Low-temperature alumina is widely employed in industrial catalysis as a catalytically active material and as a support for catalytically active components. Its utility is based on the high specific surface area and surface acidity of different crystal modifications of alumina including γ-, η-, and χ-Al₂O₃. γ-Al₂O₃ has been used most extensively. Pure, activated, and chlorinated γ-aluminas have been applied to the hydroisomerization of C₄-C₆ hydrocarbons[1-3], alkylation of isobutane by alkenes [4, 5], cracking of long-chain alkanes[6], and dehydrochlorination of chloroalkanes [7]. The activity and stability of alumina in these reactions is presumed to depend on surface acidity [2]. In our opinion, χ-Al₂O₃ deserves greater consideration owing to the greater number of defects in its structure [8] and, consequently, more sites for stabilization of catalytically active components [9].

In this paper, we focus on the use of alumina as a catalyst and as a support for the direct oxidation of hydrogen sulfide. Direct catalytic H₂S oxidation to elemental sulfur selectively removes hydrogen sulfide from natural gas without appreciable conversion of the hydrocarbons [10, 11]. The reaction takes place at a comparatively low temperature of 220-280 ℃ [11]. Although activated carbon [12-14] and zeolites [15] are commonly employed as catalysts for H₂S oxidation, individual oxides of transition metals (e.g., Fe₂O₃, CuO, MnO_x, CrO_x, and V₂O₅) or their mixtures are more promising [16-32]. Granulated and monolithic aluminas, which are widely used as a support material for this process [17, 18, 22, 25, 33], possess a natural activity for direct H₂S oxidation. However, their activity and selectivity are insufficient, because alumina accelerates both the forward and reverse directions of the Claus reaction [34, 35].

This work examines the oxidative interaction of hydrogen sulfide with different crystal modifications (γ, χ, and α) of alumina and with cation-and anion-modified aluminas. Fourier transform infrared (FTIR) spectroscopy of adsorbed CO as a probe molecule and temperature-programmed desorption of ammonia (NH₃-TPD) were used to characterize the surface acid properties of aluminas and to explain their catalytic activity in direct H₂S oxidation. The textural properties of aluminas also were used to identify structure-performance relationships.

Five spherical alumina samples, γ-Al₂O₃, (γ+15%χ)-Al₂O₃, (γ+35%χ)-Al₂O₃, (γ+50%χ)-Al₂O₃, and α-Al₂O₃ were prepared by hydrocarbon-ammonia molding [22, 36, 37]. Samples were treated with an acid peptizer [37] to impart plasticity to the aluminum hydroxide pastes. The acid mole fraction was 0.067 as in [36]. The γ-Al₂O₃ sample was prepared from aluminum hydroxide with a pseudoboehmite structure as in ref. [33]. γ-Al₂O₃ samples containing 15 and 35 wt% χ-Al₂O₃ were prepared from aluminum hydroxide produced by the thermal decomposition of gibbsite in a catalytic heat generator [36, 37]. The sample containing 50 wt% γ-Al₂O₃ and χ-Al₂O₃ was produced from a mixture of gibbsite and pseudoboehmite. The χ-Al₂O₃ sample containing trace γ-Al₂O₃ was prepared from gibbsite. The above samples were calcined at 550 ℃ for 4 h in air. The spherical α-Al₂O₃ sample was prepared by thermal calcination of γ-Al₂O₃ at 1200 ℃ for 6 h.

MgO/(g+15%c)-Al₂O₃ was produced by modifying spherical granules of aluminum hydroxide with magnesium nitrate solution, drying at 110 ℃, and heating at 550 ℃ as described elsewhere [33]. The MgO concentration in the calcined samples was 3.2 wt%. Sulfate-modified alumina samples were prepared by incipient wetness impregnation of spherical granules of (γ+15%χ)-Al₂O₃ with a sulfuric acid solution, drying at 110 ℃, and calcining at 500 ℃ for 4 h. The sulfate content determined by X-ray fluorescence analysis varied from 1 to 10 wt% SO₄^2–. The chlorinated alumina sample was prepared by impregnating spherical granules of (γ+15%χ)-Al₂O₃ with a hydrochloric acid solution (0.5 mol/L), drying at 110 ℃, and calcining at 500 ℃ for 4 h. The nominal chlorine concentration was 1 wt% Cl, but the actual Cl content determined by X-ray fluorescence (XRF) analysis was 0.35 wt%.

The chemical composition of initial and spent aluminas was examined by XRF spectroscopy using an ARL analyzer with a Rh anode. The physicochemical properties of the samples are given in Table 1. The textural properties of the catalysts including specific surface area (S_BET), pore volume (V_Σ), and pore diameter (d) were determined by nitrogen adsorption isotherms at –196 ℃ using an ASAP2400 automated adsorption analyzer (Micromeritics, United States).

The phase composition of alumina was determined from X-ray diffraction (XRD) patterns recorded for 2θ = 10°-70° at a scan rate of 1°/min using the crystallographic databases X-ray Powder Diffraction Files JCPDS-ICDD and ICSD-for-WWW (database Fachinformationszentrum (FIZ), Karlsruhe, Germany, 2003-2010). The XRD patterns were obtained on an HZG-4C diffractometer (Freiberger Prazision Mechanik, Germany) using monochromatic Cu K_α radiation. The extent of χ-and γ-modification of the alumina samples was established from XRD data employing calibrating mixtures.

The compositions of spent and sulfated alumina samples were determined by differential thermal and thermogravimetric analysis (DTA-TG) on a NETZSCH STA 449 (Тetzsch-Geratebau GmbH, Germany) instrument. A 100-mg powdered sample was placed in a crucible and heated from 20 to 1000 ℃ at a rate of 10 ℃/min in air (50 cm³/min).

A "BOMEM MB102" FTIR spectrometer was used for spectroscopic studies. Carbon monoxide adsorption was studied in a homemade low-temperature IR reactor/cell equipped with a heating chamber and a chamber with CaF₂ windows. The design allowed us to expose the sample to heat pretreatment under vacuum and then record a spectrum from –196 to 20 ℃. Hydrogen sulfide adsorption was examined in a homemade high-temperature spectroscopic reactor/cell, whose design allowed for preheating of the sample under vacuum, adsorption of H₂S, and recording of a spectrum at 20-400 ℃.

Triturated samples were pressed into 10-20 mg/cm² thick self-supported wafers. A catalyst wafer was placed in the low-or high-temperature spectroscopic reactor/cell and pretreated under vacuum (< 10^–3 mbar) at 400 ℃ for 4 h. CO or H₂S was then adsorbed on the catalyst.

Pure carbon monoxide was introduced to the spectroscopic reactor/cell at 2.5 mbar. An FTIR spectrum was recorded at room temperature at 1200–4000 cm^–1 with 4 cm^–1 resolution and 100 accumulated scans. The sample was cooled to –196 ℃ with liquid nitrogen to provide maximal CO adsorption on the sample, and a spectrum was recorded. A series of spectra were recorded as the temperature rose from –196 ℃ to ambient during nitrogen evaporation. This procedure allowed identification of different catalytic sites on the sample surface due to the continuous decrease in CO coverage during a single run. The spectra were normalized by wafer density.

The concentration and strength of Lewis acid sites (LAS) and Br nsted acid sites (BAS) were determined using methods described elsewhere [38, 39]. The integrated intensities of absorption bands in the CO vibration regions of 2180–2240 cm^–1 (LAS) and 2150–2175 cm^–1 (BAS) were calculated from alumina spectra and were divided by the molar extinction coefficients. The latter quantity (cm/μmol) was assumed to equal 1.23 for the band at 2240-2220 cm^–1, 1.1 at 2215–2205 cm^–1, 1.0 at 2200 cm^–1, 0.9 at 2190–2196 and 2177–2180 cm^–1 [38, 39, 40], and 2.6 at 2150–2175 cm^–1 [40, 41]. The BAS content of aluminas was determined using the 3650-3800 cm^–1 OH vibration region and a molar extinction coefficient of 5.3 cm/μmol [42]. The uncertainty of the quantitative measurements was 20%.

