and acted as a preheater for the inlet coal gas and the incoming water solution. Originally, the HCl was introduced into the reactor chamber through a capillary as a dilute HCl-in-water solution, metered in using a syringe pump. Severe corrosion problems in the pump seals led to a design change. The water was still metered in with a syringe pump, but the HCl was injected as a dry gas premixed with nitrogen from a pressurized cylinder. Much care had to be taken to keep the HCl-gas delivery lines absolutely free of moisture by purging liberally and completely with bonedry nitrogen before and after the test runs. The system was operated in a bubbling fluidized-bed mode. A coarse porous quartz frit at the bottom of the reactor bed was used as the distributor for the simulated HCl-containing coal gas. All the original stainless steel lines in and out of the reactor were replaced by quartz lines to prevent corrosion. Approximately 15 in. of freeboard served as a cooling zone for the gas exiting the reactor. A special Inconel alloy thermocouple constantly monitored the sorbent-bed temperature with these data being continuously collected using a stateof-the-art computerized data acquisition system. The partially cooled (200-300 °C) product gas from the reactor was sent to a glass condenser in which all

the water and HCl vapors were condensed and collected in a catch pot. A slipstream of the noncondensable gases was sent to a gas chromatograph (GC) system for analysis. The rest of the gas exiting the condenser was vented through a hood and a blower. A positive pressure of ≈5 psig was maintained using a pressure-regulating valve downstream of the GC analysis system to ensure the slipstream sample flow to the GCs. The GC system consisted of two GCs, a Carle Series 400 AGC with a thermal conductivity detector (TCD) and a Varian Model 3300 with a flame photometric detector (FPD). The GCs were each separately connected to Spectra Physics SP4270 integrators, which continuously recorded the analyses of the gas streams. Multiple GC sampling valves and dual loops in the Varian FPD provided the capability of measuring H2S and COS from 1500 ppmv to less than 0.1 ppmv every 6 min. A Chromosil-310 column was used to separate these gaseous components at 50 °C. Prior to each experimental run, the Varian GC was calibrated using a reference gas standard. The concentrations of H2S, SO2, and the other bulk gases (H2, CO2, N2, O2, CH4, and CO) were measured every 25 min with the Carle-TCD to evaluate the mass balance and the extent of the shift reaction (CO + H2O f CO2 + H2). The extent of this shift reaction was low because of the high temperature (538-650 °C) and the low catalytic activity of ZnO for this shift reaction. The chloride content of spent sorbent samples was measured by leaching the solid sample in deionized water and analyzing an aliquot by an ion chromatograph. Because the sorbent primarily contained ZnO and TiO2 and traces of alkali oxides and carbonates (because of the presence of a bentonite binder), corresponding chlorides formed by reaction of HCl with the sorbent were highly soluble in water. Therefore, it was assumed that the majority of the chloride could be leached by this procedure and analyzed to estimate the chloride content on the spent sorbent. The dry gases required for sulfidation and regeneration of the sorbent were bought premixed and available

