J. Chem. SOC., Faraday Trans. I, 1984, 80, 531-541 Infrared Study of the Interactions between NO and CO on Rh/Al,O, Catalysts BY EDWARD A. HYDE AND ROBERT RUDHAM Department of Chemistry, The University, Nottingham AND COLIN H. ROCHESTER* Department of Chemistry, The University, Dundee DD1 4HN Received 6th April, 1983 Infrared spectra of NO and NO+CO mixtures on Rh/Al,O, are reported. At ambient temperatures NO alone was adsorbed non-dissociatively and gave no detectable reaction products. At T 2 413 K NO underwent partial decomposition to N,O(g) at surface sites which were rapidly poisoned as the reaction proceeded. The addition of NO to preadsorbed CO eliminated spectral bands characteristic of linear and bridged carbonyl species, but maxima due to gem-dicarbonyl species remained.The main products of the CO + NO reaction detectable by infrared spectroscopy were N,O(g), adsorbed isocyanate and CO,, either gaseous or chemisorbed. The extent of isocyanate formation depended on the relative amounts of NO and CO present, the gas-phase pressures of NO and CO and the order of addition of NO and CO to the Rh/Al,O, catalyst. The order of addition did not affect the generation of N,O. The results are discussed in terms of reaction mechanism and types of catalytically active adsorption site. Dissociative adsorption of CO, on Rh dispersed on alumina led to an adsorbed CO species giving an infrared band at 2015 cm-'. The reaction on supported rhodium catalysts between carbon monoxide and nitric oxide to give nitrogen and carbon dioxide is incompletely understood, although of great importance with respect to vehicle emission-control catalysis.The most detailed relevant information concerns the non-dissociative adsorption of carbon monoxide which has been well characterized by infrared spectr~scopy.~-~ The formation of bridged species Rh,CO, linear species RhCO and gem-dicarbonyl species Rh(CO), depends on the preparation of the samples which in turn influences the dispersion of the metal. Significant factors are the size of the rhodium particles or two-dimensional arrays of rhodium atoms, interactions between metal atoms and the oxide support and the relative areas of different exposed crystal faces. Dissociative adsorption of carbon monoxide may also Nitric oxide is both dissociatively and non-dissociatively adsorbed on rhodium, and nitrogen, nitrous oxide and oxygen have been detected as products of subsequent desorption at elevated temperature^.^, 11-13 Infrared studies have been less detailed than for the adsorption of carbon monoxide but three bands in the 1700-2000 cm-l spectral region have been reported and are generally assigned to NO stretching vibrations of RhNO+, RhNO and RhNO- surface species.l*q l5 However, results from a study of interactions between nitric oxide and a RhY zeolite suggest that two of the bands could be attributed to the symmetric and asymmetric stretching vibrations of a gem-dinitrosyl complex, Rh(NO),, of rhodium.16 Evidence exists for the simultaneous adsorption of a nitric oxide and a carbon * probably at step or defect surface sitesg* lo 53 1532 REACTION OF co +NO ON RHODIUM monoxide molecule at the same rhodium atom to give a surface Rh(NO)(CO) species.14* l5 Mixtures of nitric oxide and carbon monoxide adsorbed on supported rhodium also lead to the appearance of infrared bands due to isocyanate species,14-17 probably adsorbed on the surface of the oxide 18* l9 A band at 2195 cm-l has been ascribed to the transitory appearance of isocyanate species liganded to rhodium.15 Excess of nitric oxide apparently inhibits the formation of isocyanate,15 possibly because dissociative adsorption of nitric oxide is increasingly unfavourable with increasing nitric oxide pres~ure.~~ l2 A previous paper7 reports a study of the adsorption of hydrogen and carbon monoxide on Rh/Al,O, catalysts prepared from dispersions of rhodium nitrate, rhodium sulphite and chloropentamminerhodium(II1) chloride on alumina.This study of the adsorption of carbon monoxide plus nitric oxide mixtures on Rh/Al,O, samples prepared by reduction of dispersions of the same three rhodium salts was aimed at probing the involvement of distinguishable forms of adsorbed carbon monoxide and nitric oxide in the formation of Rh(CO)(NO) and isocyanate species, and in assessing to what extent the various surface species were intermediates in the overall catalytic reaction to nitrogen and carbon dioxide. EXPERIMENTAL Rhodium (1.2 wt %) supported on y-alumina (Condea Chemie, surface area 132 m2 g-l) was prepared as before’ by reduction of dispersions of rhodium nitrate, rhodium sulphite and chloropentamminerhodium(m) chloride on alumina in a flow of hydrogen (1 atm,? 773 K, 1 h).Self-supporting discs compacted at 138 MN m-2 were at ca. 303 K during spectroscopic examination with a Perkin-Elmer 125 spectrophotometer. Nitric oxide was purified by passage over potassium hydroxide and repeated fractional distillation20 until the only impurity detectable by infrared spectroscopy was nitrous oxide at < 0.01%. Carbon monoxide, nitrous oxide and carbon dioxide (all research grade X) and nitrogen (spectroscopic purity) were used as received in 1 dm3 bulbs. RESULTS The spectroscopic results for the adsorption of NO and NO+CO mixtures on Rh/Al,O, prepared by reduction of Rh(NO,),/Al,O,, Rh,(SO,),/Al,O, and [Rh(NH3),C1]C1,/A1,0, were not significantly affected by the nature of the salt precursor to rhodium metal.Even for Rh,(SO,),, which underwent incomplete red~ction,~ the same infrared bands were observed as for samples prepared from the other two salts and no new bands, absent for the other salts, appeared in the spectra. In general, the infrared bands were more intense for Rh/Al,O, derived from Rh(NO,),/Al,O, and therefore it is primarily for these samples that spectra are shown here. Isotherms for the adsorption of carbon monoxide and hydrogen have established that, of the three salts, Rh(NO,), gives the most highly dispersed rhodium with the greatest number of exposed adsorption sites.7 Spectra of NO adsorbed on y-Al,O, which had been preheated in hydrogen at 773 K exhibited weak bands at 1590 and 1230 cm-l.The latter compares with a band at 1220-1 240 cm-l previously reported for NO on alumina which had been pre-evacuated at > 673 K 2 1 Three strong broad bands at 1910, 1830 and 1740 cm-l for NO on Rh/Al,O, (fig. 1) are assigned to NO stretching vibrations of NO molecules bonded to surface rhodium atoms.14-16 The exact band positions were a function of surface coverage. At low coverages a single maximum [fig. 1 (b)] appeared at ca. 1865 cm-l t 1 atm = 101 325 Pa.E. A. HYDE, R. RUDHAM AND C. H. ROCHESTER 533 shifting towards 1910 cm-l [fig. 1 (c) - (e)] with increasing pressures of NO. At high pressures the maxima at 1910 and 1830 cm-l contained contributions from the band envelopes of the R and P branches due to the fundamental stretching vibration of NO gas, which also gave a weak maximum at 1876 cm-l [fig.1 ( f ) ] . 