Reaction between Sulfur Dioxide and Limestone under Periodically

Jul 30, 1998 - Energy Fuels , 1998, 12 (5), pp 905–912. DOI: 10.1021/ .... J XIAO , M ZHANG. Energy Conversion and Management 2008 49, 3178-3187 ...
2 downloads 5 Views 107KB Size
Energy & Fuels 1998, 12, 905-912

905

Reaction between Sulfur Dioxide and Limestone under Periodically Changing Oxidizing and Reducing ConditionssEffect of Cycle Time Tobias Mattisson*,† and Anders Lyngfelt‡ Department of Inorganic Chemistry, Go¨ teborgs Universitet, S-412 96 Go¨ teborg, Sweden, and Department of Energy Conversion, Chalmers University of Technology, S-412 96, Go¨ teborg, Sweden Received December 21, 1997

To better understand the sulfur capture process during fluidized bed combustion, the reaction between sulfur dioxide and limestone has been investigated under periodically changing oxidizing and reducing conditions. All experiments were carried out in a fixed-bed quartz reactor at 850 °C, with the reactant gas consisting of 1500 ppm SO2, 10% CO2, and with the O2 and CO concentration varied between 0 and 4%. The fraction of time under reducing conditions was varied between 0 and 100% and the total cycle time (i.e., time of one oxidizing period + one reducing period) was varied between 30 and 900 s. The present study showed that alternating conditions may result in both an increase and a decrease in the final conversion depending on both the total cycle time and the fraction of time under reducing conditions. Whereas a final conversion of approximately 9% was found when the limestone was sulfated under oxidizing conditions, conversions between 1% and 25% were found under alternating conditions for the cases where only small amounts of CaS was formed. For the experiments where most of the sulfur was in the form of sulfide the conversion was higher, up to 46%. The conversion increased for all cycle times up to 33% of the time under reducing conditions, with the largest increase seen for long cycle times. When the fraction of time under reducing conditions was increased further, the conversion decreased significantly for the shorter cycle times.

Introduction The question in focus in this study is whether alternating oxidizing and reducing conditions could have a positive effect on the sulfur capture performance in circulating fluidized bed boilers (CFBBs). The question originates from the conflicting results seen when the sulfur capture performance of two Swedish limestones, the reactive porous Ignaberga and the unreactive crystalline Ko¨ping, was compared in a 40 MW CFBB.1 Although the conversion, i.e., the molar S/Ca ratio, in laboratory investigations was about 4 times larger for the reactive limestone Ignaberga, the boiler results showed an unexpectedly small difference between the two limestones. In Figure 1, a comparison of the degree of conversion as a function of particle size is made between Ko¨ping limestone sulfated in the laboratory under 100% oxidizing conditions2 and the same limestone sulfated in two Swedish boilers, a 40 MW CFBB in Nyko¨ping1 and a 165 MW CFBB in O ¨ rebro. As can

Figure 1. The conversion, i.e., molar S/Ca ratio, as a function of particle size as analyzed from boiler ashes from a (i) 40 MW CFBB and a (ii) 165 MW CFBB and from (iii) the same limestone sulfated in a fixed-bed quartz reactor under oxidizing conditions.

* Author to whom correspondence should be addressed. E-mail: [email protected]. † Go ¨ teborgs Universitet. ‡ Chalmers University of Technology. (1) Mjo¨rnell, M.; Leckner, B.; Karlsson, M.; Lyngfelt, A. Proceedings of the 11th International Conference on Fluidized Bed Combustion; Anthony, E. J., Ed.; The American Society of Mechanical Engineers: New York, 1991; pp 655-663. (2) Mattisson, T.; Lyngfelt, A. A method of evaluating the reactivity of limestone under fluidized bed combustion conditions. Can. J. Chem. Eng. in press.

be seen, a much higher conversion was found in the boilers, for all particle sizes. The laboratory results were obtained in a study with the original purpose of determining the reactivity for use in modeling of sulfur capture. For this purpose, however, the laboratory data turned out to be unusable as the maximum conversion obtained is only a third to a half of that in the boilers. This raises the question: Why is the sorbent conversion so much higher in the CFBB and what is the important

S0887-0624(97)00232-6 CCC: $15.00 © 1998 American Chemical Society Published on Web 07/30/1998

