1890 Ind. Eng. Chem. Res. 1992,31, 1890-1899 Sulfidation of Titanate and Zinc Solids

Susan Lew,?Adel F. Sarofim, and Maria Flytzani-Stephanopoulos* Department of Chemical Engineering, Massachusetts Institute of Technology, Cambridge, Massachusetts 02139

The sulfidation of bulk mixed of zinc and of various compositions and Zn-Ti-0 crystalline phases in H+3-H2-H20-N2 gas mixtures was investigated in a thermogravimetric apparatus over the temperature range of 400-800 "C. Comparative sulfidation experiments with ZnO were also performed. In comparison to ZnO, the use of Zn-Ti-0 solids allows raising the operating temperature for desulfurization of hot coal-derived fuel gas, e.g., by as much as 94 "C for solids with (Zn/Ti)aumic= 2/3. The initial sulfidation rate of Zn-Ti-0 solids with (Zn/Ti),bmic I3 was ap- proximately 1.5-2 times slower than for ZnO. Different zinc titanate phases (i.e., Zn2Ti04,ZnTi03, and Zn2Ti308)had the same initial sulfidation rate. Similar activation energies (9-10 kcal/mol) were measured for ZnO and Zn-Ti-0 sulfidation. No effect of H2 or H20 was observed on the initial sflidation rate. However, at high conversions, rates were lower with increasing H2 concentrations. Fine particles and cracks were observed in sulfided ZnO as a result of ZnO reduction and subsequent vapor-phase reaction between Zn and H2Sto form ZnS. These structural effects were largely absent in Zn-Ti-0 solids.

Introduction ratory (Lew, 1987; Flytzani-Stephanopoulos et al., 1987; The removal of hydrogen sulfide (H2S) to sufficiently Lew et al., 1989) have found that in association low levels from coal-derived fuel gases at elevated tem- with titanium dioxide is reduced more slowly to volatile peratures is crucial for the efficient and economic coal zinc than pure zinc oxide. In these earlier studies, cyclic utilization in emerging advanced power generation systems sflidation-regeneration (six to eight cycles) of various such as the integrated gasification-combined cycle (IGCC) Zn-Ti oxides was performed in a packed bed reactor. A and the gasification-molten carbonate fuel cell (MCFC). simulated coal gas mixture with a molar composition of For these technologies, highly efficient sulfur removal from 1% H2S-13% H2-19% H20-67% N2 was typically used. several thousand parts per million (pprn) down to -1 ppm H2S removal efficiency comparable to that of ZnO was for the MCFC or less than 100 ppm for the IGCC is measured for Zn-Ti4 sorbents. The same was true when needed. Commercial desulfurization processes are based 15% CO-10% COz was substituted for N2 in the sulfida- on liquid scrubbing at or below ambient temperatures, tion gas mixture. Reduction experiments were performed resulting in considerable thermal efficiency loss as well as with Hz, a stronger reducing agent than CO. A reduction costly wastewater treatment. rate lower by as much as 5 times was measured in these Previous studies (Jalan and Wu, 1980; Grindley and experiments for Zn-Ti-0 solids compared to ZnO reduc- Steinfeld, 1981; Flytzani-Stephanopoulw et al., 1985) have tion by hydrogen at 650 "C. More detailed reduction investigated the potential use of zinc oxide as a high-tem- kinetic experiments were performed recently (Lew, 1990; perature regenerable sorbent. Kinetic studies using single Lew et al., 1992a). The overall reaction scheme in sulfi- pellets of zinc oxide were also performed (Gibson and dation-regeneration of Zn-Ti-0 materials is Harrison, 1980; Ranade and Harrison, 1981). The ther- sulfidation modynamic equilibrium for sulfidation of ZnO is quite favorable, yielding desulfurization down to a few parts per ZnxTiyOx+Py(s)+ xH2S(g) - million (ppm) H2S. can be regenerated if xZnS(s) + yTi02(s) + xH20(g) (1) sufficiently high temperatures or low oxygen concentra- In addition to sflidation, some reduction can occur by the tions are used to avoid zinc sulfate formation. A major reaction drawback of zinc oxide is that in the highly reducing at- mosphere of coal-derived fuel gases, it is partially reduced reduction (>600 "C) to elemental zinc, which at high temperatures is volatile. Zn,TiyO,+zy(s) + xH&) or xCO(g) - Consequently, sorbent loss is observed at temperatures xZn(g) + yTi02(s) + xH20(g)or xCOz(g) (2) above 600 "C. More recently mixed metal oxides have been studied in regeneration an effort to improve the properties of single oxide sulfur sorbents (Grindley and Steinfeld, 1983; Flytzani-Stepha- xZnS(s) + yTi02(s)+ (3x/2)02(g) - nopoulos et al., 1985). The mixed oxide sorbent zinc Zn,Ti,0,+2y(s) + xSOz(g) (3) ferrite, ZnFe204,combining ZnO with Fe203 has been The packed bed reactor experiments provide informa- developed as an alternative to single zinc oxide sorbent tion on sulfur removal efficiency and utilization of sorbents. (Grindley and Steinfeld, 1983) because of its high sulfur However, these are not suitable for kinetic measurements. capacity, rapid reaction with H2S, and high H2Sremoval In this work, the sflidation kinetics of Zn-Ti4 materials efficiency. Zinc femte decomposes into (ZnO + Fe304)in were examined in a thermogravimetric apparatus as a the reducing coal gas atmosphere. Hence, it is similarly function of the operating conditions and sorbent compo- limited (as ZnO) to an operating temperature of approx- sition. To determine the sflidation kinetics, a gas mixture imately 600 "C. Previous studies performed in this labo- of H2S and N2 was used. A small amount of H2 (1%) was added to the gas mixture to prevent the decomposition of * To whom correspondence should be addressed. hydrogen sulfide. The effects of H2and H20on the initial Present address: ARC0 Chemical Company, Newtown sulfidation rates were examined separately. In practice, Square, PA 19073. the actual coal gas is a mixture of H2S-H2-H20-CO- 0888-5885/92/2631-l890$03.00/00 1992 American Chemical Society Ind. Eng. Chem. Res., Vol. 31, No. 8, 1992 1891