2 2 Adsorbed NO was largely desorbed by evacuation at ca. 303 K although a weak band remained at 00 1800 2000 1800 wavenum berlcm -' Fig. 1. Infrared spectra of NO on Rh/A1,0, at equilibrium pressures/kN rn-, of (a) 0 ; (6) 0.051, (c) 0.205, ( d ) 1.31, (e) 8.81 and cf) 34.3, followed by subsequent desorption to (g) 8.2 N m-, and (h) evacuation (5 min). Prolonged contact between NO gas and Rh/Al,O, for several days at room temperature failed to give any reaction products in the gas phase with detectable infrared spectra. However, heat treatment of Rh/A120, at 413 K in NO (13 kN m-,) for 30 min generated small amounts ( < 0.15 %) of N,O gas. The intensities of maxima at 2218 and 2240 cm-l due to the P and R branch band envelopes of one of the fundamental vibrations of N20 decreased in intensity after subsequent heat treatment in NO at 633 K (30 min) and became very weak after heating the disc at 823 K (30 min) in NO.The removal of N,O was accompanied by the appearance of a band at ca. 1617 cm-l characteristic of NO,, also at very low partial pressure compared with the added pressure of NO. Infrared bands due to the fundamental vibration of NO gas were unchanged in apparent intensity during the sequence of thermal treatments, confirming that an undetectably small fraction of the NO had undergone reaction. However, in accordance with previous results,14$ l5 heat treatment of Rh/Al,O, in NO at increasing temperatures progressively caused the desorption of adsorbed species responsible for the infrared bands at 1830 and 1740 cm-l and the species were not reformed when the disc was cooled in NO from the treatment temperature to ca.303 K for spectroscopic examination. In contrast, the band at 1910 cm-l became more prominent and shifted to 1920 cm-l. Fig. 2 shows spectra of Rh/Al,O, after the addition of CO to adsorbed NO. After NO adsorption the gas pressure was reduced to a level for which the bands due to534 REACTION OF co +NO ON RHODIUM gas phase NO made negligible contributions to the spectrum but the bands at 1910, 1830 and 1740 cm-l due to adsorbed NO remained prominent [fig. 2(a)]. These three bands were considerably reduced in intensity following the addition of CO. The appearance of maxima at 2095 and 2025 cm-l due to vibrations of gem-dicarbonyl contrasted with tlie corresponding result in the absence of preadsorbed NO 1900 1700 wavenum ber/cm -* Fig.2. Infrared spectra of Rh/Al,O, after (a) exposure to NO (2.03 kN m-2) and evacuation to 16 N md2, followed by the addition of CO to total gas-phase pressures (in N m-2) of (b) 33, ( c ) 430 and ( d ) 980 N m-2. where a maximum at ca. 2050cm-l due to linearly adsorbed CO dominated the spectrum at low surface coverages7 The formation of an appreciable surface population of the linear complex RhCO was apparently unfavourable in the presence of NO. With increasing pressure of CO the maxima at 2095 and 2025 cm-l grew in intensity [fig. 2(c)] and showed evidence for splitting into four bands at 2100, 2085 (sh), 2040and 2010 cm-l [fig.2(d)]. Bands whichappeared at 2260,2235 and 2185 cm-l [fig. 2(c) and (d)] may be ascribed to products of a reaction between NO and CO catalysed by rhodium. The reaction was not catalysed by alumina alone in the absence of rhodium. The band at 2235 cm-l contained a contribution from the spectrum of a low partial pressure of nitrous oxide which was formed as a reaction product in the gas phase. Similarly the spectrum of CO gas made a small contribution to the band at 2 185 cm-'. However, the three bands were primarily due to vibrations of adsorbed products of reaction. Similar bands have previously been ascribed to isocyanate species.14-17 Increasing the CO pressure to 5.5 kN m-2 had little effect beyond the result shown in fig. 2(d), and subsequent evacuation reduced the intensities of the bands due to adsorbed isocyanate to approximately one-half their initial values before evacuation.Residual gem-dicarbonyl species were also retained on the surface after evacuation and a broad weak region of absorption remained in the spectra at 1 940- 1 700 cm-l. The maximum at 2260 cm-l was absent from spectra recorded after exposure of Rh/Al,O, to NO at high pressure [fig. 3(a)] followed by the admission of CO with the NO still present in the gas phase [fig. 3(b)]. Furthermore, the band at 2185 cm-' was not discernible and the maximum at 2235 cm-l could be largely ascribed to N20E. A. HYDE, R. RUDHAM AND C. H. ROCHESTER 535 in the gas phase [fig. 3(e); the spectrum of N,O gas overlaps the spectrum of CO in the 2000-2300 cm-l spectral region].Solymosi and Sarkany have reported a similar result for NO+CO on 5% Rh/A1,0, prepared from rhodium(II1) ch10ride.l~ Non- dissociatively adsorbed NO was gradually displaced from the rhodium surface [fig. 3(b) and (c)] and reductions in the intensities of the bands due to gem-dicarbonyl - 60- E z E + 40- s 0 .