J. Chem. SOC., Faraduy Trans. I , 1982, 78, 2215-2225 Infrared Spectroscopic Study of CO and CO, Adsorption on Rh-ZrO,, Rh-Al,O, and Rh-MgO BY YUKARI TANAKA, TOKIO IIZUKA* AND Kozo TANABE Department of Chemistry, Faculty of Science, Hokkaido University, Sapporo 060, Japan Received 28th September, 198 1 The adsorption states of CO and CO, on Rh catalysts supported on ZrO,, Al,O, and MgO were studied using i.r. spectroscopy. Upon adsorption of CO on reduced Rh-ZrO,, the carbonate bands due to the reaction 2CO + C+CO,, together with the bands of twin, linear and bridge CO species were observed at moderate temperatures, whereas Rh-MgO gave no appreciable formation of CO, even at higher temperatures (> 100 "C) and Rh-A1,0, showed intermediate behaviour. On the adsorption of CO,, the linear CO band was formed at lower frequency than that on CO adsorption. The linear CO species formed from CO, showed higher reactivity toward hydrogen compared with that from CO adsorption.Infrared spectroscopy is widely used in the study of adsorbed CO on transition metals in relation to their catalytic properties for Fischer-Tropsch synthesis. Carbon monoxide is known to chemisorb in three forms on transition metals. These are the 'bridge' type, the ' linear' type and in some cases the 'twin' type. The stability of the adsorbed species follows the order bridge > linear > twin.' Eischens, reported that metals on which CO chemisorbs in the bridge form (Ni and Pd) show higher catalytic activity in the methanation of CO than metals on which CO chemisorbs in the linear form (Cu and Pt).On the other hand, Heal et aL3 showed that the temperature region for CH, formation on Ni-SiO, is close to that for a quick decrease in the adsorption band of linear CO and concluded that linear CO is involved in CH, formation. Recently, Fujimoto et al., reported that the bridge CO was hydrogenated at lower temperatures than the linear CO to form hydrocarbons consisting mainly of CH, on Rh-Al,O, and Ru-AI,O, and that the twin CO on Rh-Al,O, desorbed without being hydrogenated. On the other hand, the adsorbed states of CO, on the catalyst in the hydrogenation of CO, were not clearly elucidated. Solymosi et aL5 observed linear and bridge CO formed from CO, in the presence of H,, but could not obtain those bands clearly without H,.Dubois and Somorjai6 observed the dissociation of CO, on the surface of Rh metal crystals with LEED and ELS techniques in an ultrahigh vacuum system. In this report we discuss the support effect for CO adsorption on Rh catalysts and the difference in the adsorption states of CO and CO, in relation to their reactivity toward hydrogen. EXPERIMENTAL Supported rhodium catalysts were prepared by impregnating ZrO,, A1,0, or MgO with an aqueous solution of Rh(NO,),. After the evaporation of water, the catalysts were dried in air at 100 "C for 24 h and calcined at 500 O C for 2 h. Zirconium oxide was prepared by the hydrolysis of ZrOC1, with aqueous ammonia, followed by calcining at 500 O C . Aluminium oxide was obtained by calcining the hydroxide at 500 OC, after its preparation by the hydrolysis of Al(NO,), with aqueous ammonia.Magnesium oxide 221 52216 was obtained by heating Mg(OH), at 500 OC. The content of Rh was 2.3 wt% on all the catalysts. For measurement of the i.r. spectrum, a conventional flow-through cell was used. The catalyst was pressed into a thin wafer and placed in the cell. All spectra were recorded in the temperature range 25-250 O C with a Jasco 701G grating spectrometer. The sample in the cell was pretreated by evacuation at 300 OC for 2 h and then reduced for 2 h at the same temperature. Carbon monoxide, carbon dioxide and hydrogen were purified by passing over molecular sieve 4A. I.R. STUDY OF CO AND CO, ADSORPTION ON Rh CATALYSTS RESULTS The i.r. spectra of CO adsorbed on Rh-ZrO, at various temperatures are shown in fig.1 . Upon adsorption at room temperature, the absorption bands at 2095, 2065 and 2030 cm-l in the linear- and twin-type region and at 1870 cm-l in the bridge-type la) 2000 1900 1800 1700 1600 1500 wavenumberlcm-' FIG. 1 .