906 Energy & Fuels, Vol. 12, No. 5, 1998

Mattisson and Lyngfelt

difference between the conditions in the combustion chamber of the boiler and in the laboratory experiments? One difference is the longer residence times in the boilers, 15-30 h,3 compared to those in the laboratory, 2 h. However, the laboratory investigations compensate for these long residence times by using an SO2 concentration that is generally ca. 1 order of magnitude larger than that normally found in boilers. Furthermore, the conversion increased only moderately in long-term laboratory tests. For example, a conversion of 9% was found after sulfating Ko¨ping limestone (0.5-0.7 mm) for 2 h with a high SO2 concentration, 1500 ppm, whereas 12% was found after 40 h.4 Thus, the long residence times in boilers do not explain the large difference in conversion seen in Figure 1. Another important difference is that the sorbent is exposed to conditions alternating between oxidizing and reducing in the combustion chamber. It is the aim of this paper to see what effect a cyclic exposure of CO and O2 has on the reaction between limestone and SO2 and if the higher degree of conversion found in the boiler can be explained by this cyclic exposure to reducing and oxidizing conditions. Below, the current knowledge concerning the effect of reducing conditions on the sulfur capture during fluidized bed combustion is summarized. Sulfur Capture under Fluidized Bed Combustion (FBC) Conditions. During FBC, limestone or dolomite is added to the combustion chamber for in situ removal of SO2. The limestone consists mainly of CaCO3, which calcines to CaO at temperatures and CO2 concentrations typical in an atmospheric fluidized bed boiler (FBB):

CaCO3 f CaO + CO2(g)

(1)

The molar volume of CaCO3 is 36.9 cm3/mol while that of CaO is 16.9 cm3/mol. This results in the formation of a highly porous structure of the CaO. Under oxidizing conditions the oxide will react with SO2 and form calcium sulfate:

CaO + SO2(g) + 1/2O2(g) f CaSO4

(2)

Because of the large molar volume of CaSO4 (46.0 cm3/ mol), compared to that of the oxide, the sulfate will progressively plug the pores of the oxide, resulting in an unreacted core of CaO. The reaction will at this point be limited by the rate of diffusion of reactants through the sulfate layer, which is a much slower process than pore diffusion and the chemical reaction. It is thus not possible to achieve complete conversion of CaO to CaSO4. Several investigations have found the presence of reducing zones in the bottom bed of fluidized bed boilers. Measurements with a zirconia cell oxygen probe in a 16 MW bubbling FBB showed that conditions changed rapidly between oxidizing and reducing.5,6 Cooper and Ljungstro¨m5 found that the partial pressure of oxygen could be as low as 10-10 bar about 80-90% of the time, (3) Mattisson, T.; Lyngfelt, A. Chem. Eng. Sci. 1998, 53, 1163-1173. (4) Mattisson, T.; Lyngfelt, A. Characterization of limestone reactivity with SO2 and sulfur capture modelling under fluidized bed combustion conditions; Va¨rmeforsk Rapport 597; Stiftelsen fo¨r Va¨rmeteknisk Forskning: Stockholm, 1996.

even though the excess air ratio was high, 1.4. Similarly, Lyngfelt et al.7 found that under normal air staging reducing conditions were present 70-80% of the time in the bottom bed of a 12 MW circulating FBB. The gas phase of the reducing zones may contain such reducing agents as CO, H2, and CH4.8 The presence of these reduced compounds could influence the sulfur capture in several ways. CaSO4, formed in oxidizing regions of the boiler may be reductively decomposed, with subsequent release of SO2:

CaSO4 + CO/H2(g) f CaO + SO2(g) + CO2/H2O(g) (3) Reaction 3 has been shown to be important both in laboratory scale combustors9 and in full scale utility boilers.10 Lyngfelt and Leckner10 found direct evidence of reaction 3 in a 16 MW stationary FBB, where, at temperatures above 890 °C, a net release of sulfur was found, indicating that reaction 3 was faster than reaction 2. Similar effects were found in a 12 MW CFBB with a negative degree of sulfur capture following an increase in temperature from 850 to 930 °C.11 In addition to the reductive decomposition of the sulfate, CaS may also be formed, either directly from the sulfate

CaSO4 + 4CO(g) f CaS + 4CO2(g)

(4)

or from CaO

CaO + 3CO(g) + SO2(g) f CaS + 3CO2(g)

(5)

At highly reducing conditions SO2 may not be the most stable sulfur species thermodynamically, but instead H2S and COS may dominate and form sulfide through

CaO + H2S/COS(g) f CaS + H2O/CO2(g)

(6)