- m v) .3 8 0 / I b J - L 1 I I I I 2 0 k 2100 2000 2100 2000 1900 1700 wavenum ber/cm-' Fig. 3. Infrared spectra of Rh/A1,0, (a) in equilibrium with NO [8.12 kN m-2; spectrum ( d ) is of gas phase alone], (b) 30 min and (c) 19 h after the subsequent addition of CO to give a total pressure of NO+CO of 19.0 kN md2. (e) Spectrum of gas phase after an experiment similar to that which led to (c). species also occurred. The maximum at 2235 cm-l grew in intensity because of the continued generation of both N,O gas and some surface isocyanate contributing to the absorption intensity at that position. A new maximum also developed at 1590 cm-l When Rh/Al,O, discs were exposed to mixtures of NO and CO several usually weak bands appeared in the spectral region 1700-1200cm-1.The exact positions and relative intensities of the bands were a function of the particular experiment being carried out. Typical band positions were 1645, 1580, 1480, 1305 and 1230 cm-l, which compare with bands at 1638, 1477 and 1238 cm-I for CO, adsorption and at 1655, 1580, 1290 and 1230 cm-l for NO, adsorption on alumina which had been preheated in hydrogen at 773 K. Parkyns,' has reported that NO, on alumina gave two prominent bands at ca. 1600 and 1240 cm-l which were assigned to vibrations of nitrate ions.Exposure of Rh/Al,O, to N,O gave no infrared bands which could be ascribed to products of chemisorption on either rhodium or alumina although a band at 2242 cm-l due to weakly adsorbed N,O on alumina,, was observed. The present results suggest that trace amounts of surface carbonate and nitrate were formed on the alumina support during prolonged contact between NO plus CO and Rh/Al,O, at room temperature. Slight increases in absorption intensity at ca. 2350 cm-l, which [fig- 3 (c)l *536 REACTION OF co 4- NO ON RHODIUM were reversed by removal of the gaseous mixtures from contact with Rh/Al,O,, also provided evidence for the formation of trace amounts of CO, in the gas phase.The addition of NO to Rh/Al,O, which was already in contact with CO [fig. 4(a)] resulted in the disappearance of the infrared band at ca. 2050 cm-l due to linear RhCO species (fig. 4(b)]. Arai and Tominaga14 reported the same effect and also observed 2200 . 2000 1900 1700 wavenum berlcm-' Fig. 4. Infrared spectra of Rh/Al,O, (a) in equilibrium with CO (23 N m-2) and (b) 20 min and (c) 77 min after the subsequent addition of NO to give a total pressure of NO+CO of 158 N m-,. that heat treatment of Rh/A1,0, after the addition of CO followed by NO led to the appearance of a new infrared band at 2101 cm-l. The generation here, at ambient temperature, of the species responsible for this band was suggested by the marked increase in intensity at ca. 2100cm-l which coincided with the already existing maximum due to gem-dicarbonyl species7 A strong maximum at ca.1830 cm-l (fig. 4) due to NO stretching vibrations of adsorbed NO simultaneously appeared and dominated the 1950-1700cm-1 spectral region for low pressures of NO. At high pressures bands due to adsorbed NO also appeared at ca. 1905 and 1730 cm-l. Evidence for the displacement of bridged species Rh,CO [broad shoulder centred at 1895 cm-l in fig. 4(a)] was not always easy to identify because of overlap between broad bands due to adsorbed CO and NO. However, a maximum at 1890 cm-l due to Rh,CO species was particularly prominent in spectra of samples prepared by reduction of Rh,(SO,),/Al,O, and then exposed to C0.7 The maximum was increasingly reduced in intensity with increasing pressures of added NO, confirming14 that bridged CO was removed from the rhodium surface after exposure to NO.Gem-dicarbonyl species were more resistant than bridge-bonded or linearly adsorbed CO to removal by NO. However, reductions in the intensities of the bands at 2100 and 2035 cm-l occurred with time [fig. 4(c)] and were enhanced when the NO was in excess over the CO present. The destruction of gem-dicarbonyl species by NO was best exemplified in one experiment involving Rh/A1,0, derived from [Rh(NH,),Cl]Cl,/Al,O,. The disc was exposed to CO (29.3 kN m-,) and then evacuated at room temperature to remove bridged CO and a high proportion of linearly adsorbed CO. Bands remaining in the spectra at 2100 and 2035 cm-l [fig. 5 (a)] could be assigned primarily to Rh(CO), species although the latter also contained aE.A. HYDE, R. RUDHAM AND C. H. ROCHESTER 537 contribution due to RhC0.' The bands showed a progressive decrease in intensity with increasing added pressures of NO [fig. 5(b)-(e)] until the carbonyl species had been completely destroyed [fig. 5 ( f ) ] . At the same time maxima at 19 10,1830 and 1740 cm-l due to adsorbed NO increased in intensity, the band at 1830 cm-l being the most prominent at low pressures of NO but undergoing less growth relative to the other two bands as CO was increasingly displaced from the surface by NO. r-- ,Ot - 2300 2100 I ' / I I 1900- 1700 2200 2000 wavenum berlcm-' Fig. 5. Infrared spectra of Rh/Al,O, derived from [Rh(NH,),Cl]Cl,/Al,O, after (a) exposure to CO (29.3 kN m-,) and evacuation (16 h, 293 K), (b)-Cf) subsequent contact with NO at pressures/kN m-, of (b) 0.24 (8 min), (c) 0.24 (24 min), ( d ) 1.31 (7 min), (e) 1.31 (21 min) and cf) 2.57 (24 min), followed by evacuation and readmission of CO at pressures/kN rn-, of (g) 0.33 (10 min) and (h) 2.21 (36 min).The formation of surface isocyanate species by interactions between NO and preadsorbed CO apparently depended on the existence of linearly bonded and/or bridge-bonded CO on the rhodium surface. If the CO pressure was sufficient to ensure the presence of these species [fig. 4(a)] then the subsequent addition of NO led to the appearance of bands at 2260 and 2235 cm-' [fig. 4(b)] due to isocyanate. In contrast only a very weak broad increase in absorption intensity at 2230-2260 cm-l occurred [fig.5 ( b ) - ( f ) ] for Rh/Al,O, which had been exposed to CO (29.3 kN m-,) and subjected to evacuation before increasing pressures of NO were admitted to the cell. The displacement of gem-dicarbonyl species by the adsorption of NO was not accompanied by the formation of adsorbed isocyanate groups. However, partial removal of NO gas by brief evacuation followed by the readmission of CO led to the appearance in the spectra [fig. 5 (f)] of bands at 2260,2235 and 2 185 cm-l characteristic of isocyanate groups. An analogous result was obtained after the exposure of Rh/Al,O, to high pressures of NO followed