ef900714q

8/12/2019 ef900714q

1/17

5291r 2009 American Chemical Society pubs.acs.org/EF

Energy Fuels2009,23, 52915307: DOI:10.1021/ef900714q

Published on Web 10/16/2009

Review of Mid- to High-Temperature Sulfur Sorbents for Desulfurization of

Biomass- and Coal-derived Syngas

Singfoong Cheah,* Daniel L. Carpenter, and Kimberly A. Magrini-Bair

National Bioenergy Center, National Renewable Energy Laboratory, 1617 Cole Blvd., MS 3322 Golden, Colorado 80401

Received July 11, 2009. Revised Manuscript Received September 2, 2009

This review examines state-of-the-art mid- and high-temperature sulfur sorbents that remove hydrogensulfide (H2S) from syngas generated from coal gasification and may be applicable for use with biomass-derived syngas. Biomass feedstocks contain low percentages of protein-derived sulfur that is convertedprimarily to H2S, as well as small amounts of carbonyl sulfide (COS) and organosulfur compounds duringpyrolysis and gasification. These sulfur species must be removed from the raw syngas before it is used fordownstream fuel synthesis or power generation. Several types of sorbents based on zinc, copper, iron,calcium, manganese, and ceria have been developed over the last two decades that are capable of removingH2S from dry coal-derived syngas at mid-to high-temperature ranges. Further improvement is necessary todevelop materials more suitable for desulfurization of biomass-derived syngas because of its hydrocarbon,tar, and potentially high steam content, which presents different challenges as compared to desulfurization

of coal-derived syngas.

1. Introduction

Rising world demands for oil and a finite petroleumreservehave increased interest in alternate liquid fuel sources that arediverse, secure, and affordable. Biomass is a renewable sourceof carbon that is abundant in many regions of the world. Inthe United States it has been estimated that domestic biomassresources could be used to sustainably replace more than athird of the U.S. demand for transportation fuel (2005 basis).1

Worldwide, the International Energy Agency estimated thatthere is sufficient nonfood biomass available to support a

substantial increase in bioenergy use.2With reduction of green house gas emissions to consider,

the focus of next-generation biofuelsis to producefuel derivedfrom cellulosic biomass, especially biofuels made with wastebiomass or feedstock grown on marginal lands that generatelittle carbon today.3-5 Thermochemical conversion of bio-mass to fuels via pyrolysis or gasification is a viable route thathas the potential to be cost competitive with gasoline, as well

as biochemical biomass conversion processes.6 During gasifi-cationof biomassfeedstocks, a gasmixture comprisingmainlyH2, CO, CO2, H2O, and CH4is produced along with severalunwanted byproducts, with their concentrations dependingon feedstock, gasifier design, and process conditions. Theseunwanted byproducts include organic tars, sulfur and nitro-gen heteroatom species (thiophene, pyridine), and inorganicconstituents containing sulfur (H2S, COS), chlorine (HCl),nitrogen (NH3, HCN), and alkali metals.

6-11

Although there are quite a number of studies that measuredthe amount of sulfur in dry biomass feedstocks, data on thelevels and speciation of sulfur in the actual syngas that isderived from biomass gasification are scarce. A summary ofthe available data on sulfur concentrations in biomass syngasis presented in Table 1. Even though this review focuses onmaterials that remove sulfur, it also includes discussion of theinfluence of other gas impurities on sorbent performance andof the capability of the sorbents to remove multiple contami-nants. Therefore the concentrations of other impurities suchas HCl, nitrogen compounds, and tar are also included inTable 1.

Van der Drift et al. tested 10 different biomass feedstocks inan atmospheric air-blown gasifier and determined H2S in thebiomass-derived syngas tobe in the range of 50-230 ppmv for

several different types of wood (Table 1).

7

Sato et al. measuredH2S at the outlet of a bench-scale heated atmospheric flui-dized bed gasifier and found less than 50 ppmv H2S and

*To whom correspondence should be addressed. Telephone: (303)384-7707. Fax: (303) 384-6363. E-mail: [email protected].

(1) Perlack, R. D.; Wright, L. L.; Turhollow, A. F.; Graham, R. L.;Stokes, B. J.; Erbach, D. C. Biomass Feedstock for a Bioenergy andBioproducts Industry: The Technical Feasibility of a Billion-Ton Annual

Supply; DOE/GO-102005-

2135; Oak Ridge National Laboratory: OakRidge, TN, 2005.

(2) International Energy Agency. The availability of biomass resourcesfor energy;summary and conclusions from the IEA Bioenergy ExCo58Workshop. http://www.ieabioenergy.com/MediaItem.aspx?id=5794(accessedAugust 27, 2009).

(3) Farrell, A. E.; Plevin, R. J.; Turner, B. T.; Jones, A. D.; O0Hare,M.; Kammen, D. M.Science2006,311 (5760), 506508.

(4) Searchinger, T.; Heimlich, R.; Houghton, R. A.; Dong, F.;Elobeid, A.; Fabiosa, J.; Tokgoz, S.; Hayes, D.; Yu, T.-H. Science2008,319 (5867), 12381240.

(5) Tilman, D.; Hill, J.; Lehman, C.Science2006,314 (5805), 15981600.

(6) Phillips, S.; Aden, A.; Jechura, J.; Dayton, D.; Eggeman, T.Thermochemical Ethanol via Indirect Gasification and Mixed AlcoholSynthesis of Lignocellulosic Biomass; NREL/TP-510-41168; National Re-newable Energy Laboratory: Golden, CO., 2007.

(7) van der Drift, A.; van Doorn, J.; Vermeulen, J. W. BiomassBioenergy2001,20 (1), 4556.

(8) Sato, K.; Shinoda, T.; Fujimoto, K. J. Chem. Eng. Jpn.2007,40(10), 860868.

(9) Milne, T. A.; Evans, R. J.; Abatzoglou, N. Biomass Gasifier Tars:Their Nature, Formation, and Conversion; NREL/TP-570-25357;

National Renewable Energy Laboratory: Golden, CO, November 1998.(10) Leppalahti, J.; Koljonen, T. Fuel Process. Technol. 1995,43 (1),

145.(11) Torres, W.; Pansare, S. S.; Goodwin, J. G., Jr. Cat. Rev. -Sci.

Eng.2007,49, 407456.

8/12/2019 ef900714q

2/17

5292


30 ppmv SOxwhen wood chips were used as feedstock, andmore than 300 ppmv total sulfur when dried sewage sludgewas used as a feedstock (Table 1).8 Kuramochi et al. modeledH2S and HCl speciation during gasification of a number ofbiomass feedstocks, including several that were studied by vander Drift et al., using thermodynamic modeling.12 There are,however, disparities between experimental and model results,indicating that further research is necessary. Research con-ducted at the National Renewable Energy Laboratory(NREL) has shown that at 850 C, the syngas H2S contentranges from about 20-50 ppmv for Vermont wood to300-600 ppmv for herbaceous feedstocks such as switchgrassand wheat straw.13,14 Boerrigter summarized data of typicalgas compositions obtained with three different types of gasi-

fiers and found an H2S content of 40-

120 ppmv when woodwas used as a feedstock.15

There are a number of reasons that sulfur needs to beremoved from biomass derived productgas. Hydrogen sulfideand other sulfur compounds can cause pipeline corrosion andthus limit plant lifetime.16 In addition, there are regulations tolimit the sulfur content in fuel such as diesel to 10-15 ppm inUnited States,17 European Union, Canada, Australia, andNew Zealand.18 It is well-known that sulfur in fuel is aprecursor to SO2 that will be released to the atmosphere,resulting in downstream environmental problems. It can beexpected that fuel derived from cellulosic biomass feedstockwill have similar requirements in terms of sulfur content;consequently it is necessary to removesulfur compounds from

biomass-derived syngas/fuel/products at some point in theprocessing.

Hydrogen sulfide is also a well-known catalyst poison.Information on the effect of sulfur on tar reforming or steamreforming catalysts in biomass product gas environments is

widely available and generally shows significant negativeimpacts of sulfur on catalyst performance.8,19-23 In additionto its effect on tar reforming catalysts, sulfur also affects fuelsynthesis catalysts such as those used in mixed alcohol and inFischer-Tropsch hydrocarbon synthesis. For example, traceamount of H2S in cedar wood derived syngas poisons ruthe-nium catalysts used for Fischer-Tropsch synthesis.24 Thespecific activities of copper catalysts used in methanol synth-esis25 and cobalt catalysts used in Fischer-Tropsch fuelsynthesis decrease significantly with H2S/H2ratios as low astens to hundreds of ppb (Figure 5.6 and Chapter 6 of ref 26).In contrast, a sulfided molybdenum alcohol synthesis catalystrequires up to 100 ppm H2S to maintain sufficient activity.

6

Consequently, some level of sulfur removal will be necessary

to achieve biomass based catalytic fuel production.6

A sum-mary of syngas purity requirements for downstream catalystsis presented in Table 2.

In a technoeconomic analysis of thermochemical ethanolproduction conducted at NREL, ethanol production, tarreforming, acid gas, and sulfur removal processes togethercomprise 31% of the total fuel productioncost (second largestcost component after feedstock).6 Thus, a more efficientmethod of H2S removal will significantly reduce the overallbiofuel production cost.

In conventional treatment, H2S or other sulfur compoundsare removed via low temperature amine scrubbers. RemovingH2S from biomass-derived syngas using scrubbers thus re-quires lowering the temperature of the syngas from 850 C

(temperature of gasification) to 40-50 C for cleanup27

withconcurrent tar condensation. However, tar condensation cancause plugging and fouling of the condenser and downstreamprocess piping, presenting a significant operating challenge. Inaddition, the scrubbed gas must then be reheated for down-stream fuel synthesis, which occurs from 250-400 C(Figure 1). This cooling and reheating of the syngas is boththermally inefficient and expensive. Theoretically, the heatenergy can be partially recovered using heat exchange equip-ment, but tar build up on reactor walls during condensationcould render the heat recovery challenging. In addition to thetar condenser clogging problem, tar condensation also reducesbiomass carbon utilization and produces a large waste streamthat must be treated. New mid- (400-600 C) to high-tem-

perature (600-850 C) syngas cleaning processes being devel-oped by the coal industry, for fuel cell applications, and forbiomass syngas applications offer potential economic advan-tages by allowing gasification, gas clean up, and downstreamprocesses to be operated at similar process temperatures.

Table 1. Experimentally Determined Concentrations of Impurities in Biomass-Derived Syngas

H2S (ppmv, dry basis) HCl (ppmv, dry basis) NH3(ppmv) tar (g/Nm3)

50-230 (wood)a 1-200 (wood, verge grass, etc.)a 1,000-13,000 (wood, verge grass, etc.)a 1-150b

8/12/2019 ef900714q

3/17

5293


Even though significant progress has been achieved in thedevelopment of midtemperature desulfurization of coal de-rived syngas, desulfurization of biomass-derived syngas pre-sents both unique challenges and advantages due to its

composition. Coal-derived syngas typically has 0.1-1.5%sulfur, which is higher than that in biomass-derived syngas.28

Coal gasification is usually a direct oxygen- or air-blownprocess conducted at higher temperature than biomass gasi-fication. Consequently, the coal syngas that is producedusually has low water and hydrocarbon content.