Hydrogen sulfide was added to the spectroscopic reactor/cell, which contained the heat-pretreated sample wafer at a pressure of 40 mbar. Hydrogen sulfide adsorption on alumina was studied at 20 ℃, and H₂S transformation was examined at 100, 200, and 300 ℃ by FTIR spectroscopy. FTIR spectra were recorded at 1000–4000 cm^–1 with 4 cm^–1 resolution and 100 accumulated scans.

The surface acidity of aluminas also was studied by NH₃-TPD. NH₃-TPD experiments were carried out in a flow apparatus equipped with a thermal conductivity detector. A 100-mg sample with a fractional composition of 0.25–0.50 was mixed with 100 mg of quartz of similar granule size. The sample was pretreated in an argon stream (30 cm³/min) at 500 ℃ for 2 h to remove adsorbed water. It was cooled to 75 ℃, and ammonia was adsorbed by flowing a mixture of NH₃ (0.35 vol%) in He over the sample for 1 h. The sample was flushed with helium (30 cm³/min) to remove ammonia from the catalyst pores and cooled to room temperature. The NH₃-TPD curve was recorded by passing helium through the sample at a rate of 30 cm³/min and heating from 25 to 700 ℃ at a rate of 10 ℃/min. The total surface acidity was estimated from the amount of desorbed ammonia (μmol/g) assuming single-site ammonia adsorption.

The catalytic activity and stability of the samples was measured in a fixed-bed catalytic reactor at a stoichiometric ratio of reagents (H₂S/O₂ = 2/1) in He. The reactor was positioned vertically so that the sulfur generated in the experiment flowed down to be collected in a glass container. The H₂S concentration in the initial feed was 0.6 vol%. A 1-cm³ alumina sample with a 0.25–0.5 mm granule size was placed in the quartz reactor. The space velocity of the feed was 100 cm³/min, and the gas hour space velocity (GHSV) was 6000 h^–1. The catalytic activity of aluminas was studied at 200-400 ℃, and the stability of their catalytic behavior was examined at 300 ℃ over 50 h.

Reagents (H₂S, O₂) and reaction products (SO₂, COS) were analyzed using a "Tsvet-500M" gas chromatograph equipped with a TCD and a Teflon column (3-m length, 3-mm i.d.) with a "Hayesep C" sorbent modified with 0.5% H₃PO₄. The space velocity of the He carrier gas was 30 cm³/min, and the operating temperature of the column was 150 ℃.

Experimental data were presented in terms of hydrogen sulfide conversion and sulfur selectivity as a function of temperature. The H₂S conversion (X_H2S, %) and selectivity of H₂S oxidation to sulfur (S_S, %) were calculated from experimental concentrations of reagents and products using the following equations:

where C_H2S⁰ is the concentration of hydrogen sulfide (vol%) in the feed, and C_H2S, C_SO2, and C_COS are the concentrations of hydrogen sulfide (vol%), sulfur dioxide (vol%), and COS (vol%) at the reactor outlet. The temperature dependences of H₂S conversion and selectivities of sulfur production were used to compare the catalytic behaviors of aluminas with various crystal modifications.

Fig. 1 displays the catalytic characteristics of aluminas in the direct oxidation of H₂S according to the reaction, H₂S + 1/2O₂ = 1/n S_n + H₂O, where n = 6 or 8, at 100-300 ℃. The total H₂S conversion and selectivity of H₂S transformation to sulfur are represented. The experimental data show that the crystal modification of alumina significantly affects the catalytic activity. The total H₂S conversion over alumina at 250 ℃ increases as α-Al₂O₃ < γ-Al₂O₃ < (γ+15%χ)-Al₂O₃ < (γ+35%χ)-Al₂O₃ < (γ+50%χ)-Al₂O₃ ~ (γ+90%χ)-Al₂O₃. Maximum H₂S conversion of 80%-90% occurred over samples containing mixtures of γ-Al₂O₃ with 35-90 wt% χ-Al₂O₃. H₂S conversion over γ-Al₂O₃ and α-Al₂O₃ increased with temperature and achieved a maximum value at 375-400 ℃. Maximum H₂S conversion over samples containing χ-Al₂O₃ occurred at temperatures as low as 300 ℃ for 15 wt% χ-Al₂O₃ and 250 ℃ for 50–90 wt% χ-Al₂O₃. H₂S conversion over samples containing χ-Al₂O₃ diminished slightly at temperatures above 300 ℃.

Selectivity with regard to sulfur formation was nearly 100% at 200 ℃ for alumina samples with different crystal modifications. The main products of H₂S oxidation by O₂ are sulfur and water; SO₂ and COS are not observed under these conditions. At 250 ℃, we observed SO₂ among the reaction products and gradually diminished sulfur selectivity due to the reaction, H₂S + 3/2 O₂ → H₂O + SO_2, and the reverse Claus process, 3S + 2H₂O ↔2H₂S + SO₂. The phenomenon is more pronounced at higher temperatures, because sulfur oxidation becomes prevalent under these conditions. When the temperature was increased to 400 ℃, sulfur selectivity decreased to 90%-92% on γ-Al₂O₃ and α-Al₂O₃ and to 82% on (γ+35%χ)-Al₂O₃. The effect of temperature on the change in sulfur selectivity is enhanced with χ-Al₂O₃. A greater χ-Al₂O₃ content leads to lower sulfur selectivity at 400 ℃ as shown in Fig. 1(B). Thus, S_S was only 53% for the 90% χ-Al₂O₃ sample.

The results above are in accord with those in the literature [16, 21, 25]. The same tendencies are observed for catalysts containing Fe₂O₃ and Cr₂O₃ supported on alumina [16, 25], silica [16, 21], zirconia [16], and other oxide materials. These catalysts exhibit high selectivity and efficiency of H₂S conversion to elemental sulfur at temperatures close to or slightly above the sulfur dew point, but selectivity diminishes due to SO₂ formation at higher reaction temperatures. For example, Fe₂O₃/γ-Al₂O₃ provides 92% H₂S conversion and 96% sulfur selectivity at 230 ℃ [25], although the GHSV was half that of our experiments (3000 versus 6000 h^–1). Conversion increased to 98%-99% with increasing temperature, but sulfur selectivity decreased and did not exceed 41% at 350 ℃ [25]. On the low surface area α-Al₂O₃ modification [16], efficient H₂S conversions (97%-98%) were achieved with Fe₂O₃/α-Al₂O₃ at 280-400 ℃, although sulfur selectivity began to deteriorate (70%-85% at GHSV = 3000–12000 h^–1). Thus, 35 wt% χ-Al₂O₃ aluminas are more active and more selective in H₂S → S oxidation at 200–300 ℃. Their catalytic behavior is comparable to, and in some cases greater than, that of the supported Fe₂O₃ catalysts mentioned in the literature.

To study the catalytic stability of χ-alumina, we tested it in a H₂S/O₂/He feed at 300 ℃ for 50 h and compared the results with those for γ-Al₂O₃. The results of the stability test are illustrated in Fig. 2. Both materials catalyzed the oxidation of H₂S to sulfur without deactivation for at least 50 h under a stoichiometric H₂S/O₂ mixture at 300 ℃. The (γ+35%χ)-Al₂O₃ catalyst was more active and produced more sulfur compared with γ-Al₂O₃ in the stability tests. The mean sulfur yield over (γ+35%χ)-Al₂O₃ was 78%-82%, which is slightly less than the value allowed by thermodynamics (88%). The mean sulfur yield over γ-Al₂O₃ was nearly 50%. Ceramic supports prone to formation of sulfates (MgO, ZrO₂, TiO₂, and Al₂O₃) are generally believed [16] to exhibit poor characteristics in the oxidation of hydrogen sulfide to sulfur and to be deactivated under the reaction conditions. In contrast to [16], we observed good stability in the catalytic behavior of the (γ+35%χ)-Al₂O₃ catalyst.

