Tin Oxide Surfaces Part 8.t-Infrared Study of the Mechanism of Formation of a Surface Isocyanate Species on SnO, .055 CuO during Catalysis of the Oxidation of Carbon Monoxide by Nitric Oxide BY PHILIP G. HARRISON* AND EDWARD W. THORNTON Department of Chemistry, University of Nottingham, University Park, Nottingham NG7 2RD Received 2nd November, 1977 An experimental study has been carried out on the mechanism of formation of a surface isocyanate species on the mixed oxide catalyst SnOz. 0.55 CuO during the initial stages of catalysis of the CO+NO reaction at %470 K. Using infrared spectroscopy, the isotopic shifts of the 2189 cm-1 pseudo-antisymmetric stretching vibration have been measured for 13C, 15N and lSO substitution. The oxygen atom of the surface isocyanate has been shown to originate from NO rather than CO as was previously assumed.This observation has been interpreted in terms of a mechanism involving initial dissociative chemisorption of CO followed by the formation of a fulminate via reaction of NO with the surface carbon atom and subsequent rapid isomerisation to the isocyanate : 23 +co +]-c + 3-0 1-C+NO-+ 1 -CNO 1-CNO + 1-NCO. The presence of surface isocyanate species on catalysts during the catalysis of the CO +NO reaction has been well established by means of infrared spectros~opy.l-~ For noble metal catalysts Unland 1-3 has suggested the surface isocyanate as an intermediate in ammonia formation in a damp CO+NO reaction mixture. For copper metal catalysts, Rewick and Wise interpreted their kinetic data in terms of a strongly bound and thermally stable -NCO adspecies on the copper surface exhibiting primarily the function of a catalyst inhibitor.Similarly, London and Bell in their study of a silica-supported CuO catalyst were able to incorporate the formation of a surface isocyanate species into their reaction mechanism and interpret this as strongly supporting the postulate that nitric oxide can dissociate upon adsorption. These studies lm5 were all carried out at elevated temperatures ( 3 -420 K), and the authors did not investigate the mechanism(s) whereby the surface isocyanates were formed. In contrast, the mechanistic study of Brown and Gonzalez ti on silica-supported ruthenium was carried out at x 300 K. In this case, the proposed mechanism involves the interaction of an adsorbed NO molecule with two adjacent CO molecules to form an isocyanate and a molecule of carbon dioxide, without the dissociative chemi- sorption of either reactant.The study of Arai and Tominaga on alumina-supported rhodium at 473-573 K concluded that a reactive Rh-N species was formed from the CQ surface complex by elimination of C02 and then could subsequently react / \ Rh NO with CO to form Rh-NCO or NO to form Rh-N20. t Part 7. P. G. Harrison and E. W. Thornton, J.C.S. Faraday I, 1976, 72, 2484. 2604P . G . HARRISON AND E . W. THORNTON 2605 involved isotopically substituted reactants, only 13C-enriched CO and 15N-enriched NO have been used to demonstrate that both the C and N atoms in the product isocyanate are derived from the reactants, and that the isotope shifts are consistent with an isocyanate surface species.Con- sequently, in the present study we have extended the scope of the isotope substitution experiments to include l 8 0 and hence elucidate the origin of the oxygen atom in the surface isocyanate species, whether from CO, NO, or the oxide catalyst surface. This study has been carried out at ~ 4 7 0 K for reaction mixtures withp(CO)/p(NO) > 1.5 on a mixed oxide catalyst formed by complete ion exchange (during coprecipita- tion) of CU',~, on to hydrous tin@) oxide gel. The catalyst has been described by Fuller and Warwick for their kinetic studies during catalysis of the CO+O, and CO +NO reactions. Although some of the investigations EXPERIMENTAL APPARATUS The infrared cell and vacuum line have been described previously.O Infrared spectra