Biomass gasification is accomplished using either a directoxygen-blown process or an indirect steam-driven process.Both types of gasifiers are currently under development andbothare able to produce a syngassuitable for downstreamfuelsynthesis. For direct gasification, air is not usually considereda gasifying agent in the context of fuel production because thehigh nitrogen levels in the resulting syngas would be econom-ically unacceptable. Therefore, oxygen is often used as a

gasifying agent in direct gasification, a major disadvantagebeing the need for concurrent oxygen production. This step isexpensive andrequires large plant sizes to improve economics.For methanol production, a comparison of these two gasifica-tion processes indicates that the capital intensity, that is thecapital cost per unit product, is comparable or less expensivefor indirect gasification (Figure 1 in ref 6).

Steam gasification brings its own set of challenges, such aserosion problems inside reactor vessels due to the need tocirculate hot solids for process heat. Also, since indirectgasifiers are usually operated at low pressure, the synthesisgas requires downstreamcompression for fuel production. Onthe other hand, there are several advantages to using steam.Steam helps to produce a syngas with a H2/CO ratio that ishigher than one,29 whichis more suitable for mixed alcohol orhydrocarbon fuel (e.g., via the Fischer-Tropsch route) synth-eses. Moreover, it has been reported that the tar content of thesyngas decreases as the steam/biomass ratio increases,30 andmore recent research suggests that less tar is produced withsteam/oxygen mixtures as a gasifying agent.29,31 Steam in the

process gas also increases tar reforming and reduces cokeformation.29 Using steam as a gasifying agent results in a rawsyngas that has much higher steam content (30-65%) thanthat in syngas from coal conversion, with the low and highends from gasifiers using oxygen/steam and steam as media,respectively.6,9,11,15 Although biomass-derived syngas con-tains much lesssulfurthan coal-derivedsyngas, thisadditionalsteam content presents process challenges unique to biomassgasification.

The other process challenge unique to biomass derivedsyngas is the higher hydrocarbon and tar content comparedto coal syngas. Methane concentrations can range 10-15%by volume, and other light hydrocarbons, such as acetylene,ethylene, and ethane, can total a few percent.19 Tar may bepresent in concentrations of up to 5%.9 It is well-known thatsulfur has a detrimental effect on tar reforming catalysts;

however, the potential effect of tar on high temperaturesulfur sorbents is still not clear. In addition, hydrocarbonssuch as ethene have been known to form coke on steamreforming catalysts,32,33 and could also potentially form cokeon sulfur sorbents. Literature data on the impacts of somethese impurities on sorbent performance are summarized inSection 2.8.

1.1. Economics of Mid- to High-Temperature Desulfuriza-

tion.Because of the lack of data about sorbent performancein biomass-derived syngas, no economic analyses exist re-garding the thermal efficiency and process economics ofusing high temperature desulfurization versus conventionalsorbent processes. Several reports were published describinghot gas cleanup systems applied to coal gasification. For

example, RTI International estimated that desulfurization at370-480 C in a 600 MW integrated gasification combinedcycle (IGCC) plant could increase the overall process ther-mal efficiency by 3.6 efficiency points over conventionalSelexol sulfur removal technology,34 where thermal effi-ciency is defined as the ratio of net work done to the heatcontent of the fuel that is consumed. This gain would reducethe plant cost by $269 per kilowatt, resulting in a associated9.6% reduction of electricity cost.34 Considering that IGCCnormally has a thermal efficiency of 40-50 points,35,36 anincrease of 3.6 thermal efficiency points is significant.

The Netherlands Agency for Energyand the Environment(NOVEM)37 modeled hot gas cleanup with a 600 MWsystem operated at 250, 350, and 600 C. At each tempera-

ture a different set of processes for cleaning up NH3, halo-gens, sulfur, and dust was used. One exception was that thesame dehalogenation and denitrification processes were used

Table 2. Acceptable Levels of Impurities in Biomass-Derived Syngas for Downstream Fuel Synthesis Catalytic Processes

process sulfur (ppmv) halogen (ppbv) nitrogen (ppmv) tar

Fischer-Tropsch sum of sulfur compounds

8/12/2019 ef900714q

4/17

5294


at 600 and 350 C, which might have contributed to themarginal improvement in thermal efficiency at 600 C overthat at 350 C. One of the key conclusions of the reportwas that the gain in thermal efficiency from using mid- tohigh-temperature desulfurization is intimately tied to theavailability and choice of denitrification technology; conse-quently, research and development of denitrification pro-cesses is important.

Another factor to be taken into consideration is alkali

removal. Because of process equipment limitations and theneed to operate below alkali metal condensation tempera-tures (so that the condensed metals can be collected andfiltered),38,39 the optimum hot gas cleanup temperaturerange designed for use in the coal industry is approximately300-600 C.

38-40 Information on alkali content of biomass-derived syngas is scarce, and its effect on the ultimateoptimum biomass syngas clean up temperature and condi-tions will still need to be determined.

The processes downstream of syngas cleanup for IGCCare power generation using gas and steam turbines, whichdiffer significantly from the downstream fuel synthesis pro-cesses that are used for biomass-derived syngas. Therefore, itcannot be assumed that thermal efficiency improvements in

the biomass-based process would be similar. In addition,because of the inefficiency and cost of transporting biomassover long distances, it is likely that most biomass gasificationplants will be significantly smaller than coal gasificationplants. This will likely entail different process economicsfor heat recovery and gas cleaning in the smaller biomassgasification plants. As a result, it is critical that thermalefficiency and economics for biomass syngas cleanup beanalyzed as more data on sorbent performance in biomass-derived syngas becomes available.

In this literature review, sources related to H2S removal inthe context of IGCC, biomass gasification, and fuel cellapplications are reviewed. The criteria for choosing anappropriate sorbent material for use with biomass-derived

syngas are then discussed. In later sections, the basic sulfursorption steps and research results obtained with severalsorbent materials are discussed. The review then focuses onsorbent regeneration pathways, which are dependent bothon the inherent thermodynamics of the material and also onprocess conditions.

1.2. Criteria for Selecting Sulfur Sorption Materials.Cri-teria that the sorbent material must satisfy, for syngasdesulfurization to be both economical and operational,are: (1) High adsorption capacity for H2S. This reduces bothsorbent quantity and process equipment size. (2) Fast ad-sorption kinetics. (3) Mechanical properties: low attritionrate and able to tolerate high temperature. (4) Chemicalproperties: stable in reducing environment containing steamand hydrocarbon. (5) Regenerable through a suitable path-way while maintaining efficient sulfur sorption capacityduring repeated sulfidation-regeneration cycles. (6) It islikely that other contaminants such as HCl and NH3 will

also require removal from raw syngas. Multifunctional mid-to high-temperature sorbents for HCl, NH3, and H2S re-moval could present significant advantages in terms ofoperation and economics.41

2. Sulfur Sorption Materials

2.1. Basic Sulfidation Reactions. Much of the researchconducted to date uses metal based sorbent materials thatare presumed to remove H

2S according to the general

reaction:

MxOys yH2Sg SMxSys yH2Og 1

where M denotes a suitable metal. Equation 1 describes asolid-gas reaction. This reaction is distinct from, and pre-ceded by, a surface reaction that occurs when the time scalefor sulfidation is short.42

Westemorland conducted a pioneering thermodynamicsstudy of thestable phasesof various oxidesand theaffinity ofthese oxides for H2S adsorption and chose oxides or carbo-nates of barium, calcium, cobalt, copper, iron, manganese,molybdenum, strontium, tungsten, vanadium, and zinc asviable candidates.43 Slimane and Abbasian discussed in

detail how to link the thermodynamic results to practi-cal experimental parameters such as regeneration con-ditions.44

Zinc, manganese, copper, iron, rare earth, and calciumsorbents areamong themost promisingand most extensivelystudied, and they are discussed in detail in Sections 2.2-2.7.Selected results and experimental conditions are alsosummarized in Table 3. Several other oxides, though theoreti-cally capable of removing H2S, are not well suited fordesulfurization of syngas for varied reasons. For example,molybdenum and tungsten oxides can form carbides, whichhave poor desulfurization capabaility.45 Strontium- andbarium-based carbonate materials behave similarly to cal-cium carbonate, but calcium-based sorbents are preferable

because of their lower cost and wider operating temperaturerange.43

In addition to thermodynamic properties, kinetics ofdesulfurization (or sulfidation of the solid sorbents) is criticalto establishing sorbent performance. Several studies indi-cated the intrinsic reaction kinetics of the sulfidation of ZnO,MnO, CaO, Fe, and Cu on ceria to have first-order depen-dence in H2S.

46-48

Besides sorbent efficiency and kinetics, other side reac-tions or catalyzing properties of the sorbents would need to

(38) Abbasian, J.; Slimane, R. B.; Lau, F. S.; Wangerow, J. R.;Zarnegar, M. K. In Development of High Temperature Coal Gas Desul-

furization Systems ; An Overview, Fourteenth Annual InternationalPittsburgh Coal Conference, Taiyuan, Shanxi, P.R. China, September23-27, 1997; Pittsburgh Coal Conference, 1997; S4/21-S24/28.

(39) Fantom, I. R.; Cahill, P.; Sage, P. W. Hot gas cleaning ; Anoverview. InDesulfurization of Hot Coal Gas, Atimtay, A. T.; Harrison, D.P., Eds.; Springer: Berlin, 1998; Vol. 42, pp 103-116.

(40) Zhu, F.; Li, C.; Fan, H.; Li, Y. Meitan Zhuanhua (Coal Con-version)2000,23 (2), 1722.

(41) Huber, G. W.Breaking the Chemical and Engineering Barriers toLignocellulosic Biofuels: Next Generation Hydrocarbon Biorefineries;University of Massachusetts at Amherst. National Science Foundation,

Chemical, Bioengineering, Environmental, and Transport Systems Division:Washington D.C., 2008.(42) Flytzani-Stephanopoulous, M.; Sakbodin, M.; Zheng, W.

Science2006,312 (5779), 15081510.(43) Westmoreland, P.; Harrison, D. P. Environ. Sci. Technol.1976,

10(7), 659661.(44) Slimane, R. B.; Abbasian, J. Adv. Environ. Res.2000,4 (2), 147

162.(45) Elseviers, W. F.; Verelst, H.Fuel1999,78, 601612.(46) Tamhanker, S. S.; Bagajewicz, M.; Gavalas, G. R.; Sharma, P.

