J. Chew. SOC., Faraday Trans. I, 1989, 85(1), 79-90 Chromia/Silica-Titania Cogel Catalysts for Ethene Polymerisation Infrared Study of Nitric Oxide Adsorption Steven J. Conway, John W. Falconer and Colin H. Rochester* Department of Chemistry, The University, Dundee DDI 4HN, Scotland Infrared spectroscopy has been used to study the adsorption of NO on a chromia/silica-titania cogel (4.2 O h Ti) reduced in carbon monoxide. A mononitrosyl surface complex gave a band at 1807 cm-'. Four types of dinitrosyl complexes have been identified by pairs of infrared bands at I856/ 1742, 1868/ 1736, 1875/ 1755 and 1886/ 1766 cm-' and are ascribed to Cr'' sites of differing structural type. A Crl*I site also gave infrared bands at 1875/1755 cm-' due to a dinitrosyl complex. The effects of reduction conditions, chromia content, water, oxygen and ethene on NO adsorption have been investigated.Recent kinetic and polymer studies have indicated that the surfaces of chromia/silica gels and chromia/silica-titania cogels are heterogeneous. '* Heterogeneity could be attributed to several factors, which include the amorphous nature of the silica creating a broad distribution of adsorption site rea~tivities,~ the possible existence of chromate and dichromate structures of supported chromia4 and the formation of reduced chromia in different oxidation and coordination ~ t a t e s . ~ - ~ Infrared-spectroscopic examination of NO adsorption on reduced catalyst has proved useful for the characterisation of chromia catalysts, and several types of site which form dinitrosyl complexes have been identified.7 The present study was carried out in an attempt to gain further evidence about the oxidation and coordination states of chromium ions present on the surface, and to establish the nature of the catalytically active sites for ethene polymerisation.Experimental The catalyst, a chromia (1 YO Cr)/silica-titania (4.2 Yo Ti) cogel was identical to that used previously.' Infrared spectra of adsorbed NO (equilibrium pressure 133 N m-2) were recorded using a Perkin-Elmer model 580B spectrometer. Catalysts in the form of self-supporting discs were prepared by compacting ca. 70 mg of catalysts in a 25 mm die at ca. 10 MN m-2. The cell and apparatus used were similar to those described elsewhere.8 Following calcination in dry air (50 cm3 min-') at 873 K for 16 h and evacuation at 623 K for 30 min, the catalyst was then activated by reduction at 623 K in CO (5.33 kN mP2) for 15 min, with subsequent evacuation at 623 K for 5 min prior to cooling to beam temperature.This treatment is referred to as standard reduction. NO was purified by a distillation m e t h ~ d . ~ Results No infrared bands due to adsorbed NO were observed when either calcined catalyst with no reduction, or calcined and reduced support alone was exposed to NO. Fig. 1 ( a ) shows the infrared spectra of NO (equilibrium pressure 133 N m-2) adsorbed 7980 Modijied Phillips Catalyst : I. R. Study 10% I 7 1807 I 1 1856 ' /- I :.i I I, I I , , 1742 :a) I I I 2000 1900 1800 1700 60 3) 1742 I I 1 2000 1900 1800 1700 w avenumber/an- Fig.1. (a) Infrared spectra of NO adsorbed on a standard reduced catalyst: (-) background spectrum, (-O-) equilibrium pressure 133 N m-2, and after evacuation for 5 min (-. . - ) and 20 min (----) at beam temperature. (b) Infrared spectra showing the effect of chromium content on NO adsorption on standard reduced catalysts : (-O-) 0.4 % Cr, (--. - .) 1 .O % Cr and (----) 5.0% Cr. on a standard reduced catalyst. The predominant feature of NO adsorbed on a standard reduced catalyst is a pair of broad unsymmetrical bands with maxima at 1856 and 1742 cm-'. Bands of this type are normally assigned to the symmetric and asymmetric stretching frequencies of dinitrosyl complexes, Cr(N0),.7 Present as minor features are a weak band at 1807 cm-', which disappeared when the sample was evacuated, and a shoulder at 1868 cm-', which is responsible for the asymmetry of the 1856 cm-' band.The 1807 cm-' is assigned to a mononitrosyl complex, Cr(NO).' Fig. l(b) shows infrared spectra of NO in contact with three standard reduced catalysts containing 0.4, 1 .O and 5.0 YO chromium. Increasing the chromium con- centration resulted in an increase in intensity of the symmetric component of the dinitrosyl bands and its broadening to higher wavenumbers, a broadening of the asymmetric component and an increase in intensity