Table I. Chemical Properties of Sorbents Used in Sulfidation Experiments" (Zn/ Ti) crystalline phasesb (wt %) sorbent (atomic ratio) ZnO ZnsTiOl ZnTiOs ZnzTi3O8 Ti02(rutile) Z3T7' 317 0 0 69 0 31 Z2T3-a 213 0 0 65 16 19 Z2T3-b 213 0 0 83 0 17 ZT 111 0 20 45 35 0 Z3T2 312 0 68 18 14 0 Z2T-a 211 0 100 0 0 0 Z2T-bd 211 0 100 0 0 0 Z3T 311 28 72 0 0 0 ZnO 100 0 0 0 0 "Prepared from zinc acetate and titanium(1V) isopropoxide with 1:l mole ratio of metal ions to citric acid (unless otherwise noted). All solids calcined at 720 "C for 12 h except for ZnO, which was calcined for 4 h. *Identified by X-ray diffraction. Prepared with titanium tetrachloride. Prepared with 1:2 mole ratio of metal ions to citric acid. C02-N2. The complexity of this gas mixture makes it Micult to quantify the influence of different components. In this paper, we will report on the kinetics of sflidation and the structural changes associated with reductive sflidation. Several experiments in regeneration with OpN2were performed to determine the regenerability of the solids. These results (sulfidation and regeneration) were compared with those of ZnO to determine the level of improvement attained by the use of Zn-Ti oxides. Experimental Section Preparation and Characterization of Solids. Bulk mixed oxide solids of zinc and titanium were prepared by a known method for synthesizing highly dispersed mixed oxides from amorphous citrate precursors (Marcilly et al., 20 40 60 80 100 1970; Courty et al., 1973). The preparation of &-Ti oxides consists of mixing a 2:l volume ratio solution of glacial mol% ZnO (based on ZnO-Ti02) acetic acid (Mallinckrodt, grade) and titanium(1V) AR Figure 1. Crystalline phases formed as a function of Zn-Ti-0 isopropoxide (Strem Chemical, AR grade) with an aqueous composition of solids calcined at 720 "C for 12 h. solution of zinc acetate (Mallinckrodt,AR grade) and citric acid monohydrate (Mallinckrodt, AR grade). Typically, on the &/Ti atomic ratio and the calcination temperature. an equal mole ratio of citric acid to metal ions (zinc and The observed phase transformation with increasing titanium) is used in preparation of the solution. The fiial temperature is Zn2Ti308- ZnTiO, - Zn2TiOl. At high solution is first dehydrated rapidly (15-30 min) in a rotary calcination temperature (11000 "C), Zn2Ti04is the only evaporator at 65-75 "C under vacuum to form a viscous stable mixed oxide phase for all Zn-Ti4 solids, coexisting liquid and then dehydrated slowly (4-6 h) in a vacuum with either ZnO or TiOz phases depending on the solid oven at 70-80 "C to form a porous solid foam. The solid stoichiometry (Zn/Ti ratio). Also, Zn2Ti04is the stable foam was calcined in air in a muffle furnace at 720 OC for phase for solids with Zn/Ti 1 2 calcined at temperatures 12 h producing a porous, homogeneous mixed metal oxide. 1700 "C for long periods of time (112 h). At temperatures The solids were characterized by several bulk and sur- below 800 "C, all three zinc titanate phases may be present face analysis techniques. The elemental composition (zinc (Lew, 1990). and titanium) of the solids was verified by atomic ab- Table I shows the stoichiometry and XRD analysis of sorption spectroscopy (Perkin Elmer 360 spectrophotom- the various Zn-Ti4 materials used in sulfidation exper- eter) of the solids dissolved in a hot HF-HC1-H20 solution iments in this work. All sorbents were calcined at 720 "C (-90 OC). X-ray diffraction (XRD) for identification of for 12 h. The type of phases present depended on the crystalline phases in the mixed oxides was performed with Zn/Ti ratio as shown graphically in Figure 1. Decreasing a Rigaku RU300 instrument using Cu (Ka) radiation. the Zn/Ti ratio of the solids produced phases in the order Scanning electron microscopy (SEM) with energy-dis- ZnO - Zn2Ti04- ZnTi03 and Zn2Ti308- Ti02. There persive X-ray analysis (EDS) using a Cambridge Ster- were some variations in the relative amounts of ZnTi03 eoscan 250 MK3 instrument were used to observe the and Zn2Ti308as exemplified by the solids Z2T3-a and surface morphology, crystallite size, and compositional Z2T3-b in Table I, prepared, respectively, in flowing and variation of the solids. Surface areas were measured by static air calcination. a Micromeritics Flow Sorb I11 2300 BET apparatus using Effecta of varying the %/Ti atomic ratio on the physical N2gas, while pore volumes and pore size distribution were properties of the &-Ti4 solids, i.e., surface area and pore measured by a Micromeritics Autopore 9200 mercury po- volume, are shown in Figure 2. Solids with up to 50 mol rosimeter. % Ti02are characterized by higher surface area and pore Physicochemical Properties of Bulk Zn-Ti-0 volume than ZnO neat. Addition of small amounts of Ti02 Sorbents. Three distinct zinc titanate phases, namely into ZnO has the largest effect, with a maximum in surface Zn2Ti04,Zn2Ti308, and ZnTi03, can be formed through area and pore volume shown for (Zn/Ti)ab~c= 9/1. These solids preparation by the citric acid complexation method data indicate that Ti02 disperses ZnO, effectively pre- using zinc acetate and titanium(IV) isopropoxide precur- venting ZnO particle growth (sintering). High levels of sors followed by pyrolysis in air at different time-tem- Ti02, however, and compound formation (e.g., ZnTi03, perature conditions. The type of phases present depends Zn2Ti308)reduce the overall surface area. 1892 Ind. Eng. Chem. Res., Vol. 31, No. 8, 1992