K.; Flytzani-Stephanopoulos, M. Ind. Eng. Chem. Proc. Des. Dev.1986,25, 429437.

(47) Westmoreland, P. R.;Gibson, J. B.;Harrison, D. P. Environ. Sci.Technol.1977,11 (5), 488491.

(48) Flytzani-Stephanopoulous, M.; Li, Z. Kinetics of sulfidationreactions between H2S and bulk oxide sorbents. In Desulfurization ofHot Coal Gas, Atimtay, A. T.; Harrison, D. P., Eds.; Springer: Berlin, 1998;Vol. 42, pp 179-211.

8/12/2019 ef900714q

5/17

8/12/2019 ef900714q

6/17

5296


be considered. Iron-, cobalt-, and copper-oxide based mate-rials readily reduce in reducing environments at high tem-perature.49 Iron oxide and copper metal are both water gasshift catalysts that promote conversion of CO and H2O toCO2 andH2,at350-500 Candbelow200C, respectively.

26

Consequently the potential effect of iron (and possiblycopper) sorbent components on the resulting CO/H2 ratioin the desulfurized syngas should be considered as they mayoffer additional compositional tuning of the syngas for

downstream fuel synthesis.2.2. Zinc-Based Materials. Zinc-based sorbents have

been used in desulfurization of natural gas feedstock,typically at 370 C.

50 Consequently, it was one of the firstmaterials to be studied extensively for coal syngas desul-furization. Early work with zinc oxide investigated theeffects of product layer diffusion, pore diffusion, and gas-film diffusion on the kinetics of sulfidation. For example,Gibson and Harrison (Table 3) attributed an apparent20% utilization of the ZnO sorbent mass to be due to slowdiffusion from pore closure and formation of a densesulfide layer.51

Sulfate formation is a factor that needs to be carefullyaddressed when zinc-based sorbent is used.50 Siriwardane

and Woodruff characterized reactions of water vapor andoxygen with zinc sulfide at 550-650 C using Fouriertransform infrared spectroscopy (FTIR).52,53 Oxygen pro-motes the formation of SO2and sulfate on the surface, butthe extent of sulfate formation is less at higher tempera-tures.52 The presence of water vapor promotes the forma-tion of sulfite (SO3

2-), at the expense of sulfate (SO42-). The

authors suggested that it may be advantageous to havesteam present during regeneration of ZnS since waterpromotes formation of the sulfite species, which are easierto decompose or remove from the surface than sulfatespecies.

Sasaoka et al. studied the reaction of sulfided ZnOsorbent with COS (Table 3).54,55 They found that ZnS

catalyzes hydrolysis of COS to H2S, and the H2S formedcan be removed by ZnO downstream in a packed bedreactor.

Though zinc oxide has a high H2S sorption capacity, attemperatures above 600 C, early investigation by Gibsonand Harrison (Table 3) has clearly demonstrated thatreduction of ZnO in the highly reducing atmosphere ofsyngas followed by vaporization of elemental zinc canpresent a significant problem.51 Consequently, a significantpart of the research on zinc-based material involved mod-ifying it to improve its stability in the temperature range ofinterest.

Sasaoka et al. examined the stability of zinc oxide withZrO2, TiO2, and Al2O3addition and concluded that even

though these oxides improve the performance, they donot completely eliminate reduction and vaporization ofzinc.56 Many other research groups, however, foundTiO2 to be a very useful stabilizer of ZnO. Sections2.2.1 and 2.2.2 summarize some of the key findings inthe development of zinc ferrite and zinc titanate sulfursorbents.

Baird et al. found that transition metals coprecipitatedwith zinc oxide tend to produce mixed oxides of higher

surface areas.57

Quantification of the improvements in per-formance due to surface area increases versus other mechan-isms will be tremendously useful.

Zinc-based materials remain promising, particularly inthe low- (

8/12/2019 ef900714q

7/17

5297


Kobayashi et al. studied zinc ferrite and zinc ferrite-silicacomposite sorbents.67-71 Using in situ X-ray diffraction,Mossbauer spectroscopy, and fixed bed reactor tests, theyshowed that when the H2S concentration was less than80 ppmv, zinc was the reactive sorbent and a mixture of zincsulfide and iron oxide (Fe3O4 and FeO) formed. At higherH2S concentrations, both zinc and iron were reactive andformed zinc and iron sulfides. The authors conducted sulfi-dation and regeneration for 20 cycles and used that data to

extrapolate the loss in activity to 500 cycles of desulfuriza-tion. They concluded that verification of the long-termdurability and activity of the sorbent would require furthertesting of up to 40-50 total cycles.

71

In a separate study, Ayala and Marsh characterized andconducted long-range testing (50 cycles of sulfidation andregeneration) of zinc ferrite sorbent at approximately 550 Cand detected no undesired metal carbide, sulfate, and ele-mental iron based on XRD.61

Gangwal et al. evaluated zinc ferrite based sorbents59,72 andfound that in as little as 5% steam, soot formation, which theyhypothesized to be catalyzed by iron, was a process chal-lenge.59 Sasaoka et al. also investigated the zinc ferrite systemand found that soot formed on zinc ferrite sorbent at 500 C

under atmospheric pressure.73 Soot formation decreased asreaction temperature (450-600 C) increased, and soot for-mation was accelerated by the presence of H2 and CO, butsuppressed by H2O and CO2. On the basis of scanning electronmicroscopy and XRD data, they hypothesized that soot andiron carbide (Fe3C and FexC) formation are correlated.

Besides the binary zinc ferrite system, a number of ternarymodified zinc ferrite systems were also studied. Siriwardaneand Poston investigated the zinc copper ferrite system.74 Thestarting material was 28.5 mol % ZnO, 65% Fe2O3, 4.6%CuO, and 2% bentonite, but XRD showed that the synthe-sized material did not significantly incorporate copper andhad a primary composition of ZnFe2O4. This XRD resultsuggests that copper may exist in an amorphous phase or is

bound to zinc and iron to form dispersed species that cannotbe detected by XRD. X-ray photoelectron spectroscopy(XPS), however, showed that when this material was heatedto 550 C, most of the copper was on the surface in the 1oxidation state. Enrichment of copper on the surface wasattributed by Siriwardane and Poston to the high perfor-mance of zinc copper ferrite as sulfur sorbent. In a reducingenvironment, iron in zinc copper ferrite exists in both Fe3

and Fe2 oxidation states.Pineda et al. characterized zinc oxide and zinc ferrite

sorbents doped with titanium or copper using XRD andRaman spectroscopy.75 For the study on zinc copper ferrite,

they fixed the iron atomic content at 50% and varied thecopper content from 7 to 40%.75 They concluded that copperdoes not affect the stability of zinc ferrite. X-ray photoelec-tron spectroscopy measurements indicate thatthe ratio of Cuto Fe on the surface is always much higher than the bulk,indicating copper migration to the surface. The authorshypothesized that copper migrates to the surface duringcalcinations and regeneration steps. They attributed theimproved sorbent performance (lower H2S concentration

after sulfidation) to the enhanced level of copper on thesurface, in agreement with results from Siriwardane andPoston.74 The finding of high H2S removal with copperaddition into zinc ferrite is in agreement with results reportedby Gangwal et al.,59 who found that addition of copper intozinc ferrite improves its efficiency, and the copper modifiedzinc ferrite sorbent removes H2S to less than 1 ppmv at600 C even with 20% steam in the atmosphere. Pineda et al.also found that during sulfidation Cu-Zn-Fe mixed oxidesare completely converted into low oxidation sulfides, that is,the sulfide species on the surface are Cu2S,FeS, and ZnS andthat in the case of FeS there were a lot of Fe vacancies.75

Multicycle tests of these sorbents show that the decay inperformance is not related to structural changes but to

decreased porosity.76Pineda et al. also investigated the effect of incorporating

titanium into zinc ferrite, with zinc content fixed at 50 atom %,and titanium content varied from 7 to 40 atom %.75 At650 C, the inclusion of titanium actually impeded theformation of zinc ferrite. However, with calcination con-ducted at 1100 C, the addition of titanium to zinc ferriteprevented decomposition into its component oxides, that is,titanium addition stabilized the ferrite lattice. The stabiliza-tion of titanium is particularly effective in noniron-contain-ing sorbents.77

In summary, zinc ferrite is an efficient sorbent at approxi-mately 500 C. Efforts to produce ternary materials includedpreparation of zinc copper ferrite, which can reduce H2S to

lower levels than zinc ferrite, and manufacture of zinc ferritedoped with titanium, which shows higher stability undercertain preparation conditions. The interest in zinc ferritematerials for moderate temperature desulfurization con-tinues, and more recently the sol-gel method has beenexplored as a way to synthesize a high surface area zincferrite sorbent.78

2.2.2. Zinc Titanate. Combining zinc and titanium dioxideis one of the most successful approaches to stabilize ZnOagainst reduction at mid- to high-temperature ranges. Lewetal. conducted extensive research on the kinetics and mechan-isms of reduction and sulfidation of a series of Zn-Ti-Omaterials with different zinc to titanium ratios (Table 3).79-81

Lew et al. found that between 400 and 700 C, the activationenergies for sulfidation of Zn-Ti-O and ZnO sorbents weresimilar (9-10 kcal/mol), indicating that the sulfidation

(67) Kobayashi, M.; Shirai, H.; Nunokawa,M. Energy Fuels 1997, 11(4), 887896.(68) Kobayashi, M.; Shirai, H.; Nunokawa, M. Ind. Eng. Chem. Res.

2000,39 (6), 19341943.(69) Kobayashi, M.; Shirai, H.; Nunokawa,M. Energy Fuels 2002, 16

(3), 601607.(70) Kobayashi, M.; Shirai, H.; Nunokawa, M. Ind. Eng. Chem. Res.

2002,41 (12), 29032909.(71) Kobayashi, M.; Shirai, H.; Nunokawa,M. Energy Fuels 2002, 16

(6), 13781386.(72) Gangwal, S. K.; Stogner, J. M.; Harkins, S. M.; Bossart, S. J.

Environ. Prog.1989,8 (1), 2634.(73) Sasaoka, E.; Iwamoto, Y.; Hirano, S.; Uddin, M. A.; Sakata, Y.

Energy Fuels1995,9 (2), 344353.(74) Siriwardane,R. V.;Poston, J. A. Appl. Surf. Sci. 1993, 68 (1), 65

80.(75) Pineda, M.; Fierro, J. L. G.; Palacios, J. M.; Cilleruelo, C.;

Garcia, E.; Ibarra, J. V. Appl. Surf. Sci.1997,119 (1-2), 110.

(76) Pineda, M.; Palacios, J. M.; Alonso, L.; Garcia, E.; Moliner, R.Fuel2000,79 (8), 885895.

(77) Garcia, E.; Cilleruelo, C.; Ibarra, J. V.; Pineda, M.; Palacios, J.M.Ind. Eng. Chem. Res. 1997,36 (3), 846853.