Table 2 shows the material balance of H₂S oxidation over γ-Al₂O₃ and (γ+35%χ)-Al₂O₃ following a stability test. The catalysts provided average H₂S conversions of 55% and 85%, respectively, at 300 ℃ (Fig. 2). The main sulfur-containing products were sulfur and sulfur dioxide. The average selectivities for these products were 95% and 4%, respectively, on γ-Al₂O₃ and 92% and 7% on (γ+35%χ)-Al₂O₃. Some sulfur deposited on the catalysts. According to XRF analysis, spent γ-Al₂O₃ and (γ+35%χ)-Al₂O₃ samples accumulated 0.23 and 0.42 wt% sulfur, respectively, during the experiment. The minimum deviation of the sulfur material balance is ≤5%, which can be attributed to the uncertainty of the measurements.

The catalytic properties of aluminas modified with Mg²⁺ cations or Cl^– and SO₄^2– anions are shown in Fig. 3. The (γ+15%χ)-Al₂O₃ sample modified with Mg²⁺ produced 82% H₂S conversion at 200-400 °C. This value was 20% greater than that achieved with the unmodified sample at 250 °C and differed by only 5% at 300 °C. Moreover, the sulfur selectivity over Mg²⁺/(γ+15%χ)-Al₂O₃ equaled 93%-94% at 350 °C and above (Fig. 3, curve (2)). The results were superior to those obtained with the unmodified sample.

Modification of (γ+15%χ)-alumina by sulfate ions was accompanied by a deterioration in catalytic performance. Hydrogen sulfide conversion was significantly reduced at temperatures below 300 ℃ and slightly so above 300 ℃ (Fig. 3(A), curve (3)). The selectivity of sulfur formation was reduced by 5%-10% over the entire temperature range (Fig. 3(B), curve (3)). Modification of (γ+15%χ)-Al₂O₃ with chloride ions resulted in a dramatic decrease in H₂S conversion, although selectivity increased to 100% at temperatures above 350 ℃ (Fig. 3, curve (4)).

We used NH₃-TPD and FTIR spectroscopy of adsorbed CO as molecular probes of Lewis and Brønsted acid sites to study the surface properties of aluminas. The FTIR spectra of CO adsorbed on alumina with γ, (γ+χ), and α crystal modifications, which are presented in Fig. 4, are discussed first. Absorption bands at 2186, 2210, and 2216 cm^–1 in the CO stretching region occur in the spectrum of γ-Al₂O₃ at 20 ℃. There also are bands at 2147 and 2156 cm^–1 in the –196 ℃ spectrum. After addition of χ-phase Al₂O₃ to γ-Al₂O₃, new bands appear at 2195-2197 and 2200-2206 cm^–1, and the band at 2147 cm^–1 vanishes. The frequencies of two absorption bands increase with increasing χ-phase content: 2186 and 2195 cm^–1 for (g+15%c)-Al₂O₃, 2187 and 2197 cm^–1 for (γ+35%χ)-Al₂O₃, and 2189 and 2194 cm^–1 for (γ+50%χ)-Al₂O₃ and (γ+90%χ)-Al₂O₃. Also, when the χ-phase content equals or exceeds 50%, the bands at 2210 and 2220 cm^–1 reappear in the spectrum. According to published data [20, 40, 41], the absorption bands at 2140-2155 cm^–1 are attributed to CO hydrogen-bonded to hydroxyl groups on the alumina surface. The bands at 2210-2240 and 2180-2200 cm^–1 are assigned [38, 40-45] to surface LAS that are characterized as unsaturated aluminum cations (Al³⁺) with tetrahedral coordination [41, 43, 45] at crystallographic defects [46] and with octahedral coordination [44, 45] at the regular defects of low-index crystallite faces [45-47], respectively. Shifting of the 2143 cm^–1 frequency to greater wavenumbers indicates an increase in the heat of CO adsorption and the strength of the LAS [38, 43, 44]. Lewis acid sites are commonly classified as weak (2180-2205 cm^–1), medium (2205-2220 cm^–1), and strong (2220-2245 cm^–1) according to the heat of CO adsorption [38, 43].

Table 3 lists the Lewis and Br nsted acidities determined from the FTIR spectra of CO adsorbed on aluminas. The table shows that γ-Al₂O₃ and (γ+χ)-Al₂O₃ primarily contain weak LAS with a heat of adsorption equal to 32-38 kJ/mol [38, 44] and a CO stretching vibration at 2185-2197 cm^–1. Examination of FTIR data for samples containing χ-Al₂O₃ shows that the disordered χ-Al₂O₃ phase contributes significantly to the weak LAS content with a maximal concentration observed at 35% χ-phase. γ-Al₂O₃, (γ+90%χ)-Al₂O₃, and α-Al₂O₃ samples also have a small amount of intermediate strength LAS, which is confirmed by high frequency absorption bands at 2210, 2220, and 2212 cm^–1, respectively. Their heats of absorption are equal to 42-49 kJ/mole [38, 44]. Thus, aluminas with γ, γ+χ, and α crystal modifications differ significantly in their surface acidity. The total LAS concentrations and strengths of g-Al₂O₃ and (γ+90%χ)-Al₂O₃ samples are close to those observed for γ-and χ-aluminas [40, 48].

The integrated intensities of the 2150-2162 cm^–1 absorptions are believed [42] to give semiquantitative information on the BAS content of alumina. H-bonding of CO with surface hydroxyl groups is characterized by an absorption band at 2150-2162 cm^–1[20, 38, 43, 49] and a shift of the relevant OH-group frequencies to lower values by 80-90 cm^–1 [39-41] or 125 cm^–1 [49]. These OH-groups are generally assigned to bridging hydroxyls coordinated to two or three octahedral Al³⁺ ions [46]. The OH-groups are characterized by absorption bands at 3735, 3710, 3690, and 3670 cm^–1 [38, 46] and weakly acidic properties with a proton affinity of 1420 ± 20 kJ/mol [38]. In rare instances, assignment of the 3725-3745 cm^–1 band to terminal Al³⁺ cations with octa-[45], penta-[47] and tetrahedral[49] coordination is disputed. The BAS contents calculated from the integrated intensities of the 2150-2162 cm^–1 bands are presented in Table 3. Among the aluminas studied, γ-Al₂O₃ contains the most proton acid sites. The number of such sites decreases as the fraction of the χ-phase increases. α-Al₂O₃ has minimal BAS content. The data obtained correlate qualitatively with the results of [48], which examined the concentration of terminal and bridging OH-groups on aluminas with γ-and χ-crystal modifications using FTIR spectroscopy in the OH stretching region (3580-3800 cm^–1).