were recorded with a Perkin-Elmer model 577 spectrometer, using abscissa expansion as required over the range 2400-2000 cm-l. The ambient beam temperature was 320 K. MATERIALS The oxide catalyst sample was kindly supplied by Dr. M. J. Fuller of the International Tin Research Institute, and came from a batch which was used by Fuller and Warwick for their kinetic studies of the CO+02 reaction.8 The Cu/Sn atomic ratio was 0.55 in the calcined material, this composition corresponding to the full utilisation of the parent ion exchange capacity of the hydrous Sn02 towards Cuxl. Oxygen, nitrous oxide and carbon monoxide (B.O.C., grade X) were used as received, carbon dioxide (Distiller's) was purified by passing through a stainless steel coil 1 m in length and 2 mm diameter held at 211 K by freezing 2-ethoxyethyl acetate, nitric oxide (B.O.C.) was purified by passing through the same trap at 197K (acetone+solid C02).1802 ( ~ 6 3 atom % 13C0 (-91 atom % 13C), ClSO (-60 atom % l80), 15N0 (m95 atom % 15N) and N180 (-41 atom % l80) were obtained from B.O.C. Prochem and used without further purification. PROCEDURE For infrared experiments the oxide catalyst was ground with an agate pestle and mortar and pressed into self-supporting discs using a stainless steel die; the discs were 2.5 cm in diameter with a " thickness " of 12.2 mg cm-2. Before performing adsorption experiments, the discs were evacuated (w N m-2) in the infrared cell for 2-24 h at temperatures in the range room temperature to 630 K, usually followed by oxygen treatment at the evacuation temperature (1.3 kN m-2, 1 h) and re-evacuation for 0.5 h.The exact procedure varied for each experiment, and will be described in the Results section. For accurate measurement of the positions of the bands shown in fig. 6, the spectra were recorded with 5x or lox abscissa expansion at scanning speeds of 100 or 50 cm-l min-l. The wavenumber markers were calibrated with gaseous carbon monoxide. The positions are estimated to be accurate to _+2 cm-l. RESULTS DEHYDRATION Following evacuation at room temperature, the initially pale blue disc turned pale green, and the infrared spectrum showed a broad, intense hydroxyl stretching band (fig. 1) and a weak shoulder at = 1600 cm-l due to molecular water.The2606 TIN OXIDE SURFACES v(0H) band decreased on evacuation up to 420K and the disc became darker in colour. At higher evacuation temperatures (> 550-570 K), the v(0H) band was removed almost entirely, a weak band remaining at 3600cm-l. This band could be only partly exchanged with D,O vapour, and the surface hydroxyl groups could not be restored by exposure to water vapour (at room temperature or 570 K). At this stage, the disc was black in colour. The background spectrum in the range 100, I 1 I I 4000 3500 3000 2: wavenumber /cm-' io FIG. 1.-Infrared spectra of SnOz .0.55CuO disc after evacuation at (1) room temperature for 0.75 h ; (2) 373 K for 0.5 h ; (3) 508 K for 1 h ; (4) 513 K for 14 h ; and (5) 623 K for 2 h followed by oxygen treatment at 623 K (130 N m-2, 2 h).3500-900 cm-l was a sloping straight line showing no spurious absorption bands due to impurities. Treatment with oxygen had no effect on the colour, but at higher evacuation temperatures (> 570 K) the transmission of the disc was improved by oxygen treatment. Evacuation of the disc above ~ 7 2 0 K resulted in a severe loss in infrared transmission, which could not be restored by oxygen treatment, thus making infrared experiments impossible. ADSORPTION OF CARBON DIOXIDE The adsorption of carbon dioxide was studied at 320 K as a function of the disc pretreatment evacuation temperature in the range 320-650 K. After evacuation at 320 K, the catalyst disc showed three broad, strong infrared bands at 1530-40, 1380 and 1060cm-1 during exposure to CO, (fig.2). These decreased in intensity on evacuation at 320K and further decreased in intensity on