454
J. Phys. Chem. 1985,89,454-451
perhaps the subsequent disappearance on desorption at high temperatures, of such A1-S species owing to the strong absorption of alumina below 1000 cm-'. A technique such as X-ray photoelectron spectroscopy could perhaps be fruitfully used for this purpose. Further evidence for the presence of two different types of adsorbed H2S species is provided by the work of Lunsford et al.I3 on the EPR spectrum of N O adsorbed on y-alumina wherein it was found that H2S, if preadsorbed on the alumina, effectively blocked the sites which yielded the characteristic 27Alhyperfine structure in the nitric oxide spectrum. It was concluded that this was due to adsorption of H2S on the A1 ion sites which thereby prevented the NO from using these same sites. Lunsford et al. also found that, after degassing a sample poisoned with H2S for 1 h at 300 "C, only about 50% of the N O spectrum could be (13) Lunsford, J. H.; Zingery, L. W.; Rosynek, M. P. J. Catal. 1975,38, 179.
developed, suggesting that half of the aluminum ion sites were still blocked by the H2S, and even heating the sample at 300 "C did not remove the H2S and liberate the N O adsorption sites. These observations are consistent with our interpretation. The restoration of 50% of the nitric oxide spectrum is attributed to the unblocking of some of the Al' sites through desorption of the undissociatively adsorbed form of H2Sat 100 OC. The Al+ sites which are still blocked after degassing at 300 OC are probably those on which HIS had dissociatively adsorbed, leading to the formation of A1-S species, which, as discussed earlier, being very stable, is not expected to desorb or break up to temperatures below 300 OC.
Acknowledgment. We thank the Natural Sciences and Engineering Research Council of Canada for support of our research. We also thank R. W. Tower of the Alberta Research Council, Edmonton for preparation of the sample pellets and the A.R.C. for the loan of the infrared cell. Registry No. HIS, 7783-06-4;alumina, 1344-28-1.
Claus Catalysis. 3. An FTIR Study of the Sequential Adsorption of SO, and H,S on the Alumina Catalyst Arunabha Datta and Ronald G. Cavell* Department of Chemistry, University of Alberta, Edmonton, Alberta, Canada T6G 2G2 (Received: June 25, 1984)
The sequential adsorption of H2S and SO2on y-alumina catalyst activated at 400 OC has been studied by Fourier transform infrared (FTIR) spectroscopy. When incremental amounts of H2S were added to a sample of alumina upon which SO2had been preadsorbed, the Claus reaction took place but the species responsible for the band at 1055 cm-I (characteristic of SO, adsorbed on alumina) showed very low reactivity toward H2S. No infrared bands due to adsorbed H2S were observed, but this cannot be taken as conclusive evidence for the absence of adsorbed H 2 0 . On addition of SO2 to an alumina sample on which HzS had been preadsorbed, the Claus reaction also proceeded but to a much lesser extent probably because of the dissociative adsorption of part of the preadsorbed H2S. Also, in this case bands due to adsorbed SO, were observed throughout the reaction. The dependence of the rate of the Claus reaction on the activation temperature of the catalyst, the nature and mechanism of catalyst poisoning, and an alternative approach for carrying out the Claus reaction are also discussed.
Introduction In spite of the commercial importance of the modified Claw reaction for sulfur recovery
2H2S + SO2
+3s + 2H20 alumina
the mechanism of this reaction is still not established. Attempts have been made by various authors to understand the mechanism through infrared spectroscopic studies of the adsorption of SO, and HIS on alumina, both individually and in combination. Earlier work led to the conclusion that SO2was adsorbed on alumina in two or possibly three different one of which was very strongly adsorbed. In contrast, H2S was believed to be weakly adsorbed in only one form.'sbs Accordingly, Karge et al.4 suggested that the Claus reaction proceeded between adsorbed SO2 and gas-phase H,S. It has also been proposed4 that SO, radicals which form when SO2 is adsorbed on alumina were responsible (1) Deo, A. V.;Dalla Lana, I. G.;Habgood, H. W. J . Catal. 1971,21,270. (2) Fiedorow, R.; Dalla Lana, I. G.;Wanke, S. E. J . Phys. Chem. 1978, 82, 2474. (3) Chang, C. C. J . Catal. 1978, 53, 374. (4) Karge, H. G.;Tower, R. W.; Dudzik, Z.; George, Z. M. Stud. Surf. Sei. Caral. 1981, 7 , 643. (5) Karge, H. G.;Dalla Lana, I. G. J. Phys. Chem. 1984, 88, 1538. (6) Sager, T. L.; Amberg, C. H. Can. J . Chem. 1972, 50, 3416. (7) Lavalley, J. C.; Travert, J.; Laroche, D.; Saw, 0. C. R. Hebd. Sceances Acad. Sei., Ser. A 1977, 285, 85; Chem. Abstr. 1978, 88, 55320b. (8) Saur, 0.; Cherreau, T.; Lamote, J.; Travert, J.; Lavalley, J. C. J . Chem. SOC.,Faraday Trans. 1 1981, 77, 427.