The surface acidity of alumina also was evaluated by NH₃-TPD. The number of sites capable of ammonia chemisorption is assumed [50] to be sum of the Br nsted acid (proton donor) and moderately strong Lewis acid (electron acceptor) sites having OH and CO vibrational frequencies of 3688 and 2200-2210 cm^–1, respectively. According to published data, weak Lewis acid sites with ν(CO) = 2180-2195 cm^–1 [38, 43] and weakly acidic OH-groups with ν(OH) = 3750-3780 cm^–1 and a proton affinity of 1590 kJ/mol [38, 40, 41] are not observed by NH₃–TPD due to formation of weak physical bonds with NH₃ [50]. The latter are removed by purging with helium at 75-100 ℃ before NH₃–TPD. The NH₃–TPD profiles of unmodified aluminas are shown in Fig. 5(A). The total acid site content estimated by NH₃–TPD is indicated in Table 3. The γ-Al₂O₃ sample is characterized by ammonia desorption over a wide temperature range with pronounced maxima at 200, 355, and 530 ℃ and shoulders at 460 and 615 ℃. The peaks correspond to ammonia desorption from different acidic sites, which are divided [50] according to the energy of ammonia chemisorption into Lewis acid sites (200–400 ℃) and Br nsted acid sites (400–600 ℃). The LAS and BAS contributions to the total acidity of the sample are comparable. The proton acid sites have ammonia desorption energies greater than 150 kJ/mol. Moderately strong LAS are characterized by ammonia desorption energies of 85-110 and 110-150 kJ/mol. The (γ+90%χ)-Al₂O₃ sample has a broad asymmetric ammonia desorption peak with a maximum at 415 ℃ and a shoulder at 530 ℃. These features correspond to Lewis acid and Br nsted acid sites, respectively. Moderately strong LAS with an NH₃ desorption energy of 110-150 kJ/mol make a greater contribution to the total sample acidity. The BAS content is much lower than in the γ-Al₂O₃ sample. The NH₃-TPD experiments show that the acidic properties of (γ+15%χ)-Al₂O₃ and (γ+35%χ)-Al₂O₃ are a superposition of the γ-and χ-phase characteristics. An increase in the c-phase content leads to an increase in the number of moderately strong LAS with an ammonia desorption energy of 110-150 kJ/mol and a decrease in the BAS and weaker LAS contributions. The total acidity of the two-phase samples, which was calculated from the amount of ammonia desorbed by NH₃-TPD, is less than that of the pure γ-phase and χ-phase aluminas. For γ-Al₂O₃ and (γ+χ)-Al₂O₃, the NH₃-TPD results are consistent with the FTIR data of adsorbed carbon monoxide.

Fig. 6 shows spectra of CO adsorbed on the surface of (γ+15%χ)-Al₂O₃ modified with Mg²⁺, SO₄^2–, and Cl^– ions. CO adsorption on the MgO-modified sample yields an absorption band at 2154 cm^–1, which corresponds to the vibration of CO adsorbed at acidic OH sites [38, 43, 49], and a band at 2181-2193 cm^–1 belonging to the vibration of CO adsorbed on LAS [38, 40, 41, 43]. The frequency shifts from 2187 to 2181 cm^–1 and from 2197 to 2193 cm^–1 indicate a decrease in LAS strength [41, 43]. This is accompanied by a twofold increase in the concentration of Al³⁺ LAS. The strengths and concentrations of LAS and BAS sites on the surface of modified alumina samples are given in Table 3.

CO adsorption on (10%)sulfate-modified (γ+15%χ)-Al₂O₃ produces absorption bands at 2210 cm^–1 and 2199 cm^–1, which are attributed to CO adsorbed on intermediate and weak Al³⁺ LAS [40, 43], respectively. The LAS strength of the 2210 cm^–1 band is 35%-45% greater than that of the sites of the original (γ+15%χ)-alumina. However, the number of LAS decreases significantly. There also is a band at 2168 cm^–1, which is assigned to CO H-bonded to hydroxyl groups on the catalyst surface [38, 43, 49]. Their concentration changes little compared to unmodified (γ+15%χ)-alumina, but their proton acidity increases markedly.

FTIR spectra of CO adsorbed on Cl-modified (γ+15%χ)-Al₂O₃ exhibit two types of LAS on the catalyst surface. These are characterized by absorption bands at 2191 and 2198 cm^–1. Their strength is slightly greater (by 10%-15%) than that of unmodified alumina. The Cl-modified sample has twice as many Lewis acid (ν_CO = 2195-2198 cm^–1) and Br nsted acid (ν_CO = 2154 cm^–1) sites, while the number of LAS with ν_CO = 2186-2191 cm^–1 is comparable to that of the unmodified sample. This occurs, because interaction of HCl with an > Al-O-Al < fragment [2] forms terminal > Al-OH (band at 2154 cm^–1) and > Al-Cl (band at 2198 cm^–1) groups. The chloride ion content of the calcined Cl/(γ+15%χ)-Al₂O₃ sample was less than of the sample impregnated with hydrochloric acid and dried in air, because chloride ions are partially removed as AlCl₃ or HCl during thermal treatment at 500 ℃ [51].

The NH₃-TPD profiles of modified aluminas are shown in Fig. 5(B). Ammonia desorption from MgO-and Cl-modified samples begins at lower temperatures (200 and 250 ℃) than from unmodified (γ+15%χ)-Al₂O₃ (400 ℃) and continues to 600 ℃. The majority of NH₃ desorbs from Cl/(γ+15%χ)-Al₂O₃ and MgO/(γ+15%χ)-Al₂O₃ at temperatures of 295 ℃ (85-110 kJ/mol) and 415-430 ℃ (110-150 kJ/mol), respectively, which indicates the presence of weaker LAS sites with lower ammonia desorption energies than in the unmodified sample. NH₃ amount from unmodified and MgO-modified (γ+15%χ)-Al₂O₃ at 500–700 ℃ are the same indicating similar BAS concentrations in these samples. The BAS content is somewhat less in Cl/(γ+15%χ)-Al₂O₃. Sulfate(5%)-modified and unmodified (γ+15%χ)-Al₂O₃ desorb ammonia over the same temperature range (400–700 ℃), but the intensity of NH₃ desorption from the modified sample is greater. Because this peak can be a superposition of ammonia and sulfur dioxide desorptions, the BAS concentration (Table 3) determined in this temperature range can be overestimated. SO₂ evolution during NH₃-TPD experiments was previously reported for sulfated alumina samples and was attributed to a redox reaction between sulfate groups and adsorbed NH₃ [52].

Modification of alumina with Cl^– and Mg²⁺ ions alters surface acidity primarily by formation of more Lewis acid sites, whereas sulfate ions increase the BAS content. Chlorination of (γ+15%χ)-alumina significantly changes the strength of the LAS notably by decreasing the ammonia desorption energy from 110-150 to 85-100 kJ/mol. The trends of increasing concentration of protonated and deprotonated sites, which we observe for (γ+15%χ)-alumina modified by Mg²⁺ or sulfate/chloride ions, respectively, are consistent with the data published in [41] for modified γ-Al₂O₃ samples.