by addition of CO, brief evacuation to remove both gases but not adsorbed species, and readmission of CO alone. Bands at 2260 and 2235 cm-l were present in the spectra after the second contact with CO whereas the band at 2260 cm-l was absent (fig.3) when NO was also present and had been admitted first.538 REACTION OF CO+NO ON RHODIUM Equilibration of Rh/Al,O, with NO at high pressure before mixing with CO inhibited the formation of isocyanate species responsible for the band at 2260 cm-l (fig. 3). However, this species was formed if the order of addition of the gases was reversed. Fig. 6 compares the spectra resulting from two experiments in which the final mixtures of CO and NO were identical. When NO was added first the 2260 cm-l 22002000 2 m 2 0 0 0 1900 1 700 wavenumberlcm-’ Fig. 6. Infrared spectra of Rh/Al,O, after (a) addition of CO (9.3 kN m-2) followed by NO (9.3 kN m-2) and (b) subsequent evacuation; (c) addition of NO (9.3 kN m-2) followed by CO (9.3 kN m-2) and ( d ) subsequent evacuation.(a’Hd’) Corresponding spectra of gas-phase alone in contact with discs when spectra (a) - ( d ) were recorded. band was non-existent or very weak but the clear maximum at 2235 cm-l was observed [fig. 6(c)] and could partly be ascribed to N,O in the gas phase [fig. 6(c’)]. Removal of the gas phase established the existence of a narrow band at 2235cm-l due to adsorbed isocyanate [fig. 6(d)]. In contrast, the addition of CO first followed by NO led to the appearance of considerably stronger bands at both 2260 and 2235 cm-l [fig. 6(a)] despite the fact that the amounts of N,O formed in the gas phase were similar in the two experiments [fig.6(a’) and (c’)]. The isocyanate species were again resistant to desorption by evacuation at ambient temperature [fig. 6(b)]. Because of the time dependence of the formation of N,O and surface isocyanate (fig. 3 ) the growth in intensities of the bands at 2235 cm-l were determined as a function of time until the rates of growth were negligible (ca. 1 h) when the spectra in fig. 6(a) and (c) were recorded. The half-life for the attainment of the maximum intensity at 2235 cm-l was ca. 8 min (at ca. 303 K) in both experiments. Exact comparisons of the relative intensities of the three bands ascribed to adsorbed NO were difficult because differing scattering effects from disc to disc significantly varied the slopes of the baseline spectra. However, in experiments involving highE. A.HYDE, R. RUDHAM AND C. H. ROCHESTER 539 pressures of CO and NO a maximum at ca. 1720-1 730 cm-l was more intense, relative to the maxima at 1910 and 1830 cm-l, in spectra of Rh/Al,O, exposed to NO and CO [fig. 3 (b)] than to NO alone [fig. 3 (a)]. In general, the spectra of non-dissociatively adsorbed CO (2100 and 2035 cm-l) andNO (1910,1830 and 1740 cm-l) on Rh/Al,O, in the presence of both gases were similar whether CO or NO was added to the catalyst first (fig. 6). The apparent enhancement of the intensity of the maximum at 1740 cm-l was more pronounced when CO was in excess over NO with total pressures of ca. 16-19 kN rn-, and pressure ratios in the range 1.0 < (pco/pNo) < 2.1. P I I I I 2200 2000 1900 1700 wavenum ber/cm -' Fig.7. Infrared spectra of Rh/Al,O, (a) in equilibrium with CO (1 1.2 kN rn-,), (6) after 64 h contact with CO (1 1.2 kN m-,) plus NO (5.25 kN m-,) and (c) - (s) a consecutive series of heat treatments in vacuum at temperatures/K of (c) 303 (30 min), ( d ) 380 (30 min), (e) 448 (30 min), cf) 303 (43 h) and (g) 469 (30 min). Desorption of isocyanate species by evacuation at elevated temperatures (fig. 7) generated no products, either in the gas phase or adsorbed state, other than those formed in the reactions at ambient temperatures. The two types of isocyanate responsible for the bands at 2260 and 2235 cm-l were similarly affected by thermal activation. The form of non-dissociatively adsorbed NO giving a band at ca. 1900 cm-l was most resistant to desorption. The enhancement in band intensity at 1900 cm-l following heat treatment at 380 K [fig.7(d)] was consistent with the present and previously reported1* effects of heating Rh/Al,O, in NO alone. Residual bands at 2090 and 201 5 cm-l after exposure of Rh/Al,O, to NO + CO and evacuation at high temperature [fig. 7cf> and (g)] emphasized an important general aspect of all the present results. The maxima assigned to vibrations of gem-dicarbonyl species Rh(CO), were at 2100 and 2035 cm-l in the absence of NO.' However, in the presence of NO the maxima were broadened towards lower wavenumbers with either resulting shifts in the positions of the maxima in the overall540 REACTION OF co + NO ON RHODIUM band envelopes or the appearance of distinct shoulders (fig. 2) due to contributions to the spectra from two new bands.The desorption spectra (fig. 7) confirmed the existence of these bands at 2090 and 2015 cm-l. The former was consistent with the growth in absorption intensity at ca. 2100 cm-l and the broadening of the band at 2100 cm-l towards lower wavenumbers when NO was added to Rh/Al,O, already exposed to CO (fig. 4). Spectra of CO, (13-17 kN m-,) adsorbed on Rh/Al,O, exhibited a band at 201 5 cm-l confirming,24* 25 despite a report to the contrary,17 that dissociative adsorption of CO, which occurs on rhodium single-crystal surfaces9* l1 also takes place on supported rhodium. Iizuka and Tanaka24 attributed a similar band at 2020 cm-l in spectra of CO, on Rh/A1,0, to RhCO species resulting from dissociative adsorption of CO,.The intensities of the present band for a given CO, pressure were greatest for Rh/Al,O, prepared by reduction of Rh(NO,),/Al,O,, less for reduced [Rh(NH3),C1]C1,/A1,0, and least for reduced Rh,(SO,),/Al,O,. This sequence is consistent with the relative surface areas of the dispersed rhodium prepared from the three precursor salts.' No bands due to CO on Rh other than that at 2015 cm-l were observed following the chemisorption of CO, at ca. 303 K. DISCUSSION The existence of infrared bands at 1910, 1830 and 1740 cm-l in spectra of NO adsorbed on Rh/A1,0, was consistent with the results of previous studies of Rh/A1,0,l47l5 