for Claus conversion and that the species responsible for the band a t 1080 cm-' in the spectrum of adsorbed SO, might be an important intermediateS in the Claus reaction. Our recent studyg of the adsorption of SO, on alumina using FTIR spectroscopy has shown however that SO, may be adsorbed in as many as five different forms with varying strengths of adsorption. We also foundlo that, in contrast to earlier reports, HIS is strongly adsorbed on alumina in not one but two different forms. Herein we report the next logical step in the study of Claus catalysis, the study of the sequential adsorption and reactions of SO, and H2Son alumina undertaken with the hope of identifying which of the adsorbed SO2 and H,S species are reactive in the Claus reaction. We have studied both approaches: the reaction of SO2with preadsorbed H2Son alumina and the reaction of H2S with preadsorbed SO2. Experimental Section The experimental procedures used have been described prev i o ~ s l y .As ~ described therein, the Kaiser Alumina wafers (approximate "thickness" of 25.5 mg/cm2) were activated by heating them in the cell" under vacuum torr) for 16 h at the temperatures indicated. The BET surface area of a typical wafer (9) Datta, A.; Cavell, R. G.; Tower, R. W.; George, Z. M. J . Phys. Chem., accompanying article in this issue. (10) Datta, A.; Cavell, R. G.J. Phys. Chem., preceding article in this issue. (1 1) The Kieselev type. cell used herein is described in detail by: Karge. H. G. 2. Phys. Chem. (Wiesbnden) 1971, 76, 133.
0022-365418512089-0454$01 SO10 0 1985 American Chemical Society
Adsorption of SO2 and H2S on Alumina Catalyst
The Journal of Physical Chemistry. Vol. 89, No. 3, 1985 455 1056
TABLE I: Treatment of SO, Adsorbed on Alumina with H2S
treatment of alumina 0.5 mmol of so2 SO2 0.05
+
band position" (absorbance) 1055 (2.45)
1327 (0.49)
1189 (0.68)
1148 (1.13)
1054 (2.43)
1324 (0.38)
1190 (0.66)
1147 (1.12)
1054 (2.41)
1322 (0.28)
mmol of
H2S
SO2 + 0.1 mmol of
H2S
+ 0.3
1056 (2.43)
SO2 + 0.5 mmol of
1056 (2.01)
SO2
mmol of HIS H2S
SO2 + 1.0 mmol of
1059 (1.82)
HIS SO2 1.5 mmol of
1058 (1.80)
+
1141 (1.13)
H2S "Positions in cm-I. Range 1400-1000 cm-I. was 296 m2/g, and the average mass of a wafer was approximately 98 mg. The infrared spectra were recorded on a Nicolet 7199 Fourier transform infrared spectrometer using a Globar source, a germanium-coated KBr beam splitter, and a mercury-cadmium telluride detector operating at 77 K. The mirror velocity was 0.8 cm-'. Usually 1000 consecutive scans were summed to obtain a spectrum with a resolution of 4 cm-' over the spectral range. A spectrum of the empty infrared cell was used as the instrument background, and all single-beam spectra of the samples were ratioed to this background. The spectrum of the adsorbed gases was obtained by a 1:l subtraction of a stored spectrum of the activated alumina sample from that of the alumina sample with adsorbed gas. All spectral subtraction and subsequent plotting of the spectra were carried out by the Nicolet 1180 computer which controls the instrument. The peak positions noted on the graphs were located on the primary Fourier transformed interferogram and not derived from the plotted curves shown in the figures which were in general subject to an 11-point smoothing treatment except for the SH stretching region which was not smoothed. In some cases changes in spectral gain due to reduced transmission of the sample as a result of increased gas load rendered impossible display of the spectra on a universal scale. Details are given in the figure captions. The alumina sample was activated by heating at 400 OC for 16 h. In the first case 0.5 mmol of SO2 was added to the alumina wafer at room temperature and allowed to adsorb for 1 h to ensure saturation. This was followed by the addition of incremental amounts of H2S. For the reverse sequence, 0.5 mmol of H2S was allowed to adsorb (to saturation) at room temperature for 1 h on the activated alumina wafer and incremental doses of SO2 were then added. In both cases reactions occurred quickly upon admission of the second gas, but no quantitative rate data could be obtained from the experiments.