Table 11. Pore Size Distribution of Sorbents Used in Sulfidation Experimentsa

~~ pore size distributionb sorbent (%), diameter (pm) Z3T7' 16.0, > 25 69.5, 25-1 13.3, 1-0.1 1.2, 0.1-0.003 Z2T3-a 58.5, >25 21.8, 25-1 10.8, 1-0.1 8.9, 0.1-0.003 ZT 15.7, >25 67.2, 25-1 7.9, 1-0.1 9.2, 0.1-0.003 0' I 40 60 BO 100 Z3T2 44.2, >25 38.4, 25-1 mol% ZnO (based on ZnO-Ti02) 10.5, 1-0.1 6.9, 0.1-0.003 Z2T-a 56.2, >25 28.1, 25-1 5.8, 1-0.1 9.9, 0.1-0.003 Z2T-bd 10.0, >25 53.4, 25-1 15.3, 1-0.1 21.3, 0.1-0.003 Z3T 19.6, >25 54.6, 25-1 10.0, 14.1 15.8, 0.1-0.003 ZnO 15.0, >25 62.2, 25-1 21.1, 1-0.1 1.7, 0.1-0.003

-4 0 60 80 100 'Prepared from zinc acetate and titanium (IV) isopropoxide mol% ZnO (based on Zn0-1102) with 1:l mole ratio of metal ions to citric acid (unless otherwise noted). All solids calcined at 720 "C for 12 h except for ZnO, Figure 2. Effect of Zn-Ti4 composition on physical properties of which wae calcined for 4 h. bPore size distribution for 90-125-pm the didparticles (90-125 am): (a) surface area and (b) pore volume. particles by mercury porosimetry. Prepared with titanium btra- chloride. Prepared with 1:2 mole ratio of metal ions to citric acid. As a result of the preparation method, the sorbents are produced in a highly macroporous form. Table I1 shows the pore size distribution (based on Hg porosimetry) of t hermalqn]electrobalance Zn-Ti4 particles (90-125-pm size) after calcination in air shield at 720 OC for 12 h. Macropores (>l-pm diameter) typically - comprise more than 70% of the pores. Such pore struc- tures allow for kinetic studies in the absence of pore dif- fusion limitations even with relatively large particles ( N 100 rm). Apparatus and Procedure. Kinetic sulfidation ex- periments with solids containing various Zn/Ti atomic ratios were performed in a Cahn System 113-X thermo- gravimetric analyzer (TGA) equipped with a Cahn 2000 electrobalance, a Micricon temperature controller, and a Bascorn Turner data acquisition system. The TGA reactor system is shown in Figure 3. The TGA measured the weight gain as a function of the time required for Zn-Ti oxide sulfidation to ZnS and TiOz. The solid was pre- treated in a vacuum oven at 90 "C for 1 h to remove any absorbed HzO before it was reacted in the TGA. Gas flow rates were set by passing 6.3% HzS-in-Nz,Hz, Ix] mass flowmeter and N2 gases through Brooks Model 58503 mass flow Figure 3. Schematic of TGA reactor system. controllers. A gas flow rate of 350 cm3(STP)/minwas used in the experiments. Approximately half the gas flow threeneck flask assembly. The saturated gas stream was (containing HzS, Hz, and N,) entered the reactor (TGA) then mixed with the HzS-in-Nzgas and entered the ap- through a side arm. The other portion of the gas con- paratus side arm. A thin layer of sample (typically 1-3 taining only Nzentered the balance section of the TGA mg of 90-125-pm-size particles) was placed in a hemi- serving both to protect the balance from the corrosive HzS spherical-shaped quartz pan suspended by a quartz and as a diluent to the reactant gas. Water vapor was hangdown wire. Isothermal sulfidation experiments were added to the gas by bubbling nitrogen and hydrogen performed at temperatures between 400 and 800 OC. It through a water saturator maintained at 25 OC in a was experimentally verified by varying the gas flow rate, Ind. Eng. Chem. Res., Vol. 31, No. 8, 1992 1893

1 H2S-l%H2-bal.N2

4t ZnO: 700°C

I

Z2T-a: 600°C

I "- 20 40 60 80 100 0 1 2 3 4 5

mol% ZnO (based on ZnO-Ti02) mol% Hydrogen Sulfide Figure 4. Initial sflidation rate of several Zn-Ti-O sorbents in 2% Figure 5. Determination of reaction order for ZnO and Z2T-a re- HzS-l% Hz-97% N2 at 600 and 700 "C. action with H2S.