Ind. Eng. Chem. Res., Vol. 39, No. 3, 2000 613

in cylinders. A four-way valve, shown in Figure 1, was used to pass the appropriate gas through a precalibrated electronic mass-flow controller into the bottom of the reactor bed. A purge gas, such as nitrogen, was used during the prerun heating and the postrun cooling of the sorbent bed and also during the switching between sulfidation and regeneration cycle steps. Test Procedures and Conditions All the experiments for this study were carried out at nearly atmospheric pressure, with a 2-psig positive pressure maintained to ensure sufficient sample flow to the GC gas analysis system. A typical bench run consisted of the following steps. A known weight of the sorbent, ≈50-60 g in the 100-300-µm particle size range, was charged to the reactor. The reactor was heated to the desired gas desulfurization temperature (538-750 °C), with a continuous stream of nitrogen passing through the sorbent bed. The sulfidation of the sorbent (fuel-gas desulfurization) was performed with the simulated coal gas continuously flowing upward through the fluidized bed. The medium-Btu fuel gas contained 1.4% H2S (14000 ppmv), while the simulated low-Btu fuel gas contained 0.5% H2S (5000 ppmv). The sorbent sulfidation cycle step continued until the H2S concentration in the reactor exit gas reached 500 ppmv, which was arbitrarily defined as the breakthrough value. Following the sulfidation, the reactor was switched to the sorbent regeneration mode and heated to the desired temperature under nitrogen flow. Generally, the initial regeneration temperature was set at a value from 50 to 150 °C higher than the sulfidation temperature. Once the desired temperature was attained, the flow of regeneration gas was started. The typical regeneration gas contained 3 vol % O2 in nitrogen. This regeneration reaction continued until the outlet SO2 concentration dropped below 500 ppmv. At both the end of the sorbent sulfidation and the end of the regeneration step, samples of the sorbent particles were withdrawn and analyzed for chlorine content. Representative samples of the condensate and other gas-stream scrubbing solutions were also collected and analyzed to determine the chlorine-ion material balance during both the sulfidation and the regeneration steps. At the end of the run, the total quantity of sorbent solids was removed from the reactor and weighed to determine any loss because of elutriation and attrition. Post-test characterization tests of the sorbent solids and condensate-liquid samples were performed primarily to obtain chlorine and sulfur material balances. Additional tests were performed on the reacted sorbent to identify changes in chemical reactivity, surface area, particle size distribution, and Zn-to-Ti contents. The effects of the following specific operating variables were investigated to determine the influence of the chloride on the performance of the zinc titanate sorbents in hot-gas desulfurization service: (1) HCl concentration in the coal gas; (2) sulfidation gas composition; (3) sulfidation temperature; (4) sorbent composition. The first block set of experiments was conducted with a medium-Btu fuel gas simulating the product from the Texaco entrained-bed, oxygen-blown gasifier. The second block set of experiments was performed to evaluate the behavior of HCl on the desulfurization of a low-Btu

Table 1. Composition of Simulated Fuel-Gas Streams fuel-gas type

medium BTU

low BTU

simulating production from gasifier

“Texaco” gasifier (vol %)

“U-Gas” gasifier (vol %)

hydrogen, H2 carbon monoxide, CO carbon dioxide, CO2 steam, H2O nitrogen, N2 hydrogen sulfide, H2S

28.0 39.0 13.0 18.6 0.0 1.4

13.0 24.0 5.0 6.0 51.5 0.5

gas simulating that being generated in the U-Gas airblown gasifier (developed by the Institute of Gas Technology, Chicago). Both of these gasifiers have been developed at nearly commercial scale and are ideally suited for gasifying the high-sulfur Illinois coals. The compositions of the two laboratory fuel-gas streams simulating these two gasifier products are shown in Table 1. Two zinc titanate sorbents were originally selected for this study. These sorbents were prepared using a patented granulation technique described in Gupta et al. (1993). The sorbent ZO-1 contained essentially pure ZnO with 2-3% TiO2 and a suitable binder. This sorbent was selected to establish the role of TiO2 in the zinc titanate on chloride removal reactions; however, this sorbent proved chemically unstable at the higher desulfurization operating temperatures, and its use was discontinued. The second sorbent evaluated was ZT-4, which contained ZnO and TiO2 in a molar ratio of 1.5. In previous studies comparing several candidate sorbents with varying Zn/Ti ratios, the zinc titanate sorbent with this particular molar ratio proved to be the most efficient in high-temperature coal-gas desulfurization service (Gupta and Gangwal, 1992). The test matrixes for both the Set 1 (medium-Btu Texaco gas) and the Set 2 (low-Btu U-Gas gas) experiments were designed to examine the zinc titanate sorbents (ZO-1 and ZT-4, described above) at two sulfidation temperatures, 538 °C (1000 °F) and 650 °C (1202 °F) and at three HCl concentrations (zero, medium, and high). For example, the HCl concentrations for the medium-Btu gas study, Experiment Set 1, was selected as zero, 200, and 1500 ppmv. The HCl concentrations for the low-Btu fuel-gas desulfurization were at zero, 200, and 800 ppmv. Experiments at the higher sulfidation temperature of 750 °C (1382 °F) were later added to probe the expected upper temperature limit of sorbent activity. After the completion of each set of single-cycle tests, a 10-cycle test was made to determine the long-term effect of HCl on the chemical reactivity and the attrition resistance of the zinc-titanate sorbents. The conditions for this 10-cycle test were based on the results of the foregoing single-cycle tests. Results and Discussion Zinc Titanate Sorbent Performance in MediumBtu Gas. Single-Cycle Tests. A total of 11 experiments (Runs I-1-I-11) were performed involving a single sulfidation of the sorbent followed by the regeneration step. In these experiments, the two sorbents were separately sulfided at three temperatures (538, 650, and 750 °C) and with three concentrations of HCl in the fuelgas stream (0, 200, and 1500 ppmv). Table 2 lists the pertinent test conditions for runs I-1-I-11. Also included in this table are the percentage of chloride retentions by the sorbent. As expected, the