-1.r. spectra of CO adsorbed on Rh-ZrO, previously reduced at 300 "C: (a) background, (6) room temperature, and after heating at (c) 100 "C, ( d ) 150 OC, (e) 200 "C. region were observed. In addition to those bands, the broad signals of carbonate species at ca. 1560 and 1300 cm-l were observed at the beam temperature (ca. 40 O C ) . Upon heating at higher temperatures in the presence of CO, the bands at 1560 and 1300 cm-l and the twin bands at 2095 and 2030 cm-l increased in intensity and the band at 1870 cm-l shifted slightly to lower frequency. After heating at 200 O C in CO, the twin bands almost disappeared, as shown in fig.I (e). When Rh-ZrO, was reduced at 500 OC and evacuated at the same temperature, the development of carbonate and twin bands was also observed, as shown in fig. 2, but the intensity of these bands was slightly less than those in the previous case. With Rh-Ai,O,, the formation ofY. TANAKA, T. I I Z U K A AND K. TANABE 2217 carbonate species was not significant at room temperature, as shown in fig. 3, and only the linear- and bridge-type CO species were observed. Upon heating at higher temperatures ( 2 100 "C), carbonate bands at ca. 1550 and 1300 cm-l appeared, along with twin bands at 2105 and 2035 cm-l, but the intensity of these bands was very weak compared with the case of Rh-ZrO,. On the other hand, the adsorption states of CO on Rh-MgO were very different from those on Rh-ZrO, and Rh-Al,O,, as shown I I I I I 1 2000 1900 1800 1700 1600 1500 wavenumber/cm-' FIG.2.-1.r. spectra of CO adsorbed on Rh-ZrO, previously reduced at 5OOOC: (a) background, (b) room temperature, and after heating at (c) 100 O C , ( d ) 150 O C , (e) 200 O C . in fig. 4. Upon adsorption of CO at room temperature, only the linear-type band at 2020 cm--l and the bridge-type band at 1870 cm-l were observed and the bands of the carbonate species were not observed. After heating the sample in gaseous CO at 100-200 "C, the band at 1870 cm-l increased in intensity and shifted to lower frequency (1825 cm-l after heating at 200 "C), but the carbonate species was not observed after heating at 200 "C.The adsorption of CO, on the Rh catalysts was also examined by i.r. spectroscopy. As shown in fig. 5 (A), carbon dioxide adsorbed on Rh-Al,O, which had been reduced at 300 "C gave the linear CO species at ca. 2020 cm-l. The band intensity increased with the adsorption temperature, but the twin and bridge species were not observed even after heating at higher temperature (> 200 "C). The linear CO band was also observed for CO, adsorption on Rh-ZrO, and Rh-MgO with reduced intensity. Upon adsorption of a mixture of CO, and H,, a strong enhancement of the CO band was observed, as shown in fig. 5(B), but the band frequency was almost the same as that in the case of pure CO, adsorption. 72 F A R 12218 I.R.STUDY OF CO AND CO, ADSORPTION ON Rh CATALYSTS 1 2000 1800 1600 1400 wavenurnberlcm- ' FIG. 3.-1.r. spectra of CO adsorbed on Rh-A1,0, previously reduced at 500 O C , (a) background, (b) room temperature, and after heating at (c) 100 O C , ( d ) 150 "C, ( e ) 200 OC. 1 1 1 1 I I I 2000 1800 1600 1400 wavenumber/cm -' FIG. 4.-1.r. spectra of CO adsorbed on Rh-MgO previously reduced at 300 O C , (a) background, (b) room temperature, and after heating at ( c ) 100 OC, ( d ) 150 OC, (e) 200 "C, (f) CO, adsorption at 200 O C .Y. TANAKA, T. I I Z U K A A N D K. TANABE 2219 When CO was adsorbed on Rh-A1,0, which had been oxidized at 5OO0C, a spectrum similar to that of CO, adsorbed on the reduced surface was observed. The spectra are shown in fig.6. The only difference was the existence of twin species in the case of CO adsorption on the oxidized catalyst. Upon adsorption of CO at room temperature, a sharp band at 2 1 15 cm-l and a small band at 2030 cm-l were observed, 2035 ,2065 , 2105 , I l l 00 2000 1800 1600 : I I 1 I I l 00 2000 1800 1600 wavenum ber/cm -' FIG. 5 . 