quantity of sample, and particle size that these experi- Table 111. Arrhenius Constants for the Sulfidation ments were performed in the absence of both mass transfer Reactions of ZnO and Z2T-a Sorbents and pore diffusional limitations. Thus, the measured rate Arrhenius const was due only to the intrinsic sulfidation kinetics and product layer diffusion. Typically, each sflidation ex- . . , I . , I periment was repeated at least once to verify reproduc- ZnO + HzS - ZnS + HzO 400-800 1.31 10.3 ibility. In regeneration, the sulfided solids were reacted 1/7Zn7TiOA+ H,S ZnS + 400-700 0.40 9.3 in 21% 02-79% N2 at 650 "C. Two cycles of sflidation- - regeneration were performed. transfer resistances, the irreversible surface chemical re- Results and Discussion action can be described as Initial Sulfidation Rate of Bulk Zn-Ti-O Sorbents. (4) Initial rate experiments in the TGA were performed under isothermal conditions to determine the sflidation re- where Ro is the initial molar rate of ZnS formation per unit activity of the solid listed in Table I. The results of these surface area of the solid reactant [mmol/(cm2-s)],k is the experiments at 600 and 700 "C in 2% H2S-1% H2-97% intrinsic rate constant, CHp is the molar concentration of N2are shown in Figure 4. The initial rate was expressed hydrogen sulfide [mmol/cm3], and n is the reaction order. in mmol of ZnS formed/ (cm2-s). The rate was normalized The reaction orders (n)for ZnO and Z2T-a sulfidations with the initial surface area of the sorbent. No bulk ti- were both determined to be 1. Figure 5 illustrates the tanium sulfide was formed. The initial rate of sulfidation results obtained for ZnO and Z2T-a sflidation. The lin- for ZnO at 600 and 700 "C was approximately 1.5-2 times earity of the variation in initial sulfidation rate with higher than for all Zn-Ti-0 sorbents containing 125 mol changing hydrogen sulfide concentration indicated a re- % Ti (based on Zn-Ti stoichiometry). action order of 1. This agrees with the kinetic data re- The initial sulfidation rate was similar for different zinc ported by Westmoreland et al. (1977) for zinc oxide. titanate phases despite the fact that sorbents with different huation 4 was used to describe the initial sulfidation rate. %-Ti compositions formed different zinc titanate phases As discussed in the previous section, the sulfidation ex- as shown in Table I. For example, a sorbent with 30 mol periments were peformed in the absence of pore diffusional % ZnO-70 mol % Ti02contained a mixture of ZnTiO, and and mass-transfer resistances by using small particles and Ti02,while a sorbent with 75 mol % ZnO-25 mol % Ti02 high gas flow rates. An Arrhenius relationship can be used was composed of Zn2Ti04and ZnO. However, both of to express the intrinsic rate constant as these sorbents had approximately the same initial sulfi- = exp[-E/RT] (5) dation rates (Figure 4). k ko Since all Zn-Ti-0 sorbents containing 125 mol 9% Ti In eq 5, ko is the Arrhenius frequency factor [cm/s], E is (based on Zn-Ti stoichiometry) had approximately the the activation energy [kcal/mol], R is the gas constant same initial sflidation rate, further kinetic experiments [1.987 X kcal/(mol*K)],and Tis the temperature [K]. were performed with just two different types of Zn-Ti-0 The Arrhenius dependence for ZnO, Z2T-a, and Z2T3-a sorbents. Experiments were performed with Z2T-a, a sflidation reactions was determined by measuring the sorbent prepared with 2:l atomic ratio of Zn:Ti, and initial sulfidation rate as a function of temperature. The Z2T3-a, which was prepared with 2:3 atomic ratio of Zn:Ti. experiments were performed in 2% H2S-1 % H2+7 ?% N2. Zn2TiOl was the only crystalline phase identified by XRD Figure 6 shows the resultant Arrhenius plots for ZnO, in Z2T-a, while Z2T3-a contained a mixture of ZnTiO,, Z2T-a, and Z2T3-a sulfdations. The intrinsic sulfidation Zn2Ti308,and Ti02. For the purpose of comparison and rate for ZnO was greater than Z2T-a and Z2T3-a at all in order to establish a base-line sulfidation performance temperatures between 400 and 800 "C,while the Arrhenius level to meet, similar kinetic experiments were also per- plots for Z2T-a and Z2T3-a virtually overlapped. The formed with ZnO. kinetic constants obtained for ZnO and Z2T-a sflidation On the basis of SEM micrographs of the unreacted are listed in Table 111. The activation energies for both solids, the sorbents are composed of nonporous particles ZnO and Z2T-a sulfidation were approximately the same (or grains). In the absence of both diffusional and mass- (10.3 kcal/mol and 9.3 kcal/mol, respectively). The major 1894 Ind. Eng. Chem. Res., Vol. 31, No. 8,1992

2XH25-1 WH2-97XN2 '" I 0 zno

T=KSO"C 2XH25-1 XH2-97KN2 I

I I 0.0 0.0 1.0 1.1 1.2 1.3 1.4 1.5 0' 20 40 KO 80 100

Iff (K) * I000 lime (mln) Figure 6. Comparative Arrheniw plots for ZnO, m-8,and ZTl3-a Figure 7. Comparative aultidation conversion profiles for mrbents sulfidation reactions. ZnO, Z2T-8, and ZZT3-a. difference was in the frequency fadors. The frequency factor for ZnO sulfdation was approximately 3 times greater than that for Z2T-a. The lower frequency factor for the latter was probably caused by fewer reaction sites on the readant surface due to the presence of titanium on the surface. In contrast to the results obtained in this work, Woods et al. (1990) reported the sulfdation kinetics of Zn-Ti oxides to be independent of temperature in the range of 65Cb760 OC. This discrepancy is due to the fact that Woods et al. measured global sulfidation rates of Zn-Ti oxide pellets (3/16-in. diameter and LID = 2.5) under conditions not free from mass-transfer and pore diffusion limitations. Such measurements with pellets provide ap- parent reaction rates, are pellet-specific, and should not be used to obtain values for intrinsic kinetic parameters such as the activation energy. The initial sulfdation rate reported in the present study for ZnO powders was approximatelytwice as fast as that reported by Westmoreland et al. (1977). This difference in sulfidation rate is believed to be due to differences in crystallinity. Better agreement of the sulfidation rate with Westmoreland et al. (1977) was obtained using a com- mercially purchased ZnO (EM Science, AR grade). Com- parison of SEM micrographs of both ZnO types shawed different crystal geometries present in each sample. The (EM Science) ZnO had a majority of rectangular crystals, while the ZnO prepared by the amorphous citrate tech- nique consisted mainly of spherical crystals. It is known from the literature that exposure of different ZnO crystal faces can be obtained by varying the precursor (Krebs and Littbarski, 1981; Hindermann et al., 1988) during prepa- ration. Variations in exposed crystal faces can cause differences in reactivity in structure-sensitive reaction systems. Profiles of Bulk Zn-Ti4 Conversion Sorbents. Figure 8. SEM micrographs showing (a, top) nonuniform grain sizee The sulfidation reactions of three sorbents, ZnO, Z2T-a, of sorbent Z2T-a and (h, bottom) uniform grains of sorbent Z2T-b. and Z2T3-a, were studied in detail. Figure 7 shows the conversion profiles for these sorbents at 650 'C. Con- in the grain size distribution of the solids. SEM micro- version was defined as follows: graphs of Z2T-a and Z2T3-a showed the presence of both small = (Wi - W)/(Wi - w:, (6) spherical grains and larger plate-like grains. In the x early portion of the conversion profiles for Z2T-a and where W is the weight, Wi is the initial weight, and W: Z2T3-a (Figure 7). the small grains were mainly reacting is the fmal weight at complete conversion assuming that while at higher conversion all the small grains had reacted the sorbent reacts completely to form ZnS and TiOz (for and reaction was due only to the larger plate-like grains. the Zn-Ti4 sorbents). The effect of nonuniform grain size was verified by sul- For Z2T-a and Z2T3-a, the decrease of reaction rate at fidation of the solid Z2T-b, which was prepared by using high conversion is attributed primarily to nonunifomity a 21mole ratio of citric acid to metal ions instead of the Ind. Eng. Chem. Res., Vol. 31, No. 8, 1992 1896