The textural properties of aluminas with γ, (γ+15%χ), and (γ+35%χ) crystal modifications, including alumina modified by Mg²⁺ cations, have been described in detail by Ismagilov et al. [33, 36, 37]. Some of these data are discussed below. The isotherms of low temperature nitrogen adsorption on aluminas with γ, γ+χ, and α crystal morphologies and aluminas modified by Mg²⁺ or sulfate/chloride ions are shown in Fig. 7. It is known [53, 54] that a phase transition from γ-Al₂O₃ to α-Al₂O₃, which are low-and high-temperature modifications of alumina, respectively, leads to a sharp change in textural characteristics (Fig. 7, curves (1) and (7)). Modification of γ-Al₂O₃ by addition of the low-temperature χ-form up to 35 wt% does not lead to a significant change in texture. Low-temperature nitrogen adsorption isotherms of γ-Al₂O₃ and (γ+15%χ)-Al₂O₃ are type Ⅳ, which is characteristic of 2-to 50-nm microporous or mesoporous materials having some pores filled with capillary condensed nitrogen as the saturation vapor pressure is approached. A hysteresis loop is present in the adsorption isotherms of γ-Al₂O₃ and its mixture with χ-Al₂O₃. In the low-temperature nitrogen adsorption isotherms (Fig. 7(a)), there are tendencies for increased nitrogen adsorption with increasing χ-alumina content and for saturation of the pores at somewhat greater relative pressures. This trend usually indicates a change in the pore size and shape of porous materials, which we observe in pore-size distribution curves (Fig. 7(b)). The samples based on γ-Al₂O₃ and (γ+15-35%χ)-Al₂O₃ are biporous, and the fraction of ≤ 4-nm pores decreases while that of 7-to 20-nm pores increases with increasing χ-phase content. The prevailing pore size also increases from 7 nm for γ-Al₂O₃ to 11-12 nm for (γ+15%χ)-Al₂O₃ and to 14-15 nm for (γ+35%χ)-Al₂O₃. In addition, the specific surface areas calculated by Brunauer-Emmett-Teller (BET) and Barrett-Joyner-Halenda (BJH) for low-temperature alumina samples differ by approximately 30%. This means that the pore-size distribution curves indicate ~70% of the material to be mesoporous. The remaining pores are less than 3-nm in size and are not represented in the distribution curves (Fig. 7(b)). The pore-size distribution of α-Al₂O₃ is markedly heterogeneous. The distribution maximum occurs at 4 nm, and larger 5-, 8-, and 40-nm pores also are present. However, the contribution of micropores and mesopores is insignificant, because the ratio of micro/mesopore to macropore volumes is about 0.07. The α-Al₂O₃ sample possesses a very small surface area (9.3 m²/g).

The low-temperature nitrogen adsorption isotherms for modified (γ+15%χ)-Al₂O₃ shown in Fig. 7(a) (curves (4)-(6)) are type Ⅳ, as are the isotherms of the unmodified sample. However, at the same relative adsorbate pressure, the Mg²⁺-and Cl^–-modified samples bind less adsorbate and reach saturation at lower pressures relative to the unmodified sample. The textural characteristics of modified MgO/(γ+15%χ)-Al₂O₃ and Cl/(γ+15%χ)-Al₂O₃ are close to those of the unmodified sample. The specific surface area is about 150-165 m²/g, and the micro/mesopore volume is about 0.49-0.58 cm³/g. These values indicate a homogeneous distribution of modifying ions over the alumina surface. A comparison of nitrogen adsorption isotherms of MgO/(γ+15%χ)-Al₂O₃, Cl/(γ+15%χ)-Al₂O₃, and (γ+15%χ)-Al₂O₃ reveals a slight increase in the fraction of fine (3-to 4-nm) pores and a decrease in size (from 11-12 to 9 nm) of large pores, which indicates near-surface localization of the modifying ions.

The textural characteristics of (γ+15%χ)-Al₂O₃ modified with sulfate ions depend on the sulfate ion content. For example, the sample texture changes slightly when 1-2.5 wt% of the modifier is added. This amount corresponds no more than of that required for monolayer coverage of sulfate ions on the (γ+15%χ)-Al₂O₃ surface. An increase in sulfate ion loading (up to 7.5-10 wt%) leads to creation of a uniformly porous sample with a large pore size (14-15 nm instead of 11-12 nm; Fig. 7(b), curve (5)), but with a small specific surface area (120-135 versus 167 m²/g) and a reduced pore volume (0.40-0.46 versus 0.55 cm³/g) relative to the unmodified sample. These changes may be due to the surface dissolution of alumina to aluminum sulfate during impregnation with sulfuric acid and to precipitation of Al₂(SO₄)₃ into the larger pores upon calcination. Some mesopores may be transformed into ≤3-nm micropores and not observed in the pore-size distribution curves. This possibility is suggested by the difference in the BET and BJH specific surface areas by a factor of 1.25-1.40. The data obtained on the texture of modified alumina samples are in good agreement with the results of [41, 52] from one perspective. However, a monolayer coating by four sulfate groups per nm² [55] usually is achieved at concentrations of 10-12.5 wt% depending on the specific Al₂O₃ surface [52]. A monolayer coating was achieved at 5 wt% SO₄^2– with our (γ+15%χ)-Al₂O₃ sample. The apparent contradiction can be explained by the presence of NO₃^– groups in the initial alumina sample, which decreases the number of centers for sulfate ion stabilization. A decrease in BAS concentration and an increase in LAS concentration were observed for γ-Al₂O₃ obtained by peptizing pseudoboehmite with nitric acid [56]. We used pastes peptized with nitric acid to obtain granular samples of γ-Al₂O₃, (γ+15%χ)-Al₂O₃, and (γ+35%χ)-Al₂O₃.

Many aluminas are used as a catalyst support for the selective oxidation of hydrogen sulfide [18, 34, 35]. Thus, it is interesting to note that alumina also contributes to the activity and selectivity of the overlying catalyst. Hydrogen sulfide oxidation data for aluminas with different crystal modifications indicate that samples containing a mixture of γ-Al₂O₃ with 35-90 wt% χ-Al₂O₃ exhibit the greatest catalytic activity. These samples provide 94%-99% selective sulfur formation at temperatures below 300 ℃, but differ in their selectivity above 300 ℃. Differences in sulfur selectivity can be explained by both the textural (surface area, pore volume, and pore-size distribution) and acidic properties of the aluminas. Pure γ-Al₂O₃ and samples modified with χ-Al₂O₃ are characterized by large specific surface areas of 214 and 160-166 m²/g, respectively. Although these values differ by 25%, the disparity does not clarify the differences in catalytic behavior. Greater H₂S conversion and sulfur selectivity occur with small specific surface area of low-temperature crystal modifications of alumina. The effect of textural properties on sulfur selectivity is more pronounced for α-Al₂O₃, whose textural characteristics differ greatly from those of low-temperature modifications of alumina. Unlike the low-temperature examples of γ-Al₂O₃ and (γ+χ)-Al₂O₃, the sample of α-Al₂O₃ is a wide-pore material with a small surface area. Thus, sulfur selectivity over α-Al₂O₃ decreases from 100% to 90% even at 250 ℃. The effect of catalyst pore structure on the selectivity of hydrogen sulfide oxidation is known [57-59] to be associated with the capillary condensation of sulfur vapor in narrow pores. Capillary condensation was observed in zeolites and on activated carbon [16, 60], where H₂S is oxidized primarily to elemental sulfur. The latter is effective in the selective oxidation of hydrogen sulfide [61].

According to literature data [62], adsorption of hydrogen sulfide on BAS does not activate H₂S, but Lewis acid and basic sites activate and decompose H₂S to SH and H groups. Thus, we decided to investigate the correlation between LAS content and the catalytic oxidation of H₂S by aluminas. Fig. 8(a) displays hydrogen sulfide conversion as a function of the weak LAS content that is characterized by heats of CO adsorption of 32-35 kJ/mol and FTIR bands at 2186-2195 cm^–1. These LAS prevail in low-temperature aluminas. The concentration of weak LAS on the alumina surface unambiguously influences hydrogen sulfide oxidation activity. H₂S conversion increases linearly with the weak LAS content at 200 ℃. H₂S conversion also increases with LAS content at 250 and 300 ℃, reaches a maximum at 800-830 μmol/g, and changes only slightly thereafter. Modification of (γ-15%χ)-Al₂O₃ with MgO confirms this supposition by increasing H₂S conversion from 50% to 62% at 200-250 ℃ accompanied by an increase (from 630 to 1250 μmol/g) in the concentration of LAS. At 300 ℃, the difference between the activity of the MgO-modified and unmodified samples (g-15%c)-Al₂O₃ becomes inconsequential. Surface modification of (γ+15%χ)-Al₂O₃ with sulfate ion (~10 wt%) completely suppresses the formation of weak LAS based on FTIR evidence. Thus, the activity of the sulfate-treated sample is diminished relative to that of the unmodified sample at temperatures below 250 ℃. When the temperature is greater than 300 ℃, the effect of sulfate ion on H₂S conversion and sulfur selectivity becomes less pronounced.