4 . r . spectra of CO, adsorbed on Rh-Al,O, previously reduced at 300 O C . (A) Adsorption of pure CO, : (a) background, (b) room temperature, and after heating at (c) 100 O C , ( d ) 150 O C , (e) 200 O C , (f) 250 "C, ( g ) CO adsorption at room temperature. (B) Adsorption of CO, and H,: (a) background, (b) room temperature, and after heating at ( c ) 100 OC, ( d ) 150 O C . but the band of the bridge species was not observed. After heating the sample at 100 O C , the linear CO band appeared as a shoulder at 2080 cm-l and the carbonate band due to the reduction of the surface with CO increased in intensity.The linear band shifted to 2055 cm-l upon evacuation at room temperature. The activity of adsorbed CO species on Rh-A1,0, toward H, was examined spectroscopically. When CO species adsorbed on the reduced catalyst were heated in H,, the linear band shifted slightly to lower frequency and the bridge species disappeared at 120- 150 OC, then the linear species disappeared at ca. 175 O C , as shown in fig. 7. The twin-type CO species was stable up to 190 O C . Since these species were observed at 180-200 O C in the absence of H,, the disappearance of bridge and linear CO bands in H, can be ascribed to the reaction with H,.The formation of CH, in the gas phase was detected at ca. 150 OC. In the case of adsorbed CO, in H,, the linear CO band disappeared completely a t 150 OC, as shown in fig. 8. In the presence of CO, 72-22220 I.R. STUDY OF CO AND CO, ADSORPTION ON Rh CATALYSTS and H, in the gas phase, the linear band still existed at 150 OC, as shown in fig. 5 . In this case, the formate species appeared at 1590 and 1390 cm-l at temperatures > 100 OC, but this species was stable after heating at 150-200 "C in gaseous H,. On the oxidized Rh-A1,0,, the linear CO species at 2050-2055cm-l disappeared at 150-175 "C in H,, but the twin species was stable at this temperature. I I 1 I I 2000 1900 1800 1700 1600 wavenum ber/cm-' FIG. 6.-1.r. spectra of CO adsorbed on Rh-A1,0, previously oxidized at 500 O C : (a) background, (b) room temperature, and after heating at (c) 100 O C , ( d ) 150 O C , (e) 200 O C . DISCUSSION ADSORPTION OF co As shown in fig.1, the formation of adsorbed CO, (carbonate) species over reduced Rh-21-0, was observed along with the ordinary CO adsorbed species such as the bridge, linear and twin species upon the introduction of CO. This suggests that CO disproportionates to carbon and CO, on the surface. Upon the adsorption of CO at room temperature, the twin-species bands were weak and the higher-frequency band was observed as a shoulder of the linear band. After heating the sample in CO at 100-150°C, the twin-species band increased in intensity with the increase of the carbonate bands. Since the twin species is known to form on the oxidized Rh site, probably Rh', the development of this species upon heating in CO would indicate oxidation of the Rh site by the dissociation of CO, as reported by Primet.' AfterY.TANAKA, T. I I Z U K A A N D K. TANABE 222 1 the sample had been heated at 200 O C , the twin species almost disappeared and the carbonate band increased in intensity on cooling to room temperature. In this case, the disappearance of the twin-species band was not due to desorption, because gaseous CO was still present in the i.r. cell. The site responsible for the formation of the twin species might be reduced by CO to form CO,. Moreover, since carbonate 1 1 1 I I I 2000 1900 1800 1700 1600 1500 wavenumberlcm -' FIG. 7.