614

Ind. Eng. Chem. Res., Vol. 39, No. 3, 2000

Table 2. Test Conditions and Chloride Retentions during Single-Cycle Runs Made in Set 1

run. no.

sorbent formulation

HCl concn in feed gas (ppmv)

temp (°C)

HCl retention by sorbent (% of feed)

I-1 I-2 I-3 I-4 I-5 I-6 I-7 I-8 I-9 I-10 I-11

ZT-4 ZT-4 ZO-1 ZO-1 ZT-4 ZT-4 ZO-1 ZO-1 ZT-4 ZT-4 ZT-4

-0-0-0-01500 1500 1500 1500 1500 200 200

650 538 650 538 650 538 650 538 750 538 650

-0-0-0-02.57 11.37 8.82 14.35 2.26 69.16 18.36

Figure 2. H2S breakthrough curves at various temperatures for ZT-4 sorbent in Texaco Gas containing 1500 ppm HCl.

Figure 3. H2S breakthrough curves at various HCl levels in Texaco Gas at 650 °C.

percentage of HCl retention decreased with an increase in temperature. When exposed to the Texaco gas containing 1500 ppmv HCl, the ZT-4 sorbent formulation retained 11.3%, 2.57%, and 2.27% chloride at the sulfidation temperatures of 538, 650, and 750 °C, respectively. A similar trend was observed for the ZO-1 sorbent formulation, where 14.35% chloride was retained at 538 °C and 8.82% was retained at 650 °C. Figure 2 compares the breakthrough behavior of the ZT-4 sorbent in the Texaco gas containing 1500 ppmv HCl at the three temperatures 538, 650, and 750 °C. Except for the 538 °C test, the prebreakthrough H2S concentration was consistently below 20 ppmv. As expected, the sulfur capacity increased with an increase in temperature. The effect of the HCl concentration in the coal gas is examined in Figure 3, which illustrates the H2S breakthrough behavior of the sorbent ZT-4 at 650 °C in the medium-Btu gas stream, each curve representing a different concentration of HCl. The presence of HCl in the medium-Btu coal gas enhanced