and a RhY zeolite.ls For NO on Rh/A1,0, similar bands have been ascribed to surface RhNO+, RhNO and RhNO- species, re~pective1y.l~~ l5 However, isotopic labelling experiments for NO on a RhY zeolite strongly suggested that maxima at 1860 and 1780 cm-l could be assigned to vibrations of a gem-dinitrosyl complex Rh(NO), in which two NO molecules were simultaneously liganded to a single Rh' ion in the zeolite.ls Evidence for the assignment of the present bands at 1830 and 1740 cm-l to Rh(NO), species is provided by a comparison of results for the adsorption of CO and NO on Rh/Al,O,. Uptakes of CO by Rh/Al,O, prepared by the reduction of Rh(NO,),/Al,O, showed that a high proportion of the exposed Rh atoms were each able to act as a site for the adsorption of a pair of CO molecules to give a gem-dicarbonyl complex Rh(CO),.7 Bands at 2100 and 2035 cm-l in the infrared spectrum of adsorbed CO confirmed7 the presence of Rh(CO), after CO adsorption.s Yao et a1.26 have reported (NO/Rh) and (CO/Rh) values for NO and CO adsorption on alumina-supported Rh.The data showed that for samples similar to those studied here [(CO/Rh) z 1.6817 the values of (NO/Rh) exceeded those of (CO/Rh), providing strong evidence that NO also undergoes multiple adsorption on highly dispersed Rh. The assignment of bands at 1830 and 1740cm-l to vibrations of a rhodium dinitrosyl complex would be consistent with infrared spectra of dinitrosyl complexes of transition metals. Typically two bands are observed in the approximate ranges 1862-1 742 and 1764-1 699 cm-l and with a separation of ca. 41-107 cm-1.27-29 By analogy with proposals for gem-dicarbonyl species on supported Rh, the gem- dinitrosyl complex probably exists at individual Rh atom sites5 or edge sitess with cationic character generated through interactions between metal atoms and the alumina s ~ p p o r t .~ Cationic dinitrosyl complexes give i j (NO) bands at higher spectral positions than neutral complexes29 and this possibly explains why the bands for NO on Rh/Al,O, are at lower positions than for NO liganded to Rhl in a RhY zeolite.ls The edge sites for Rh dispersed on alumina have less cationic character than would be consistent with a formal + 1 oxidation state. However, this conclusion must beE. A. HYDE, R. RUDHAM AND C. H. ROCHESTER 54 1 tentative because the structure and total coordination at the adsorption site would also influence the infrared band positions.28 In contrast to the results for CO adsorption' no bands were observed which could be ascribed to bridging nitrosyl groups.3o Sites for the adsorption of bridged CO were perhaps active for the dissociative adsorption of The infrared band at 1910 cm-l is ascribed to RhNO+ species in accordance with previous assignmentsl47 l5 and with spectra of transition-metal complexes containing NO+.,' The formation of this species might be associated with the occurrence of both non-dissocative adsorption and dissociative adsorption of NO on the same surface.32 Loss of electronic charge from a chemisorbed NO molecule would be expected to be induced by an oxygen adatom liganded to the same or an adjacent Rh atom.Dissociative adsorption of NO on Rh occurs at temperatures at least as low as 330 K.12 The enhancement of the band at 1910 cm-l and the shift to 1920 cm-l with increasing temperature of contact between NO and Rh/A120, could be due to an increased amount of dissociative adsorption and hence a higher surface population of adjacent NO molecules and 0 atoms.The concomitant loss of the bands at 1830 and 1740 cm-l shows that reactions at edge sites were at least partly responsible for the results. A plausible scbheme would be that the reactions l1, l3 NO Rh + NO g Rh/ NO Rh/No + NO Rh/ (1 830 and 1740 cm-l) \NO occurred at ambient temperatures and led to the gem-dinitrosyl complex at edges sites. At higher temperatures pairs of NO molecules reacted at the metal surface in accordance with Rh + 2N0 + Rh/O + N,O (3) probably via dissociative adsorption of the first molecule and subsequent reaction between the resulting N adatom and the second NO molecule.Further addition of NO to vacant ligand positions gave Rh(O)(NO) species 0 0 Rh/ +NO + Rh( (1920 cm-l) (4) \NO . particularly when the sample was cooled for spectroscopic examination. The oxygen adatoms blocked the formation of the dinitrosyl complex and therefore the absorption bands at 1830 and 1740 cm-l did not reappear on cooling. Eventually the formation of the Rh(O)(NO) complex would also be hindered by total oxidation of the rhodium surface resulting from NO dissociation at high temperatures. This effect would be compatible with the growth of a band at 1920 cm-l in the spectra of Solymosi and Sarkany15 after heat treatment of Rh/A120, in NO at increasing temperatures up to 473 K but a decrease in intensity after subsequent heating at 673 K.Bands at 1838 and 1740 cm-l due to adsorbed NO were absent from spectra after heat treatment at 523 K.15 Reaction (3) accounts for the formation of N20. The small extent of conversion of NO to N20 arose because the reactive sites were simultaneously poisoned by an oxygen adatom which prevented further reaction. Oxygen is not readily desorbed from542 REACTION OF co -t- NO ON RHODIUM Rh below the maximum temperature (823 K) studied he~e.~q l2, l3 Nitrogen was probably also a product of reaction between N ad atom^.