(78) Zhang, R.; Huang, J.; Zhao, J.; Sun, Z.; Wang, Y. Energy Fuels2007,21 (5), 26822687.

(79) Lew, S.; Sarofim, A. F.; Flytzani-Stephanopoulos, M. Chem.Eng. Sci.1992,47(6), 14211431.

(80) Lew, S.; Sarofim, A. F.; Flytzani-Stephanopoulos, M.Ind. Eng.Chem. Res.1992,31 (8), 18901899.

(81) Lew, S.; Sarofim, A. F.; Flytzani-Stephanopoulos, M. AlChE J.1992,38 (8), 11611169.

8/12/2019 ef900714q

8/17

5298


mechanism on these two types of solids is likely the same.80

However, for sorbents containing more than 25 mol % Ti,the initial sulfidation rate of Zn-Ti-O sorbents was 1.5-2times slower than that of ZnO, suggesting the frequencyfactor in the rate expression for Zn-Ti-O is smaller, that is,there are fewer reaction sites on Zn-Ti-O than ZnO. Thesmaller number of reaction sites on Zn-Ti-O was presumedto be due to nonreactive titanium on the surface.80 Lew et al.conducted studies on the kinetics of Zn-Ti-O reduction

and found that in H2-N2 gas mixtures, Zn-Ti-O solidshave a lower reduction rate than ZnO in the temperaturerange550-1050 C and the associatedactivation energies forZn-Ti-O and ZnO reduction were calculated at 37 and24 kcal mol-1, respectively.79 Modeling of the results withthe overlapping grain model suggests that the primarylimitation to reaction rate on both Zn-Ti-O and ZnO isdiffusion through a ZnS product layer.81

Gangwal et al. studied various binary oxides of zinc andfound that zinc titanate had excellent mechanical strengthand capacity retention at 700 C.

59,72 The reduction,sulfidation, and regeneration kinetics of zinc oxide-tita-nium oxide single pellet sorbents were studied and mod-eled by Woods et al.82 (Table 3) and Jothimurugesan and

Harrison.83 In agreement with Lew etal.,79 they found thatthe addition of titanium oxide increased the maximumsorbent operating temperature range by stabilizing zincoxide against reduction and subsequent volatilization.Mass transfer and product layer diffusion resistance werethe main parameters that controlled the global reactionrate.

Pineda et al.75 found spectroscopic evidence that theaddition of titanium increases the stability of ZnO againstreduction through the formation of Zn2TiO4, in agreementwithLew etal. and Woods etal.79,82Because of theinterestinZnO-TiO2 materials, Yangand Swisher studied the stabilityof Zn2Ti3O8 in detail and provided additional informationon the ZnO-TiO2phase diagram.

84

Siriwardane and Poston studied the sulfidation of zinctitanate in the presence of H2and CO.

85 Using XPS to studythe surface species after H2S, CO, and H2 exposures at670-800 C, they deduced that when Zn2TiO4 reacts withH2S, zinc is sulfided while titanium is not, with oxygenrelease according to the reaction:

Zn2TiO4 2H2S S2ZnS TiO2 2H2 O2 2

The authors also concluded that some of the O2 canoxidize H2S to SO2 and also ZnS to ZnSO4. The reactionofSO2 with Zn2TiO4 canalso resultin ZnSO4 formation. In areducing syngas, both H2 and CO can reduce ZnSO4 toZnO.85

Hatori et al. studied the role of TiO2 in oxidative regen-eration of a ZnO-TiO2 sorbent.86 They found that TiO2

increased the oxidative regeneration rate by accelerating thereaction between ZnSand O2 aswellas thatbetweenZnS andH2O. The rate increase could be explained by activation of

H2O over TiO2 and subsequent spillover of the activatedH2O species to react with ZnS. The overall reaction is

ZnS3H2OSZnOSO23H2 3

3H23=2O2S3H2O 4

A few research groups reported on modifying or dopingZn-Ti based materials further with other oxides, and thisapproach yielded some promising results. Sasaoka et al.

investigated modification of a 50% ZnO-

TiO2 mixture, amixed oxide that primarily has the Zn2Ti3O8 structure.

87

Addition of 5-10% ZrO2 improved its reactivity for H2Sremoval and regenerability. They authors hypothesized thataddition of ZrO2improved the pore structure and helped tomaintain a large surface area after regeneration.87

Siriwardane et al. studied molybdenum-containing zinctitanate sorbents.88 They found evidence of sulfate forma-tion, and that changes in the structure due to the largevolume difference between sulfate and oxide during sulfida-tion and regeneration may have contributed to sorbentspalling. In addition, they also found that when regenerationwas conducted at 649-760 C, there was incomplete refor-mation of the titanate structure, and that the degree of

incomplete reformation increased (amount of pure TiO2increased) with an increasing number of sulfidation andregeneration cycles.

Liu et al. investigated the effect of V2O5, B2O3, andtungsten doping on sintering and phase transition of zinctitanate ceramics.89-91 Jothimurugesan and Gangwal eval-uated the coprecipitation of a series of transition metaloxides with zinc-titanum oxides as a means of loweringthe regeneration temperature of zinc titanate from 650 C.

92

They found that a combination of a small weight percent ofnickel and cobalt added to zinc titanate effectively reducedthe regeneration temperature by at least 100 C withoutimpacting sorbent performance in subsequent cycles.92

Jun et al.93 investigated cobalt oxide-doped zinc titanate

further and found it to have higher sulfur sorption capacityand better regenerability (particularly at 480 C) than un-doped zinc titanate (Table 3). They hypothesized that themodified materials have better regenerability partly becauseaddition of cobalt into the structure minimized volumeexpansion and contraction during sulfidation and regenera-tion.93 A separate report by the same research group foundthat the doped material was capable of decomposing NH3.

94

2.3. Copper-Based Materials. Copper-based sorbents havebeen widely investigated because of the favorable equilibri-um between copper oxides and H2S. Copper oxide is able toachieve low levels of H2S in the clean fuel gas provided thesorbent is not reduced to elemental copper. This is becausecopper oxide readily reduces in high temperature reducing

(82) Woods, M. C.; Gangwal, S. K.; Jothimurugesan, K.; Harrison,D. P.Ind. Eng. Chem. Res. 1990,29 (7), 11601167.

(83) Jothimurugesan, K.; Harrison, D. P.Ind. Eng. Chem. Res. 1990,29(7), 11671172.

(84) Yang, J.;Swisher,J. H. Mater. Charact. 1996, 37(2-3),153159.(85) Siriwardane, R. V.; Poston, J. A. Appl. Surf. Sci. 1990, 45 (2),

131139.(86) Hatori, M.; Sasaoka, E.; Uddin, M. A. Ind. Eng. Chem. Res.

2001,40 (8), 18841890.

(87) Sasaoka, E.; Sada, N.; Manabe, A.; Uddin, M. A.; Sakata, Y.Ind. Eng. Chem. Res. 1999,38 (3), 958963.

(88) Siriwardane,R. V.;Poston, J. A.;Evans, G. Ind. Eng. Chem. Res.1994,33 (11), 28102818.

(89) Liu, X. C.; Gao, F.; Zhao, L. L.; Tian, C. S. J. Alloys Compd.2007,436 (1-2), 285289.

(90) Liu, X. C.; Gao, F.; Zhao, L. L.; Tian, C. S. J. Mater. Sci. -Mater. Electron.2007,18 (8), 863868.

(91) Liu, X. C.; Gao, F.; Zhao, L. L.; Zhao, M.; Tian, C. S. J.Electroceram.2007,18 (1-2), 103109.

(92) Jothimurugesan, K.; Gangwal, S. K. Ind. Eng. Chem. Res. 1998,37(5), 19291933.

(93) Jun,H. K.; Lee,T. J.; Ryu,S. O.; Kim,J. C. Ind. Eng. Chem. Res.2001,40 (16), 35473556.

(94) Jun, H. K.; Jung, S. Y.; Lee, T. J.; Ryu, C. K.; Kim, J. C. Catal.Today2003,87, 310.

8/12/2019 ef900714q

9/17

5299


atmospheres, and elemental copper is an order of magnitudeless active for sulfidation than Cu2O and CuO. Stabilizingcopper oxide with a variety of metal oxides has been quitesuccessful.

Tamhankar et al. studied several pure and mixed oxidesorbents (ZnO, CuO, ZnO-Fe2O3, CuO-Fe2O3, CuO-Al2O3, and CuO-Fe2O3-Al2O3) prepared as porous solidswith very high pore volume.46 Both Fe2O3and Al2O3play arole in stabilizing Cu in the oxidized state in the temperature

range 538-600 C. However, even in the mixed oxides,copper is reduced stepwise from oxidation state 2 to1 to 0, particularly at and above 650 C. The sorbentCuO-Fe2O3-Al2O3performs better than the binary mixedoxides.

Patrick et al. found sulfidation of CuO-Al2O3 sorbentproduced digenite (Cu9xS5) and that alumina stabilizesCuO against complete reduction to Cu at 550-800 C,contributing to sulfidation of copper at the more desirableoxidation states of1 or2 (Table 3).95 They found sulfateformation during regeneration step, which they attributed toeither copper sulfate or surface aluminum sulfate. Sick andSchwerdtfeger studied desulfurization with copper particlesembedded in alumina pellets and were able to lower H2S

levels from 3000 to 300 and 150 ppmv at 750 and 650 C,respectively.96 Yoo et al. found that in an oxidizing environ-ment, sulfation of-Al2O3 support to form Al2(SO4)3 oc-curred,97 and the extent of aluminum sulfation (surfaceor bulk sulfate formation) depends on temperature andcopper loadings.Their results confirm thefindingsof Patricket al.,95 and both copper and aluminum sulfation need to betaken into account in regeneration of copper-on-aluminasorbents.

Two other metals that have been investigated exten-sively for stabilization of copper are chromium and man-ganese. Li and Flytzani-Stephanopoulos studied thebinary Cu-Cr-O and Cu-Ce-O oxides (Table 3) fordesulfurization, including their kinetics of reduction, sul-

fidation, and regeneration under a variety of conditions at650-850 C.

98 They found that in the CuO-Cr2O3 mix-ture, the stable compound copper chromite (CuCr2O4)wasformed, which has the lowest reducibility of all copperoxide-containing compounds in the literature. Thus,this compound, which contains copper in the 1 or 2oxidation states, has very high H2S removal efficiency,resulting in H2S levels of less than 5 ppmv prior to break-through.

Abbasian and Slimane researched copper chromiummixed oxide and copper supported on alumina(Table 3).99 They used thermodynamic analysis to ratio-nalize the selection of chromia (Cr2O3), for the stabiliza-tion of copper oxide (Cu2O) against complete reduction toelemental copper. The Cu-Cr-O mixed sorbent, mostefficient at 600 C, was able to reduce H2S levels toless than 5 ppmv. Regeneration with a dilute O2-N2mixture at 750 C proved to be an efficient protocol

without sulfate formation or reactivity deterioration over15 cycles.99

Slimane and Abbasian developed formulations of coppersupported on alumina and manganese oxide matrix(MnAl2O4 wasalso detected in theXRD), which was suitablefor use over the temperature range 350-600 C.