evacuation at higher temperatures, being removed altogether at about 570 I<. The positions of the absorption bands are quite different from the carbonate bands on tin(1v) oxide,l O neither do they closely resemble the complex set of bands observed by Amerikov and Kasatkina during C 0 2 adsorption on silica-supported Cu0.l The three bands may be assigned to the stretching bands of a coordinated carbonate where the 1060 cm-l band is the v2(A,) [v,(XY) for planar XYJ vibration, and the 1530-40 and 1380 cm-l bands are due to the v,(A,) and v,(B,) [the split va(XY) for planar XY3] vibrations.12P . G . HARRISON AND E . W. THORNTON 2607 The splitting of the degenerate vibrations (Av = 150cm-') is more typical of a unidentate than a bidentate carbonate ; this range of splittings has been reported as approximately AY = 230-100 cm-' for a unidentate carbonate and Av = 370-320 cm-' for a bidentate ~arb0nate.l~ Further support for the assignment to a unidentate wavenumber /crn-' FIG.2.-Infrared spectra of a Sn02. 0.55CuO disc exposed to carbon dioxide. (1) Background after evacuation and treatment with DzO (s.v.P., 0.5 h) at 393 K. (2) During subsequent exposure to C02 (400 N m-2) at 320 K, and (3) after evacuation at 320 K for 1 h. carbonate comes from the position of v2 band at 1060 cm-l ; this lies in the middle of the reported range of frequencies for unidentate carbonates (1080-1040 cm-l), whereas the corresponding bands for bidentate carbonates are found at 1030- 1020 crn-l.l3 2.0 n 8 1.0 k 0 a\- evacuation temperature/K FIG.3.-PIot of the intensity of carbonate bands produced by adsorption of carbon dioxide on SnOz .0.55CuO as a function of pretreatment evacuation temperature. F(C0,) is the sum of the optical densities of the 1535 and 1380 cm-I bands. Points refer to the intensity of carbonate bands after evacuation ( x ) and during subsequent exposure to C02 at 320 K (0).2608 TIN OXIDE SURFACES As the pretreatment evacuation temperature was increased, the intensity of the carbonate bands both in equilibrium with C 0 2 at 320 K and after evacuation at 320 K decreased. This is shown graphically in fig. 3, where the intensity of the carbonate bands is measured by F(CQ,) which is the sum of the optical densities of the 1530-40 and 1380 cm-l bands.Both curves intersect the abscissa at ~ 7 2 0 K, which implies that no surface carbonate should form during exposure to C 0 2 of samples evacuated above this temperature. None of the catalyst sample discs showed a band at ~ 2 3 5 0 cm-I due to linear, adsorbed CQ2 during exposure to CQ2. This contrasts with silica-supported CuO where a band has been observed in this region in two previous studies.5’ l1 ADSORPTION OF NITROUS OXIDE Following evacuation in the range 320-650 K (and oxygen treatment above ~ 5 7 0 K), exposure to N20 at 320 K failed to produce any absorption bands due to adsorbed species in the range 4000-900 cm-l. 1 21 20 rl E 2. 3 2110 - I ; 21 00 / / / / I I I I 400 500 600 700 L 300 calcination tempera t ure/K FIG.4.-Plot of the stretching frequency of carbon monoxide adsorbed on SnO, .0.55CuO at 320 K against the pretreatment temperature for evacuation (0) and evacuation and oxygen treatment ( x ). ADSORPTION OF CARBON MONOXIDE Carbon monoxide was adsorbed to produce an intense, sharp band in the range 2100-2120cm-l together with weak carbonate bands at 1530 and 138Ocm-l. The latter two bands showed a progressive decrease in intensity as the pretreatment calcination temperature was increased, and extrapolation showed that no carbonate formation should occur for samples pretreated by evacuation on oxygen treatment above ~ 7 0 0 K, similar to the observation for CO, adsorption above. The band at ~ 2 1 0 0 cm-I can be readily assigned as due to the stretching vibration of adsorbed carbon monoxide.The exact position of the band was sensitive to pretreatment conditions : as the evacuation temperature