^, l2, l3 For Rh/Al,O, it is proposed that reactions (1)--(4) occur on Rh atoms at apex, edge or step sites where each atom has at least two vacant coordination positions.Stepped Rh (755) and (33 1) surfacesg* l3 and polycrystalline Rh wire1, adsorb NO dissociatively with the formation of N,, N,O and 0 adatoms. The results for NO adsorption and reaction over Rh/Al,O, were unhelpful in establishing whether dissociative adsorption of NO occurred on exposed single-crystal planes at the surface of dispersed particles. A LEED study of NO adsorption on Rh (1 1 1) and (100) surfaces showed that adsorption took place but did not establish whether dissociation Occurred.ll However, there is evidence for the dissociative adsorption of NO on Rh( 110) planes3, which suggests that some of the N,O formed could have resulted from reaction between N adatoms and NO molecules on single-crystal planes or two-dimensional arrays of Rh atoms.Sites available for the formation of dinitrosyl and dicarbonyl complexes should also be capable of adsorbing one CO plus one NO molecule to form a surface Rh(CO)(NO) complexls Rh/'O ( 5 ) Rh/No (+CO, -NO) , Rh/No (+CO,-NO) \NO ' (+NO,-CO) \ c O . (+NO,-CO) ' 'co . Bands at 2101 or 2105 and 1755 or 1760 cm-l have previously been ascribed to the Vc0 and vN0 vibrations, respectively, of the Rh(CO)(NO) species.l49 l5 The similarity between the spectra after addition of CO followed by NO or of NO followed by CO and the growth of the band at 2090 cm-l after addition of NO to a CO-covered surface were compatible with the establishment of equilibria (5) between Rh(NO),, Rh(NO)(CO) and Rh(CO), at edge or step sites. A contribution to the overall absorption intensity of the band with a maximum at ca.1740 cm-l could have resulted from the vN0 vibration of Rh(NO)(CO). However, the predominant effect which accompanied the growth of the band at 2090 cm-l involved a strong maximum at ca. 183Ocm-l [fig. 4(b) and 5(c)] which may have contained a contribution due to Rh(CO)(NO) as well as contributions due to the rhodium dinitrosyl complex and possibly carbonate species formed as reaction products on alumina by the NO+CO reaction on rhodium. Spectra of CO, adsorbed on Rh/A1,0, discs exhibited a broad maximum centred at 1815 cm-l with shoulders at 1850 and 1775 cm-l. The concomitant appearance of the present bands at 2090 and 1830 cm-l [fig. 4(b)] was analogous to the existence of bands at 2090 and 1810 cm-l in spectra of a RhY zeolite exposed to mixtures of CO and NO.ls The latter bands, together with a maximum at 2165 cm-l were assigned to vibrations of a [Rhl(CO),(NO)]+ complex.Here the 2165 cm-l band was absent. However, typical spectra of transition-metal complexes containing one NO and two CO molecules suggest that 2165 cm-l is a rather high spectral position for a CO vibration of such a complex.34 The present spectra for Rh/Al,O, do not provide unambiguous evidence for the existence of detectable equilibrium concentrations of [Rh(CO),(NO)] although it would not be unreas~nable,~ to assign bands at 2090, 2015 and 1830 cm-l to such a complex involving rhodium atoms of cationic character.' The bands at 2090 and 2015 cm-l generally appeared together in support of this assignment. However, the coincidence between the latter band and a maximum at 2015 cm-l in spectra of CO, adsorbed on Rh/Al,O, suggested the alternative possibility that the co-adsorption of NO and CO on Rh gave an adsorbed CO species similar to that resulting from the dissociative chemisorption of CO,.The band at 2185 cm-l (fig. 2) occurs in the same spectral region as bands due toE. A. HYDE, R. RUDHAM AND C. H. ROCHESTER 543 dinitrogen liganded to Rh atoms.35 Since nitrogen is a product of CO plus NO reactions over Rh/Al,O, the attribution of this band to RhNN species is conceivable. Spectra of nitrogen adsorbed on Rh/A1,0, exhibited a maximum at 2205 cm-I which may be ascribed to the NN stretching vibrations of surface RhNN species. The nitrogen was desorbed by evacuation at ambient temperatures.A similar band at 2236 cm-l has been reported for N, on Rh/Si0,.3sv 37. The present bands at 2185,2235 and 2260 cm-l cannot be ascribed to adsorbed nitrogen unless the band positions and/or strengths of adsorption for the adsorbed nitrogen were appreciably affected by the co-adsorption of CO, NO and other possible products of the CO+NO reactions. A more likely attribution of these bands is to surface i~ocyanate.'~-~~ Similar maxima in spectra of CO plus NO adsorbed on supported chromia were due to species containing both carbon and nitrogen atoms,18 which further precludes the asignment of the bands to dinitrogen species. Evidence, particularly from results for platin~m,'~ suggest that the band at 2185 cm-l should be ascribed to RhNCO speciesI5 but that the bands at 2235 and 2260 cm-l resulted from the migration of isocyanate groups from rhodium to the surface of the alumina upp port.^^^ l8 Variations in the relative intensities of the bands after different treatments may be compared with variations in the positions of band maxima reported elsewhere.In two studiesI59 l7 the predomi- nant band due to isocyanate was at 2269 or 2272cm-l whereas a third report14 described a single maximum at 2235 cm-l. These differences reflect the effects of the sequence of addition of NO and CO and the relative amounts of NO and CO present on the formation of surface isocyanate groups. The band positions reported here were independent of whether Rh(NO,),, [Rh(NH,),Cl]Cl, or Rh,(SO,), was the precursor salt in the catalyst preparation.The relative band intensities were also approximately independent of salt under similar experimental conditions, showing that variations in Rh dispersion' had little effect on the character of the isocyanate groups generated by the NO+CO reactions. The reactions N Rh/No + Rh/ +CO, (6) 'co NCO R h I N +CO + Rh/ (7) represent one proposed mechanism for the formation of isocyanate from NO + CO over Rh/Al,O, ~atalysts.'~q l5 Several factors suggest that this mechanism was not significant in the present reactions at ambient temperature. The intermediate species Rh(NO)(CO) would be expected to be formed at edge-site Rh atoms which give gem-dicarbonyl species in the presence of CO alone.s* The results in fig. 6 exemplify the mutual displacement of non-dissociatively adsorbed CO and NO at edge sites and the establishment of the equilibria represented by reaction (5).Accordingly the generation of Rh(NO)(CO) should have been independent of whether NO or CO was added to the system first. Hence, in contradiction to the experimental result, the extent of formation of isocyanate should also have been independent of the sequence of addition of the two gases. Furthermore, if the isocyanate group formed in reaction (7) migrated to the alumina support then the Rh atom would again be available for further reaction. The outcome would be build-up of the same amount of isocyanate whether CO or NO was added first. Finally, the absence of isocyanate after the gradual displacement of adsorbed gem-dicarbonyl species by excess NO (fig. 5) suggests that Rh atoms responsible for the formation of the dicarbonyl did not constitute active sites for the generation of adsorbed isocyanate.544 REACTION OF co +NO ON RHODIUM Bands at ca.2050 and ca. 1890 cm-l in spectra of CO adsorbed on Rh are ascribed to linear (RhCO) and bridge-bonded (Rh,CO) carbonyl species, respectively, on exposed crystal planes or two-dimensional arrays of rhodium atoms.'