100 Thesehighly attrition resistant sorbents were able to remove H2Sconcentrations to

8/12/2019 ef900714q

10/17

5300


Atimtay et al. andGasper-Galvin et al. explored incorpor-ating copper, manganese, and molybdenum into silica-richzeolite to stabilize these metal sorbents (Table 3).106-108

Those formulations with Cu had the highest sulfur capacityand copper was found to be the primary desulfurizationagent with manganese and molybdenum acting as promo-ters.

2.4. Calcium-Based Materials. Calcium-based sorbentssuch as limestone (CaCO3) or dolomite (CaCO3 3 MgCO3)

are relatively low cost and can be used in both reducing andoxidizing conditions. Equilibrium calculations and experi-ments by several research groups have shown that calcium ismost active for H2S removal near 880 C, the temperature atwhich CaCO3 decomposes to CaO(lime) and CO2.

43,109 Thisproperty makes it suitable for systems operating at highertemperatures and thus attractive from a thermal efficiencystandpoint. However, other research suggested that H2Sremoval by limestone was restricted to conditions far awayfrom the calcination equilibrium and that the simultaneouspresence of CO2and H2S would inhibit both calcination andsulfidation reactions.110 More importantly, due to thermo-dynamic constraints, H2S removal efficiencies with calcium-containing materials are only on the order of 90% under

typical gasification conditions, resulting in residual H2Slevels of 100 ppm or greater. This may limit the sorbentsusefulness strictly to bulk H2S removal, requiring an externalbed to further polish the gas if an application requires morestringent sulfur cleanup.

Fenouil and Lynn studied the kinetics of H2S sorption byuncalcined and calcined limestone.109,111,112They found thatfor the uncalcined material, the kinetics are reaction limitedbelow 660 C, and diffusion limited above 660 C. Theconversion of uncalcined CaCO3 to CaS was limited to10% (Table 3).112 For precalcined limestone, the conversionfrom CaO to CaS was 100% in an hour, and the sulfidationrate for calcined material was largely insensitive to tempera-ture (Table 3).109

Yrjas et al. and Zevenhoven et al. studied reactions oflimestone (CaCO3) and dolomite (CaMg(CO3)2) under ele-vated pressure conditions.113-115 They found that calcinedlimestone, half calcined dolomite (CaCO3 MgO), andfullycalcined dolomite (CaO MgO) all have faster sulfidationkinetics and higher percentages of sorbent conversion thanuncalcined limestone (Table 3).114 The unreacted shrinkingcore model with variable effective diffusivity was used tomodel the data.115 The results of Fenouil and Lynn as well asthose of Yrjas et al. and Zevenhoven et al. indicate that

precalcination of limestone and dolomite leads to fastersulfidation kinetics and higher sorbent conversion.

Very few studies considered regenerating spent calcium-based sorbents largely due to the fact that limestone anddolomite are relatively soft and easily broken up,116 leadingto high attrition rates in fluidized applications. In addition,they tend to form a stable sulfate layer during regeneration,leading to a loss in active material. As a result, calcium-basedsorbents are usually considered for once-through applica-

tions, requiring stabilization and disposal of large amountsof solid CaS, although this can be partially offset by conver-sion to salable gypsum (CaSO4 3 2H2O). Some attempts havebeen made to improve the attrition resistance and reactivityof limestone and dolomite by forming agglomerates of thepulverized material with various binders.117

Akiti and co-workers reported promising results usingpelletized calcium sulfate hemihydrate (plaster of Paris)coated onto a porous shell of powdered alumina and lime-stone.116 When heat treated, the resulting core-in-shellmaterials were reported to be mechanically strong, activefor H2S removal, andregenerable for10 cycles with no loss inactivity (Table 3). However, several oxidation/reductioncycles were required to regenerate these sorbents.

Abad et al. investigated desulfurization of coal gas usinglimestone and dolomite in a moving bed reactor.118 Theyfound both countercurrent and concurrent configurations tohave high desulfurization levels but the countercurrent con-figuration was more effective (Table 3).

Bjorkman and Sjostrom found that dolomite can alsocatalyze ammonia decomposition but the catalytic activitywas inhibited in the presence of hydrocarbons and steam.119

They hypothesized that the hydrocarbons formed carbonac-eous materials on dolomite that likely inhibited ammoniadecomposition.

2.5. Manganese-Based Materials. In a reducing gas envir-onment, manganese oxides of higher oxidation states arelikely reduced to MnO.120 The thermodynamics of MnO

sulfidation is not as favorable as some other metal oxidessuch as zinc oxide and copper oxide. However, MnO doesnot decompose to elemental manganese readily in reducinggas environment. Thus, manganese-based materials offer theadvantage of stability at high temperatures, which bettermatches biomass gasification and tar reforming processtemperatures. The potential disadvantage is that manga-nese-based sorbents are prone to sulfate formation and haveto be regenerated at very high temperature.101,121

Ben-Slimane and Hepworth studied manganese-basedsorbents for desulfurization and modeled the kinetics ofthe sulfidation, obtained through TGA studies, using theshrinking core model.120 In this model the reaction proceedsat a narrow front which moves into the solid particle. Threedifferent diffusion controlled rate models, gas film diffusioncontrol, product layer diffusion control, and surface reactioncontrol, were used to fit the data. Product layer (MnS)diffusion control was found to best fit the experimental data.

(106) Atimtay, A. T. Development of supported sorbents for hydro-gen sulfide removal from fuel gas. In Desulfurization of Hot Coal Gas,Atimtay, A. T.; Harrison, D. P., Eds.; Springer: Berlin, 1998; Vol. 42, pp

315-

329.(107) Atimtay, A. T.; Gasper-Galvin, L. D.; Poston, J. A. Environ.Sci. Technol.1993,27(7), 12951303.

(108) Gasper-Galvin, L. D.; Atimtay, A. T.; Gupta, R. P. Ind. Eng.Chem. Res.1998,37, 41574166.

(109) Fenouil,L. A.;Lynn,S. Ind. Eng. Chem. Res. 1995, 34 (7),23342342.

(110) de Diego, L. F.; Abad, A.; Garcia-Labiano, F.; Adanez, J.;Gayan, P.Ind. Eng. Chem. Res. 2004,43 (13), 32613269.



(113) Yrjas, K. P.; Zevenhoven, C. A. P.; Hupa, M. M. Ind. Eng.Chem. Res.1996,35 (1), 176183.

(114) Yrjas, P.; Iisa, K.; Hupa, M. Fuel1996,75 (1), 8995.(115) Zevenhoven, C. A. P.; Yrjas, K. P.; Hupa, M. M. Ind. Eng.

Chem. Res.1996,35 (3), 943949.

(116) Akiti, T. T.; Constant, K. P.; Doraiswamy, L. K.; Wheelock, T.D.Ind. Eng. Chem. Res. 2002,41 (3), 587597.

(117) Voss,K. E. Limestone-Based Sorbent Agglomerates for Removalof Sulfur Compounds in Hot Gases and Method of Making. U.S. Patent4,316,813, Feb 23,1982.

(118) Abad, A.; Adanez, J.; Garcia-Labiano, F.; de Diego, L. F.;Gayan, P.Energy Fuels2004,18 (5), 15431554.

(119) Bjorkman, E.; Sjostrom, K. Energy Fuels1991,5 (5), 753760.(120) Ben-Slimane, R.; Hepworth, M. T.Energy Fuels1994,8, 1175

1183.

8/12/2019 ef900714q

11/17

5301


The thermodynamic data for the Mn-S-O system haveconsiderable variability. Turddogan as well as Ben-Slimaneand Hepworth evaluated the data to determine which ther-modynamic information should be given more credence.121,122

Ben-Slimane and Hepworth then used the information tohelp designregeneration conditions for a manganese sorbentthey had synthesized. They concluded that oxidative regen-eration is more favorable and that a temperature as high as900 C is necessary to avoid sulfate formation.123 The

sorbent was tested for 25 cycles and showed no degradationin performance.121

Atakul et al. studied the sulfidation and regeneration ofMnO/-Al2O3sorbent.

124,125 Both sulfidation and regenera-tion were conductedat 600C. Steamwas more efficientthanhydrogen for regeneration, and the rate of sulfur removalwas found to be proportional to steam content. Steamconsumption for regeneration increased slowly during theinitial 60-70% of regenerationand rose rapidly afterward toreach complete regeneration, that is, regeneration was sig-nificantly slower after about 60-70% regeneration.

Bakker et al. developed both regenerable monolith- andparticle-based sorbents consisting mostly of crystallineMn3O4.

126 The material is able to remove H2S from gases

with inlet concentrations of 6400 ppm down to 5-50 ppm.After sulfidation, XRD and TEM detected mostly MnA-l2O4, MnS, andsome amorphous phases. After regeneration,there was still considerable MnAl2O4, together with MnOand amorphous phases. The researchers were able to use SO2to directly regenerate elemental sulfur from the sulfidedmaterial and demonstrated more than 100 regenerationcycles with no significant loss of sorption capacity. Accord-ing to the theoretical estimates of the authors, MnAl2O4canalso be used to remove HF and HCl.126

Wakker et al. studied manganese and iron supported on-Al2O3 at 400-800 C (Table 3).

28 It is one of the fewlaboratory-scale studies where the sorbents have gonethrough several hundred sulfidation and regeneration cycles

(100 cycles for iron-, and more than 300 cycles for manga-nese-based sorbents, respectively). The deactivation profileindicates that most of the deactivation occurred in the first10 cycles.

