J . Chem. SOC., Faraday Trans. I , 1982, 78, 1457-1463 Infrared Study of CO, Adsorption on ZnO Adsorption Sites BY JACQUESAUSSEY, JEAN-CLAUDE LAVALLEY* AND CLOTILDE BOVET Laboratoire de Spectrochimie, ERA 824, I.S.M.R.A., Universite de Caen, 14032 Caen Cedex, France Received 18th May, 1981 Adsorption of carbon dioxide on a ZnO Kadox-15 powder has been studied by Fourier transform infrared spectroscopy. The following surface species are formed : bidentate carbonates, polydentate carbonates which appear with time or heating, hydrogenocarbonates and linear CO, species. Moreover, a band at 1546 cm-l could correspond to carboxylates reversibly adsorbed at room temperature. Attention is paid to the effect of CO, addition which splits the w,(O=C=O) and d(C0,) bands due to linear species and shifts the bidentate carbonates band from I595 to 161 5 cm-'. Taking account ofthe band u,(O='~C=O) (in natural abundance), we deduce that the splitting is due to a coupling between two linear species held by the same Zn2+ ion.We propose that such Zn2+ ions that are two-fold coordinate are situated on the edges formed by the (0001) and (1010) planes. Infrared spectroscopy has been widely used to study CO, adsorption on various oxides. Knozingerl has reviewed the studies carried out on ZnO and reported that the species formed on this oxide are not so well-defined as those for, e.g., alumina, or a-chr~mia.~ It is the purpose of this paper to examine in detail the structure of the species in order to gain some information on the nature of active sites on ZnO. We have used a Fourier transform interferometer with spectral ratioing facilities and adsorbed CO, on 100 mg discs to study the 1100-600 cm-l range, which has not been investigated previously and which could provide interesting information on the structure of the species.EXPERIMENTAL The zinc oxide powder used was Kadox- 15 from the New Jersey Zinc Co. ; its surface area was 9 m2 gel. The samples were pretreated with oxygen at 723 K as described in ref. (4). The carbon dioxide used was obtained from Air Liquide with a stated purity of 99.998%. Infrared spectra were recorded on a Perkin-Elmer 580 or a Nicolet FT MX-1 at room temperature. RESULTS The spectra of chemisorbed CO, were recorded at increasing coverage. Table 1 lists the positions and the relative intensities of the bands.The bands formed at low coverage grow in intensity until ca. 20 pmol g-' of CO, are chemisorbed. Note the frequency shift with the coverage of the two main bands, initially at 1582 and 1335 cm-l. Addition of another dose (total amount: 36 pmol g-l) increases the intensity of the bands, except for those at 1000 and 848 cm-l, but they become broader. Moreover, new bands appear (table I). Addition of larger quantities (equilibrium pressure: 60 N m-,) shifts the band at 14571458 CO, ADSORPTION ON ZnO TABLE 1 .-WAVENUMBER AND RELATIVE INTENSITIES (ABSORBANCE) OF BANDS RESULTING FROM THE ADSORPTION OF SUCCESSIVE DOSES OF CO, ON ZnO equilibrium evacuation pressure (P = 2.6 x 10-4 18 pmol g-' 36 pmol g-l 60 N m-2 N m-,) - 1665 (0.04) 2358 (0.24) 2290 (vvw) 1665 (0.08) 1595 (0.58) 1546 (0.14) 1520 (sh) - 1339 (0.40) 1325 (sh) 1303 (sh) 1040 (0.02) 1000 (0.16) 848 (0.07) 841 (sh) 680 (0.01) - 1597 (1.10) 1546 (0.23) 1520 (sh) 1424 (0.03) 1370 (sh) 1340 (0.55) 1325 (sh) 1300 (sh) 1265 (sh) 1229 (0.01) 1040 (0.03) 1002 (0.18) 848 (0.09) 841 (sh) 835 (sh) 680 (0.02) 638 (0.04) -~ ~ 2369 (0.46) 2353 (0.59) 2290 (0.02) 1650 (sh, b) 1615 (1.53) 1580 (sh) 1547 (0.32) 1520 (sh) 1425 (0.06) 1369 (sh) 1346 (0.95) 1325 (sh?) 1300 (sh?) 1265 (sh) 1229 (0.03) 1039 (0.04) 999 (0.28) 852 (sh) 841 (0.18) - - - - - 1665 (sh) 1593 (0.58) 1580 (sh) 1520 (0.25) - 1338 (0.56) - - 1002 (0.15) 848 (w) - - 679 (w) I 2350 2000 1650 1300 950 600 wavenumberlcm-' FIG.1.-Infrared spectrum of adsorbed species formed by addition of CO, (equilibrium pressure: 60 N m-,) on ZnO (spectrometer: Nicolet MX-1).Note scale change at 2000 crn-'.J. SAUSSEY, J-C. LAVALLEY A N D C. BOVET 1459 1597 cm-l to 16 15 cm-l and splits the bands at 2358 and 638 cm-l (table 1). Moreover, shoulders near 1585 and 1520 cm-l appear (fig. 1). Note that the spectrum of CO, chemisorbed on ZnO pretreated with D, to exchange the hydroxyl groups is similar except that it does not exhibit the 1229 cm-l band. Room-temperature desorption allows the characterization of the bands due to irreversibly adsorbed species (table 1). Spectra relating to degassing at temperatures 1550 ' 1350 ' 1150 ' 950 ' 750 ' wavenum ber/cm-' FIG. 2.