Mattisson and Lyngfelt12 investigated the presence of CaS in the bottom bed of a 12 MW circulating FBB as a function of the degree of air staging and temperature. They found that only minor amounts of CaS were present under normal operating conditions. However, some calcium sulfide was present in all retrieved samples, and it was suggested that CaS may act as an intermediate during the sulfur capture process. Further, large amounts of CaS were found during periods (5) Cooper, D.; Ljungstro¨m, E. The influence of bed temperature on the in-bed O2 partial pressure in a 16 MW AFBC fired with petroleum coke. Internal Report OOK A87 002; Department of Inorganic Chemistry, Chalmers University of Technology: Go¨teborg, 1987. (6) Ljungstro¨m, E. Proceedings of the 8th International Conference on Fluidized Bed Combustion; The U.S. Department of Energy: Morgantown, 1985; pp 853-864. (7) Lyngfelt, A.; Bergqvist, K., Johnsson, F.; Åmand, L.-E.; Leckner, B. Dependence of sulphur capture performance on air staging in a 12 MW circulating fluidized bed boiler. In Gas cleaning at high temperatures; Cliff, R., Seville, J. P. K., Eds.; Blackie Academic & Professional: Glasgow, 1993; pp 470-491. (8) Lyngfelt, A.; Åmand, L.-E.; Leckner, B. Proceedings of the 26th International Symposium on Combustion; The Combustion Institute, 1996; pp 3253-3259. (9) Khan, W. U. Z.; Gibbs, B. M. Inst. Chem. Eng. Symp. Ser. 1991, 123, 193-203. (10) Lyngfelt, A.; Leckner, B. Chem. Eng. Sci. 1989, 44, 207-213. (11) Lyngfelt, A.; Leckner, B. Fuel 1993, 72, 1553-1561. (12) Mattisson, T.; Lyngfelt, A. Proceedings of the 13th International Conference on Fluidized Bed Combustion; Heinschel, K. J., Ed.; The American Society of Mechanical Engineers: New York, 1995; pp 819830.

Reaction between Sulfur Dioxide and Limestone

Energy & Fuels, Vol. 12, No. 5, 1998 907

Figure 2. A phase diagram of the solid phases CaO-CaSO4CaS as a function of the SO2 and O2 partial pressures at 850 °C.

of high SO2 concentrations. Similar observations were made by Lyngfelt et al.13 in a 16 MW stationary FBB. Any CaS formed in the bottom bed, reactions 4-6, may be oxidized in regions of high concentrations of O2 to form either the oxide

CaS + 3/2O2(g) f CaO + SO2(g)

(7)

or the sulfate

CaS + 2O2(g) f CaSO4

(8)

In addition, it has recently been shown that CaS and CaSO4 react at a surprisingly rapid rate at temperatures common in FBC, probably through the formation of a melt, with the release of SO2:14

CaS + 3CaSO4 f 4CaO + 4SO2(g)

(9)

To alleviate the interpretation of the sulfur capture in FBB, a stability diagram of the solid sulfur species is helpful. Figure 2 shows such a stability diagram of the stable solid sulfur species at 850 °C as a function of the O2 and SO2 partial pressures. The “reducing potential”, i.e., PCO/PCO2, is shown as well and is calculated from the equilibrium of the reaction, 1

/2O2(g) + CO(g) f CO2(g)

(10)

CaSO4 is the stable compound at highly oxidizing conditions and high partial pressures of SO2, while CaS is stable at strongly reducing conditions. CaO is the stable species at “intermediate” reducing conditions and normal SO2 concentrations. These intermediate conditions where CaO is favored thermodynamically can be expected during the shifts between oxidizing and reducing conditions. This is also consistent with results from a laboratory study under alternating conditions, which show that SO2 is released during the shifts between oxidizing and reducing conditions.15 (13) Lyngfelt, A.; Langer, V.; Steenari, B.-M.; Puroma¨ki, K. Can. J. Chem. Eng. 1995, 73, 228-233. (14) Davies, N. H.; Hayhurst, A. N.; Laughlin, K. M. Proceedings of the 25th International Symposium on Combustion; The Combustion Institute, 1994; pp 211-218.