Table IV. XRD Analyses of Sulfided SorbentsD crystdine phases (wt %) sample ZnO ZnlTiO, Zn2Ti,08 ZnTiOa u-ZnS 8-ZnS Ti02(rutile) Z2T-a 48% sulfded 0 23 35 0 16 26 negligible 95% sulfided 0 0 6 0 17 53 24 Z2T3-b 35% suhided 0 0 0 48 15 0 37 ZOO 100% sulfided 0 0 0 0 33 67 0 'Sulfded at 650 OC in 2% H&l% H2-97% N,.

lime (min) Figure 9. Sulfdation eonversion profiles for Z~T.~ ZZT-~, Figure IO. SEM micrograph of unreacted sorbent Z2T3-a. the Zn-Ti4 sorbents were examined. The samples were 1:l. A solid with a more uniform, small grain size usual subided in the TGA at 650 "C with 2% H2S-1% H2-97% was, thu8,pduced. Figure 8 shows the SEM micrographs of the solids Z2T-a and Z2T-b, while Figure 9 shows the N,. To obtain sufficient quantity of samples for accurate (40-60 corresponding conversion profiles obtained at 600 OC. A measurements of the surface area, a large amount uniform reaction rate was measured for sorbent Z2T-b up mg) of sample was used. The solids were sulfided to ap- proximately 90% conversion, and then their surface areas to very high conversions (Figure 9). The formation of a were measured. product layer of ZnS and TiO, around the unreacted solid Typically, all Zn-Ti4 materials showed an increase in core also contributed to the slower reaction rate at higher conversion. surface area upon sulfdation. This in an agreement with A more detailed discussion of the effect of our previous reports with this type of sorbent (e.g., Flyt- product layer diffusion on zinc titanate sulfidation is mi-Stephanopouloa et al., 1987). For Z2T3-a after 90% presented in Lew et al. (1992b). subidation, the surface area increased from 2.3 to 2.8 m2/g. XRD analysea were performed on samples reacted in the By constrast, the surface area of sulfided ZnO was 1.7 TGA at 650 "C in 2% H2S-1% H2-97% Nz. To provide m2/g, lower the value for the fresh sorbent (2.4 m2/g). sufficient amount of samples for analysis, 20-30 mg was than On the basis of SEM micrographs of the sulfded sam- used. This was a much larger quantity than what was used ples, it is apparent that while ZnS tends to sinter, TiOz' for the hetic experiments (1-3 mg). On the basis of XRD inhibits this sintering by acting a physical barrier analyses of partially sulfided sorbents (Table IV), the as to prevent growth of ZnS particles. Thus, TiO, acta as a different zinc titanate compounds are believed to react dispersant of ZnS. SEM photographs of unreacted and according to 9(t95% subided Z2T3-a are shown, respectively, in Figures ZnTi03 + H2S + ZnS + Ti02 (7) 10 and 11. On the basis of EDS elemental analms of the Zn2Ti04+ %H2S y3Zn2Ti308+ y3ZnS + %H20 (8) surface, more sintered regions contained lower levels of - titanium. The sulfided region in Figure lla contains Zn,Ti308+ 2H2S - 2ZnS + 3Ti0, + 2H20 (9) (Zn/Ti)abmic= 0.32 and (S/Zn)ah~c= 1.0 and preserves No si&icant amount of free Ti02was detected for the small grain sizes. In contrast, the more sintered region Z2T-a sample sulfided to 48%. This indicates that Zn,- shown in Figure llb contains (Zn/Ti),,, = 1.8 and Ti308reacted much slower than Zn,TiO,. However, sor- (S/Zn).~<7 1.0. bent ZT which contained 20 wt % Zn2Ti0,, 35 wt % To determme how different Zn-Ti4 compositions af- ZnzTi30s,and 45 wt % ZnTiO, (Table I) did not have an fected the surface area of the sulfided solids, solids with initial sulfidation rate significantly different than sorbent initially similar surface area were chosen for sulfidation. Z2T-b (100 wt % Zn,TiO,). The reason that no free TiO, All solids except for Z2T3-b were sintered (lo00 OC, 1 h) was detected in the partially reacted Z2T-a sorbent must until they reached approximately the same surface area then be attributed to a rapid reaction of any TiO, formed (-0.8 m'/g). ZZT3-b was not subjected to any further heat with Zn2Ti04to produce Zn2Ti30,: treatment beyond its initial calcination. The solids were reacted at 650 OCin 2% H2S-l% H,-97% for 90 min Zn2Ti04 2Ti02 ZnzTi30s N2 + - (10) or until -90% sulfidation was reached. The surface areas Physical Changes of Sorbents during Sulfidation. obtained for the sulfided solids are shown in Figure 12. Surface area changes along with morphological changes of After sulfidation, the surface area of all solids increased. 1896 Ind. Eng. Chem. Res., Vol. 31, No. 8,1992

Sulfldallon: ZXHZS-98XNZ rl R.~YCI~~~:~O%H~.~PHZO.S~%NZ

Z2T-a Reduction 10.'. 2213-a Reduction

1fl ( K) * 1000 FSgura 13. ComparativeArrhenim plots of the initid reduction and suhidation rates for ZnO, ZZT-a, and Z2T3-a.