-Infrared spectrum of adsorbed species formed by heating at 723 K an activated disc of ZnO under oxygen ( P = lo3 N m-2) with a small amount of CO, in it ( P = lo2 N m-2) (spectrometer: Nicolet MX-1). higher than 360 K show only two main bands, at 1520 and 1335 cm-l.These were very strong in Taylor and Amberg's study5 when ZnO was outgassed for 18 h at 593 K then treated with oxygen at 623 K. We were able to obtain the same bands by heating a ZnO disc at 723 K in oxygen (lo3 N m-,) containing a small amount of CO, (lo2 N rn-,). In addition to the 1522 and 1327 cm-l bands (fig. 2), sharp bands at 1030 and 876 cm-l appear, which will allow us to identify the structure of the corresponding species as shown below. In one experiment, CO, was adsorbed on a ZnO sample pretreated with H,O (25 pmol g-l). Bands of medium intensity appear at 1635, 1513, 1468, 1415, 1390, 1341 and 1222 cm-l.In the final experiment, we studied the room-temperature evolution with time of the spectrum reported in fig. 1. We observe (i) an increase in the intensity of the bands at 1580, 1520 and 852 cm-l, (ii) the appearance of new bands at 1007 and 875 cm-l, and (iii) a decrease in the intensity of the bands at 1615,999, 841 and 1547 cm-l, the latter almost disappearing. DISCUSSION ASSIGNMENT CARBONATES, HYDROGENOCARBONATES A N D CARBOXYLATES Our assignments are based on the work of Little6 and Filimonov and coworkers.' In a previous experiment8 involving addition of water to ZnO treated with CO,, two of us showed that the bands at 1590, 1525 and 1335 were not assignable to a single1460 CO, ADSORPTION ON ZnO species.* From the present study, we can deduce that the bands at 1595, 1000 and 848 cm-l are connected, as they shift or broaden when a large amount of CO, is introduced.We assign them to bidentate carbonates. Desorption experiments at higher temperatures show that the 1 339 cm-l band is also connected with the other three bands. A part of this species is irreversibly adsorbed at room temperature. When large amounts of CO, are added, the 1595 cm-l band shifts to 161 5 cm-l (table I), showing a slight change in its structure. These bands tend to disappear gradually with time giving rise to another type of bidentate carbonate (bands at 1580, 1007 and 852 cm-l). The band at 1665 cm-' can be ascribed to a different bidentate carbonate with another band near 1300 cm--l (table 2). TABLE 2.-TENTATIVE ASSIGNMENT OF SURFACE SPECIES POlY- dentate hydro- linear CO, bidentate carbonates carbon- geno- species ates carboxyl- carbon- (1) (2) (3) (4) ( 5 ) ates ates (6) (7) 1595 1615 1580 1665 1522 1546? 3605 2358 (2353 2369 1339 1346 1348 1303 1327 1635 1370 1370 64 1 1030 1424 638 (636 1000 999 1007 848 841 852 876 1229 680 677 720 835 660 (1) Formed by adsorption of 18 pmol g-l of CO,; (2) formed by the transformation of species (1) and carboxylates when larger amounts of CO, are introduced; (3) appearing after a longer contact time (2 h); (4) formed by adsorption of 18 pmol g-l of CO,; the intensity of these bands does not grow noticeably with the amount of CO, added; (5) appearing after heating; (6) formed when 36 pmol g-l of CO, were added; (7) formed when the equilibrium pressure of CO, is 60 N mP2.Bands (table 2) associated with species formed at high temperature may be assigned to polydentate carbonate^.'^ They tend also to appear with time at room temperature and were assigned to carboxylates in a previous study.g It is difficult to determine the origin of the 1546 cm-l band, which characterizes species reversibly adsorbed at room temperature and which tends to disappear with time. As we were not able to correlate it with any band in the 1100-900 cm-l range, it could be assigned to carboxylates. Hydrogenocarbonates (bands at 1424 and 1229 cm-l) are formed when more than 20 pmol g-l of CO, are added. They are reversibly adsorbed at room temperature, as Borello has suggested.lo Adsorption of CO, on a sample of ZnO pretreated with H,O shows that a band near 1635 cm-l is connected with these.It is difficult to observe the corresponding v(0H) mode. However, from fig. 3 we deduce that it absorbs at 3600 cm-l and, as in the case of alumina,, that the formation of hydrogenocarbonates proceeds through an attack of the OH groups which have the highest wavenumber band (3670 cm-l). These assignments are summarized in table 2. * The spectrum and the wavenumbers were different from those given here due to the temperature effect which favoured the appearance of the more stable species (beam temperature of the Perkin-Elmer 225 spectrometer, ca. 340 K).J . SAUSSEY, J-C. LAVALLEY A N D C. BOVET 1461 wrlvenum ber/cni- ' FIG. 3.