was increased from 320 to 673 K the position of v(C0) changed continuously from 2100 to 21 10 cm-l, but when the sample was oxygen treated after evacuation above ~ 5 0 0 K, the frequency increased still further at 2121 cm-I (fig. 4). The absorption band became noticeably weaker and sharper for higher pretreatment temperatures, and could be removed completely from all samples by evacuation at 320 K for a few minutes.P . G . HARRISON AND E. W. THORNTON 2609 The position of the CO stretching band is more typical of CO adsorbed on copper metal, where the band has been variously reported at abound 2110 cm-1,14-20 than of CO adsorbed on copper(I1) oxide, where the vibration has been observed at 2130-2140 21* 22 On oxidised copper (where the adsorption site was presumed to be Cu+), the band has been reported at 2113 cm-l,19 whilst the CO complex with copper(1) chloride in water has v(C0) = 2112 Thus it would seem more probable that the adsorption site in the present study is a coordinatively unsaturated copper atom or ion in the 0 or +1 oxidation state, but which can be oxidised to Cu" at higher temperatures by molecular oxygen.On heating in CO above ~ 5 0 0 K the transmission of the sample disc decreased to such an extent that infrared experiments were no longer possible. Similarly, following evacuation or evacuation followed by oxygen treatment above M 730 K, the transmission of pressed discs was too low for infrared experiments to be carried out.wavenumber /cm-' FIG. 5.-Infrared spectra of nitric oxide adsorbed on to a SnOz .0.55CuO disc. (1) Background after evacuation at 508 K for 3 h ; during subsequent exposure to NO at 320 K at (2) 660 N m-2 and (3) 3.3 kN m-2. (4) Adsorption of NO (320 K, 4.0 kN m-2) on a disc after evacuation and oxygen treatment at 602 K. ADSORPTION OF NITRIC OXIDE No new infrared absorption bands were detected during exposure of catalyst discs preevacuated at -= M 380 K to nitric oxide at 320 K. For sample discs evacuated above x380 K NO was adsorbed to produce a weak band at 1873 cm-l (403 K evacuation), shifting as the pretreatment temperature was increased to 1880 cm-' (643 K evacuation and oxygen treatment) (fig.5). The absorption band was removed by pumping off the gas phase at 320 K, and was also displaced readily and completely by addition of CO to the gas phase at 320 K. The position of the band is close to that reported for NO adsorbed on silica-supported CuO (1890 cm-1).22 The position of the v(N0) absorption band may be taken as indicative of the type of bonding of the adsorbate to the surface, and from the classifications of Terenin et uZ.24-27 (reviewed more recently by Shelef and Kummer),28 it can be surmised that the bonding in the present case is either double bond ionic : -l=&=o or coordinative : J 1- :NO2610 TIN OXIDE SURFACES where ] represents the surface site, probably a coordinatively unsaturated copper ion or atom since separate experiments showed that no new bands were formed in this region during exposure of pure tin@) oxide discs (pretreated by evacuation and oxygen exposure at ~6623 K) to NO at 320 K.No bands due to oxidised species (such as NO2, NO; or NOT) were observed. ADSORPTION OF CO+NO MIXTURES After exposure of a 623 K evacuated and 0,-treated disc to a CO +NO mixture at 320 K for 1 h, new absorption maxima appeared at 1500 and 1350 cm-l due to a carbonate species and at 2120 cm-l due to adsorbed CO [fig. 6(a)]. Following 2500 r,,,, 21 0 2 wavenumber /cm-' (6) 0 K) FIG. 6.-Spectra recorded during the adsorption of CO+NO mixtures on SnOz .0.55CuO after activation in oxygen at 623 K. (a) (1) Background. After exposure to CO (6.7 kN m-2, 320 K) (2) and after exposure to 23 % NO in CO (8.65 kN m-2) at (3) 320 K for 1 h ; (4) 388 K for 0.25 h ; and (5) 473 K for 0.25 h.