Laboratory Investigations of the Effect of Alternating Oxidizing and Reducing Conditions. Jonke et al.16 suggested in 1972 that the presence of local reducing zones enhances the sulfur penetration into added limestone particles due to the continual release and uptake of sulfur. Hansen17 conducted the first extensive study of the effect of alternating conditions using an integral fixed-bed reactor. The effects of reducing agent, particle size, temperature, gas composition, and type of limestone were studied. In contrast to Jonkes theory, Hansen found no major difference between the final sulfur capacity of sorbent sulfated under alternating conditions and that of sorbent sulfated under oxidizing conditions at 850 °C. However, the total cycle time and fraction of time under reducing conditions was not varied extensively, and most experiments were performed with 50% reducing conditions and a total cycle time of 60 s. In the experiments where the mentioned parameters were varied, the highly porous and reactive limestone Stevns Chalk was used. For this limestone a clear effect of the fraction of time under reducing conditions on the final conversion was seen, with a significantly higher degree of conversion found when using a higher fraction of time under oxidizing conditions. Okamoto et al.18 conducted similar experiments to Hansen’s but found much higher conversions when the limestone was sulfated during alternating conditions, compared to those when the same limestone was sulfated in oxidizing conditions (34.4% versus 23.9%). Yrjas and Hupa19 performed alternating experiments using a pressurized thermogravimetric analyzer (PTGA) and with a reducing gas with a CO/ CO2 ratio of 0.013 and an SO2 concentration of 0.02%, i.e., the region where CaO is stable in Figure 2. Only a few samples were sulfated at atmospheric pressure, and these showed no significant difference in final conversion compared to limestone sulfated under oxidizing conditions. From the literature investigations it is clear that there is a need for further research into the effects of alternating oxidizing/reducing conditions on the sulfur capture process. In the present study the effects of cycle time and fraction of time under reducing conditions were varied extensively. The Swedish limestone Ko¨ping coarse was a logical choice due to the large difference in conversion between samples withdrawn from boilers and samples sulfated in laboratory investigations under oxidizing conditions, see Figure 1. Experimental Section Experimental Setup. All experiments were performed in a fixed-bed reactor constructed entirely in quartz, Figure 3. The reactor was built in a fashion that made it possible to (15) Hansen, P. F. B.; Dam-Johansen, K.; Østergaard, K. Chem. Eng. Sci. 1993, 48, 1325-1341. (16) Jonke, A. A.; Vogel, G. J.; Carls, E. L.; Ramaswami, D.; Anastasia, L.; Jerry, R.; Haas, M. AIChE Symp. Ser. 1972, 68, 241251. (17) Hansen, P. F. B. Sulphur capture in fluidized bed combusters. Ph.D. Thesis, Technical University of Denmark: Lyngby, 1991. (18) Okamoto, T.; Sakaue, T.; Nakamichi, J.; Suzuki, Y.; Nishimura, M.; Moritomi, H. Proceedings of the 2nd SCEJ Symposium on Fluidization; The Society of Chemical Engineers, Japan (SCEJ): Tokyo, 1996; pp 398-405. (19) Yrjas, P.; Hupa, M. Proceedings of the 14th International Conference on Fluidized Bed Combustion; Preto, F. D. S., Ed.; The American Society of Mechanical Engineers: New York, 1997; pp 229236.

908 Energy & Fuels, Vol. 12, No. 5, 1998

Mattisson and Lyngfelt Table 1. Reaction Conditions limestone particle size inlet SO2 CO2 CO O2 temperature gas flowa duration a

Ko¨ping 0.5-0.7 mm 1500 ppm 10 vol % 0-4 vol % 0-4 vol % 850 °C 1000 mL/min 2h

0 °C, 1 bar. Table 2. Ko1 ping Limestone

Figure 3. The fixed-bed quartz reactor used in the present study.

size (mm) chemical composition (wt %) Ca Si Mg Fe Al Mn Zn Cu Na K structural parameters average pore radiib,c (nm) BET areab(m2 g-1)

0.5-0.7 36.6a 1.73 0.683 0.523 0.246 0.138 0.016 0.013 0.070 0.167 33.5 15.5

a Corresponds to 91.5% CaCO . b Samples were calcined for 30 3 min in an atmosphere of 4% O2 and 10% CO2 at 850 °C. c As determined from Hg porosimetry data.

Figure 4. The experimental setup. introduce the O2 and CO separately from the rest of the reactant gas just above the quartz filter where the limestone was placed. This construction minimized the degree of backmixing between the CO and O2 in the reactor. The experimental setup is shown in Figure 4. The gases SO2, CO2, and N2 were led from gas tubes containing either the concentrated gas component (N2, CO2) or a diluted mixture (SO2) through mass flow controllers (Brooks 5850E) where the gas flows were adjusted to obtain the right concentrations and flows. These gases were mixed prior to entering the inlet located at the bottom of the reactor; see Figure 3. O2/N2 and CO/N2 were, similarly, led through mass flow controllers but were then directed through two programmable three-way magnetic valves which, alternatingly, directed either O2 or CO through a 4 mm i.d. quartz capillary at the top of the reactor or to a fumehood. The gas component that was led to the reactor was distributed a distance of 10 cm from the top of the sintered quartz plate through five small holes. By always having a constant flow of both CO and O2, the total flow through the reactor was always constant and no noticeable deviation occurred, even when switching between O2 and CO. The concentrations of SO2 and O2 from the outlet of the reactor were measured using a URAS 10E gas analyzer, and the data were logged to a file with 1 s intervals. The SO2 was measured using nondispersive infrared analysis, while O2 was measured paramagnetically. In several of the experiments a quadropole mass spectrometer (Baltzers QMG 421) was used for measurement of COS.