the desulfurization efficiency of the ZT-4 sorbent formulation. Possible reasons for this enhancement are discussed later. Observations and experimental data indicated that excessive vaporization occurred when the low Ti/Zn sorbent (ZO-1) was exposed to the Texaco gas at the higher temperatures. Because of this, further evaluation of this ZO-1 sorbent was discontinued. 10-Cycle Tests. After the 11 single-cycle experiments were completed, a 10-cycle sulfidation-regeneration sequence was performed. During this experiment, the sorbent sulfidations were performed at 650 °C with 1500 ppmv HCl in the simulated medium-Btu “Texaco” fuel gas. Some difficulties occurred when unexpectedly high corrosion interfered with the pumping of the concentrated aqueous-HCl solution into the reactor chamber. Figure 4 shows the H2S breakthrough curves for cycles 1-10 of the 10-cycle test. The breakthrough sulfur capacity during the first cycle was slightly higher than the capacity values produced during the following nine cycles. Unfortunately, unexpected mechanical problems were encountered during cycle 2 and are believed to be primarily responsible for a rather early breakthrough of the sorbent. However, as shown in Figure 4, during the continuation of the experiment the breakthrough curves for cycles 3-10 were almost identical. It should also be noted that the prebreakthrough H2S level was consistently below 20 ppmv. The results of this experiment indicate that the presence of as much as 1500 ppmv HCl in the medium-Btu coal gas does not appear to have significant deleterious effect on the longterm chemical reactivitfy of the zinc titanate sorbent. Regarding the fate of the chloride, 367 ppmw (defined as µg of Cl- per g of sorbent) chloride was detected in the regenerated ZT-4 sorbent after the 10th cycle of the 10-cycle test. The chloride retained in the regenerated sorbent after the first cycle was 359 ppmw. On the basis of these values, it appears that, at 650 °C, 350-370 ppmw of chloride has been permanently absorbed by this sorbent. No deleterious effect of HCl during the 10 cycles of sulfidation-regeneration was observed on the physical and chemical properties of the sorbent, as is evident from Table 3. The particle and pore size distribution remained essentially unchanged. X-ray diffraction (XRD) patterns taken on the reacted samples indicated some traces of TiCl4. The attrition resistance of the sorbent significantly increased after the cycling, as indicated by the 14.2% attrition-loss test value of the reacted sorbent solids as opposed to a 32.2% attrition-loss value for the fresh sorbent material. These results indicate that the presence of the chloride does not significantly deteriorate the mechanical strength or the chemical reactivity of the zinc titanate sorbents. Zinc Titanate Sorbent Behavior in Low-Btu Fuel Gas. Following the evaluation of the behavior of the zinc titanate sorbents in the desulfurization of an HCl-laden medium-Btu coal gas, attention was turned to the behavior of the sorbents when reacting with a low-Btu fuel gas, simulating the product gas from a U-Gas fluidized-bed, air-blown coal gasifier. Single-Cycle Tests. Nine experiments (three temperatures and three HCl concentrations) were performed involving a single sulfidation followed by a regeneration cycle of the ZT-4 sorbent. In these experiments, batches of the ZT-4 sorbent were sulfided at three temperatures (538, 650, and 750 °C) and with three concentrations of

Ind. Eng. Chem. Res., Vol. 39, No. 3, 2000 615

Figure 4. Breakthrough behavior of ZT-4 during 10-cycle testing in Texaco Gas containing 1500 ppmv HCl at 650 °C. Table 3. Properties of Sorbent before and after 10-Cycle Test ZT-4 Sorbent, 650 °C, Texaco Gas, 1500 ppmv HCl)

average particle size (µm) BET surface area (m2/g) pore size distribution mercury pore volume (cm3/g) median pore diameter (Å) attrition resistance, 5-h loss (%) Zn/Ti ratio (atomic) TGA sulfur capacity (% of stoichiometric) chloride level (ppmw)

fresh

10-cycle regenerated

176 3.53

175.4 2.74

0.22 2157 32.2 1.52 100 0

0.2186 2436 14.2 1.51 91 367

HCl in the fuel-gas stream (0, 200, and 800 ppmv). This series of experiments evaluating the desulfurization behavior of the sorbent at 538 and 650 °C revealed no harmful effect traceable to the presence of HCl. However, excessive zinc vaporization occurred during the 750 °C experiments, which could be attributed to the relatively low steam content of the incoming fuel-gas stream. Low values of the steam concentration in the coal gas made this gas highly reducing, which in turn led to instability in the zinc/zinc oxide/zinc sulfide solids reactions. Figures 5 and 6 show the H2S breakthrough curves at 538 and 650 °C, respectively, for the three levels of HCl in the coal gas (i.e., with 0, 200, and 800 ppmv). As shown in these figures, the sorption performance of the zinc titanate is enhanced because of the presence of the HCl in the coal gas at both temperatures. The higher concentration of HCl led to significantly better H2S removal efficiencies at both of these temperatures. A comparison of the breakthrough behaviors at 538 and 650 °C for two HCl concentration levels, namely, the 200 and 800 ppmv concentrations, indicated a significantly higher reactivity at 650 °C as compared to that at 538 °C. Chloride Material Balance. During each single-cycle test, five samples were collected for chloride analysis. These samples included the sulfided sorbent, the regenerated sorbent, the condensate sample after sulfidation, and the impinger solution samples after each sulfidation and after each regeneration. These samples