preserving ZnS dispersion and total surface area in the Zn-Ti-0 sorbents. Effect of HI. Hydrogen reduces Zn-Ti4 and ZnO solids at the temperatures of interest in hot gas deaul- furization (2600 "C). In particular, ZnO is reduced ap- proximately 3-10 times faster than Zn-Ti-0 solids at temperaturea in the range of 6W700 OC (Lew, 1990; Lew et aL, 199%). Small amounts of H20(-1% in the reactant gas stream) were shown (Lew, 1990; Lew et al., 1992a) to inhibit reduction of both Zn-Ti-0 and ZnO solids, while further increasing the HzO content to 8% did not have any additional effect on the reduction rate. From the reduction of partially sulfided Zn-Ti4 and ZnO solids, it appeared i that H,S had an inhibitory effect on reduction similar to Figure 11. SEM micrographs of Z2T3-a sulfided -90% in 2% HzO. Comparative Arrhenius plots of the initial reduction H&1% H247% N2at 650 "C (a, top) region with (Zn/Ti),- = and sulfidation rates are shown in Figure 13 for ZnO, 0.32 and (b, bottom) more sintered region with (Zn/Ti)ammic= 1.8. Z2T-a, and Z2T3-a. The reduction rate is shown for a gas containing 10% H,-3% Hz0-87% NP,while sulfidation is for 2% HzS-9S% Nz. The point at which the initial reduction rate of ZnO became faster than its initial sul- fidation rate was at 848 OC, while for Z2T3-a it was 942 OC. Consequently Zn-Ti-0 solids can be used for the desulfurization of coal-derived fuel gas at higher temper- ature than ZnO. The desulfurization temperature can be 94 "C higher with Z2T3-a than with ZnO. This higher operating temperature largely compensates for the lower rate of sulfdation of Z2T3-a relative to ZnO. No change in the initial sulfidation rate was observed with the addition of various amounts of H, (0, 1, and 10%) in the gas stream at temperatures in the range of 400-800 zn.3 z:, z:, 727.. 'C. Typical sulfidation profiles for ZnO and Z2T-a at 800 .ZT "C with various hydrogen concentrations (1,10, and 20%) 0 0 20 40 60 80 are shown in Figure 14. For both ZnO and Z2T-a, it appears that little if any zinc loss took place based on the mol% Ti02 (based on ZnO-TiO2) fmal weight of the solids. Both sorbents were completely Figure 12. Surface area of presintered (lo00 'C) sorbents before sulfded (based on the initial sorbent weight). The absence and aftm sulfdation; solid ZZT3-b not presintered. of zinc loss is due to the particular configuration of our system (i.e., a hemispherical pan with a thin layer of The largest surface area increase after sulfidation was reactant solid). Although little noticeable zinc loss took found with solids containing (Zn/Ti)abc = 1. Increasing place, reduction did occur. Zinc vapor which formed as the amount of titanium in the solids beyond a result of reduction diffused toward the reactant gas = 1produced no further increase in the surface area The subsequently reacting with HzS to form a solid product formation of free TiOz with higher surface area cannot be of ZnS which was deposited on the pan. After sfidation invoked to explain the surface area incrase of Zn-Ti4 was completed, a white film was observed on the sample sorbents upon sadation. In fact, when TiOzwas dcined pan. Analysis of this film by atomic absorption spec- for the same length of time and at the same temperature troscopy identified the presence of zinc. as ZnO, the surface area of ZnO was 1.3 times higher than At high conversion, the reaction rate dropped faster TiOP Therefore, low levels of TiO, are most effective in when sulfidation was performed with a gas containing Ind. Eng. Chem. Res., Vol. 31, No. 8,1992 1897

ZnO T.800"C x' 0.2 2%H2S-bal.N2

0.0 1 I 0 10 20 30 40 50 lime (min) Figure 15. SEM micrograph showing fine ZnS particles on the sdace of ZnO after complete sulfdstion at 800 "C in 2% H&20?& n,-im N~. only zinc loss but also weakening of the strength of the material by crack formations and the formation of fine particles. In contrast to the dramatic gross structural changes observed with ZnO, minor structural changes were ob- served for Z2T-a sulfded at 800 "C in the presence of various hydrogen concentrations (1-20%). This is in agreement with the earlier finding that %-Ti4 sorbents were more resistive to reduction than ZnO (Lew, 1990; Lew et al., 1992a). No fme crystals or agglomerates were seen on the surface of the reacted solids.