Addition of Cl^– (1 wt%) to (γ+15%χ)-Al₂O₃ significantly deactivates the catalyst. The maximum H₂S conversion is only 35%. According to FTIR spectroscopy, the acidic properties of the alumina surface change relative to those of the unmodified sample. The number of weak LAS increases some and their strength increases slightly (10%-15%), whereas the intermediate LAS content (38 kJ/mol) doubles. Chlorination of γ-alumina by hydrochloric acid is known [2, 41] to enhance the Br nsted acidity of the surface, whereas Al³⁺ LAS become bound by chloride anions [2, 4, 5]. On the one hand, correlation between the surface acidic properties of Cl-modified alumina and its diminished catalytic activity does not support our hypothesis of the key role of LAS in H₂S oxidation. On the other hand, the decrease in catalytic activity is likely caused by the negative effect of Cl^- ion plus the increased concentration of BAS. The presence of chloride ions on the alumina surface and their interaction with Al³⁺ inhibit adsorption of hydrogen sulfide, because the latter is a weaker acid than HCl. BAS do not activate H₂S during adsorption [62].

Fig. 8(b) illustrates the relationship between hydrogen sulfide conversion and BAS content in aluminas. A threefold increase in BAS content (from 60 to 160 μmol/g) on the surface of low-temperature aluminas has no practical effect on the catalytic oxidation of hydrogen sulfide at 250-300 ℃. Although an increase in BAS content of only 6% in (γ+15%χ)-Al₂O₃ is critical for H₂S conversion at 250 ℃, it does not influence conversion at 300 ℃. The large BAS content in γ-alumina and its small content in α-Al₂O₃ both lead to a decrease in catalytic activity. The properties of modified (γ+15%χ)-Al₂O₃ cannot be described using this acidity-performance relationship.

No correlation was established between the ability of aluminas to oxidize hydrogen sulfide and the total concentration of acid sites determined by NH₃-TPD. This is explicable, because intermediate and strong LAS and proton sites are NH₃-TPD visible, whereas weak LAS are NH₃-TPD silent due to weak bond formation with NH₃ [50]. According FTIR, the latter sites are active in the selective oxidation of hydrogen sulfide to sulfur.

To strengthen our hypothesis concerning the effect of weak LAS on low-temperature aluminas on their selective catalytic oxidation of H₂S, a series of the sulfate-modified (γ+15%χ)-Al₂O₃ samples were prepared and characterized by FTIR spectroscopy. The sulfate content was varied from 1.25 to 10 wt%. Assuming that four sulfate groups can be stabilized on 1 nm² of alumina, the 10 wt% SO₄^2– concentrations should be close at monolayer coverage.

According to FTIR data (Table 3 and Fig. 4), two types of weak Al³⁺-LAS (bands at 2195 and 2186 cm^–1) are observed for unmodified (γ+15%χ)-Al₂O₃. The LAS detected in sulfate-modified samples (Fig. 6) can be classified into three groups. The first are weak LAS, which are characterized by a heat of CO adsorption less than 38 kJ/mol (ν_CO = 2180-2198 cm^‑1). The second are those with a heat of CO adsorption close to 39 kJ/mol (n_CO = 2199-2200 cm^–1), and the third are intermediate sites with a heat of adsorption greater than 40 kJ/mol (ν_CO = 2205-2216 cm^–1).

The FTIR-detected concentrations of weak and intermediate LAS on aluminas as a function of sulfate loading are shown in Fig. 9. An increase in SO₄^2– loading of modified (γ+15%χ)-Al₂O₃ results in a sharp decrease in the concentration of weak LAS (ν = 2180-2198 cm^–1). At the same time, the number of intermediate LAS (ν = 2199 and 2210 cm^–1) increases and reaches a maximum value when the SO₄^2– content is near 5 wt%. Compared to the unmodified material, the total concentration of LAS sites decreases significantly in 10% SO₄^2–/(γ+15%χ)-Al₂O₃. The SO₄^2– coverage on the alumina surface (θ_SO4) was evaluated to explain this phenomenon. q_SO4 was calculated on the assumption that an SO₄^2‑ anion occupies an area of 7 × 7 Å. Fig. 9 shows that the maximum concentrations of intermediate LAS (ν = 2199 and 2210 cm^–1) are observed at θ_SO4 near 1.0–1.1 monolayer. Prior to reaching monolayer coverage, sulfate ions on an alumina surface have been shown [55, 63, 64] to exist as isolated, multicentered, and bridging groups. Note that the monolayer on our (g+15%c)-Al₂O₃, unlike γ-Al₂O₃[63, 64], is achieved when the sulfate ion content is about 5 wt%. Polymeric sulfate species, which form at sulfate loadings above 5 wt% [64], predominate in multilayers [63, 64]. There could be trace amounts of Al₂(SO₄)₃·xH₂O crystallites, which can be identified by XRD [52, 64]. We have not detected a crystalline aluminum sulfate phase by XRD even in 10%SO₄^2–/(γ+15%χ)-Al₂O₃, although the aluminum sulfate particles could be amorphous. Therefore, the decrease in concentration of deprotonated surface sites with increasing SO₄^2– can be accounted for by a blocking of the surface sites by sulfate anions via formation of polymeric sulfate species and amorphous aluminum sulfate. This conclusion is consistent with Yang's results [63]. Their analysis showed that the acidity of sulfate species on an alumina surface decreases as surface sulfates > multilayer sulfates > aluminum sulfate. Surface sulfates possess greater acidity due to the induction of Lewis superacid sites, multilayer sulfates have weaker acidity due to the increased concentration of Br nsted acid sites, and crystallized aluminum sulfate is neutral [63].

We next seek to find a relationship between H₂S conversion and the concentration of LAS sites of different strength. Fig. 10 shows that the correlation between H₂S conversion and the concentration of weak LAS (ν = 2186 cm^–1) is similar to that obtained for alumina with different crystal modifications (Fig. 8). Activity increases with weak LAS content and reaches an optimum value near 250-300 μmol/g at 200 ℃ and 400-500 μmol/g at 250 ℃ (Fig. 10(A)). Further increase in the weak LAS concentration does not result in a noticeable change in catalytic activity. Nonlinear trends are observed for intermediate LAS characterized by ν = 2199 and 2210 cm^–1 (Fig. 10(B), and (С)). H₂S conversion decreases linearly at first with increasing LAS concentration, but then reaches a minimum and begins to grow with increasing LAS content. Careful analysis of these dependences (Fig. 10(B), and (C)) shows that the minimum values of H₂S conversion correspond to modified aluminas with a sulfate content of 7.5 wt% and above. This sulfate coverage amounts to more than a monolayer. If these points are removed from the experimental curves and only 1.25-5 wt% sulfate concentrations are considered, H₂S conversion is practically independent of the intermediate LAS concentration at 200 ℃ and changes only slightly at 250 ℃. Therefore, we assume that the low activity of highly concentrated (7.5–10 wt%) SO₄^2–-modified alumina samples is connected with a change in phase composition, specifically formation of a multilayer sulfate film and highly dispersed X-ray amorphous aluminum sulfate. The latter has no acidity [63].

The experimental data confirm that weak LAS enable the catalytic oxidation of H₂S by unmodified aluminas with different crystal modifications (γ, χ, α) and aluminas modified by Mg²⁺, Cl^-, and SO₄^2- ions. The effect depends strongly on temperature with the greatest effect observed below 300 ℃.