were analyzed for their chloride content using an ion chromatograph as described previously. Table 4 lists the HCl distribution as determined by chloride material balance in the sorbent, the condensate, and the impinger. As shown in Table 4, greater than 90% material balance is obtained for runs II-4-II-6, whereas for run II-7, only 61.1% of the HCl (relative to the total amount fed) could be accounted for. The reasons for the poor material balance for this run are attributed to the loss of chlorine resulting from the vaporization of ZnCl2 from the sorbent. The vapor pressure of ZnCl2 at 650 °C is ≈0.3 bar compared to that of ≈0.04 bar at 550 °C (Knacke et al., 1991). It is therefore reasonable to assume that some of the chloride left the sorbent as ZnCl2 during the test, thus resulting in a 39% difference in the material balance for run II-7 (Table 4). A number of interesting observations can be made from the data reported in Table 4. It is clear that, at 538 °C, some of the HCl reacts with ZnO to form ZnCl2, which converts back into ZnO during the subsequent regeneration. At 650 °C, however, the absorption of HCl by the sorbent appears to be permanent in a concentration of about 300 ppmw, and very little, if any, HCl absorption takes place via ZnCl2 formation. The observations are in direct agreement with the thermodynamic predictions discussed later. 10-Cycle Test. Following the completion of nine singlecycle tests as described above, a 10-cycle test was performed with the low-Btu U-Gas at 650 °C with 800 ppmv of HCl in the gas. This test was designed to identify any deleterious effect of HCl on the long-term chemical reactivity of the sorbent. Figure 7 shows the breakthrough curves for cycles 1-10. As with the 10cycle test in the medium-Btu gas (shown in Figure 4), the breakthrough sulfur capacity during the first cycle was higher than those for the other nine cycles. The prebreakthrough H2S level was consistently below 10 ppmv in all of the 10 cycles. Despite the excellence performance of the sorbent for H2S removal, a number of operational problems were encountered during this test. Severe plugging of reactor lines and the GC sampling system was observed most

616

Ind. Eng. Chem. Res., Vol. 39, No. 3, 2000

Figure 5. H2S breakthrough curves in U-Gas system at 538 °C.

Figure 6. H2S breakthrough curves in U-Gas system at 650 °C. Table 4. HCl Distribution for U-Gas Runs HCl content (ppmw) HCl distribution (% of feed)

run no.

temp (°C)

HCl content of feed (ppmv)

sulfided sorbent

regenerated sorbent

sorbent

condensate

impinger

total

II-4 II-5 II-6 II-7

538 650 538 650

200 200 800 800

734 303 1053 323

571 356 289 304

22.1 5.2 5.1 2.0

72.3 82.4 76.6 50.8

3.3 6.1 11.1 8.3

97.7 93.7 92.8 61.1

likely because of vaporization of ZnCl2 (as discussed previously) from the sorbent and its subsequent condensation downstream of the reactor. As a result of these problems, this test had to be stopped several times to unplug the lines and then had to be restarted. The erratic behavior of H2S breakthrough profiles can be partially attributed to the loss of zinc from the sorbent, thus reducing the active metal oxide for H2S capture. Furthermore, post-test characterization of the sorbent

material was not conducted to compare physical and chemical property changes due to the reaction. The changes in the sorbent’s physical and chemical properties associated with start-up/shutdown/restart, zinc vaporization, and other operational problems were thought to be more significant than those due to sulfidation-regeneration. There was no easy way to decouple these two sets of changes. Nonetheless, the important finding was to avoid operating zinc titanate

Ind. Eng. Chem. Res., Vol. 39, No. 3, 2000 617

it will tend to escape from the sorbent. However, the thermodynamics of the above reaction is not very favorable. Estimates were made of the equilibrium HCl concentration as a function of the steam content in the simulated U-Gas composition for two cases, 200 and 800 ppmv HCl in the inlet gas. For the 5% steam case used in this study, the extent of possible HCl removal was fairly low. For example, at 900 K (623 °C), the equilibrium HCl level is 681 ppmv for a feed gas containing 800 ppmv HCl. It is hypothesized that the ZnCl2 formed by reaction 1 subsequently reacts with H2S via the following reaction:

ZnCl2 + H2S h ZnS + 2HCl

w

Figure 7. Breakthrough curves for ZT-4 sorbent in U-Gas at 650 °C and 800 ppmv HCl.