Results and Discussion Adsorption on Alumina of SO2 Followed by H2S. The adsorption behavior of H2S on an alumina sample on which SO2 has been preadsorbed is shown in Figure 1. On initial adsorption of 0.5 mmol of SO2,all bands due to different adsorbed species present on alumina9 are observed. The stoichiometry of the Claw reaction requires 1.0 mmol of H2Sto react with the adsorbed SO2. On the addition of the first dose of 0.05 mmol of H2S there is an observable decrease in the intensities of the bands at 1327 and 1145 cm-'. The fact that the absorbance value (Table I) for the 1145-cm-I band does not initially decrease however is probably because this band also contains contributions from the 1141- and 1134-cm-' bands and it is only the 1145-cm-' component of the
1400
1300
1200
1100
1000
WAVENUMBERS (cm-1)
Figure 1. FTIR spectrum (S-O stretching region) illustrating the treatment of SO2adsorbed on alumina with H2S. The first spectrum (1) shows alumina saturated with 0.5 mmol of adsorbed SO2. Subsequent spectra illustrate the effect of addition of (2) 0.05, (3) 0.15, (4) 0.3, (5) 0.5, (6) 1.0, and (7) 1.5 mmol of H2S. The absorbance scale is arbitrary, and the spectra have been displaced relative to one another for clarity of representation. The intensity of the major band (1054-1056 cm-I) is virtually constant throughout the series of H2S treatments as shown by the absorbance values in Table I.
band that decreases initially. The 1189- and 1055-cm-' bands on the other hand are unaffected. On addition of the next dose of H2S, the 1189-cm-' band disappears and the 1145-band shifts to 1141 cm-I. The 1055-cm-' band is not significantly reduced in intensity (Table I) until nearly 0.5 mmol of H2S has been added, and this band, although reduced, is still present even when an excess (1.5 mmol) of H2S has been added. The spectrum of the region containing the bending mode of water (Figure 2) indicates that with the addition of progressive amounts of H2S the Claus reaction takes place, leading to the formation of increasing amounts of water as shown by the steady increase of absorbance through the sequence. No infrared bands corresponding to the stretching or bending modes of adsorbed H2S were observed during this experiment. It appears from the above observations that physically adsorbed SO2 with bands at 1327 and 1145 cm-' is the first to react with H2S. This is followed by the SO2species responsible for the band at 1189 cm-I, and finally the chemisorbed species with bands a t 1322 and 1141 cm-' are consumed in the Claus reaction. It is significant that the 1055-cm-' band is still present even after the addition of an excess of HIS. Similar results have been also recently reported by Karge et al.,5 who found low reactivity of the species responsible for the 1055-cm-l band at room temperature, but in that case the reactivity increased on heating to around 200 OC. The 1055-cm-' band appears to be due to species that , ~ it would appear that are very strongly held on a l ~ m i n a and heating to around 200 "C is necessary to "loosen it up" and allow it to interact with H2S. The absence of infrared bands due to the stretching and bending modes of H2S indicates that undissociatively adsorbedlo H2S is not present. The possibility of dissociative adsorption of HIS however cannot be ruled out since the A1-S species resulting from such adsorption cannot be monitored by infrared spectroscopy.
456
The Journal of Physical Chemistry, Vol. 89, No. 3, 1985
1.60
4
Datta and Cave11
1633
040/III.5-3 0.00 1800
1600
2
1400
WAVENUMBERS (cm-')
Figure 2. FTIR spectrum (water bending region) illustrating the treatment of SOz adsorbed on alumina with H2S. The first spectrum (1) shows alumina saturated with 0.5 mmol of adsorbed SO2. Subsequent spectra were taken after addition of (2) 0.05, (3) 0.15, (4) 0.3, (5) 0.5, (6) 1.0, and (7) 1.5 mmol of H2S.