(b) Continued from (a) (5). Spectrum recorded after evacuation at 320 K for 5,10,15,20,25,30 and 35 mh. (c) (1) Background. After exposure to CO (6.7 kN m-2, 320 K) (2) and after exposure to 16.6 % NO in CO (8.0 kN m-2) at (3) 320 K for 0.25 h ; (4) 473 K for 0.5 h ; and 473 K for 1.5 h. subsequent heating in the mixture at 388 K for 0.3 h, the carbonate bands weakened slightly and a further band appeared at 2189 cm-l. On further heating at 473 K for 0.25 h, the carbonate bands shifted to 1480 and 1360 cm-' and weakened still further, whilst the band at 2189 cm-l grew stronger and the v(C0) band shifted to 2112 cm-l. On evacuation at 320 K the 2112 cm-1 band was removed, but the 2189 cm-l band was unchanged [fig.6(b)]. During the reaction, absorption bands due to N20 and C02 appeared in the spectrum of the gas phase. On heating in a CO+NO mixture for longer periods such that the gas phase became depleted in NO, the overall transmission of the disc decreased and the spectra at 2000-2200cm-l decreased in intensity considerably [fig. 6(c)].P. G . HARRISON AND E. W. THORNTON 261 1 The 21 89 cm-l band was formed readily on heating the catalyst discs in CO +NO mixtures [p(CO) > p(NO)] at ~ 4 7 0 K. In order to investigate the nature of the surface species responsible for this band, further experiments were carried out with isotopically substituted reactants. The spectra obtained are illustrated in fig.7(a)-( f ). The full lines are spectra recorded after heating in the appropriate reaction mixture at 463-483 K for 0.3 h and cooled without evacuation. During the reaction the CO invariably underwent a decrease in stretching frequency. Typical observed fre- quencies and shifts were : v ( C 0 a ) before Av during spectrum gas phase composition reactionlcm-1 reaction Icrn-1 7(a) normal mixture 2120 -8 7(b) (91 % 13C)CO+N0 207 1 -8 7(4 (95 % ~~N)NO+CO 2120 -6 7(d) (91 % 13C)C0+(95 % ISN)NO 2071 -7 7(e) (63 % 180)CO+N0 2070 -7 21 14) -41 The dashed lines show spectra recorded after removal of the gas phase at 320 K. The magnitudes of the AI3C, AISN and A13C+ 15N shifts for the 2189 cm-I band (table 1) are consistent with a surface isocyanate species by comparison with the shifts observed in other studies.When l*O-enriched CO was used, no isotopic shift was observed [fig. 7(e)], the band remaining sharp and symmetrical at 2189 cm-l. However, when 40 % l 8 0 - enriched NO was employed [fig. 7 ( f ) ] the band shifted to 2184 cm-l and developed TABLE 1 .-INFRARED ABSORPTION BANDS AND ISOTOPIC SHIFTS (cm-') FOR ISOCYANATE SPECIES ADSORBED ON CATALYST SURFACES AND FOR MODEL COhlPOUNDS catalyst or model compound Ru/AI,O~ Rh/A1903 PtIA1203 PtJA1203 CuO/Si02 d cu Ru/Si02 f Rh/A1203 g Si+NCO h Sn02 .0.55 CuO Si+CN Si+NC NCO- in KCI Hg(CN0)2 j vas(NC0) 2238 2264 2200 2380 21 80 2235 2315 21 89 221 8 2100 2195.8 2200 2220) 60 17 60 10 13 62 11 17 73 51 30 39 36 58.7 17.4 7.9 76.3 30 4 Shift to lower wavenumber compared with figure in column 2.b Ref. (2) and (3). Ref. (1). d Ref. (5). e Ref. (4). f Ref. (6). g Ref. (7). h B. A. Morrow and I. A. Cody, J.C.S. Farday I, 1975, 71, 1021. i V. Shettino and I. C. Hisotsuma, J. Phys. Chem., 1970, 52, 9. j Ref. (29).2612 TIN OXIDE SURFACES wavenumber /cm-I (4 (4 FIG. 7.Tdrared spectra recorded after exposure of SnOz .0.55CuO discs to isotopically-substituted CO+NO mixtures at 463-483 K for 20-30 min. Full = before evacuation, dotted = after evacua- tion at 320 K. The gas mixture compositionsand pressures were 16.6 % NO, 8.0 kNm-2 for (a)-@) and (f), 21.7 % NO, 6.12 kN m-2 for (e). (a) Normal abundances. (b) CO = 91 atom % 13C, normal NO. (c) NO = 95 atom % 15N, normal CO. (d) NO = 95 atom % 15N, CO = 91 atom % 13C. (e) CO = 60 atom % "0, normal NO.(f) NO = 41 atom % l*O, normal CO. a shoulder on the low wavenumber side. In order to investigate this band further, the spectral trace and background were digitised at 2cm-l intervals from 2100 to 2300 cm-l using a 10 x expanded abscissa scale. The resultant data were converted to an absorbance spectrum and analysed by adapting the data to a form suitable for input to a least-squares curve-fitting program. The envelope was thus resolved into %Po L I I 21 00 2200 2300 wavenumber /cm-l FIG. 8.