of the 1807 cm-' band. Thermal desorption of the adsorbed NO species responsible for the bands at 1856 and 1742 cm-' bands was incomplete even at 473 K [fig. 2(a)]. The asymmetry of the 1856 cm-' band towards high wavenumbers was probably due to a band at 1870 cm-' [fig.1 (a)] which is assigned to an overtone of the Si-0 stretching mode for the catalystS. J. Conway, J. W. Falconer and C. H. Rochester 81 lO0/0 I 1742 la) 50 'i 1 I I , 4 1856 1742 - 1736 (b) 2000 1900 1800 1700 2000 1900 1800 1700 1600 wavenumber/cm-' Fig. 2. (a) Infrared spectra showing the effect of thermal desorption of NO adsorbed on a standard reduced catalyst: (-O-) after evacuation at beam temperature, and at (-- - - ) 433 K and (----) 473 K for 15 min. (b) Infrared spectra showing the effect of water on NO adsorbed on a standard reduced catalyst: (-O-) after evacuation at beam temperature for 15 min; (-a - .) to (-0-) after exposure to increasing water vapour pressures, and subsequent evacuation.support." High-temperature evacuation promoted loss of the shoulder at 1868 cm-' and a narrowing of the 1742 cm-l band, particularly on the low-wavenumber side. Adsorption of water on an NO-treated standard reduced catalyst removed the bands at 1856 and 1742 cm-l, revealing a second pair of bands at 1868 and 1736 cm-' [fig. 2 (b)]. A weak band indicative of adsorbed water was present at 1630 cm-l. Spectra recorded during a series of exposures to oxygen of an NO-treated standard reduced catalyst, and after evacuation for 5 min are shown in fig. 3(a) and (b). This resulted in the disappearance of the bands at 1868 and 1856 cm-', the appearance of a band at 1875 cm-' and a progressive shift of the maximum at 1742 cm-l to 1755 cm-' [fig. 3(a)]. This was accompanied by the appearance of a band at 1625 cm-l, the intensity of which increased with increasing exposure to oxygen [fig.3(b)]. When an NO-treated standard reduced catalyst was allowed to stand in NO for 12 h a shift in maxima from 1856 and 1742 cm-l to 1875 and 1755 cm-l [fig. 3(c)] and the appearance of an intense band at 1625 cm-' (not shown) was observed. Infrared spectra of NO adsorbed on catalyst reduced for various times [fig. 4(a)]1 0 O/e I 1875 Modified Phillips Catalyst : I.R. Study I 1742 a) 0 1900 1800 1700 I I I 50 50 50 50 50 50 50 1 i 0 1600 60 2c 1 I I . , 1856 1755 3) 0 1900 1800 1700 I I I wavenurnberIcn-' Fig. 3. (a) and (b) Infrared spectra showing the effect of oxygen on NO adsorbed on a standard reduced catalyst: (-O--) after evacuation at beam temperature for 15 min; (a) (-- - a ) to (-@-) and (6) (-a - - ) to (--..) after increasing time exposure to oxygen (1 atm) and subsequent evacuation. (c) Changes in the infrared spectra with time of NO adsorbed on a standard reduced catalyst : (-O-) immediately after contact with NO and (-- . .) after 720 min. exhibited only intensity differences of the 1856 and 1742 cm-' bands for reduction times up to 60 min. Increasing reduction periods (> 60 min) resulted in a shift of the maximum dt 1856 cm-' to 1870 cm-'. Exposure of standard reduced catalysts to ethene at 293, 343 and 413 K for 3 min before adsorption of NO [fig. 4(b)] resulted, except for catalyst treated at 41 3 K, in lower intensities of bands due to adsorbed NO than the corresponding intensities for catalyst which had not been exposed to ethene.The fall in band intensity was greatest for the catalyst treated at 293 K. The band at 1807 cm-' gave the biggest fractional decrease in its intensity. Spectra measured during a series of oxygen treatments of an ethene-treated standard reduced catalyst which had been exposed to NO are shown in fig. 4(c). Initial oxygen treatments resulted in loss of the 1856 cm-' band to give a pair of bands at 1868 and 1739 2 cm-'. Further exposure to oxygen reduced the intensity of both bands. Standard reduced catalysts further reduced in CO at high temperature (873 K) for various times and exposed to NO exhibited changes in band positions. Reduction for 2 h shifted the maxima from 1856 and 1742 cm-' to 1873 and 1746 cm-'.After furtherS. J . Conway, J . W. Falconer and C. H. Rochester 83 6( h $5 3 B 3 b 10% I 7; f 1870 - Ip 1742 1856 4 2000 1900 1800 1700 60 1 856 1742 I I I 2000 1900 1800 1700 wavenumber/an- ' 60 1868 1856 741 .. . 2000 1900 1800 1700 Fig. 4. (a) Infrared spectra of NO adsorbed on a catalyst reduced (CO, 623 K) for (-O-) 3, (- . . .) 15, (----) 60, (-@-) 180 and ( * - .) 