Yasyerli prepared a series of sorbents (equimolar ratios ofvanadium-manganese, zinc-manganese, iron-manganese,and several different ratios of cerium-manganese) andfound that a cerium-manganese mixed oxidehad the highestH2S sorption rate constant. Regeneration of the sulfidedsorbent converted 90% of the sorbed sulfur to elementalsulfur.127

Alonso and Palacios studied Zn- and Cu-doped manga-nese oxide as a high temperature, regenerable sorbent(Table 3).128,129 These sorbents were prepared by calcining

the extrudate formed from mixed powders of pure oxideswith a Mn/dopant ratio of 10:1. For zinc-doped materials,XRD and transmission electron microscopy (TEM) studiesindicated that in the freshly prepared sorbent Zn was in-corporated into the Mn3O4lattice, and the resultant mixedoxide was the dominant phase (77%) of the prepared sor-bent. After sulfidation, most of the sulfide species was MnS(89%), with mixed sulfide (Mn,Zn)S comprising the remain-ing 10% of the material. Copper doping, on the other hand,

mainly impacted the degree of dispersion of the fresh sorbentwith only a small amount of Mn-Cu mixed oxide form-ing.129 After sulfidation the dominant phases detected byXRD were MnS and MnO. Any copper oxide or coppersulfide phase, if present, was not detectable by XRD, poten-tially due to poor crystallinity. After regeneration, only onephase was detected in each of the doped materials, with zincor copper incorporated into the Mn3O4 tetragonal struc-ture.129 In a long-term, 70-cycle test, with regenerationconducted at 800 C in air, MnSO4was detected using FTIRspectroscopy.128 Alonso and Palacios hypothesized that theformation and decomposition of some MnSO4 during re-generation could be advantageous as it may help maintainsorbent pore structure.128

Yoon et al. investigated desulfurization by and regenera-tion of manganese ore (Table 3).130,131 The manganeseore used consisted primarily of-MnO2 (pyrolusite). Theyfound sulfate formation could cause a decline in break-through time in subsequent cycles. Adding 0.5% ammoniagas (NH3) in the regeneration gas stream prevented sulfateformation, and (NH4)2SO4 was detected as one of theproducts.

One recent study investigated the performance of MnO2supported on a monolith as a catalyst for decomposing NH3and found it to be very effective.132 However, it is not clearwhether this material can be used for both NH3 and H2Sremoval because the experiment was not conducted in thepresence of H2S or steam.

2.6. Iron-Based Materials.Iron-based and iron-modifiedmaterials (e.g., zinc ferrite that was discussed in Section2.2.1) have received considerable attention. They are rela-tively low cost and have considerable sulfidation efficiencyeven when reduced.

Tamhankar et al. studied the kinetics of desulfurizationusing an iron oxide sorbent (45 wt. % Fe2O3, 55 wt. %SiO2).

133 In syngas (CO and H2), the iron oxide rapidlyreduced to metallic iron, which then reacted with H2S.X-ray diffraction, Mossbauer spectroscopy, and TGAweight change analysis revealed that the product is FeS1.1,a form of pyrrhotite that has the general formula Fe1-xS.The kinetic studies also showed that hydrogenconcentrationhad no effect on the sulfidation rate over a wide range ofconcentrations, and that the sulfidation reaction was firstorder with respect to H2S concentration. The activationenergy for sulfidation, 3.3 kcal/mol, was low, which theauthors attributed to the adsorption mechanism, with H2Sdecomposition as the rate limiting step. For small and

(121) Ben-Slimane, R.; Hepworth, M. T. Energy Fuels 1995, 9 (2),372378.

(122) Turkdogan, E. T. Ironmaking Steelmaking 1993, 20 (6), 469475.

(123) Ben-Slimane, R.; Hepworth, M. T. Energy Fuels1994,8, 11841191.

(124) Atakul, H.; Wakker, J. P.; Gerritsen, A. W.; van den Berg, P. J.Fuel1996,75 (3), 373378.

(125) Atakul, H.; Wakker, J. P.; Gerritsen, A. W.; Vandenberg, P. J.Fuel1995,74 (2), 187191.

(126) Bakker, W. J. W.; Kapteijn, F.; Moulijn, J. A. Chem. Eng. J.2003,96 (1-3), 223235.

(127) Yasyerli, S.Chem. Eng. Process.2008,47(4), 577584.(128) Alonso, L.; Palacios,J. M. Energy Fuels2002, 16 (6), 15501556.(129) Alonso, L.; Palacios, J. M.Chem. Mater.2002,14(1), 225231.

(130) Yoon, Y. I.; Chun, B. H.; Yun, Y.; Kim, S. H. J. Chem. Eng.Jpn.2004,37(7), 835841.

(131) Yoon, Y. I.; Kim, M. W.; Yoon, Y. S.; Kim, S. H. Chem. Eng.Sci.2003,58 (10), 20792087.

(132) Ismagilov, Z. R.; Shkrabina, R. A.; Yashnik, S. A.; Shikina, N.V.;Andrievskaya, I. P.;Khairulin,S. R.;Ushakov, V.A.; Moulijn,J. A.;Babich, I. V.Catal. Today 2001,69 (1-4), 351356.

(133) Tamhankar, S. S.; Hasatani, M.; Wen, C. Y. Chem. Eng. Sci.1981,36 (7), 11811191.

8/12/2019 ef900714q

12/17

5302


large particles, the reaction rate was controlled by chemicalreaction and pore diffusion, respectively.

Tseng et al. studied the kinetics of regeneration of sulfidediron oxide sorbent, using O2 and SO2 separately, in thetemperature range 600-900 C.

134 At equilibrium, oxygenoxidizes the FeS1.1 that was formed during sulfidation toFe2O3, with the simultaneous liberation of sulfur as SO2(reaction 5). The activation energy for this reaction wasdetermined to be 15.63 kcal/mol.

2FeS1:1 3:7O2 SFe2O3 2:2SO2 5

With SO2, the oxidation of FeS1.1 proceeds to magnetite,with the formation of elemental sulfur; this reaction can bewritten as

3FeS1:1 2SO2 SFe3O4 2:65S2 6

The activation energy for this reaction was determined to be17.5 kcal/mol.

In a separate publication, Tamhankar et al. studied theoxidation of FeS with steam and steam-air mixture usingTGA and Mossbauer spectroscopy.135 They determined thatoxygen and steam oxidized FeS in parallel to Fe2O3 and Fe3O4,respectively. Further oxidation of Fe3O4 by O2 also occurred in

the reactor, but for as yet unknown reasons it is likely to havenegligible impact since Fe3O4is detected even in samples thathave almost completely reacted. They also deduced that ele-mental sulfur is formed from H2S decomposition.

White, et al. further studied the regeneration of FeS insteam-O2mixtures with the H2O/O2ratio varied from 7.7 to200 at 550-700 C.

136 Building on the reactions proposed byTamhankar et al.,135 they interpreted their experimentalresults on the basis of four simultaneous reactions: (1)oxygen reacts rapidly with FeS to produce SO2and Fe2O3,(2)steamreacts relatively slowlywith FeS to form Fe3O4 andH2S, (3) further oxidation of Fe3O4 formed in the secondreaction by O2, and (4) SO2and H2S react to form elementalsulfur and water. The route for elemental sulfur formation is

the main difference between Tamhankar et al. and Whiteet al. in their interpretation of reactions that occur betweeniron sulfide and steam-air mixture. On the basis of their data,White et al. concluded that to achieve high yields of ele-mental sulfur, the experimental conditions need to be ad-justed to have a large fraction of the FeS convert to H2S (therelatively slow second reaction) and that both SO2and H2Sbe formed at positions within the bed with sufficient time toreact. Among the experimental parameters investigated bythe authors, the maximum yield of elemental sulfur wasapproximately 75% of the theoretical, which was achievedusing a H2O/O2ratio of 200 and a temperature of 600 C.

2.7. Rare Earth-Based Materials. Rare earth-based sor-bents show very good promise. There are several advantagesof rare earth-based sorbents including their use at hightemperature (>700 C), their favorable sulfidation equilib-rium in reducing gas environments, and their potential to beregenerated via a route that directly produce elementalsulfur, which reduces the complexity of the regeneration step(Section 3.3).137

The thermodynamic, oxidation-reduction, and electronicproperties of cerium oxides, ceria-zirconia mixed oxides, andlanthanum oxides have been widely investigated.138-143

These studies yield detailed information on the reductionof CeO2 to CeO2-x, which further reduces to Ce2O3. Thisthermodynamic information laid a foundation for using rareearth oxides as sulfur sorbents. In addition, sulfidationthermodynamics of ceria and reduced ceria, as illustratedin (reaction 7 and 8) have also been well studied.144,145

2CeO2s H2Sg H2g SCe2O2Ss 2H2Og 7

Ce2O3s H2Sg SCe2O2Ss H2Og 8

In contrast to copper, where the reduced form is not veryefficient in H2S removal, the reduced forms of ceria andlanthanum are more active. This makes rare earth-basedmaterials suitable for desulfurization in highly reducingsyngas.

Wheelock et al. used rare earth metals, with particularfocus on lanthanum-containing sorbents, to remove H2SandCOS.146 Kay et al. recognized that thethermodynamic of thereactions between rare earth oxides and sulfur is particularlyfavorable in a reducing environment. They developed meth-

ods of using rare earth oxides, in particular cerium oxide, todesulfurize gases as well as molten iron and steel that havehigh sulfur content.144,147-150 Use of the sorbents and meth-ods for regenerating the sorbents and for SOx and NOxremoval were also developed.

Li and Flytzani-Stephanopoulos studied binaryCu-Cr-O and Cu-Ce-O oxides (Table 3) for desulfuriza-tion.98 For the CuO-CeO2 system, they concluded thatcopper remains in a dispersed state in the sorbent and thatthe reduced ceria explains the desulfurization efficiency.Kobayashi and Flytzani-Stephanopoulous studied the sulfi-dation kinetics of cerium oxide and Cu-modified ceriumoxide.151 By obtaining experimental data of the initial re-duction rate of CeO2 and Cu-CeOx (where Cu content

ranged from 5 to 40 atom %) as a function of temperature,the authors clearly demonstrate that it is easier to reduce the

(134) Tseng, S. C.; Tamhankar, S. S.; Wen, C. Y. Chem. Eng. Sci.1981,36 (8), 12871294.

(135) Tamhankar, S. S.; Garimella, S.; Wen, C. Y. Chem. Eng. Sci.1985,40 (6), 10191025.

(136) White, J. D.; Groves, F. R., Jr; Harrison, D. P. Catal. Today1998,40 (1), 4757.

(137) Zeng, Y.; Zhang, S.; Groves, F. R.; Harrison, D. P. Chem. Eng.Sci.1999,54 (15-16), 30073017.

(138) Zhou, G.; Shah, P. R.; Montini, T.; Fornasiero, P.; Gorte, R. J.Surf. Sci.2007,601 (12), 25122519.

(139) Zinkevich, M.; Djurovic, D.; Aldinger, F. Solid State Ionics2006,177, 9891001.

(140) Kim,T.; Vohs, J.M.; Gorte,R. J. Ind. Eng. Chem. Res. 2006, 45(16), 55615565.

(141) Xiao, W.; Guo, Q.; Wang, E. G. Chem. Phys. Lett. 2003,368(5-6), 527531.

(142) Huang, S.; Li, L.; Van der Biest, O.; Vleugels, J. Solid State Sci.2005,7(5), 539544.

(143) Andersson, D. A.;Simak, S. I.;Johansson, B.;Abrikosov, I. A.;Skorodumova, N. V. Phys. Rev., B, Condens, MatterMater. Phys. 2007,75,(3), 35109-35101 to 35109-35106.

(144) Kay, D. A. R.; Wilson, W. G.; Jalan, V. J. Alloys Compd.1993,193(1-2), 1116.(145) Ferrizz, R. M.;Gorte, R. J.;Vohs,J. M. Appl. Catal., B 2003, 43

(3), 273280.(146) Wheelock, K. S.; Aldridge, C. L.Sulfide Removal Process. U.S.