-' The absence of these bands after exposure of Rh/A1,0, to CO followed by NO, or to NO then CO, confirmed that NO adsorption took place on the same exposed planes. Arai and TominagaL4 have reported a similar result. The non-dissociative and dissociative adsorption of NO9? 1 1 - 1 3 9 33 may be represented by (8) 2Rh+NO -+ Rh-N+Rh-0 (9) (10) Rh + NO -+ Rh-NO reaction (9) producing N adatoms which could undergo the reaction Rh-N + CO -+ Rh-NCO to give adsorbed isocyanate groups.The present results do not preclude the occurrence of these reactions on exposed planar Rh surfaces. After the adsorption of NO at low pressures, or at high pressures followed by partial desorption, the plane surfaces probably contained adsorbed NO molecules, N and 0 adatoms and some vacant sites. The infrared band at 1910 cm-l may at least in part, particularly after adsorption at ambient temperatures [cf. reaction (4)], be due to adsorbed NO+ species liganded to Rh atoms influenced by the electron-withdrawing effects of adjacent 0 adatoms. Bands at 1860-1880 cm-l for NO on ruthenium have been interpreted in the same way.,, Subsequent addition of CO led to isocyanate formation [reaction (1 O)] possibly via a Langmuir-Hinshelwood mechanism involving adsorption of CO on vacant sites and reaction between N adatoms and adsorbed CO molecules.12 The pronounced appearance of the band at 2185 cm-l under these conditions [fig.2(c) and 5(h)] may be attributable to the existence of a high population of adjacent N atoms and vacant sites on the surface before the addition of CO. The inhibiting effect on isocyanate formation of NO preadsorbed at high pressure would then be ascribed to the absence of vacant sites for CO adsorption and subsequent reaction. This explanation would require that non-dissociatively adsorbed NO was not readily displaced by CO from the surface and that N adatoms did not undergo direct reaction with gaseous CO molecules to give isocyanate via a Rideal-Eley mechanism.The present proposals are supported by a suggestion that surface isocyanate possibly results from reaction between N adatoms and adsorbed CO molecules on a polycrystalline Rh wire.12 A further factor contributing to the inhibiting effect of NO on isocyanate formation might be that the extent of dissociative adsorption of NO is decreased by non-dissociative adsorption in dominant competition for sites at high NO press~res.~~ l2 The considerable enhancement in isocyanate formation for Rh/Al,O, exposed to high pressures of CO followed by NO (fig. 6) is explicable in terms of adsorption on planar surfaces if, in accordance with the removal of infrared bands at ca. 2050 and ca.1890 cm-l, linear and bridged CO were displaced from adsorption sites by the adsorption of NO. A plausible reaction scheme (RhCO sites) would be as follows: CO CO CO(+N:co) CO NO CO Rh Rh Rh Rh Rh Rh Rh, Rh Rh NCO I l l - 1 1 I - + I + co, . (1 1) An important factor could be that a non-dissociatively adsorbed NO molecule is initially surrounded by several adsorbed CO molecules, at least one of which promotes the dissociation of adsorbed NO and one of which becomes incorporated into theE. A. HYDE, R. RUDHAM AND C. H. ROCHESTER 545 isocyanate group. A possibility is that a concerted reaction led to the simultaneous formation of CO,, or a dissociatively adsorbed CO, species (band at 201 5 cm-l), and isocyanate. This would be consistent with a proposed mechanism for the generation of CO, and adsorbed isocyanate by the reaction of CO and NO on ruthenium catalysts.38 The present proposals support a previous conclusion15 that interactions between NO and CO molecules adsorbed on Rh at low temperatures enhance the dissociation of NO.The correct stoichiometry of one NO molecule and two CO molecules required for the formation of CO, and an isocyanate group would be contained in a [Rh(CO),NO] complex1* which might be formed on edge or apex Rh atoms with a high degree of coordinative unsaturation. The involvement of this complex as a precursor of isocyanate was unlikely to be the main reason for the appearance of isocyanate in the present systems although reaction at an edge site might be advantageous for the subsequent migration of isocyanate to the surface of the alumina support. The formation of N,O from NO at ambient temperature cannot be simply attributed to the occurrence of reaction (9) followed by Rh-N + NO 4 Rh + N,O (12) because no N20 appeared in the absence of CO.The generation of N,O must have required CO to be involved in one or both of two ways. First, the dissociation of NO was enhanced by the presence of CO. Secondly, the formation of N,O(g) via reaction ( 12) was promoted possibly by the displacement of reaction product from the surface by the adsorption of CO. The generation of carbon dioxide in the step involving NO dissociation would be consistent with the appearance of a weak band due to CO,(g) and of bands due to adsorbed carbonate species on the alumina support.However, N,O(g) formation became negligibly slow after ca. 1 h (fig. 6) even though the added pressures of CO and NO had not been appreciably depleted. Sites responsible for catalysing the N,O reaction were progressively poisoned either by isocyanate or by the retention of 0 adatoms formed by the dissociation of NO. Poisoning by isocyanate alone was inconsistent with the contrast between the similar amounts of N,O formed and the dissimilar amounts of isocyanate formed in the experiments for which