-Effect of addition of CO, (equilibrium pressure: lo3 N m-*) on the hydroxyl groups of ZnO (spectrometer: Perkin-Elmer 580; dotted line: background).LINEAR SPECIES Bands close to the gaseous CO, frequencies can be assigned to CO, weakly adsorbed onto cationic sites. Their linear structure is preserved. Due to the interaction with the surface, the CO, symmetry is lowered, which explains the appearance of the v, mode in the infrared (band at 1370 cm-l, table 2). Note that the use of a Fourier transform spectrometer allows us to observe the d(C0,) mode. It is situated 29 cm-l below the wavenumber of the gas-phase band, showing an interaction with ZnO Lewis sites. This result confirms the presence of such sites on ZnO activated at 723 K and agrees with a recent study using CD,OCD,H as a probe molecule.ll Note that, when the bands at 2358 and 638 cm-l split due to the further addition of CO,, the band due to bidentate carbonates also shifts (1595 1615 cm-l).The splitting of v,(CO,) of linear species can be explained by looking at the v(O=13C=O) band, situated at 2290 cm-l. Whatever quantity of CO, introduced, it appears sharp and unique. Its intensity increases with the amount of CO, added, even when this amount is high enough to induce the splitting of the v,(O=~~C=O) band. So we deduce that the double bands at 2369 and 2353 cm-l and at 641 and 636 cm-1 are not due to different adsorption sites but to coupling between linear species, ADSORPTION SITES ON ZnO The sharpness of the bands reported in the present study is certainly due to the presence of well-defined adsorption sites. As in the case of H,-CO interaction on Zn0,12 we could distinguish two effects due to the addition of CO,: a discrete effect, which causes the bands due to linear species to split and the band at 1595 cm-1 to shift abruptly to 1615 cm-l, and a continuous effect responsible for the continuous shift of some bands (that at 1339 cm-l for instance) with coverage.In the discussion below, we will consider only the former, which could be associated with interactions in adjacent positions; the latter presumably involves interactions at more distant positions. l2 The morphology of zinc oxide, especially that prepared by the combustion of metallic zinc, is well known:13 the surface corresponds to the (OOOl), (OOOi), (101 1) and (1010) cleavage planes of the wurtzite crystal structure. Monodentate carbonates could be formed on the (OOOT) plane, and carboxylates and linear species on the (0001) plane.Bidentate carbonates show the presence of Zn2+02- ion-pair sites. The results1462 CO, ADSORPTION ON ZnO of the present study can determine whether they are situated on the (1010) plane or not. In a recent study, Runge and Gopel14 have compared the reactivity of polycrystalline and single-crystal ZnO surfaces using 0, and CO, adsorption. A surprising corre- spondence has been observed in results on the ZnO (1010) surface and chemically clean ZnO powder under u.h.v. conditions. Therefore, the authors concluded that the Oo- 0 I - 0 co2 0" linear carbonate 0" (1595 cm-' 1 0" 2369 cm-' 2353 cm-' 641 cm-' 638 cm-l bident ate carbonate (1615 cm-'1 FIG.4.-Proposed model for CO, adsorption. properties of 0, and CO, complexes on ZnO powder surfaces are determined by the properties of ZnO (lOi0) surfaces. On these surfaces, Gopel and coworkersL5 suggest the formation of carbonate-like surface complexes. Moreover, they found that oxygen vacancies act as specific sites for strong CO, chemisorption. The present study does not confirm the formation of strongly bound carboxylates. However, our activation mode is quite different from Gopel's as we heat our powder under oxygen, which prevents the formation of point defects VOs. From the infrared study of H,-CO interaction on ZnO powder at room temperature, Zecchina and coworkersL2 deduced the presence of sites which are formed by a triplet of exposed zinc ions and at least one reactive oxygen ion.Such sites would be situated on the (0001) face, reconstructed