720 min. (b) Infrared spectra showing the effect of ethene adsorption/polymerisation on NO adsorption on a standard reduced catalyst : (-O-) NO on a standard reduced catalyst, and after exposure to ethene at ( - - .) 293, (----) 343 and (--. . .) 413 K for 3 min. (c) Infrared spectra showing the effect of oxygen on NO adsorbed on a standard reduced catalyst pretreated with ethene (343 K) : (-O-) after evacuation at beam temperature for 15 min; (--.. - ) to (-@-) after increasing time exposure to oxygen, and subsequent evacuation. reduction at 873 K the bands moved to 1868 and 1739 cm-' and also suffered losses in intensity [fig. S(a)]. Fig. 5(b) shows spectra of NO adsorbed on a standard reduced catalyst which had been evacuated at high temperature (873 K) before admission of NO. Band intensities decreased with increasing treatment times at 873 K. The maxima also shifted to higher wavenumbers. For short treatment times the intensity of the band at 1807 cm-l was greater than the corresponding intensity for standard reduced catalysts. The changes with time in the infrared spectrum of NO adsorbed on a high temperature (873 K) evacuated catalyst are shown in fig.6(a) and (b). The spectra show the intensity of the 1807 cm-' band to decrease with time and the two major bands to shift to high wavenumbers. After 12 h two new bands with maxima at 1886 and 1766 cm-' were observed.84 Modijied Phillips Catalyst : I. R . Study I 1746 a) 10 1900 1800 1700 60 2c '- /--\ 1865 I 1745 3) 0 1900 1800 1700 I wavenumber/cm- Fig. 5. (a) Infrared spectra showing the effect of high-temperature reduction (873 K) in CO of a standard reduced catalyst on NO adsorption: (-O-) 120, (--. . a ) 360 and (----) 720 min reduction. (b) Infrared spectra showing the effect of high-temperature evacuation (873 K) of a standard reduced catalyst on NO adsorption: (-O--) 15, (--. a ) 180 and (----) 360 min evacuation. Discussion Infrared spectra of NO adsorbed on standard reduced catalysts [fig.l(a) and (b)] are characteristic of NO adsorbed on reduced silica-supported ~ h r o m i a . ~ The weak central band at 1807 cm-', easily removed by evacuation at beam temperature, is commonly assigned to a mononitrosyl specie^.^ Infrared and calorimetric investigations have shown that the sites characterised by the two more intense bands are the result of two- coordinate complex f~rrnation.~ Analogues of these complexes are also formed on reduced molybdena-alumina catalysts. l1 The infrared spectra of these surface complexes have been assigned to both dinitrosyl', '' and dimeric12* l3 complexes, with the balance of the present evidence currently favouring dinitrosyl complexes. l1 For a dinitrosyl complex the two bands are interpreted as being the symmetric and asymmetric stretching modes.7 The broad and unsymmetrical nature of the present infrared bands suggest NO was adsorbed on several types of different site.This conclusion is substantiated by the band resolution and enhancement and shifts induced by various catalyst treatments which were carried out.S. J. Conway, J. W. Falconer and C. H. Rochester 85 I 1744 4 60 1766 b) 2000 1900 1800 1700 2000 1900 1800 1700 Fig. 6. (a) Changes with time in the infrared spectra of NO adsorbed on a standard reduced catalyst evacuated for 15 min at 873 K: (-O--) immediately after contact with NO, and after (--. . .) 30, (----) 60, (-@-) 120 and (. * .) 240 min. (6) Changes with time in the infrared spectra of NO adsorbed on a standard reduced catalyst evacuated for 180 min at 873 K: (-O-) immediately after contact with NO and (--.- .) after 720 min. Thermal desorption of NO adsorbed on a standard reduced catalyst showed the complex characterised by the 1856/1742 cm-' band pair was thermally stable in comparison to other NO complexes [fig. 2(a)]. Krauss and H0pfl14 have shown, using temperature-programmed desorption, that NO is desorbed partially as NO, partially as N, and partially as N,O with accompanying oxidation of Cr". A desorption mechanism involving the decomposition of the Cr(NO), species and partial oxidation of chromium has also been rep~rted.'~ The interaction of water with NO adsorbed on a standard reduced catalyst revealed a pair of bands at 1868 and 1736 cm-l [fig.2(b)]. The interaction of propionitrile, pyridine, H,O, CO and NH, with NO adsorbed on reduced chromia-silica catalysts has been investigated previously. 