Patent 3,974,256, Aug. 10, 1976.(147) Kay, A. R.;Wilson, W. G.Methods of Desulfurizing Gases. U.S.

Patent 4,604,268, Aug. 5,1986.(148) Kay, D. A. R.; Wilson, W. G. Methods of Desulphurizing Iron

and Steel and Gases, Such As Stack Gases and the Like . U.S. Patent4,084,960, Apr 18,1978.

(149) Kay, D. A. R.; Wilson, W. G. J. Metals1988,40 (11), 57.(150) Kay, D. A. R.; Wilson, W. G. Method for the Regeneration of

Sulfided Cerium Oxide Back to a Form That Is Again Capable ofRemoving Sulfur from Fluid Materials. U.S. Patent 4,857,280, Aug.15,1989.

(151) Kobayashi, M.; Flytzani-Stephanopoulos, M.Ind. Eng. Chem.Res.2002,41 (13), 31153123.

8/12/2019 ef900714q

13/17

5303


composite Cu-CeOxstructure than the nonmodified ceria.In addition, the authors also determined that the sulfidationrate of all Cu-containing ceria samples is much higher thanthat of pure ceria. This canbe explainedby thehigher affinityof reduced Ce2O3 for sulfur and the increased number ofreduced Ce2O3sites in copper doped ceria.

Wang and Flytzani-Stephanopoulos studied the activityand stability of lanthanum- and copper-containing ceriumoxide sorbents (Table 3).152 Sorbents were tested using

simulated hot reformate gas at 650 and 800 C and regener-ated at the same temperature (details of gas composition arein Table 3). Lanthanum doping (up to 50 atom %) waseffective against crystal growth and sintering, thus moderat-ing the surface area loss of ceria in H2S-free reformate gas at800 C.

152 Onthe basis of XRD, copper oxide exists in highlydispersed form in ceria and is not effective in moderatingsintering. However, copper-doped ceria had the best sulfida-tion kinetics among the ceria-based sorbents tested.152 Insamples containing lanthanum, La2O2S was detected aftersulfidation. However, in samples containing ceria, Ce2O2Swas not detected via XRD, which the authors interpreted asdue to the oxidation of Ce2O2S in air to Ce2O2.5S.

152,153 Intheir study, Wang and Flytzani-Stephanopoulos determined

that, by using very high space velocities, sulfidation could belimited to the sorbent surface only.

Using presulfided CeO2, La2O3, and lanthanum-substi-tuted ceria, Flytzani-Stephanopoulos et al. were able reduceH2S concentrations to sub-ppm levels and to limit surfacearea loss to just 4% after 15 regeneration cycles, indicatingthat fast sulfidation-regeneration may reduce surface arealoss and prolong the overall sorbent lifetime.42 The authorssuggested that by using two reactors that each rapidlyalternates its function between a sorber and a regenerator,and conducting the sulfidation and regeneration both underhigh space velocities, the time required for sulfidation andregeneration would be similar and the reactor design couldbe compact.152

Zeng et al. investigated the effect of H2O and H2/H2O onceria sorbent efficiency in thetemperature range 700-850 Cand interpreted the results based on equilibrium O2pressureexerted by the product gas in the gas composition.154 Nocarbon monoxide or carbon dioxide was used in the experi-mental gas. They also investigated the use of SO2to regen-erate sulfided ceria as described in Section 3.3.137

Other studies found that addition of ZrO2 increases theoxygen mobility within the CeO2 crystal lattice

155 and theresulting CeO2-ZrO2 solid solution has 3-5 times thereversibly stored oxygen than pure CeO2.

156 Assuming thatthese properties are related to enhanced performance as asulfur sorbent, Yi et al. studied the desulfurization propertiesof CeO2-ZrO2.

157 They found that ZrO2 doping increasedthe reducibility of ceria, increased the specific surface area ofthe sorbent produced, and improved its sintering resistance.

Addition of small amounts of CO2 in the feed gas signi-ficantly reduced the sorbent capacity, resulting in muchshorter prebreakthrough time.

2.8. The Effect of Hydrocarbon, Tar, and Steam on Sulfur

Sorption. Most sorbent tests using simulated coal-derivedsyngas were conducted with relatively dry syngas and in theabsence of tar and other contaminants. Although the effectof some of these contaminants on sorbent performance orstability can be modeled using thermodynamics (e.g., effect

of steam), other effects (e.g., soot formation) cannot bemodeled easily. Experimental verification of the effect ofthese other syngas components would be tremendouslyimportant for efficient pilot and demonstration planttesting.

2.8.1. Steam.Thermodynamics predicts that water vaporhas a negative effect on the equilibrium between H2S andmetal oxide sorbents. Novochinskii et al.158 hypothesizedthat in addition to shifting the thermodynamic equilibriumshown in reaction 1, the competitive adsorption of H2O onsurfaces also plays a role in water vapors effect on desulfur-ization.

Actual experimental results indicate varying levels ofwater vapor impact. This may be because reported data wereobtained in different conditions with different factors con-

trolling the reaction rates. In general, the effect of steam tosulfur sorbent performance is expected to be more severe athigher temperatures, although there are a lot fewer studies ofthe effect of steam at higher temperatures. According tothermodynamics, zinc-based sorbents should perform effec-tively even with high steam content below 500 C. Experi-mental results generally confirm that. For example, Guptaetal. found that a zinc titanate sorbent efficiently removed H2Sfrom simulated syngas containing 50% steam (other gascomponents comprising CO2, CO, H2, no hydrocarbons) at493 C.

159 Sanchez et al. used Z-Sorb III, a sorbent contain-ing less than 50% ZnO and 10% nickel oxide, and found nodetrimental effects on sorbent performance from H2, CO, or10-15% H2O.

160

Kim at al. conducted a study of the effect of steam (0-45vol %, in nitrogen) on H2S removal and found that thepresence of 45% steam reduced the H2S breakthrough timeby almost half (from approximately 460 to 280 min, at363 C).

161 Before breakthrough though, the ZnO sorbentwas able to reduce an inlet H2S concentration of 2000 ppmvto 1 ppmv even in 45% steam.161

Jun et al. compared the performance of Zn-Ti-basedsorbents doped by cobalt and nickel oxide as a function ofwater vaporcontent (5-20%byvolumeinCO,CO2, H2,andN2). They found that at 480 C the reaction rate of zinctitanate with H2S is decreased in water vapor whereas thoseof the doped materials are not affected.162

2.8.2. Gas Components Such As CO and CO2

.In the highlyreducing environment of coal- and biomass-derived syngas,H2S should be the dominant sulfur species. Nevertheless,here are a number of studies that indicate COS can be a non-negligible species. Yang et al. studied the effect of individual(152) Wang, Z.; Flytzani-Stephanopoulos, M. Energy Fuels2005,19

(5), 20892097.(153) Sourisseau, C.; Cavagnat,R.; Mauricot,R.; Boucher, F.; Evain,

M.J. Raman Spectrosc. 1997,28 (12), 965971.(154) Zeng, Y.;Kaytakoglu, S.;Harrison, D. P. Chem. Eng. Sci. 2000,

55(21), 48934900.(155) Colon, G.; Pijolat, M.; Valdivieso, F.; Vidal, H.; Kaspar, J.;

Finocchio, E.; Daturi, M.; Binet, C.; Lavalley, J. C.; Baker, R. T.;Bernal, S.J. Chem. Soc., Faraday Trans.1998,94 (24), 37173726.

(156) Hori, C. E.; Permana, H.; Ng, K. Y. S.; Brenner, A.; More, K.;Rahmoeller, K. M.; Belton, D.Appl. Catal., B 1998,16 (2), 105117.

(157) Yi, K. B.; Podlaha, E. J.; Harrison, D. P. Ind. Eng. Chem. Res.2005,44 (18), 70867091.

(158) Novochinskii, I. I.; Song, C.; Ma, X.; Liu, X.; Shore, L.;Lampert, J.; Farrauto, R. J.Energy Fuels2004,18, 576583.

(159) Gupta, R. P.; Turk, B. S.; Portzer, J. W.; Cicero, D. C. Environ.Prog.2001,20 (3), 187195.

(160) Sanchez, J. M.;Ruiz,E.; Otero,J. Ind. Eng. Chem. Res. 2005, 44(2), 241249.

(161) Kim, K.; Jeon, S. K.; Vo, C.; Park, C. S.; Norbeck, J. M. Ind.Eng. Chem. Res.2007,46, 58485854.

(162) Jun, H. K.; Jung, S. Y.; Lee, T. J.; Kim, J. C. Korean J. Chem.Eng.2004,21 (2), 425429.

8/12/2019 ef900714q

14/17

8/12/2019 ef900714q

15/17

5305


pressure PO2) and the inherent thermodynamics of the sorbentmaterial, SO2, H2S, elemental sulfur (S2), or a combination ofthese gases is also produced. Another important aspect toconsider is the regeneration scheme, which defines the numberof reactors or vessels needed, and whether any byproducts ofeconomic value are generated.

3.1. Regeneration with Production of H2S. One potentialregeneration scheme is the reverse of reaction 1, that is,

Mx

Sy

s

yH2

O

gSM

xO

ys

yH2

S

g

11

From a process engineering perspective, assuming station-ary packed beds are used, a dual bed reactor system isrequired for this regeneration process. One of the reactorscontains active sorbent used during the sulfidation step,while the other contains spent material being regenerated(Figure 2). Regenerated process gas would contain H2S, aprecursor to SO2and a regulated emission. In addition, H2Sis itself an odor nuisance, with an odor threshold of 0.5-2 ppb by volume.167 Depending on process conditions, theoutlet gas may exceed the emission limit for H2S and/or SO2,requiring further scrubbing before it is discharged to theatmosphere. Another potential approach that could be usedin a biomass-derived syngas application, if a sulfided mo-lybdenum catalyst was chosen for alcohol synthesis, wouldbeto sendsome oftheH2S to themixed alcohol fuel synthesisunit as the sulfided molybdenum catalyst requires up to100 ppm H2S to maintain catalyst activity.

6 However, thismight entail sulfur removal from the synthesized fuel.

Even though it is normally assumed that the role of H2O isto regenerate the metal oxide with release of H2S, severalstudies have shown that H2O can also cause SO2 release.Sasaoka et al. studied the role of H2O in the oxidation ofsulfided ZnO, Fe2O3 and CuO sorbents using TPR andisotopic H2

18O.168,169 They concluded that H2O directlycontributes to theoxidation of ZnS, FeS, andCu2S, resultingin the release of SO2.