in order to satisfy the charge compensating criterion, as suggested by Nosker.16 We could explain our results with successively (i) formation of a bidentate carbonate on the Zn-0 pair site, (ii) formation of a linear species on an adjacent Zn2+ ion and (iii) formation of a second linear species involving the thirdJ. SAUSSEY, J-C. LAVALLEY A N D C. BOVET 1463 Zn2+ ion of the site, in such a way that when a large amount of CO, is added, the bidentate carbonate species would be adjacent to two linear species held by two adjacent Zn ions. The coupling between two adjacent linear species would explain the splitting of the bands at 2358 and 638 cm-l.However, we think that three argu- ments can be put forward against this model: (i) the reconstruction of the (0001) face does not seem to have been proved,17 (ii) the three underlying Zn ions which become coordinatively unsaturated by removing one of four negative ions from the (0001) face do not seem accessible enough to adsorb three molecules of CO, and (iii) the coupling between two linear species favours two linear species held by the same Zn2+ ion rather than two linear species held by two adjacent ions. Hence we propose another model involving two-fold coordinated cations. Such Zn2+ ions could correspond to oxygen vacancies V& on the (1010) face, but we have already mentioned that such sites are very improbable due to our activation process. Zn2+ ions with two vacancies are also present on the (1011) face, which contains equal numbers of surface cations that are three- and two-fold coordinated to lattice 0~ygen.l~ However such a face, without any reconstruction, does not present any 0,-; therefore, it is difficult to correlate the splitting of the linear species with the shift of the band at 1 595 cm-l, which characterizes bidentate carbonates and thence necessitates adsorption on an 02- ion neighbouring a Zn2+ cation.We prefer to explain our results according to the scheme reported on fig. 4, involving Zn2+ ions carrying two unshared coordinative vacancies with a reactive oxygen ion in an adjacent position. In powders, the number of steps and edge positions might be extremely high. The sites which we propose from this work could correspond to the edges formed by the (0001) and (1010) planes.LEED studies have shown that both polar faces are steppedls with a step height of one unit cell (two double layers) along the c-axis, in the (1010) directions, exposing non-polar faces. According to our model, some bidentate carbonates would be formed on the non-polar faces of these steps. Gravimetric measurements have shown that ca. 1.4 CO, molecules are chemisorbed per nm2? Unfortunately, due to the multiplicity of species formed, it is not possible to determine separately the number of bidentate carbonates giving rise to the 1595 cm-I band, i.e. the number of the corresponding sites. We thank Prof. N. Sheppard and Prof. W. Gopel for helpful discussions. H. Knozinger, Adu. Catal., 1976, 25, 184. C. Morterra, A. Zecchina, S. Coluccia and A. Chiorino, J. Chem. SOC., Faraday Trans. I , 1977, 73, 1544. A. Zecchina, S. Coluccia, E. Guglielminotti and G. Ghiotti, J. Phys. Chem., 1971, 75, 2790. J. H. Taylor and C. H. Amberg, Can. J. Chem., 1961, 39, 535. L. H. Little, Infrared Spectra of Adsorbed Species (Academic Press, New York, 1966). Ya. M. Grigor’ev, D. V. Pozdnyakov and V. N. Filimonov, Russ. J. Phys. Chem., 1972, 46, 186. * J. Saussey and J. C. Lavalley, J . Chim. Phys., 1978, 75, 505. F. Bozon-Verduraz, J . Catal., 1970, 18, 12. lo E. Borello, Discuss. Faraday SOC., 1971, 52, 44. J. C. Lavalley and J. Caillod, J. Chim. Phys., 1980, 77, 373. F. Boccuzzi, E. Garrone, A. Zecchina, A. Bossi and M. Camia, J . Cural., 1978, 51, 160. * T. T. Nguyen, J. C. Lavalley, J. Saussey and N. Sheppard, J. Catal., 1980, 61, 503. l 3 K. Atherton, G. Newbold and J. A. Hockey, Discuss. Faraduy SOC., 1971, 52, 33. l4 F. Runge and W. Gopel, 2. Phys. Chem., in press. l5 W. Gopel, R. S. Bauer and G. Hansson, Surf. Sci., 1980,99, 138; W. Hotan, W. Gopel and R. Haul, l6 R. W. Nosker, P. Mark and J. D. Levine, SurJ Sci., 1970, 19, 291. l7 S. C. Chang and P. Mark, Surf. Sci., 1974, 46, 293. l a M. Henzler, Surf Sci., 1970, 22, 12; 1973, 36, 109; D. Kohl, M. Henzler and G. Heiland, Surf. Sci., l9 V. Lorenzelli and G. Busca, personal communication. Sur- Sci., 1979, 83, 162. 1974, 41, 403. (PAPER 1 /796)