-1.r. spectra of CO adsorbed on Rh-Ai,O, in H, at various temperatures: (a) background, (b) room temperature, (c) 100 O C , ( d ) 125 O C , (e) 150 O C , (f) 175 O C .species were also formed on Rh-ZrO, which had been reduced and evacuated at temperatures > 500 O C , we excluded the possibility of CO, formation in the water-gas shift reaction, because the residual water on the surface would be minimized in the evacuation at high temperature. Thus, we concluded that the formation of CO, in CO adsorption was due to the disproportionation of CO to carbon and CO,. In the case of Rh-Al,O,, the formation of CO, was not observed in the adsorption of CO at room temperature, but carbonate bands appeared along with an increase in intensity of the twin-species bands at higher temperatures (2 100 "C).Since the intensities of the carbonate and twin-species bands at 100 O C on Rh-A1,0, were weak compared with those of Rh-ZrO,, the dissociation of CO on Rh-A1,0, was less favourable than on Rh-ZrO,. On the other hand, over the Rh-MgO surface the formation of carbonate and twin species was not observed, even at higher temperatures. This is a marked contrast with the case of Rh-ZrO,. We recently reported that Rh-ZrO, was an excellent catalyst for the hydrogenation reactions of CO and CO, to form hydrocarbons.s The order of activity was Rh-ZrO, > Rh-A1,0, > Rh-2222 I.R. STUDY OF CO AND CO, ADSORPTION ON Rh CATALYSTS c 0 .- in .- E E? c Y b u 1 I 1 I 2000 1900 1800 1700 1600 1500 wavenumber/cm - I FIG. 8.-1.r. spectra of CO, adsorbed on Rh-A1,0, in H, at various temperatures: (a) room temperature, (6) 75 OC, (c) 100 OC, ( d ) 125 O C , (e) 150 OC.SiO, 9 Rh-MgO for the reactions of both CO and CO,. Though we failed to obtain a clear i.r. spectrum of CO adsorption on Rh-SO,, the tendency for CO dissociation corresponded very well with the activity. Recently, experimental evidence has been presented supporting the idea that CO dissociation to form carbon on the surface is an essential initial step in the synthesis of hydrocarbon^.^ Thus, it would be useful to study the surface characteristics of supported Rh catalysts affecting the dissociation of CO. Ichikawa and Kawai recently reported a partial electron transfer from TiO, or ZrO, oxide supports to Rh,(CO),, on the basis of X.p.s. experiments10 They stated that partially reduced states of TiO, or ZrO, stabilize the low oxidation state of Rh and hence the reactivity of CO is enhanced in the synthesis of alcohol from CO and H,.Probably this electron-transfer effect is also operating in our case to stimulate the dissociation of CO over Rh-ZrO,. On the other hand, the reason for the difficulty of CO bond dissociation on Rh-MgO, which probably leads to the inactivity for the hydrogenation of carbon oxides, is not clear yet. One reason for the inactivity of Rh-MgO would be a low dispersion of Rh on MgO. The dispersion values of Rh measured on the basis of H, adsorption were 0.51,0.60 and 0.27 for Rh-ZrO,, Rh-A1,0, and Rh-MgO, respectively.s Since MgO is familiar as a substrate for the epitaxial growth of metal films with low-index orientations on its surface,ll Rh atoms will exist as a metal crystallite of low-index planes mainly on MgO, even in this case.Moreover, the absence of twin CO bands on Rh-MgO in the i.r. spectrum is very similar to that of CO adsorption on Rh metal crystal.12 It is known that the adsorption of CO on Rh metal crystal is classified as non-di~sociative.