Phase Analysis. Because the reaction conditions were such that CaS may be formed, many of the sulfated samples were analyzed using both X-ray powder diffraction (Siemens D5000 powder diffractometer) and wet chemical analysis. The amount of CaS was determined by converting the sulfide to gaseous H2S which was absorbed as ZnS and quantitatively determined by titration. This method has been described previously in more detail.12 Experimental Procedure. Initially the gas mixture was made to bypass the reactor, and the inlet gas composition was determined. This was always done with O2 present. At this time the limestone sample was heated to the reaction temperature in an atmosphere of 100% CO2. The CO2 prevents any precalcination of the sample before the actual experiment. As the temperature reached steady state, the reaction gas mixture was led through the reactor. All of the experiments were initiated and terminated with an oxidative period. To obtain a valid measurement of the final degree of conversion, the sample was carefully weighed before and after sulfation. Table 1 shows the reaction conditions used in this work, while Table 2 contains the chemical composition and structural parameters of Ko¨ping limestone. The degree of final conversion was calculated by two methods: (i) gravimetrically, i.e., from the difference in mass of the sorbent sample before and after the sulfation experiment, and (ii) integration of the absorbed SO2. The former method assumed that the only phases present in the sample were CaO, CaSO4, and CaS, which was confirmed by X-ray powder diffraction. The total conversion is calculated from

Xtotal ) XCaS + XCaSO4

(11)

where Xtotal is the total degree of conversion and XCaS and XCaSO4 are the conversion to calcium sulfide and calcium sulfate, respectively. The conversion to CaS can be calculated from

XCaS ) WfyS2-MCaCO3/WifCaCO3MS

(12)

where yS2- is the mass fraction of sulfide present in the sulfated sample, Mi is the molecular weight of component i, fCaCO3 is

Reaction between Sulfur Dioxide and Limestone

Energy & Fuels, Vol. 12, No. 5, 1998 909

Figure 5. The measured outlet concentrations of SO2, COS, and O2 versus time for a trial with a total cycle time of 900 s and with 50% reducing conditions. the fraction of CaCO3 in the limestone sample, Wi is the initial weight of the limestone sample, and Wf is the final weight of the limestone sample. The conversion to CaSO4 can be calculated with the expression

MCaCO3Wf,adj XCaSO4 )

WifCaCO3

- MCaO - XCaS(MCaS - MCaO) MCaSO4 - MCaO

(13)

where Wf,adj is the final weight of the limestone adjusted for the amount of inerts in the sorbent, i.e.:

Wf,adj ) Wf - Wi(1 - fCaCO3)

(14)

The method whereby the conversion is determined from integration of the absorbed SO2 has been described earlier and will not be covered in this paper.2 The conversion as determined by eqs 11-14 is believed to be more accurate than the conversion calculated with integration and, unless stated otherwise, the method used in this work. Initial integration of the absorbed SO2 from experiments conducted entirely under reducing conditions yielded considerably higher degrees of conversion compared to when the conversion was estimated from change of mass (assuming that the only products were CaO and CaS). By measuring the outlet gases with the mass spectrometer it was found that substantial amounts of COS were formed during the experiments with a large fraction of time under reducing conditions. Subsequently some of the experiments were conducted with continuous measurement of COS. No other reduced sulfur species were found with the mass spectrometer. However, it was discovered that under experiments conducted with a high fraction of time under reducing conditions, a thin layer of sulfur was deposited at the bottom of the reactor, which indicated that elemental sulfur may have been formed.

Results and Discussion For comparison purposes Ko¨ping limestone was sulfated under oxidizing conditions for 2 h. A conversion of approximately 9% was found which agrees well with earlier investigations of Ko¨ping limestone under similar conditions.1 X-ray diffraction confirmed that the only major phases present in this sulfated sample were CaO and CaSO4. Figure 5 shows the outlet SO2, COS, and O2 concentration as a function of time for a sulfation experiment with 450 s under 4% O2 followed by 450 s