In the catalytic reactions of hydrogen sulfide oxidation [12, 18, 20] and decomposition [65], adsorption of H₂S on the catalyst surface is thought to be necessary for H₂S activation and its transformation to sulfide or sulfite/sulfate complexes [62, 66]. Although H₂S adsorption on g-Al₂O₃ has been studied in detail [62, 66-68], there are debates concerning the nature of the sites involved in H₂S adsorption. H₂S adsorbs in molecular form on proton sites of γ-Al₂O₃ [62, 67]. It adsorbs dissociatively on basic sites such surface oxygen [67, 69]. H₂S also is thought to adsorb both dissociatively and in molecular form on strong LAS. Physical adsorption of hydrogen sulfide on the surface of oxide catalysts can be accompanied by its oxidation to surface sulfite/sulfate complexes [22, 24, 70]. The bridging lattice oxygen of the catalyst or oxygen adsorbed on the surface is consumed in H₂S oxidation [71]. However, oxidation on g-Al₂O₃ is hindered by the low mobility of surface oxygen [68].

Fig. 11 displays FTIR spectra of H₂S adsorbed on the surface of the aluminas. The results of the experiments are summarized in Table 4. The spectra of H₂S-treated aluminas at 20 ℃ exhibit absorption bands at 1330-1333, 1557-1575, and 2555-2588 cm^–1 (the last is not shown in Fig. 11). Free H₂S has three vibrations in the IR region: 2722 (ν₁), 1215 (ν₂), and 2739 (ν₃) cm^–1 [72]. These shift due to interaction with the surface of a heterogeneous catalyst [68-75]. The bands at 1330-1333 and 2555-2588 cm^–1 are the bending and stretching vibrations of physically adsorbed hydrogen sulfide or chemisorbed HS, respectively [68-71, 73-75]. The band at 2588 cm^–1 in the spectrum of α-Al₂O₃ treated with H₂S indicates that the broad band at 1330-1368 cm^–1 belongs to vibrations of physically adsorbed hydrogen sulfide. The band at 1557-1575 cm^–1 is not unambiguously assigned to a specific vibration of any S-containing species, and its origin is still hotly debated [68, 71-73]. We repeated the H₂S adsorption experiment several times and always observed the band at 1557-1575 cm^–1 in the FTIR spectra of H₂S-treated aluminas at 20 ℃. The band decreases in intensity or disappears with increasing temperature. The band at 1550-1568 cm^–1 accompanied by 1341, 1625, 2556, and 3400 cm^–1 bands was observed early in the spectra of γ-Al₂O₃ [73], MoS₂/γ-Al₂O₃ [73], and Cu-ZSM-5 [75] exposed to H₂S at room temperature. Dalla Lana et al. [71] reported bands at 1330 and 1568 cm^–1 in the spectra of γ-Al₂O₃ pretreated with H₂S or SO₂ and heated at 400 °С. The FTIR spectra of oxidized and reduced aluminas exhibited bands at 2560, 1568, and 1345 cm^–1 after alumina treatment by H₂S [69]. Bands at 1330 and 1565 cm^–1 were assigned in [69, 71] to the S-O vibration of chemisorbed sulfur atoms, which formed SO₂-like species with oxide ions of the alumina lattice. Slager and Amberg [73] assigned a band at 1568-1575 cm^–1 to Al-O vibrations, whose frequency changed upon adsorption of different molecules (H₂S, CO, HCOOH, and C₂H₆). Saur et al. criticized both points of view [68] in assigning the 1570 cm^–1 band to a carbonate mode [76]. It should be noted that the FTIR spectra in [76] did not display a 1570 cm^–1 band for CO₂-treated alumina, but did show this feature after H₂S adsorption on CO₂-treated alumina. Despite disagreement as to the origin of the 1565-1575 cm^–1 band, most authors have reached the conclusion [68-76] that surface structures with a coordination bond between Al and S are formed during treatment of alumina with H₂S, CH₃SH, and (CH₃)₂S. Saur et al. [68, 76] have suggested that strong and weak Lewis acid-base couples can break the H-S bond in H₂S and that dissociative adsorption of H₂S on the former and latter leads to formation of Al-S and O-H bonds. The reactivity of weak Lewis acid-base couples predominates [68] when the adsorbate has acid properties. Thus, the bands at 1330-1333, 1557-1575, and 2555-2588 cm^–1 are assigned to H-bonded and chemisorbed H₂S.

In addition to the features at 1330, 1557, and 2550 cm^–1, the spectrum of H₂S-treated γ-Al₂O₃ at 20 ℃ shows absorption bands at 1120, 1180, 1262, and 1350 cm^–1. These bands correspond to the vibrations of sulfite/sulfate surface complexes and indicate that partial oxidation of H₂S adsorbed on γ-Al₂O₃ occurs at temperatures as low as 20 ℃. The absorption bands at 1120 and 1350 cm^–1 are attributed to the symmetric and antisymmetric vibrations of bonds in a multicentered sulfate-containing structure (AlO)₃SO [20, 55, 64, 70, 77]. The bands at 1180 and 1260 cm^–1 are attributed to the symmetric and antisymmetric vibrations of bridging surface sulfur atoms within (AlO)₂SO₂ centers [55, 70, 77] that contain free SO₂ unperturbed by hydrogen bonds. Other aluminas do not show bands at 1120/1350 and 1180/1262 cm^–1 in their FTIR spectra after H₂S adsorption. Monomeric bi-and tridentate sulfates (1120/1350 cm^–1) form on the surface of χ-modified aluminas after heating at 200-300 ℃, whereas inorganic bridging sulfate (1180/1262 cm^–1) is not observed even after thermal treatment at 300 ℃.

As the H₂S-treated aluminas were heated at 100 ℃, the 1332 and 1557 cm^–1 bands vanished from the spectra of γ-Al₂O₃ and α-Al₂O₃, whereas these bands did not disappear completely in the case of χ-modified samples. For example, the 1332 cm^–1 band diminished in intensity by a factor of 10 after heating H₂S-treated (γ+35%χ)-Al₂O₃ and MgO/(γ+15%χ)-Al₂O₃ at 100 ℃, but the band was still observed in the spectra of these samples after heating at 300 ℃. The disappearance or decrease in intensity of the 1332 and 1557 cm^–1 bands may indicate the oxidation of physically adsorbed hydrogen sulfide and formation of sulfur and water via the reaction, H₂S + [O] → S_n + H₂O. As discussed below, water was found in the FTIR spectra (as a band at 1650 cm^–1 [71]) of g-Al₂O₃ and χ-modified samples after heating at 300 and 100 ℃, respectively.

The intensity of the bands at 1350 and 1262 cm^–1 become more intense in the spectrum of H₂S-treated γ-Al₂O₃ after heating at 200 ℃. The bands become stronger at higher temperatures indicating an increase in the amount of multicentered and bidentate sulfate species, respectively. We have identified two new pairs of bands in these spectra. The first pair at 1160-1170 and 1315-1320 cm^–1 is characterized by asymmetric band shapes. Their frequencies are assigned to symmetric and asymmetric stretching vibrations of a second type of bridging sulfur [77]. The second pair is located at 1232 and 1430-1445 cm^–1. Bensitel and co-authors [78] suggested that bands with ν(S=O) ≥ 1400 cm^–1 are typical of a polynuclear sulfate complex such as S₂O₇^2–. The FTIR spectrum of sulfated zirconia is characterized by a band at 1400-1405 cm^–1 at high sulfate loading [78]. However, the spectrum of SO₂ adsorbed on alumina was found by Bensitel et al. [79] to display ν(S=O) near 1390 cm^–1. Polymeric sulfate species, which are characterized by absorptions at 1538 and 1488 cm^–1, were found by Mekhemer [64] in alumina containing 10 wt% sulfate. We assume that polynuclear sulfate complexes exhibit characteristic absorption bands at 1405-1445 and 1230-1237 cm^–1 and that their formation is possible during heating of alumina with adsorbed hydrogen sulfide at 200 ℃.