~~ Thus, the inactivity ofY. TANAKA, T. I I Z U K A AND K. TANABE 2223 Rh-MgO for CO dissociation is ascribed to the segregation of metal which probably has low-index planes on the surface. ADSORPTION OF co, As for the adsorption of CO, over Rh catalysts, previous worker^^^-^^ could not detect the CO band, indicative of the dissociation of CO, on the surface. However, Dubois and Somorjai found the dissociation of CO, on several faces of Rh single crystal by means of LEED or ELS techniques in an ultrahigh vacuum system.6 Owing to the difficulty of detection, Dubois and Somorjai suggested the low sticking probability of CO, and the high rate of the association reaction CO(ads)+O(ads)+ CO,(g) requiring an approximately five- to ten-fold higher exposure for the adsorption of CO, compared with CO to obtain the spectra.6 On the other hand, Primet' and Solymosi et al.5 observed weak CO bands in CO, adsorption on Rh-Al,O,, but this was not reported in their work.In our case on Rh-A1,0,, the dissociation of CO, to CO and 0 atom was observed even at a moderate pressure of CO, (2-15 Torr) and at room temperature. The CO band was also observed on Rh-ZrO, and Rh-MgO, but the intensity of the CO band was weak over those catalysts compared with that on Rh-Al,O,.Probably the dissociation of CO, depends on the nature of the support used, the preparation of the catalyst and the dispersion of the Rh. Note the difference in CO and CO, adsorption on Rh-MgO. The surface of Rh-MgO was inactive for the dissociation of CO to carbon but showed comparable activity to Rh-ZrO, for the dissociation of CO, to CO and 0 atom. This would be the reason why CO was the main product in the reaction of CO, and H, at > 200 O C on Rh-MgO.s Solymosi et al. observed the formation of a CO band from CO, in the presence of H,, but could not obtain the clear band without H, under the same condition^.^ Concerning the main difference between the spectrum of adsorbed CO and that obtained after coadsorption of the H,+CO, mixture, they concluded that (i) the doublet due to twin CO was completely missing and (ii) the linearly bonded CO appeared at lower frequency, 2020-2039 cm-l.They suggested that the adsorbed hydrogen prevents the formation of a twin structure, the hydrogen adsorbed on the metal atom of the carbonyl might donate an electron to CO and consequently the vibrational frequency of CO would shift to lower freq~ency.~! 1 7 7 However, in this work, the twin CO species was also completely missing and the band frequency of the linear species was still lower than that of pure CO adsorption in the absence of H, after a comparable band intensity with that in H, was obtained, as shown in fig. 5. Moreover, since the adsorption of CO was very strong compared with that of hydrogen,8 it would be unlikely that the adsorbed hydrogen prevented the formation of twin species.Thus, it is rather difficult to ascribe the reason for the absence of twin bands and the shift of the linear band to lower frequency to the coadsorption of hydrogen with CO on the same Rh atom. In the case of CO, adsorption on Ni/Aerosil, van Hardeveld et al. reported the shift of CO bands formed from CO, to lower frequency compared with CO adsorption. They explained the shift of the CO bands as follows: the CO, decomposes into CO + 0, which are most probably adsorbed onto adjacent atoms. The adsorption of 0 on Ni results in the oxidation of Ni. The positive charge on the Ni atom polarizes the CO molecule adsorbed on it and/or forces more electrons into the Ni-C bond, thereby weakening the C-0 bond and lowering its stretching frequency.However, in our case with Rh-Al,O,, after the oxidation of the catalyst the linear CO band appeared at higher frequency (ca. 2080 cm-l) than that on the reduced surface, and shifted to 2055 cm-l upon evacuation. Yang and Garland have reported the high-frequency shift of2224 I . R . STUDY OF CO AND CO, ADSORPTION ON Rh CATALYSTS the linear CO band upon the oxidation of Rh.14 Thus, it is also unlikely that the low-frequency shift is due to the oxidation of metal in our case. Since it is well known that the band frequency of CO stretching is a strong function of CO c ~ v e r a g e , l ~ > ~ ~ we are inclined to think that the appearance of CO at a lower frequency in the CO, adsorption compared with pure CO is due to a small covering of CO formed from CO, on the