under 4% CO. There are several regions in this figure that can be interpreted as follows. (a) The initial oxidation period is characterized by an immediate drop in the SO2 concentration, followed by an increase during the rest of the period. This is due to the formation of CaSO4 which progressively plugs the pores of the CaO, resulting in a higher degree of intraparticle diffusion resistance and subsequently a lower rate of reaction. (b) After the O2 has been replaced by CO, there is an immediate SO2 peak attributed to the reductive decomposition of CaSO4, reaction 3, followed by a rapid decrease in the SO2 concentration due to the formation of CaS, reaction 5. There is a steady increase in the COS concentration during this period. (c) The second oxidizing period is initiated by a massive SO2 peak which exceeds the limit of the gas analyzer (5000 ppm). This peak is due to the oxidation of CaS formed in the prior period, reaction 7. Due to the highly exothermic nature of reaction 7 these oxidation peaks are always accompanied by a temperature increase, as high as 15 °C for some of the experiments in this work. This rapid oxidation of CaS is followed by the formation of CaSO4 as indicated by a lower but progressively increasing SO2 concentration, much like that in region a. It is interesting to note that the sorbent is much more reactive just following the CaS oxidation peak compared to the end of the prior oxidation step, even though the degree of conversion is higher at this point. This is seen for all eight cycles but is most visible during the first few periods. The last few reducing periods, such as region d in the figure, are characterized by a rather slow rate of CaS formation initially, which gradually increases toward the end of the period. This was seen for all experiments with a long total cycle time (900 s). Zadick et al.20 found that calcium sulfide autocatalytically favored its own rate of formation, which could be one explanation for the behavior seen here. The described sequence is repeated for a total sulfation time of 2 h with the sequence of SO2 peaks similar for each cycle. The COS concentration increases progressively during the reducing period, but never exceeds 100 ppm for this experiment. It is believed that the CaS formed during the reducing period catalyzes the formation of COS according to (see Appendix):

3CO(g) + SO2(g) f COS(g) + 2CO2(g)

(15)

The final conversion of the trial shown in Figure 5 was 25% with 3.4 wt % CaS present in the sulfated sorbent sample, or 15% of the total captured sulfur present as S2-. However, it should be noted that this experiment was terminated only a few seconds into an oxidation period. In a similar experiment where the trial was stopped after an entire oxidizing period, a much lower CaS content (1.8 wt % with a similar total conversion) was found. X-ray diffraction of these samples confirmed that the only major phases were CaO, CaSO4, and CaS. Though most trials were qualitatively similar to the one presented in Figure 5, the extent of both CaS (20) Zadick, T. W.; Zavaleta, R.; McCandless, F. P. Ind. Eng. Chem. Process Des. Dev. 1972, 11, 283-287.

910 Energy & Fuels, Vol. 12, No. 5, 1998

Mattisson and Lyngfelt

Figure 7. The conversion as a function of time for experiments conducted with 50% reducing conditions but different total cycle times.

Figure 6. The SO2 concentration versus time for 6 different experiments with a constant total cycle time of 240 s but with a different fraction of time under reducing conditions as follows: (a) 90%, (b) 67%, (c) 50%, (d) 33%, (e) 25%, (f) 10%. Table 3. Summary of Results Obtained with 240 Seconds Total Cycle Time reducing conditions (%)

Xtotala

S2-/Stot (%)

major phasesb

100 90 67 50 33 25 10 0

0.46 0.34 0.11 0.17 0.16 0.14 0.10 0.09

100 90 5 0.8 0.4 0.4 0.1 0.0

CaO, CaS CaO, CaS, CaSO4 CaO, CaS, CaSO4 CaO, CaSO4, CaSc CaO, CaSO4 CaO, CaSO4 CaO, CaSO4 CaO, CaSO4

a As calculated with eq 11. b As determined with X-ray powder diffraction. c Only trace amounts were seen.

oxidation and the reductive decomposition of CaSO4 varied extensively. This can be seen in Figure 6 a-f, which illustrates the difference in the outlet SO2 concentration for different lengths of the reducing period, between 10-90%, with a cycle time of 240 s. In Figure 6a, at 90% reducing conditions, there are no observable peaks due to reductive decomposition of CaSO4 but only SO2 release through the oxidation of CaS, which appears to occur at a constant rate throughout the experiment. In Figure 6b, at 67% reducing conditions, there is SO2 release both through reductive decomposition and through oxidation of CaS. The rate of CaS oxidation and CaSO4 decomposition are approximately constant throughout the experiment. As the time under oxidizing conditions is increased further, the rate of CaS oxidation decreases during the experiment as does the extent of the reductive decomposition of CaSO4. Table 3 summarizes the results of the experiments shown in Figure 6. The decrease in conversion seen at 67% reducing conditions was seen