3.2. Regeneration with Production of SO2

. This regenera-tion scheme works for several different oxides. A zinc oxide-zinc sulfide system will be used as an example in the discus-sion of this regeneration scheme. The regeneration of thesulfided material, ZnS, in an atmosphere containing O2proceeds as

2ZnSs 3O2g S2ZnOs 2SO2g H600C -779kJ=mol 12

Because this reaction is highly exothermic, a low O2content of the regeneration gas must be maintained tocontrol the regeneration temperature and to avoid sinteringand the corresponding loss of sorbent effectiveness.137 Theother reason for controlling O2 content is to avoid formation

of sulfate, an undesirable side reaction. Again, a zinc basedsystem is used to illustrate the reaction:

ZnSs 2O2g SZnSO4s 13

2ZnOs 2SO2g O2g S2ZnSO4s 14

The formation of sulfate leads to potential physical da-mage to the sorbents as there is a large molar volume

difference between ZnO or ZnS (15 and 24 cm3 3 mol-1

respectively) and ZnSO4 (46 cm3

3 mol-1).170 The presence

of some oxygen during regeneration promotes formation ofSO2, but an oxygen concentration that is too high formsundesirable ZnSO4, with the suitable level of oxygen beingpressure dependent. To prevent formation of ZnSO4, a suf-ficiently high temperature and adjustments to pressure andoxygen concentration in the process gas are necessary. (See

Elseviers and Verelst

45

for modeling results, and Guptaet al.159 for regeneration conditions used in their experiments).Thedisadvantage of a regenerationschemethat produces SO2

(g) isthat SO2 is a regulatedpollutant. Thepresenceof SO2 in theatmosphere can (1) cause health problems, (2) reduce visibility,and (3) form acid rain.171,172 As in the case of a regenerationscheme that produces H2S, it would be necessary to determinewhether theoutlet gas meetsthe required emission standardsandwhether further processing of the outlet gas is necessary.

The conventional Claus process171,173 can be used torecover elemental sulfur from the SO2gas that is produced.Another method to recover pure sulfur is the direct sulfurrecovery process (DSRP) patented by a research group atRTI International.174,175 This process uses a slipstream of

Figure 3. Simplified schematic diagram of regeneration with pro-duction of SO2followed by DSRP to produce elemental sulfur.

Figure 4. Regeneration with direct production of elemental sulfur(adapted from Figure 10 in Zeng et al.137).

(167) Gabriel, D.; Deshusses, M. A. Proc. Nat. Acad. Sci. U.S.A.2003,100 (11), 63086312.

(168) Sasaoka, E.; Hatori, M.; Yoshimura, H.; Su, C.; Uddin, M. A.Ind. Eng. Chem. Res. 2001,40 (11), 25122517.

(169) Sasaoka, E.; Hatori, M.; Sada, N.; Uddin, A. Ind. Eng. Chem.Res.2000,39 (10), 38443848.

(170) Armstrong, T. R.; Carneim, R. D.; Berry, D. A. InA Review ofCurrent State-of-the-Art Materials for Hot Gas Desulfurization, Nine-teenth Annual International Pittsburgh Coal Conference, Pittsburgh, PA,September 24-26, 2002; PittsburghCoal Conference, 2002; pp 1238-1254.

(171) Stirling, D. The Sulfur Problem: Cleaning up Industrial Feed-stocks; Royal Society of Chemistry: Cambridge, UK, 2000.

(172) Environmental Protection Agency. Sulfur Dioxide;Health andEnvironmental Impacts of SO2. http://www.epa.gov/air/urbanair/so2/hlth1.html(accessed August 27, 2009).

(173) Kohl, A. L.; Nielsen, R. B. Gas Purification, 5th ed.; GulfPublishing Company: Houston, TX, 1997.

(174) Gangwal, S. K.; McMichael, W. J.; Dorchak, T. P. Environ.Prog.1991,10 (3), 186191.

(175) Dorchak, T. P.; Gangwal, S. K.; Harkins, S. M. Method forproducing elemental sulfur from sulfur-containing gases. US 5,798,088,Aug 25, 1998.
http://pubs.acs.org/action/showImage?doi=10.1021/ef900714q&iName=master.img-003.jpg&w=231&h=85http://pubs.acs.org/action/showImage?doi=10.1021/ef900714q&iName=master.img-002.jpg&w=218&h=144

8/12/2019 ef900714q

16/17

5306


coal-derived syngas containing CO and H2 to produce ele-mental sulfur from the SO2 released during regeneration(Figure 3). A skid-mounted, 6.0 in. diameter reactor wasdesigned and fabricated for conducting pilot tests of thisprocess.176 According to RTI International, DSRP is moreeconomical than the Claus process.

3.3. Regeneration with Direct Production of Elemental

Sulfur. Regeneration using O2 orSO2 with the direct produc-tion of elemental sulfur was reported in a number of studies

of sulfided iron sorbents (Figure 4).134,136,177

More recently,this approach was studied for manganese- and ceria-basedsorbents. The economic advantage of directly producingelemental sulfur is that pure sulfur can be condensed andsold. The challenging issue for this regeneration scheme is toeliminate elemental sulfur condensation in other parts of thereaction system.

Zeng at al.137 conducted research with ceria and modifiedceria and their ceria regeneration scheme consisted of:

2Ce2O2Ss 2O2g S4CeO2s S2g 15

Ce2O2Ss SO2g S2CeO2s S2g 16

Bakker et al. regenerated MnS with an SO2/O2 mixtureand with SO2alone to produce elemental sulfur (Table 3).

126

Regeneration of this sorbent with SO2 needs to be conductedat over 600 C to avoid sulfate formation. At 850 C, thekinetics of regeneration using SO2 is 2 times faster thanregeneration with steam. During regeneration, a mixture ofMnO and MnAl2O4 was formed, and greater than 90% ofthe adsorbed sulfur was recovered as elemental sulfur.

4MnSs 2SO2g S4MnOs 3S2g 17

4MnSs 4Al2O3s 2SO2g S4MnAl2O4s 3S2g

18

3.4. Factors that Affect Regenerability. The factors thataffect whether a material can be regenerated over manycycles are not well understood. Several research groupsbelieve that reactivity is directly related to available sorbentsurface area or pore volume and that sintering during hightemperature regeneration causes loss of surface area andsubsequent loss of reactivity. However, Jun et al. noted thattheir regenerated zinc titanate had a higher surface area thanfresh zinc titanate though the material steadily lost perfor-mance over ten cycles.93 Ryu et al. conducted a 100-cycle testof a ZnO-TiO2 based sorbent modified with cobalt oxide,nickel oxide, molybdenum oxide, and iron hydroxide.178 Thetotal pore area of the sorbent increased, while the medianpore diameter decreased after 100 cycles. These results

demonstrate that higher surface area does not always corre-spond to improvedperformance. The authors of both studieshypothesized that repeated expansion and contraction dur-ing sulfidation and regeneration caused spalling or crackingof the sorbent.93,178 The volume expansion and contractionproblem, specifically when sulfate species formed, was also

discussed by Armstrong et al. as a mechanism for failure ofZnO-based sorbent after regeneration.170

5. Conclusions

This review has identified several different sorbent systemsthat can potentially be used to effectively remove H2S fromcoal- or biomass-derived syngas at mid- to high-temperatureranges. For zinc, copper, and iron, reduction to metallicelements will occur at high temperature in a reducing envir-onment. Much of the research in the past two decades hasfocused on stabilization of these materials through formationof mixed oxides. The optimum sorbent depends on processconditions and target sulfur levels. Rare earth-based sorbentsshow significant promise for high temperature applica-tions (700-850 C). Manganese- and zinc titanate-basedsorbents also have potential for applications in the 700 Crange if care is taken to limit sulfate formation. Both manga-nese- and ceria-based materials may be regenerated with thedirect production of elemental sulfur. Copper-based and zincferrite materials are well suited for approximately 500 C.Unmodified zinc oxide-based materials are generally moreapplicable to the low temperature ranges. Kinetic studies in

general indicate product layer diffusion as the most usefulmodel to describe sulfidation kinetics. Some of these materialsadditionally may have the ability to remove NH3 and/or HCl,contaminants also found in biomass-derived syngas.

For high-temperature desulfurization of biomass-derivedsyngas, further research is necessary to determine the perfor-mance of sorbents in a high steam environment containinglight hydrocarbons and tar, which is quite different fromsulfur removal in coal-derived syngas. An understanding ofthe extent of removal of other sulfur species such as COS isalso essential.

There is increasingly more research on binary oxides andpromoted binary oxides to determine whether these mod-ified materials provide better attrition resistance, higher sulfi-

dation equilibrium constants, and the ability to removemultiple gas contaminants. Molecular understanding andmodeling, which has been very effective in aiding catalystdesign, could be a powerful tool for the design of nextgeneration sorbents or combined sorbent/catalyst systems.Spectroscopic and imaging characterization tools could pro-vide some of the mechanistic information that is needed forrational design and iterative testing and modification of suchmaterials.

Key remaining questions include whether multiple con-taminants in syngas can be removed simultaneously andwhether one contaminant might interfere with the removalof another. A needed and ancillary requirement is an accuratedetermination and measurement of the concentration and

speciation of biomass-derived syngas contaminants (chlorine,sulfur, and ammonia) as a function of feedstock (corn stover,switch grass, hard and soft wood, etc.) in varied gasificationenvironments, ideally at the operating temperature of thereactor,so that in situspeciation information canbe obtained.With information on syngas contaminant concentrations andspeciation, the appropriate sorbents and multifunctionalmaterials can be better designed and tested in the labora-tory. At a minimum, the effect of other gas components(e.g., hydrocarbons) and impurities (e.g., chlorine, ammonia)

(176) Portzer, J. W.;Turk,B. S.;Gangwal,S. K. In Durability Testingof the Direct Sulfur Recovery Process, Advanced Coal-Fired PowerSystems Review Meeting, Morgantown, WV, July, 1996.

(177) Patrick,V.; Gavalas, G. R.;Sharma, P. K. Ind. Eng. Chem. Res.1993,32 (3), 519532.

(178) Ryu, S.O.; Park, N.K.; Chang,C. H.; Kim,J. C.; Lee,T. J. Ind.Eng. Chem. Res.2004,43 (6), 14661471.

(179) Twigg, M. V.; Spencer, M. S.Appl. Catal., A 2001,212 (1-2),161174.

8/12/2019 ef900714q

17/17

5307


on the performance of sulfur sorbents needs to be modeledand experimentally verified.

Pilot scale testing of such sorbents will provide results andexperience for eventual commercialization. As more scientificdata on mid- to high-temperature sulfur sorbent becomesavailable, technoeconomic analysis and systems modeling todetermine the potential improvements in thermal efficiency,

environmental benefits, andprocesseconomics throughuse ofsuch sorbent systems will be critical.

Acknowledgment. Funding for this research was provided bythe Office of the Biomass Program, U.S. Department of Energy,under contract DE-AC36-99GO10337 with the National Renew-able Energy Laboratory.

ef900714q

Documents