-Pseudo-antisymmetric stretching bands of isotopically normal and1 80-substituted isocyanate species formed on the surface of SnOz .0.55CuO in the conditions as for fig. 7 ( f ) ; experimental points, fitted band profile and calculated components.P .G . HARRISON A N D E. W. THORNTON 2613 two overlapping Lorentzian line shapes centred at 2189 and 2172cm-l, with the low wavenumber band accounting for 43.5 % of the area under the curve (fig. 8). The final sum of squares was rather high, but attempts to fit the data to one band only resulted in a much worse fit, and inclusion of more than two bands resulted in unreal computed parameters (e.g. negative width or depth) for one of the bands. In different calculations the position of the higher wavenumber component was nearly constant (+2 cm-I), but both the position and intensity of the isotopically shifted band were sensitive to the estimated starting parameters used. Consequently the uncertainty in the band position is about &5 cm-l. The calculated half-height widths of the two bands [r+(2189) = 51 cm-l and r,(2172) = 42crn-l-J are similar to the half-height widths of the other isocyanate bands in fig.6 which all have in the range 35-55 cm-l. The isocyanate band was stable to evacuation up to ~ 4 4 0 K, but was removed entirely by evacuation at 480 K. It was also stable to oxygen and nitric oxide (1.3 kN m-2) at 320-440 K, but could be removed by either reagent at higher tempera- tures, and was removed entirely by exposure to water vapour (130 N m-,, 320 K, 0.5 h). Exposure of a disc exhibiting a preformed isotopically normal band at 2189 cm-l to 180-enriched 0, (63 % or 80-enriched NO (41 % at 383 K for 1 h did not shift the band or produce any detectable asymmetry. In order to investigate possible incorporation of oxygen from the oxide surface into the isocyanate, a sample disc was evacuated and then treated with 180-enriched O2 (63 % l80), all at 623 K.After subsequent reaction using isotopically normal NO and 80-enriched CO (60 % at 473 K, a single symmetrical band remained at 2189 cm-l on cooling and evacuation at 320 K. DISCUSSION Fuller and Warwick have studied the kinetics of the oxidation of carbon monoxide catalysed by coprecipitated SnO, + CuO gels both by oxygen Briefly, for near optimum composition catalysts (SnO, xCu0, 0.5 < x < 0.55) increasing calcination pretreatment temperature increased the catalytic activity for both reactions up to ~ 7 5 0 K, followed by a decrease in activity and surface area for further increase in calcination temperature.This decrease was attributed to the onset of CuO crystallisation as a separate phase above ~ 7 2 0 K, as confirmed by powder X-ray crystallography. The present infrared results show that a number of changes occur on calcination in vacuo. After evacuation at 320 K the (green) disc adsorbs CO with a stretching frequency more characteristic of CO adsorbed on a partially reduced CU' site than a Cu" species. Sites suf3ciently active to adsorb nitric oxide are only produced by evacuation above x400 K, and the oxide surface is essentially completely irreversibly dehydroxylated by evacuation above z 600 K, although some hydroxyl groups remain in the bulk or trapped in closed pores. The change in properties on heating above 720-50 K (either in vacuo or in oxygen) was also manifest in the present study, both as a loss in infrared transmission and, as indicated by lower temperature experiments, with the extent of C 0 2 adsorption (fig.3). The nature of the surface and bulk rearrangement accompanying CuO crystallisation is undoubtedly complex and out of the scope of the present study. Whether the large loss in infrared transmission is associated with the CuO crystallisation