Figure 8. Chloride retention during 10-cycle test in U-Gas at 650 °C and 800 ppmv HCl.

sorbents with a low-steam-containing coal gas at 650 °C and above, containing 800 ppmv of HCl vapor. During the 10-cycle test, after each regeneration a small sorbent sample was withdrawn from the reactor for chloride analysis. Figure 8 shows that the chloride content retained by the regenerated sorbent averaged around 300 ppmw for the 10 samples. XRD analysis of the 10-cycle regenerated sorbent did not indicate any presence of ZnCl2. Therefore, it is believed that this chloride adsorption is due to the formation of alkali chlorides. The zinc titanate sorbent was prepared using sodium-containing bentonite as the binder, and this may be the source of the chloride-scavenging alkali in the reacted sorbent. Preliminary estimates of chloride absorption by the sorbent via the reaction of Na indicate that up to 650 ppmw of chloride can be retained just by NaCl formation. Thus, it is strongly believed that, at 650 °C, the primary mechanism of chloride retention by the sorbent involves the reaction between sodium oxide and HCl. NaCl will not convert to Na2O during regeneration because of the absence of steam, as discussed later. Thermodynamic Analysis of Chloride Retention The presence of HCl in coal gas may lead to the following chemical reaction:

ZnO + 2HCl h ZnCl2 + H2O

(1)

Because ZnCl2 in the 500-750 °C temperature range is liquid and therefore has significant vapor pressure,

(2)

The Gibbs free energy (∆G) and equilibrium constant (K1) values for reaction 2 at various temperatures were calculated using the thermochemical data from Knacke et al. (1991). The K values ranged from 15 387 at 800 K to 309 at 1100 K. These relatively high K values strongly favor depletion of ZnCl2 by reaction with H2S to form ZnS and allow formation of more ZnCl2 by reaction 1. Because the formation of ZnCl2 is more favorable at 538 °C than at 650 °C, the reaction of H2S with ZnCl2 at 538 °C will be favored more as compared to the reaction at 650 °C. This is clearly illustrated in Figure 5, which shows the effect of HCl concentration on the H2S breakthrough behavior at 538 °C. ZnCl2, being in the liquid phase in the temperature range of 500-731 °C, is more accessible for reactions with H2S than is the ZnO, which is a solid. A surface layer of ZnCl2 is always present for the H2S to react with. This mechanism, therefore, explains the enhancement in the breakthrough capacity of the sorbent. Thus, higher levels of HCl in the coal gas lead to higher ZnCl2 concentrations and hence to more rapid removal of the H2S via reaction 2. The thermodynamics of the HCl reaction with sodium oxide, which may be present in the sorbent through bentonite, is extremely favorable. The reaction

Na2O + 2HCl h 2NaCl(s) + H2O

(3)

almost goes to completion. The equilibrium constant (K) values calculated using the thermochemical data of Knacke et al. (1991) for this reaction range from 6.7 × 1024 at 800 K to 5.5 × 1016 at 1100 K, making this reaction almost irreversible. Because of this, the sodium oxide-containing materials have a potential to remove HCl down to very low levels. Consequently, the use of nahcolite, a sodium bicarbonate mineral, for HCl removal is based on this principle (Krishnan et al., 1996). Higher chloride retentions at lower temperatures observed from the results reported in Tables 2 and 4 can be explained by accessibility of the ZnCl2 layer to H2S. At 538 °C, the molten layer of ZnCl2 will react with H2S to form ZnS and HCl via reaction 2 and this HCl formed will subsequently react with sodium oxide present in bentonite to form NaCl. At 650 °C, however, a significant portion of ZnCl2 will leave the sorbent because of its high vapor pressure (as discussed previously) and will not get converted into HCl for its subsequent reaction with sodium oxide. This mechanism explains that, with increased temperature, HCl retention by the sorbent decreases. During regeneration of the sulfided zinc titanate sorbent, regeneration of NaCl is technically possible in the presence of SO2 under strong oxidizing conditions via the following reaction:

618

Ind. Eng. Chem. Res., Vol. 39, No. 3, 2000

NaCl + SO2 + 0.5O2 + H2O f Na2SO4 + 2HCl (4) Sodium chloride can thus be converted into sodium sulfate in the presence of SO2, O2, and steam, and the HCl is evolved in the regenerator off gas. Sodium sulfate is readily reduced by H2 present in the coal gas into Na2O via the reaction

Na2SO4 + 4H2 f Na2O + H2S + 3H2O

(5)

No data are available on the kinetics of reactions 4 and 5, and it is suspected that these reactions are slow, with no significant regeneration taking place. In that case, the NaCl would remain permanently in the sorbent structure. Although use of sodium-based alkali minerals is widely practiced in the chemical industry for removal of HCl from syngas between 100 and 500 °C, use at higher temperatures may present serious problems with respect to vaporization of alkali compounds. Presence of alkali species in IGCC gas streams is a major concern for turbine manufacturers because of the potential for alkali-induced corrosion in gas turbines and deposit formation on turbine surfaces. Therefore, this is another reason for conducting the desulfurization of syngas below 550 °C if the syngas contains HCl vapors and the sorbent used for sulfur removal has any alkali-containing binders. Conclusions The results of this experimental study indicate no harmful effect caused by the HCl on the performance of the zinc titanate sorbent materials at temperatures below 550 °C. In fact, the results indicate that the presence of the HCl actually enhances the H2S-removal capability of the sorbent. This phenomenon is believed to be caused by the HCl reacting with the sorbent’s internal pore structure and creating new reactive surface area. The following conclusions can be drawn from this study. (1) The HCl in the coal-derived fuel gas did not have any long-term adverse effect on the zinc titanate sorbent’s sulfur capture ability, its chemical reactivity, its regenerability, its resistance to attrition, or any other of its structural properties. (2) The presence of the HCl significantly enhanced the desulfurization efficiency of the sorbent. (3) An increase in the sorbent’s sulfidation temperature led to higher sulfur-retention capacities and reduced HCl retention by the sorbent. At higher temperatures (e.g., at 650 °C), some chloride was adsorbed on the sorbent surface and was subsequently released during regeneration. (4) Some permanent chloride absorption into the sorbent structure was detected. (5) The mechanism for permanent HCl absorption is believed to be a reaction between the Na2O present in the binder portion of the sorbent. (6) Operation of zinc titanate sorbents with a lowsteam-containing coal gas at 650 °C and above, containing 800 ppmv of HCl vapor, must be avoided to minimize zinc loss due to the vaporization of ZnCl2. (7) A sorbent formulation, composed of nearly pure ZnO, exhibited excessive zinc vaporization. (8) Desulfurization of syngas with zinc titanate-based materials (containing alkali-based binders) should be