Polynuclear sulfate complexes characterized by bands at 1230 and 1405-1450 cm^–1 were observed on heating χ-modified aluminas at a lower temperature (100 ℃) than γ-Al₂O₃ (200 ℃). The polynuclear sulfate content in α-Al₂O₃ was smaller than that in other crystal modifications of alumina. This may be due to differences in the exposed crystal facets of the γ-Al₂O₃ and χ-Al₂O₃ structures, which influence the adsorption/desorption pathways of H₂S. Morterra et al. [82] showed in sulfated zirconia that species with ν(S=O) equal to 1350-1375 cm^-1 are localized at crystallographic defects (edges), whereas those with ν(S=O) at 1380-1400 cm^–1 are located mainly on regular patches of low-index crystal planes. This result can be extended to aluminas. γ-Al₂O₃ is known [45-47] to contain tetrahedrally coordinated Al³⁺ at crystallographic defect sites as well as octahedrally coordinated Al³⁺ ions at regular defects on low-index crystal faces. Based on FTIR data, these tetrahedral and octahedral Al³⁺ configurations are LAS, which are characterized by high-(2210-2240 cm^–1) and low-frequency (2180-2205 cm^–1) bands, respectively, in the spectra of CO adsorbed on alumina [38, 40-45]. According to Table 3, tetrahedrally coordinated Al³⁺ ions belong to γ-Al₂O₃, while octahedrally coordinated Al³⁺ predominates in both γ-Al₂O₃ and samples modified with 15-35 wt% χ-Al₂O₃. Thus, formation of polynuclear sulfate species (1405-1445 cm^–1) during heating of H₂S-treated aluminas is promoted by weak LAS, which are octahedrally coordinated Al³⁺ ions.

A further increase in temperature to 300 ℃ did not result in significant changes in sulfur-containing species in H₂S-treated aluminas. There are different types of isolated sulfate groups and polynuclear sulfate complexes as well as tridentate surface SO_x compounds (1160-1180/1350 cm^–1). A greater polynuclear sulfate content (1230/1445 cm^–1) is observed at higher heating temperatures. The spectrum of (γ+35%χ)-Al₂O₃ is characterized by an unstructured absorption in the low frequency region (1100-1200 cm^–1), which is associated with the vibration of surface sulfate complexes.

The spectra of H₂S-treated γ-Al₂O₃ samples evacuated at room temperature illustrate the behavior of bands connected with sulfate groups and water. The results show that the sulfate complexes formed over H₂S-treated aluminas during heating at 100–300 ℃ are stable. A broadening of the vibrational frequencies of SO_x surface compounds (1159 cm^–1) and inorganic sulfate complexes (1170 and 1260 cm^–1) in the a-Al₂O₃ spectrum can be caused by water adsorption that accompanies a decrease in temperature.

We conclude that physically adsorbed hydrogen sulfide is oxidized on alumina surfaces to stable surface sulfate groups including multicentered SO_x compounds (bands at 1120/1350 cm^–1), isolated inorganic sulfates (1180/1260 and 1160/1314 cm^–1), and polynuclear sulfates (1230-1237/1405-1445 cm^–1) at elevated temperatures (> 100 ℃). In contrast to the χ-modified samples, oxidation of hydrogen sulfide over γ-Al₂O₃ involving formation of surface SO_x compounds begins at room temperature. Physically adsorbed SO₂ has been postulated by Prette [80] and Steijns [81] to be an intermediate in H₂S oxidation on γ-Al₂O₃ at 100 ℃. Moreover, H₂S oxidation to adsorbed SO₂ proceeds slowly on γ-Al₂O₃ even when H₂S is oxidized by molecular oxygen. Although further recovery of adsorbed SO₂ with H₂S is rapid, the products are elemental sulfur and water. Water forms preferentially at 300 ℃.

H₂S oxidation on γ-Al₂O₃ and α-Al₂O₃ results in two types of isolated inorganic sulfates with different bridging structures, which can be assigned to intermediate and weak strength LAS on the alumina surface. For 15%-35% χ-containing samples, H₂S adsorption leads to formation of a single type of inorganic bridging sulfate probably due to weak LAS. The χ-modified samples are more active, because polynuclear sulfates form on their surface at only 100 ℃, and the intensity of absorption bands is greater than is true for γ-Al₂O₃ and α-Al₂O₃. The presence of polynuclear surface sulfates also was detected by DTA as a mass loss at 920-960 ℃ in the TG/DTG curves of spent (γ+35%χ)-Al₂O₃ tested for H₂S oxidation at 300 ℃ for 50 h (Fig. 12(a)). Polynuclear surface sulfates were found by Yang [63] to decompose at 950 ℃ unlike crystallized aluminum sulfate (800-860 ℃) and multilayer sulfate (630 ℃). The latter species are typical of (γ+15%χ)-Al₂O₃ modified by 10 wt% sulfate (Fig. 12(b)). Surface sulfates are characterized by a wide peak near 950 ℃, which broadens at lower temperatures. It is due to different interactions between polynuclear sulfate and the alumina surface.

The phase composition of alumina samples (α, γ, and (γ+χ)-Al₂O₃) strongly influences their activity in hydrogen sulfide oxidation. The catalytic activity of γ-Al₂O₃, including samples modified with Mg²⁺, increases in the presence of χ-Al₂O₃.

The acidic properties of α-Al₂O₃, γ-Al₂O₃, and (γ+χ)-Al₂O₃ surfaces were studied by FTIR spectroscopy with CO adsorption and NH₃-TPD. For α-and γ-Al₂O₃ samples, FTIR spectra reveal the presence of weak and intermediate strength LAS with CO vibrational frequencies greater than 2186 and 2200 cm^–1, respectively. Only weak LAS characterized by an absorption band at 2186-2200 cm^–1 are found on (γ+χ)-Al₂O₃. Introduction of MgO into (γ+15%χ)-Al₂O₃ reduces the strength of the weak LAS, as observed by a band shift from 2186 to 2181 cm^–1, but significantly increases their number.

Interaction of hydrogen sulfide with α, γ, and γ+χ crystal modifications of Al₂O₃ at 100–400 ℃ results in the appearance of surface SO_x compounds (1120/1350 cm^–1 absorption bands), inorganic sulfates (1160-1180 and 1260-1314 cm^–1 bands), and polynuclear sulfates (1230-1237 and 1400-1445 cm^–1 bands). For α-Al₂O₃ and γ-Al₂O₃, which possess weak and intermediate LAS, additional inorganic sulfate species with high frequency bands at 1265 and 1300-1315 cm^–1 are observed.

Based on data from FTIR spectroscopy and a comparison of the catalytic activity and selectivity of alumina samples, we suggest that hydrogen sulfide is adsorbed predominantly on weak LAS. Alumina samples containing χ-phase Al₂O₃ and/or modified with Mg²⁺ are characterized by a large concentration of weak LAS and exhibit enhanced catalytic activity in H₂S oxidation. When alumina is treated with a solution of sulfuric acid, strong LAS appear and the LAS population decreases significantly. Modification of alumina with hydrochloric acid has a limited effect on LAS strength. Weak LAS are retained and double in number relative to unmodified material, but the chloride-treated sample has Al-Cl rather than Al-O(OH) bonds. Alumina samples modified by sulfate and chloride anions exhibit poor catalytic activity in hydrogen sulfide oxidation.

This work was conducted within the framework of Program of the Russian Academy of Sciences and the Federal Agency for Scientific Organizations (state-guaranteed order for BIC, Project АААА-А17-117041710086-6).

The authors are grateful to T. Ya. Efimenko for the aid with estimation of adsorption characteristics, V.A. Ushakov for investigation of aluminas by XRD, and G.S.Livak for DTA-TG experiments.