for all cycle times below 240 s, as will be shown below, and may be explained by the high rate of SO2 release through both reductive decomposition of CaSO4 and CaS oxidation, as seen in Figure 6b. The constant rate of release throughout the experiment indicates that the reactions are taking place only at the surface of the particle with no buildup of product layer, which would cause intraparticle resistance and a subsequent decrease in reactivity. To illustrate the change in conversion as a function of time, the outlet concentration of SO2 was integrated. Figure 7 shows the conversion versus time for several experiments conducted with 50% reducing conditions but with varying total cycle times. It should be noted that only in the experiment conducted with 900 s cycle time was COS included in the integration. Only minor amounts of COS were seen in the other experiments and had only a small effect on the conversion. For comparison, an experiment conducted under 100% oxidizing conditions is shown in this figure as well. The degree of conversion increases with increasing cycle time and, with the exception of the 30 s experiment, is greater than that for constant oxidizing conditions. Figure 8 summarizes the results of the experiments conducted with different fractions of time under reducing conditions and different total cycle times. The extent of conversion after 2 h of experiment is shown as a function of the fraction of time under reducing conditions for five different total cycle times. It is clear that the degree of conversion increases for all cycle times up to 33% reducing conditions, with a significantly higher rate of increase for the experiments performed with 900 s total cycle time. With the exception of the long cycle time experiment, there is a rapid drop in the conversion when the fraction of reducing conditions approaches 67% with a minimum conversion level of 1% for the experiment with 30 s total cycle time. The maximum conversion is found for the sample sulfated under entirely reducing conditions, reaction 5, with a conversion of 46%. X-ray powder diffraction found no CaSO4 in this sample, only CaO and CaS. The high degree of conversion compared to when CaSO4 is the only product is attributed to the smaller molar volume of CaS (28.9 cm3 mol-1) compared to CaSO4 (46.0 cm3 mol-1) which may result in less transport resistance

Reaction between Sulfur Dioxide and Limestone

Energy & Fuels, Vol. 12, No. 5, 1998 911

Figure 8. The degree of conversion, i.e., molar S/Ca ratio, after 2 h as a function of the amount of reducing conditions for different total cycle times.

Figure 10. The correlation between the conversion calculated with eqs 11-14 and the integration of absorbed SO2.

Figure 9. The ratio of S2-/Stot as a function of the fraction of time under reducing conditions for different total cycle times.

Figure 11. The conversion as a function of time for experiments conducted with different concentrations of CO.

caused by the product accumulation in the pores of the CaO particle. Figure 9 shows the ratio of the sulfur absorbed as sulfide (S2-), as determined from wet chemical sulfide analysis, to the amount of sulfur absorbed as both CaS and CaSO4 (Stot) as a function of the fraction of time under reducing conditions. It is evident that more sulfur is absorbed as sulfide as the time under reducing conditions is increased, especially for long cycle times. Except for the cases with long cycle times or a large fraction of time under reducing conditions, the ratio S2-/Stot is small, below 5%. For comparison, the S2- to Stot ratio for the bed sample (fraction of size 0.5-0.7 mm) extracted from the 40 MW CFBB was found to be 5%. Figure 10 shows a comparison of the final conversion as calculated (i) by integration of the absorbed SO2 and (ii) gravimetrically, eqs 11-14. The different trials were divided into three groups of varying amount of reducing conditions: (a) e50%, (b) 67%, and (c) 90%. It is clear that the correlation of the two methods is very good for the experiments conducted with 50% or less reducing conditions, with a somewhat higher conversion found with integration. The good agreement between the two procedures serves two purposes: (i) because the methods are nearly independent of each other, it confirms the use of the methods in calculating the conversion at

moderate to low amounts of reducing conditions, and (ii) it is an indication that only small amounts of reduced sulfur species, such as COS and S2, were formed during these experiments. As was seen in Figure 5, some COS was clearly formed during the experiments with long cycles, which may explain the higher deviation seen for experiments with high final conversions. As the amount of reducing conditions was increased, a much higher deviation was found, with a higher conversion found with the integration method. This deviation can be explained by (i) the formation of reduced sulfur species such as COS and S2 during the reducing periods, which was confirmed for several of the experiments, and (ii) the rapid release of SO2 during the oxidation of CaS that exceeds the limit of the gas analyzer, see Figure 6b. Both of these reasons would cause an overestimation in the conversion as calculated with integration. Effect of CO Concentration. Two trials were performed with a CO concentration of 0.5 and 2% but with all other parameters constant and with a total cycle time of 240 s and 50% of the time under reducing conditions. Figure 11 shows the degree of conversion as a function of time for three experiments with 0.5%, 2%, and 4% CO and one experiment under constant oxidizing conditions. The sulfur capacity clearly increases with increasing CO concentration, in contrast