Y." Effsct of HzO.Coal-derived gas streams contain steam 0 20 40 60 BO in amounts depending on gasifier type and extent of b) lime (min) quenching (l(t50% H20). The effects of H20 on the sulfidation kinetics of the sorbent were addressed in this Figure 14. Ef€& of HIon the sulfidetion mnvetaion profiles at 800 'C for (a) ZnO and (b) ZZT-a. work. Experiments were performed in a gas mixture wnsisting of 2% H&1% H20-1% H2-96% N2and 0.5% either 10 or 20% H, as shown in Figure 14. A possible H,S-1.9% H,&l% H,-96.6% N,. Water was introduced explanation for the faster decrease of the reaction rate with in saturated amounts at 25 "C in flowing Hz-N,. In the 10 or 20% H, is pore blockage due to the deposit of solid absence of water vapor from the reactant gas stream, the product from the vapor-phase reaction between Zn(g) and calculated concentration of H20(produced during sulfi- H,S(g). Small jumps seen occasionally in the sulfidation dation) in the external film of the particle was approxi- profiles (e.g., Figure 14) are believed to be due to the mately zero. Therefore, although the 1% H20used here formation of cracks in the product layer. As the sulfidation was lower than what is typically found in wal-derived fuel product layer is formed, zinc and water vapor from re- gases (e.g., -20% from a fluidized-bed KRW gasifier), it duction will accumulate in the particle until their pressure was much higher than what is found in the external film. is sufficient to cause the product layer to crack. At that Thus, any potential effect (e.g., poisoning) of water vapor point, reaction will occur rapidly, and the conversion jump on the sulfidation kinetics would be apparent in a gas will be observed. Since this effect (i.e., the jump) is rela- containing 1% HzO. No such effect was found on the tively small, SEM micrographs of solids reacted prior to sflidation rates of ZnO, Z2T-a, and Z2T3-a under the the jump and beyond it could not conclusively verify this conditions used in this work. explanation. Regeneration of Bulk Zn-Ti-0 Sorbents. In order With the gas mixture containing 2% H&20% H24% for Zn-Ti-0 solids to be used commercially for the de- N,, large spherical agglomerates (0.7-1.3-pm radius) sulfurization of hot fuel gases, the sulfided solids must be formed as a result of the reaction between Zn(g) and regenerable. To determine the regenerability of Zn-Ti4 Ha(g) were deposited on the original swface (Fie 15). solids and their sulfidation performance after regeneration, The agglomerates were composed of smaller crystals two cycles of consecutive sulfidation-regeneration were (-0.04-pm radius). The high zinc vapor-phase concen- performed with ZnO and Z2T-a. The results are shown tration during sulfidation reaction with 20% H, in the in Figure 16 in terms of the dimensionless weight W/Wo reactant gas is believed to lead to this enhanced sintering. (where W is the instantaneous and Wothe initial weight Sainamthip and Amarakoon (1988) have also reported of the solid) versus the reaction time. Prior to the first enhanced grain growth in zinc vapor for manganese zinc cycle, the solids were reduced in 10% H,-90% Nzfor 7 ferrite. A lower level of hydrogen (10%) did not produce min. The partially reduced solids were then sulfided in these agglomerates although the small crystals were seen. 2% H2S-98% N2. After sulfidation, regeneration was In addition, some large cracks were evident in several performed in air (21% 02-79% Nz). High oxygen level waa particles. These cracks were probably caused by interior used in the regeneration to determine if any ZnSO, was pressure buildup leading to Zn(g) and H20(g)escaping formed. Incomplete regeneration is due to the formation from the particle. Thus, for ZnO, reduction can cause not of ZnSOI. All reactions were performed at 650 OC. 1898 Ind. Eng. Chem. Res., Vol. 31, No. 8, 1992 of the sorbents above 25 mol % (based on Zn-Ti T=650"C stoi- 1.10 complete sulfidation chiometry) does not lead to a further decrease in the fre- quency factor. No differences were measured in the suHdation kinetics z of different zinc titanate phases (i.e., Zn2Ti04,ZnzTi308, cr- and ZnTi03). .-OI The presence of titanium in ZnO inhibited sintering of s Zns. In fact, Zn-Ti4 Sorbents showed an increase in their surface area after sulfidation. No effect of hydrogen on the sflidation kinetics of ZnO and Zn-Ti-0 sorbents was observed for temperatures in the range of 400-8800 "C. No zinc loss was evident (by weight change in the TGA) for either ZnO or sorbent Z2T-a. However, zinc vapor was produced at 800 "C as evidenced by cracks and fine particles observed on the surface of ZnO and deposita in the pan. These morpho- time (min) logical features can be explained by ZnO reduction to zinc a) and water vapor followed by evolution of the product gases T-650"C and subsequent reaction of Zn(g) with HzS(g) to form fine ,.os} particles of ZnS(s). Such structural effects were absent complete sulfidation from the surface of Z2T-a particles after reaction at the t I same conditions as those for ZnO. Zn-Ti-0 sorbenta allow raising the operating tempera- cr- .-0) ture for fuel gas desulfurization as a result of their lower ZnO reduction rate. For example, a solid with (Zn/Ti)abdc s = 2/3 can operate at a temperature 94 "C higher than ZnO. The choice of sorbent composition for desulfurization should be based on the amount of reduction tolerated and the sulfur loading deaired. A higher percentage of titanium is needed in order to decrease the reduction rate (Lew et al., 1992a). This will be at the expense of sulfur loading. For example, a solid with (Zn/Ti)abmic= 2/3 which has 0.92 I ' 200 a 9 times lower reduction rate than ZnO, has a sulfur 0 50 I00 150 loading of only 0.16 g of sulfur/g of sorbent (based on b) time (min) complete conversion to ZnS). In contrast, ZnO has a sulfur Figure 16. Sulfidation-regeneration cycles of (a) ZnO and (b) loading of 0.39 g of sulfur/g of sorbent. Z2T-a. 1, reduction in 10% H2-90% N2; 2, sulfidation in 2% Ha- 98% N,; 3, regeneration in air. Acknowledgment This research was supported by the U.S. Department As discussed previously, the more rapid decrease in of Energy/University Coal Research Program under sflidation rate of sorbent Z2T-a, seen in Figure 16b, was Contract No. DEFG22-88PC88927. due the presence of nonuniform grain size distribution to Registry ZnO, 1314-13-2; zinc titanate, 12651-25-1. in the solid. In the first cycle, regeneration of Z2T-a and No. ZnO was essentially complete after 40-50 min. On the Literature Cited basis of the the final weight after regeneration, little if any zinc sulfate (ZnS04)was formed during regeneration. In Courty, P.; Ajot, H.; Marcilly, C.; Delmon, B. Oxydea Mixtea ou en Solution Solide sous Forme Tres Divisee Obtenus par Decompo- the second cycle, the sflidation performance of ZnO be- sition Thermique de Precurseurs Amorphea. Powder Technol. came worse when compared with the first cycle. As shown 1973, 7,21-38. in Figure 16a, the sflidation rate in cycle 2 decreased Flytzani-Stephanopouloa,M.; Gavalaa, G. R.; Tamhankar, S. S.; noticeably after approximately 90% conversion ( W/W, = Sharma, P. K. 'Novel Sorbents for High-Temperature Regenera- 1.06). In contrast, the sulfidation of Z2T-a (Figure 16b) tive HaRemoval"; Final Report DOE/MC/20417-1898, October in cycle 2 was the same as in the first cycle. In the second 1985. Flytzani-Stephanopouloa,M.; Gavalaa, G. R.;Jothimurugesan, K.; regeneration, the performace of both Z2T-a and ZnO be- Lew, S.; Sharma, P. K.; Bagajewicz, M. J.; Patrick, V. 'Detailed came slightly worse. After 40-50 min, regeneration of both Studies of Novel Regenerable Sorbents for High-Temperature Z2T-a and ZnO was about 95% complete. Coal-Gas Desulfurization"; Final Report DOE/MC/22193-2582, October 1987. Gibson, J. B.; Harrison, D. P. The Reaction Between Hydrogen Summary/Conclusions Sulfiide and Spherical Pellets Zinc Oxide. Ind. Eng. Chem. Pro- The initial sflidation rate of Zn-Ti-0 sorbents mea- cess Des. Deu. 1980,19, 231-237. Grindley, T.; Steinfeld, G. "Development and Testing of Regenerable sured with 90-125-pm particles in a TGA was 1.5-2 times Hot Coal Gae Desulfurization on Sorbents"; Final Report DOE/ slower than that of ZnO. At temperatures between 400 MC/16545-1125, October 1981. and 700 "C, similar activation energies (9-10 kcal/mol) Grindley, T.; Steinfeld, G. Zinc Ferrite aa Hydrogen Sulfide Absor- were found for both Zn-Ti-0 and ZnO sorbents, while bent. 'Third Annual Contractors' Meeting on Contaminant lower frequency factors were measured for the Zn-Ti-0 Control in Hot Coal-Derived Gaa Streama";Report No. DOE/ sorbents. Thus, it appears that the sulfidation of Zn-Ti4 METC/84-6,1983. proceeds by the same mechanism ZnO sflidation, but Hmde&, J: P.; Idrise, H.; Kiennemann, A. Adsorbed Speciea on as ZnO in CO-H2and C02-H2Reactions. Mater. Chem. Phys. 1988, the presence of titanium serves to eliminate reaction sites. 18,513-532. However, only a certain number of reaction sites are Jalan, V.; Wu, D. High Temperature Desulfurization of Fuel Gaaes eliminated by titanium. Increasing the titanium content for Molten Carbonate Fuel Cell Power Planta. Paper presented Ind. Eng. Chem. Res. 1992,31, 1899-1906 1899