However, dissociation of HIS is unlikely because the AI ion sites necessary for such adsorption would most probably be totally blocked by the preadsorbed SO,. Adsorption on Alumina of H2S Followed by SO,. On preadsorbing H2S on alumina followed by the addition of incremental amounts of SO2,it can be seen that the S-H stretching band (Figure 3), which was initially present on the adsorption of 0.5 mmol of H2S, gradually disappears. Correspondingly, the bending vibration of adsorbed HIS a t 1333 cm-' (Figure 4) decreases initially and then increases because, while the adsorbed H2S is consumed in the Claus reaction with progressive addition of SO,,bands due to adsorbed SO2 begin to appear and hence the 1331-cm-' band which arises from the antisymmetric S-0 stretch of adsorbed SO? appears and contributes to the intensity in this region. After the addition of 0.5 mmol of SOz (which is in excess of the 0.25 mmol stoichiometrically required to react with the 0.5 mmol of H2S initially adsorbed) the bending mode of adsorbed HIS has completely disappeared and only the antisymmetric S-0 stretching mode of physically adsorbed SO, is observed at 1326 cm-I. At the same time it can also be seen that even on the addition of only 0.1 mmol SO,,bands at 1060 and 1132 cm-I due to adsorbed SO, begin to appear. In the water bending region of the spectrum (Figure 5 ) the band which develops on initial adsorption of H2S is due to dissociative adsorption.1° On subsequent addition of SO2,the intensity of the water band increases gradually due to the formation of (adsorbed) water in the Claus reaction. However, a comparison with Figure 2 shows that the increase in intensity of the water band is less pronounced, indicating that in this case the extent of the Claus reaction is reduced. The appearance of bands at 1060 and 1132 cm-' due to adsorbed SO2even on the addition of only 0.1 "01 of SO2 indicates that the preadsorbed H2Sdoes not occupy all the sites available on alumina. It has been suggestedLothat H2S is adsorbed on either isolated metal sites or on metal sites adjacent to a hydroxyl and an oxide ion. The 1060- and 1132-cm-l bands on the other hand are believed9to be due to a sulfite-like species resuking from initial adsorption of SO2on an oxide site followed by migration to an adjacent metal ion. Such sites must therefore not be suitable for the adsorption of H,S.
2800
'
2600
2400
t
WAVENUMBERS (cm-')
Figure 3. FTIR spectra (SH stretching region) illustrating the sequential treatment of HIS adsorbed on alumina with SO2. The first spectrum (1) shows alumina saturated with 0.5 mmol of adsorbed HIS. Subsequent spectra illustrate the effect of the addition of (2) 0.05, (3) 0.1, (4) 0.2, (5) 0.3, and (6) 0.5 mmol of SO2. The absorbance scale is arbitrary, and the curves are displaced with respect to one another to clearly display the decrease in the 2579-2576-cm-' absorption band as treatment proceeds.
2'60
1
1050
2.104
/
1050
0.60
0.10 1400
1200
1000
WAVENUMBERS (cm-1)
Figure 4. FTIR spectra (SOstretch and H2S bending region) illustrating the sequential treatment of H2Sadsorbed on alumina with SO2. The first spectrum (1) shows alumina saturated with 0.5 mmol of adsorbed H2S. Subsequent spectra show the effect of the addition of (2) 0.05, (3) 0.1, (4) 0.2, (5) 0.3, and (6) 0.5 mmol of SO2.