or with the conduction electrons of the partially reduced Sn02 support [tin-1 19m Mossbauer spectroscopy of a 740 K-evacuated disc showed no evidence for a tin(@ species, in contrast with the previously reported results for partially reduced (by carbon mon- oxide) SnO, where a Sn" species was detected]l* is uncertain. The surface changes and by nitric oxide.2614 TIN OXIDE SURFACES continuously with increasing calcination temperature in the range 300-700 K.In particular, the sites for carbonate formation during carbon dioxide adsorption (presumably oxide ions in the ion-exchanged Cu-containing surface layer) decrease in concentration and eventually disappear at about the CuO-cry stallisation temperature. This decrease and eventual loss of surface oxide ions may be a factor responsible for CuO crystallisation, the oxide ions helping to stabilise the surface structure at lower temperatures. The environment of the copper ions responsible for CO adsorption also changes [as reflected by the change in v(CO,,,)]. After evacuation, the copper ions are in a partially reduced state, but can be returned to the +2 oxidation state by oxygen treatment above x550 K.During the initial stages of catalysis of the CO +NO reaction by a 623 K evacuated disc, a carbonate species is produced on the surface as well as adsorbed carbon monoxide exhibiting a stretching frequency (2 120 cm-I) characteristic of adsorption on an oxidised copper (Cu2+) site. As the reaction proceeds, however, the carbonate is removed and the v(CO,,,) band shifts to lower frequency, characteristic of a partial reduction of the surface and a lowering in the oxidation state of the copper ions. This corresponds to the partial reduction reported during the kinetic studies of the CO + O2 and CO +NO reactions catalysed by this oxide. The present infrared spectroscopic results also show the existence of an isocyanate surface species, formed rapidly during the initial stages of the reaction at 670 K.Previous studies reporting this surface species have assumed that it was formed by initial dissociative chemi- sorption of NO : 21 +NO -N 1-N+]-0 where ] represents some general surface site, followed by chemisorption of CO : 1-N+CO 1-NCO. (2) The 1-N surface species also provide a route to N20 formation as a product : 1-N+NO 4]+N20. (3) Alternatively, mechanistic studies at 300 K on supported ruthenium and at 473- 573 K on supported rhodium ' both invoked more complex reaction schemes. The latter study still postulated the formation of the 1-N intermediate and subsequent reaction via routes (2) and (3) above. If reaction (2) were the route to isocyanate formation in the present study, then the oxygen atom of the -NCO species would be expected to be derived from the carbon monoxide reactant.This, however, cannot be the case since reaction with 80-enriched CO resulted in an isotopically normal surface species. In contrast, when the reaction was carried out with 180-enriched NO, the absorption band was shifted slightly and developed a shoulder on the low-wavenumber side. Total isotopic shift of the band is not realised due to the paucity of the l80-enrichment in the NO (41 atom % Resolution of this envelope into two components gave an isotopic shift A l s o = 17 cm-l, in reasonable ageement with the result reported for a -Si-NCO species on silica (table 1). Separate experiments failed to show any isotopic exchange of the isocyanate species either with 80-enriched O2 or 80- enriched NO.