912 Energy & Fuels, Vol. 12, No. 5, 1998

to the results obtained by Hansen.17 The effect of the CO concentration is probably dependent on the cycle time and the fraction of time under reducing conditions. For instance, the effect of CO is likely to be different for a case where alternating conditions give a high conversion, compared to a case where alternating conditions give a low conversion. Conclusion The study involves an investigation of the effect of alternating oxidizing and reducing conditions on the sulfation process of a low-reactive limestone. The major parameters varied in the study was the total cycle time and the fraction of time under reducing conditions. The following general conclusions can be drawn: (1) Alternating conditions may both increase and decrease the total conversion depending on both the total cycle time and the fraction of time under reducing conditions. (2) The total conversion can be increased significantly under alternating conditions without much CaS formation, e.g., from 9% under constant oxidizing conditions to 25% for long cycle times; see Figures 8 and 9. (3) For a low fraction of time under reducing conditions, 10-33%, the conversion was always higher under alternating conditions, independent of cycle time; see Figure 8. (4) For a time fraction under reducing conditions of 67%, the conversion was very dependent on the total cycle time. For long cycle times the conversion was very high, whereas it was almost zero for short cycle times; see Figure 8. (5) When the fraction of time under reducing conditions was very high, g90%, the conversion was also very high, but almost all sulfur was in the form of CaS; see Figures 8 and 9. With respect to the introductory question of this work, i.e., the large difference in conversion between tests in boilers and laboratory, it is clear that a higher conversion can be obtained under alternating conditions. However, conversions as high as that obtained in the boiler were only obtained for long cycle times. These long cycle times do not seem realistic for the conditions in the bed of a CFBB. On the other hand, the gas composition used in the experiments was chosen to show the general effect of alternating conditions and differs significantly from that of a boiler, where species such as H2, hydrocarbons, H2S, and COS are likely to be present during the reducing periods. However, the results from this work indicate that alternating conditions are likely to be part of the explanation of the high conversion of Ko¨ping in the CFBBs. Acknowledgment. The financial support of the Swedish Board of Technical Development and the Carl Tryggers Foundation is gratefully acknowledged. We also thank Dongmei Zhao for her help with the sulfide analyses.

Mattisson and Lyngfelt

Appendix COS Formation. A calculation of the equilibrium concentrations of gaseous sulfur species for the reducing conditions used in this work (1500 ppm SO2, 4% CO, 10% CO2 and 850 °C) using a minimization of Gibbs free energy method21 found an equilibrium composition of 867 ppm COS, 310 ppm S2, and 8 ppm SO2. There are two feasible routes for the COS formation seen, either reduction of SO2, reaction 15, or oxidation of CaS:

CaS + CO2(g) f COS(g) + CaO

(A1)

However, reaction A1 can be ruled out from thermodynamic considerations. At 850 °C and with 10% CO2 over CaS, the equilibrium concentration of COS is only about 5 ppm. Reaction 15 was tested by running a mixture of 1500 ppm SO2 and 4% CO through the empty quartz reactor at 850 °C and measuring the outlet concentrations of SO2, CO2, and COS. Less than 10 ppm COS was formed at these conditions. A second experiment was then performed by passing the reactant gas mixture over 0.6 g of CaS powder (99.9% pure) from room temperature to 850 °C. Already at about 200 °C the SO2 concentration dropped rapidly, followed shortly by a corresponding increase in the COS and CO2 concentrations. At 850 °C the outlet concentrations were ∼0 ppm SO2, 1100 ppm COS, and 0.3% CO2. It is clear that the presence of CaS catalyzes the formation of COS through reaction 15 and is probably the main route whereby COS is formed in the experiments in the present study. The fact that the COS concentration increased throughout the reducing periods, see Figure 5, as more CaS was formed also agrees with this interpretation. Though not analyzed, the formation of elemental sulfur was seen from deposits at the bottom of the reactor and is also expected thermodynamically. S2 could be formed through

SO2(g) + 2CO(g) f 1/2S2(g) + 2CO2(g)

(A2)

Glossary fCaCO3 Mi Wf Wf,adj Wi XCaS XCaSO4 Xtotal YS2-

mass fraction of CaCO3 in limestone molecular weight of component i, g mol-1 weight of sorbent following sulfation, g final weight of sulfated limestone, adjusted for inerts, g initial weight of limestone sample, g conversion of CaO to CaS conversion of CaO to CaSO4 total conversion mass fraction of sulfide present in the sulfated lime EF970232S

(21) White, W. B.; Johnson, S. M.; Dantzig, G. B. J. Chem. Phys. 1958, 28, 751-755.