results are shown in fig. 6. The retention of 0 adatoms was consistent with the relatively small amounts of CO, apparently formed by the reaction and with the presence of the maximum at ca. 201 5 cm-l similar to a corresponding maximum in spectra of CO, dissociatively adsorbed on Rh/A1203.The generation of CO,(g) by the reaction of 0 adatoms and adsorbed CO was apparently unfavourable at ambient temperatures. The infrared spectra do not enable an unambiguous decision to be made concerning which Rh atom sites were active for the formation of N,O. The production of isocyanate and N,O on the same sites in an exposed planar surface could only be rationalized if reaction (1 1) giving isocyanate did not lead to poisoning of the active sites but the reaction giving N,O simultaneously led to poisoning. The suggestion that isocyanate groups migrate to the alumina is in accordance with this possibility since the Rh atoms would then be available for further reaction to isocyanate or to N,O. The formation of N,O would continue until all the sites were poisoned and therefore the final amount of N,O would be, in accordance with the experimental result (fig.6), independent of the amount of isocyanate formed at the same time. However, this proposal seems unlikely because the secondary products of the two reactions are identical [CO,, CO(ads)+O(ads) or O(ads)] and therefore the poisoning effects of the products should be the same. With a limited total number of sites an increase in the extent of formation of isocyanate would then be expected 19 FAR 1546 REACTION OF CO+NO ON RHODIUM to cause a corresponding decrease in the amount of N,O product. This is not compatible with the present results (fig. 6) and therefore it appears that N,O and isocyanate formation must have involved different types of surface site.Iizuka and Lunsfordls reported that N,O was readily formed from NO+CO over a RhY zeolite and proposed that a complex containing one CO and two NO molecules liganded to a single Rh atom was an intermediate in the reaction mechanism. Arai and Tominaga14 suggested that a Rh(NO)(CO) complex [reaction (S)] was a precursor of N20 over Rh/Al,O, catalysts. Edge, apex or step sites may have been active for the formation of N,O. The reactive sites must have been progressively poisoned presumably by the retention of 0 adatoms which were resistant to desorption as oxygen or to reaction with CO and desorption as CO,. It is relevant to note that a dinitrosyl complex of iridium has been shown to undergo a ligand exchange reaction with CO with the resulting elimination of N,O and the concomitant formation of C0,.40 It is here proposed that a similar reaction occurs for Rh/Al,O, at Rh atoms which offer at least two vacant ligand positions for the adsorption of CO or NO.However, not all such sites were active since the formation of N20 was severely retarded whilst spectra still contained strong bands due to surface dicarbonyl and dinitrosyl species. We thank Drs B. Harrison and D. T. Thompson for helpful discussions, the S.E.R.C. for a studentship (E. A. H.) and Johnson Matthey for financial support. A. C. Yang and C. W. Garland, J. Phys. Chem., 1957,61, 1504. M. Primet, J. Chem. SOC., Faraday Trans. I , 1978,74, 2570. H. C. Yao and W. G. Rothschild, J. Chem. Phys., 1978,68, 4774. N. Sheppard and T. T. Nguyen, Adu. Infrared Raman Spectrosc., 1978, 5, 67.J. T. Yates, T. M. Duncan, S. D. Worley and R. W. Vaughan, J. Chem. Phys., 1979, 70, 1219. D. J. C. Yates, L. L. Murrell and E. B. Prestridge, J. Catal., 1979, 57, 41. E. A. Hyde, R. Rudham and C. H. Rochester, J. Chem. Soc., Faraday Trans. I , 1983,79, 2405. F. Solymosi and A. Erdohelyi, Surf. Sci., 1981, 110, 630. D. G. Castner and G. A. Somejai, Surf. Sci., 1979,83, 60. lo D. G. Castner, L. H. Dubois, B. A. Sexton and G. A. Somerjai, Surf. Sci., 1981, 103, 134. l1 D. G. Castner, B. A. Sexton and G. A. Somejai, Surf. Sci., 1978,71, 519. l2 C. T. Campbell and J. M. White, Appl. Surf. Sci., 1978, 1, 347. l3 L. H. Dubois, P. K. Hansma and G. A. Somerjai, J. Catal., 1980, 65, 318. l4 H. Arai and H. Tominaga, J. Catal., 1976, 43, 131. l5 F. Solymosi and J. Sarkany, Appl. Surf. Sci., 1979, 3, 68. l6 T. Iizuka and J. H. Lunsford, J. Mol. Catal., 1980, 8, 391. l7 M. L. Unland, Science, 1973, 179, 567; J. Catal., 1973, 31, 459. 19 J. Rasko and F. Solymosi, J. Catal., 1981, 71, 219. 2o R. E. Nightingale, A. R. Downie, D. L. Rotenberg, B. Crawford and R. A. Ogg, J. Phys. Chem., 21 N. Parkyns, Proc. 5th Znt. Congr. Catalysis, 1972, p. 255. 22 A. L. Smith, W. E. Keller and H. I. Johnston, J. Chem. Phys., 1951, 19, 189. 23 C. Mortena, F. Boccuzzi, S. Coluccia and G. Ghiotti, J. Catal., 1980, 65, 231. 24 T. Iizuka and Y. Tanaka, J. Catal., 1981, 70, 449. 25 F. Solymosi and A. Erdohelyi, J. Catal., 1981, 70, 451. 26 H. C. Yao, S. Japan and M. Shelef, J. Catal., 1977,50,407; H. C. Yao and M. Shelef, 7th Int. Congr. 27 W. Hieber and H. Fiihrling, 2. Anorg. Allg. Chem., 1971, 381, 235. 28 J. H. Enemark and R. D. Feltham, Coord. Chem. Rev., 1974, 13, 339. 29 F. J. Regina and A. J. Wojicicki, Znorg. Chem., 1980, 19, 3803. 30 H. Brunner and S . Loskot, 2. Nuturforsch., Teil B, 1971, 26, 757. 31 F. Bottomley, W. V. F. Brooks, S. G. Clarkson and S-B. Tong, J. Chem. SOC., Chem. Commun., 1973, 32 A. T. Davydov and A. T. Bell, J. Catal., 1977, 49, 332. J. Rasko and F. Solymosi, J. Chem. SOC., Faraday Trans. I , 1980,76, 2383. 1954,58, 1047. Catalysis, Tokio, 1980, paper A2 1. 919.E. A. HYDE, R. RUDHAM AND C. H. ROCHESTER 33 R. J. Baird, R. C. Ku and P. Wynblatt, Surf. Sci., 1980, 97, 346. 34 P. S. Braterman, Metal Curbonyl Spectra (Academic Press, London, 1975), chap. 7. 35 G. A. Ozin and A. Vander Voet, Can. J . Chem., 1973,51,3332. 36 Y. G. Borod’ko and V. S. Lyutov, Kinet. Catul., 1971, 12, 202. 37 V. S. Lyutov, V. A. Redosimov and Y. G. Borod’ko, Russ. J. Phys. Chem., 1972,46, 973. 38 M. F. Brown and R. D. Gonzalez, J. Cutal., 1976, 44, 477. 39 C. T. Campbell, S-K. Shi and J. M. White, Appl. Surf. Sci., 1979, 2, 382. 40 B. F. G. Johnson and S. Bhaduri, J. Chem. SOC., Chem. Commun., 1973,650. 547 (PAPER 3/524) 19-2