\ /P . G . HARRISON AND E . W. THORNTON 2615 These results imply that the mechanism described in eqn (1) and (2) above cannot be operating in this case. The most probable alternative route for the formation of surface isocyanate is one involving initial dissociative chemisorption of CO : 2]+co +]-c+]-0 (4) followed by reaction of the adsorbed carbon atom with nitric oxide : 1-C + NO -+ 1-CNO (5) and subsequent isomerisation of the resultant fulminate to the stable isocyanate structure, probably via a cyclic oxaziranyl-type intermediate : 1-CNO + +]+ This type of rearrangement is well known to occur for alkali metal fulminates as well as nitrile oxides and covalent organometallic fulminates on cautious heating2 The identity of the surface site involved in these reactions is not known, but it is reasonable to assume that it is a coordinatively unsaturated copper ion in the 0 or + 1 oxidation state.Dissociative chemisorption of CO on such a site at 470 K is not unreasonable especially as the lowering of the vads(CO) frequency below that of the gas phase implies some weakening of the C=O bond. The mechanisms proposed in eqn (4)-(6) do not mean that eqn (1) and (3) are not operating since these seem by far the most likely reactions leading to N20 formation? and N20 and C02 were detectable in the gas phase by infrared spectroscopy during the reaction. Kinetic results on this reaction have shown that this catalyst is very selective towards NO reduction to N2, but any comparison between the present infrared study and kinetic results should be made with caution, since the kinetic results refer mainly to a steady state partially reduced catalyst while the infrared results deal with the freshly activated or only slightly reduced catalyst.Since the steady state catalyst, like the thermally sintered oxide, exhibits negligible infrared transmission, the surface species present on the steady state catalyst are inaccessible to infrared spectroscopy. Thus, the mechanism found for isocyanate formation during the initial stages of this reaction is one which previous authors have apparently not considered and have not tested for. We thank the International Tin Research Institute and the S.R.C. for support in the form of a CASE Award (to E. W. T.). M.L. Unland, J. Phys. Chenz., 1973, 77, 1952. ’ M. L. Unland, J. Catalysis, 1973, 31, 459. M. L. Unland, Science, 1973, 179, 567. R. T. Rewick and H. Wise, J. Catalysis, 1975, 40, 301. J. W. London and A. T. Bell, J. Catalysis, 1973, 31, 96. M. F. Brown and R. D. Gonzalez, J. Catalysis, 1976, 44, 477. ’ H. Arai and H. Tominaga, J. Catalysis, 1976, 43, 131. * M. J. Fuller and M. E. Warwick, J. Catalysis, 1974, 34, 445. M. J. Fuller and M. E. Warwick, J. Catalysis, 1976, 42, 418. lo E. W. Thornton and P. G. Harrison, J.C.S. Faraday I, 1975, 71,461. 1-8 3261 6 TIN OXIDE SURFACES l 1 V. G. Amerikov and L. A. Kasalkina, Kinetics and Catalysis, 1971, 12, 137. l2 K. Nakamoto, Infrared Spectra of Inorganic and Coordination Compounds (Wiley, London, l3 L. H. Little, Infrared Spectra of Adsorbed Molecules (Butterworth, London, 1966). l4 R. P. Eischens, W. A. Pliskin and S. A. Francis, J. Chem. Phys., 1954, 22, 1786. l5 A. W. Smith and J. M. Quets, J. Catalysis, 1965, 4, 163. l6 A. W. Smith, J. Catalysis, 1965,4, 172. l7 J. Pritchard and M. L. Sims, Trans. Faraday SOC., 1970,66,427. l8 A. M. Bradshaw and J. Pritchard, Proc. Roy. SOC. A, 1970,316, 160. l9 H. G. Tomkins and R. G. Greenler, Surface Sci., 1971, 28, 194. 2o M. A. Chesters, J. Pritchard and M. L. Sims, Chem. Comm., 1970, 1454; and references 21 D. A. Seanor and C. H. Amberg, J. Chem. Phys., 1965,42,2967. 22 J. W. London and A. T. Bell, J. Catalysis, 1973, 31, 32. 23 J. 0. Alben, L. Yen and N. J. Farrier, J. Amer. Chem. SOC., 1970,92,4475. 24 A. N. Terenin and L. M. Roev, Acres du Deuxizme Congds International de CataZyse (Technip, 25 L. M. Roev and A. N. Terenin, Optika i Spektroskopia, 1959, 7,759. 26 L. M. Roev and A. V. Alekseev, Elementary Photoprucesses in Molecules, ed. B. S. Neoprent 27 A. N. Terenin and A. V. Alekseev, J. Catalysis, 1965,4,440. 28 M. Shelef and J. T. Kurmner, Chem. Eng. Progr. Symp., 1971, 67, 74. 29 W. Beck, Organometal. Chem. Rev. A, 1971, 7,159. 1970). therein. Paris, 1961), vol. 2 ; and references therein. (Consultants Bureau, New York, 1968). (PAPER 7/1924)