surface.The shift of the CO linear band on the oxidized surface to lower frequency upon evacuation would be due to the removal of weakly adsorbed CO around the linear species having a dipolar interaction with strongly chemisorbed species; however, this linear species still appears at a higher frequency than CO formed from CO,, in spite of the similarity of surroundings (except for the presence of the twin species in the case of CO adsorption on the oxidized surface). Thus, it would be reasonable to ascribe the frequency shift to lower coverage of CO in CO, adsorption. The CO species formed from CO, would be isolated on the surface, and it did not show a shift to higher frequency at higher coverage in the presence of hydrogen.REACTION OF ADSORBED SPECIES WITH HYDROGEN In the reaction of adsorbed CO with H, over Rh-Al,O,, the bridge species reacted first at 120-150 OC, and then the linear species reacted at ca. 175 O C , as reported by Fujimoto et aL4 Although we could not detect a clear bridge band in CO, adsorption, the linear CO species formed from CO, reacted with hydrogen at lower temperatures than pure CO. This corresponds well with the fact that CO, reacted with H, to form methane at a temperature lower than that for the hydrogenation of CO over Rh catalysts.* This phenomenon can be explained as follows: the coverage of CO from CO, is very small compared with that in pure CO adsorption, so the adsorption of H, is more favourable in CO, hydrogenation than the case of CO, which acts as a poison for H, adsorption.For the same reason, the linear CO species in CO adsorption on the oxidized surface showed a slightly higher reactivity with H, compared with CO on the reduced catalyst, corresponding to the fact that the rate of hydrocarbon formation on the oxidized surface was higher than that on the reduced catalyst.* In the reaction CO+H,, the oxidized Rh catalyst showed higher selectivity to higher hydrocarbons (C,-C,) than the reduced catalyst, but only methane was formed in the reaction CO, + H,.* Comparing the adsorbed states of CO on the oxidized surface with those of CO, on the reduced surface, the main difference was the absence of twin species in the case of CO,.Moreover, the twin species was less intense on the reduced surface in CO adsorption. Thus, the CO in a weak adsorption state, such as twin species, would have an important role in the propagation of hydrocarbons in CO hydrogenation on the oxidized surface. H. C. Yao and G . W. Rothschild, J . Chem. Phys., 1978, 68, 4774. R. P. Eischens, in The Surface Chemistry of Metals and Semiconductors, ed. H. C. Gatos (Wiley, New York, 1960). M. J. Heal, E. C. Leisegang and R. G. Torrinton, J. Catal., 1976, 42, 10. K. Fujimoto, M. Kameyama and T. Kunugi, J. Catal., 1980, 61, 7. F. Solymosi, A. Erdohelyi and M. Kocksis, J . Catal., 1980, 65, 428. L. H. Dubois and G. A. Somorjai, Su$. Sci., 1979, 88, L13. ' M. Primet, J . Chem. SOC., Faraday Trans. I, 1978, 74, 2570. T. Iizuka, Y. Tanaka and K. Tanabe, J . Catal., in press. M. Araki and V. Ponec, J. Catal., 1976, 44, 439. lo M. Ichikawa and M. Kawai, Shokubai (CataZysf), 1981, 23, 55. l 1 J. Prichard, T. Catterick and R. K. Gupta, Surf. Sci., 1975, 53, 1 . l2 R. R. Cabanah and J. T. Yates Jr, J . Chem. Phys., 1981, 74,4150. l 3 G. Broden, T. N. Rhodin, C. Brucker, R. Benbow and Z. Hurich, Surf. Sci., 1976, 59, 593.Y. TANAKA, T. I I Z U K A A N D K. TANABE l4 A. C. Yang and C. W. Garland, J. Phys. Chem., 1957, 61, 1504. l5 C. T. Campbell and J. M. White, J. Catal., 1978, 54, 289. l 6 A. C. Collings and B. M. W. Trapnell, Trans. Faraday SOC., 1957, 53, 1436. l7 F. Solymosi, A. Erdohelyi and M. Kocksis, J. Chem. SOC., Faraday Trans. I , 1981, 77, 1003. F. Solymosi, A. Erdohelyi and T. Bansagi, J. Chem. SOC., Faraday Trans. I, 1981, 77, 2645. A. 34. Bradshaw, Surf. Sci., 1979, 80, 215. 2225 (PAPER 1 / 1504)