J. Phys. Chem. 1985,89,457-459
451
700 "C as compared to one activated at 400 "C although the surface area of both the samples is similar. This could perhaps be due to the formation of what appears to be a fifth form of adsorbed SO? on the 700 "C activated sample. Moreover, dissociative adsorption of H2Sappearsl0 to take place to a much lesser extent on the sample activated at 700 "C. Since the AI-S species resulting from such dissociation are not expected to take part in the Claus reaction and may in fact be responsible for the poisoning of sites, the rate of the reaction would be higher for the 700 "C activated sample. (3) Contrary to an earlier suggestion: there is evidence that the Claw reaction also takes place between SO2 and adsorbed H2S, and it is therefore not essential that SO2 be present in an adsorbed form on the catalyst. I 1622 ( 4 ) The poisoning of Claw catalyst by sulfation is well established,12but the mechanism of sulfate formation is still not clear.13 It has been reported" however that, in laboratory sulfation studies, 2 a mixture of HIS and O2at 240 "C on alumina produced maximum sulfation. It is therefore likely that the AI-S species possible 1 formed during dissociative adsorption of HIS on alumina may be responsible for the formation of aluminum sulfate since the Claus I 1800 1600 1400 reaction is generally carried out at around 250 "C in the presence of atmospheric oxygen. WAVEN UMBERS (cm-l) (5) Previous infrared studies of simultanqus addition of Figure 5. FTIR spectra (water bending region) illustrating the sequential stoichiometric amounts of H2S and SO2 to an alumina catalyst treatment of Hfi adsorbed on alumina with SO2. The first spectrum (1) yielded no evidence4 of adsorbed H2S. However, as mentioned shows alumina saturated with 0.5 mmol of adsorbed H2S. Subsequent earlier, it is not possible to monitor the dissociative adsorption spectra show the effect of the addition of (2) 0.05, (3) 0.1, (4) 0.2, ( 5 ) of H2S on alumina by IR spectroscopy. It is likely therefore that, 0.3, and (6) 0.5 mmol of SO1. The curves are displayed on an arbitrary during the operation of the Claw process when both H2S and SO2 absorbance scale and displaced with respect to one another. The increase are present, some dissociative adsorption of H2Soccurs with both of the signal in the 1618-1626-cm-l region is shown by the increase of reactants competing effectively for the metal sites. Our studies peak area above the background. indicate that to avoid dissociate adsorption of H2Swhich appears to both reduce the yield of the Claw reaction and be responsible The fact that the Claus reaction also takes place to a much lesser for the poisoning of the catalyst, it may be beneficial to carry out extent when H2S is preadsorbed on alumina is probably due to the Claus reaction in such a way that SO2is preadsorbed on the the dissociative adsorption of a portion of the HIS, and the recatalyst, allowing subsequent stoichiometric reaction with H2S. sultant AI-S species do not participate in the Claus reaction. Only In this way sites necessary for the dissociative adsorption of H2S the undissociatively adsorbed H2Smolecules appear to react with would be blocked by the preadsorbed SOz. the SO2. 1626
E
Conclusions (1) Of all the adsorbed species formed during the adsorption of SOz on alumina, the species responsible for the band in the neighborhood of 1060 cm-' is the least reactive toward H2S. The reactivity of these species a t higher temperature is probably one of the reasons why the Claw reaction is most efficiently carried out a t around 250 "C. (2) It has been reported4 that the initial rate of the Claw reaction is higher in the case of an alumina sample activated at
Acknowledgment. We thank the Natural Sciences and Engineering Research Council of Canada for financial support of this work. We thank R. W. Tower for preparation of the catalyst wafers and the Alberta Research Council for the loan of the infrared cell. Registry No. H2S, 7783-06-4; SOz, 7446-09-5; alumina, 1344-28-1. (12) George, 2. M. PhosphorusSulfur 1976, I, 315. (13) George, Z. M. Can. J . Chem. Eng. 1978,56, 711.
The Importance of f Functions and 3d Electron Correlation Effects in the Bonding in cu2 K. K. Sunil, K. D. Jordan,* Department of Chemistry, University of Pittsburgh, Pittsburgh, Pennsylvania 15260
and Krishnan Raghavachari AT& T Bell Laboratories, Murray Hill, New Jersey 07974 (Received: September 14, 1984)
The bonding in the ]Eg+ground state of Cu2is studied by both the multireference CI and Moller-Plesset perturbation theoretical procedures. It is found that triple and quadruple excitations are important for describing the binding and that diffuse f functions contribute about 0.14 eV to the binding energy.
Over the past few years several theoretical studies of the bonding in Cu2 have appeared.'+ Since the ground state of the Cu atom
has a zS (3d1°4s) configuration, one might expect the bonding in the molecule to be governed by the 4s electrons and to be alka-
0022-3654/85/2089-0457$01.50/00 1985 American Chemical Society