at the National Fuel Cell Seminar, San Diego, CA, 1980. Marcilly, C.; Courty, P.; Delmon, B. Preparation of Highly Dispersed Krebs, S.; Littbarski, R. Preparation and Crystal Growth. Current Mixed Oxides and Oxide Solid Solutions by Pyrolysis and Topics in Material Science; Kaldia, E., Ed.; North-Holland Pub- Amorphous Organic Precursors. J. Am. Ceram. SOC.1970,53 (l), lishing: New York, 1981; pp 170-198. 56-57. Lew, S. High-Temperature Regenerative HA Removal by ZnO-Ti02 Ranade, P. V.; Harrison, D. P. The Variable Property Grain Model Systems. M.S. Thesis, Massachusetts Institute of Technology, Applied to the Zinc Oxide-Hydrogen Sulfide Reaction. Chem. Cambridge, 1987. Eng. Sci. 1981,36,1079-1089. Lew, S. The Reduction and Sflidation of Zinc Titanate and Zinc Sainamthip, P.; Amarakoon, V. R. W. Role of Zinc Volatilization on Oxide Solids. Ph.D. Dissertation, Massachusetts Institute of the Microstructure Development of Manganese Zinc Ferritee. J. Technology, Cambridge, 1990. Am. Ceram. SOC.1988, 71 (a),644-648. Lew, S.; Jothimurugesan, K.; Flytzani-Stephanopoulos,M. High- Westmoreland, P. R.; Gibson, J. B.; Harrison, D. P. Comparative Temperature HA Removal from Fuel Gases by Regnerable Zinc Kinetics of High-Temperature Reaction Between H2S and Se- Oxide-Titanium Dioxide Sorbenta. Znd. Eng. Chem. Res. 1989, lected Metal Oxides. Enuiron. Sci. Technol. 1977, ll, 488-491. 28,535-541. Woods, M. C.; Gangwal, S. K.; Jothimurugesan, K.; Harrison, D. P. Lew, S.; Sarofh~,A. F.; Flytzani-Stephanopoulos,M. The Reduction Reaction between H&3 and Zinc Oxide-Titanium Oxide Sorbents. of Zinc Titanate and Zinc Oxide Solids. Chem. Eng. Sci. 1992a, 1. Single-Pellet Kinetic Studies. Znd. Eng. Chem. Res. 1990,29, 47 (6),1421-1431. 1160-1167. Lew, S.;Sarofim, A. F.; Flytzani-Stephanopoulos,M. Modeling of the Sflidation of Zinc-Titanium Oxide Sorbents with Hydrogen Received for reuiew March 9, 1992 Sulfide. AIChE J. 199213, in press. Accepted May 15,1992

MATERIALS AND INTERFACES

Mechanisms for Lowering of Interfacial Tension in Alkali/Acidic Oil Systems: Effect of Added Surfactant

Jeff Rudin and Darsh T. Wasan* Department of Chemical Engineering, Illinois Institute of Technology, Chicago, Illinois 60616

Experimental studies are conducted in order to determine the physicochemical mechanisms re- sponsible for lowering of interfacial tension in alkali, surfactant, and surfactanbenhanced alkali/acidic oil systems. A well-defined model oil is chosen to examine the influence of various surfactants and surfactant mixtures, such as oleic acid and ita ionic counterpart, sodium dodecyl sulfate, petroleum sulfonate, and isobutanol, on equilibrium interfacial tension. With added surfactant alone, the interfacial tension goes through an ultralow minimum with increasing acid concentration. This proves for the first time that the un-ionized acid species plays a major role in affecting interfacial tension, and appears to be the key element in the synergistic process taking place between the added surfactant and the ionized acid species. The un-ionized acid species partitions the added surfactant out of the aqueous phase, and the minimum in interfacial tension occurs when the partition coefficient is about unity. When alkali is added, the low interfacial tension is not lost, but actually shifts to different acid concentrations in a systematic way.

Introduction region, the ultralow interfacial tensions are due to the It is well-known that interfacial tension between a presence of a saturated monolayer at the interface (Chan surfactant solution and a hydrocarbon can become ultra- and Shah, 1980, Pouchelon et d,1981). In the three-phase low. A number of variables, such as salinity (Chan and region, the ultralow interfacial tensions are due to a critical Shah, 19801, oil chain length (Chan and Shah, 1980), al- behavior (Fleming and Vinatieri, 1979, 1981; Fleming et cohol concentration and type (Miller and Neogi, 1985), al., 1980). Chan and Shah (1980) also found that an in- surfactant concentration (Chan and Shah,1980) and type crease in salinity causes partitioning of the surfactant from (Doe et al., l977,1978a,b), and temperature (Miller and the aqueous to the oil phase, and interfacial tension is Neogi, 198!5), have been found to affect the position of the ultralow when the partition coefficient is about unity. interfacial tension minimum. Chan and Shah (1981) ob- Chan and Shah (1981) also showed that the interfacial served two regions of ultralow interfacial tension as sur- tension goes through an ultralow minimum when the oil factant concentration is increased. One region is at low chain length is increased, and the minimum again occurs surfactant concentrations of about 0.1 wt ?% (a two-phase when the partition coefficient is about unity. region), and the other is at high surfactant concentrations With the addition of a small amount of surfactant to the of about 4 wt '3% (a three-phase region). In the two-phase alkaline solution, the interfacial tension can become lower than either surfactant or alkali alone (Schuler et al., 1986). The reason for this synergism is not well-known. In this * To whom correspondence should be addressed. paper, we investigate the interaction of the un-ionized acid 0888-5885/92/2631-1899$03.00/00 1992 American Chemical Society