restricted to 550 °C or below if the syngas contains significant amounts of HCl vapors. Acknowledgment This work was prepared with the support, in part, by grants made possible by the Illinois Department of Energy and Natural Resources (IDENR) through its Coal Development Board and the Illinois Clean Coal Institute (ICCI), and by the U.S. DOE (Grant DE-FG2291PC91334). However, any opinions, findings, conclusions, or recommendations expressed herein are those of the authors and do not necessarily reflect the views of IDENR, ICCI, or the DOE. Authors acknowledge contributions of Drs. Santosh K. Gangwal and Brian S. Turk of RTI, Mr. Suresh C. Jain of DOE/NETL, and Dr. Frank Honea of ICCI. Literature Cited Chou, C.-L. Distribution and Forms of Chlorine in Illinois Basin Coals. In Proceedings of International Conference on Chlorine In Coal; Stringer, J., Banerjee, D. D., Eds.; Elsevier Science: New York, 1991: pp 11-29. DOE. Hot Gas Cleanup For Electric Power Generating Systems; Report No. DOE/METC-86/6038 (DE86006607); Morgantown Energy Technology Center, U.S. Department of Energy: Morgantown, WV, May 1986. DOE. Clean Coal Technology Demonstration Program, Program Update 1992; Publication No. DOE/FE-0272; U.S. Department of Energy: Washington, D.C., 1993. Engelhard, J.; Adhoch, W. The High-Temperature Winkler Processs Operational Experience And New Developments. In Report No. EPRI GS-6485; Eighth Annual EPRI Conference On Coal Gasification; Electric Power Research Institute: Palo Alto, CA, 1989; pp 10-1-10-12. Gangwal, S. K.; Harkins, S. M.; Stognet, J. M.; Woods, M. C.; Rogers, T. N. Bench-Scale Testing Of Novel High-Temperature Desulfurization Sorbents; Report No. DOE/MC/23126-2662 (DE89000935); Morgantown Energy Technology Center, U.S. Department of Energy: Morgantown, WV, 1988. Gluskoter, H. J.; Rees, O. W. Chlorine In Illinois Coal; Circular No. 372; Illinois State Geological Survey, Urbana, IL, 1964. Gupta, R. P.; Gangwal, S. K. Enhanced Durability of Desulfurization Sorbents for Fluidized-Bed Applications. DOE Report No. DOE/MC/25006-3271; Morgantown Energy Technology Center, U.S. Department of Energy: Morgantown, WV, 1992. Gupta, R. P.; Gangwal, S. K.; Jain, S. C. Development of Zinc Ferrite Sorbents For Desulfurization Of Hot Coal Gas In A Fluid-Bed Reactor. Energy Fuels 1992, 6 (1), 21-27. Gupta, R. P.; Gangwal, S. K.; Jain, S. C. Fluidizable Zinc Titanate Materials with High Chemical Reactivity and Attrition Resistance. U.S. Patent No. 5,254,516, Oct 19, 1993. Jain, S. C.; Grindley, T.; Goldsmith H. Effects of Chlorine on HotGas Desulfurization Sorbents. In Proceedings of International Conference on Chlorine In Coal; Stringer, J., Banerjee, D. D., Eds.; Elsevier Science: New York, 1991; pp 85-108. Knacke, O.; Kubaschewski, O.; Hesselman, K. Thermochemical Properties Of Inorganic Substances, 2nd ed.; Springer-Verlag: New York, 1991. Krishnan, G. N.; Gupta, R. P.; Shelukar, S. D.; Canizales, A.; Ayala, R. E. Removal of Hydrogen Chloride from Hot-Gas Streams. In High-Temperature Gas Cleaning; Institut fu¨r Mechanische Verfahrenstechnik und Mechanik der Universitat Karlsruhe (TH), Germany, 1996: pp 405-414. O’Brien, W. S.; Gupta, R. P. Desulfurization of Hot Fuel Gas Produced From High-Chlorine Illinois Coals; Final Technical Report submitted to the Illinois Clean Coal Institute (formerly the Center for Research on Sulfur in Coal), Carterville, IL, 1992. Rib, D. M. Cool Water Environmental Performance Utilizing Four Coal Feedstocks; Report No. EPRI GS-6485; Eighth Annual EPRI Conference On Coal Gasification; Electric Power Research Institute: Palo Alto CA, 1989; pp 18-1-18-17.

Ind. Eng. Chem. Res., Vol. 39, No. 3, 2000 619 Wen, C. Y. Production of Electricity Via Coal And Coal-Char Gasification; R & D Report No. 66, Interim Report No. 3; Office of Coal Research, U.S. Department of the Interior: Washington, D.C., 1973. Wen, C. Y.; Bailie, R. C.; Lin, C. Y.; O’Brien, W. C. Production Of Low-Btu Gas Involving Coal Pyrolysis And Gasification. Coal Gasification;. Massey, L. G., Ed.: Advances In Chemistry Series 131; American Chemical Society: Washington, D.C., 1974; pp 9-28.

Westinghouse Research & Development Center. Clean Power Generation From Coal; R & D Report No. 84; Office of Coal Research, U.S. Department of the Interior: Washington, D.C., 1973.

Received for review July 21, 1999 Revised manuscript received November 4, 1999 Accepted November 29, 1999 IE990533K