15, l6 The original bands assigned to Cr(NO), were replaced by new pairs of bands which were attributed to the coordination of an extra ligand forming three-coordinate complexes. L86 Modified Phillips Catalyst : I.R. Study Such a mechanism appears unlikely for water, as evidence of the 1868 and 1736 cm-' pair was noted in the absence of water [fig. l(b), 4(c) and 5(a)]. An alternative explanation is a ligand displacement reaction. However, the formation of three coordinate complexes with molecules other than water cannot be discounted. Treatment of NO adsorbed on a standard reduced catalyst with oxygen resulted in the formation of a pair of bands at 1875 and 1755 cm-', with accompanying loss of all other bands [fig.3(a)]. In a similar experiment carried out by Krauss and Weis~er,'~ oxidation of adsorbed NO to adsorbed NO, occurred at low temperature, while at ambient temperature reoxidation of the catalyst resulted. As an absorption band at 1625 cm-' was observed [fig. 3(b)] similar to the band resulting from NO, adsorption on a reduced catalyst, it can be assumed that some NO oxidation had occurred at beam temperature. These absorption bands were found for a standard reduced catalyst after exposure to NO for 12 h (band at 1625 cm-l not shown). The ingress of oxygen into the system, which would be capable of oxidising chromium as well as adsorbed NO, may account for the shift in maxima.8 In contrast to oxygen-treated standard reduced catalyst exposed to NO no 1875 and 1755 cm-' bands were observed for such catalysts pretreated with ethene [fig.4(c)]. Absorption bands corresponding to the 1868/1736 cm-l were observed, suggesting the site attributed to these bands was oxidised to the site characterised by the 1875/ 1755 cm-' pair. An adsorption site giving rise to a mononitrosyl species has been shown to coordinate an additional NO, forming a dinitrosyl which gives infrared bands at 1887 and 1765 cm-l.' Similar bands (1 886/ 1766 cm-') attributable to this fourth site were observed when a catalyst heated in vacuo at 873 K was exposed to NO for 12 h [fig. 6(b)]. Three types of adsorption sites which form dinitrosyl species exhibiting average stretching frequencies [$( vsym + vasym)] at 1806, 18 17 and 1826 cm-' have previously been identified.7 These values are close to these observed in this study (i.e.1799, 1802, 1815 and 1826 cm-l). This is normally taken to indicate that the formal charge is increasing along the series, which is in disagreement with the evidence that CO-reduced catalysts contain primarily Cr" with small amounts of unreactive Cr''' as a-chromia.6 This was explained by assuming that the 'actual' charge depends on the number of surface oxygens coordinated to the divalent chromium, the oxygen ligands acting as strong electron acceptors . The relationship between average stretching frequency, chromium electron density and dinitrosyl behaviour can be discussed in terms of a metal-nitrosyl bond consisting of a 0 bond between the nitrogen lone pair and an empty orbital of the chromium, together with a n back donation from the occupied metal d orbitals to the unoccupied n* orbitals of N0.16 A high electron density on the chromium would increase the n back- donation to the n* orbital of NO, weakening the N-0 bond and consequently shifting the frequency of the infrared bands to lower wavenumbers, while increasing the strength of the M-N bond.This could account for NO adsorbed on the 18561 1742 cm-' site being resistant to desorption as a consequence of a strong M-N bond and apparently easily oxidised due to a weakening of the N-0 bond, while NO adsorbed on the 1875/1755 cm-' site is more resistant to oxidation. The small variation in average stretching frequencies between the 1856/ 1742 cm-' and 1868/ 1736 cm-' species suggest a similarity in electron density on the chromium. Two possible alternatives are Cr(NO), complexes based on chromate (1) and dichromate-like (2) structures in oxidation state two.ON NO ON NO ON NO \ /S. J. Conway, J. W. Falconer and C. H. Rochester 87 The proposition that the 1868/ 1736 cm-' site can be oxidised to the 1875/ 1755 cm-l site could be envisaged as dichromate oxidation from oxidation state two to three. ON NO 9 ON NO ON NO \ / Cr-0- '0' ON-!h<I>Ir-NO - 0 V - 0- I -0 I I /////// /A///// This assignment of oxidation states is plausible since the average stretching frequency of the proposed Cr"' species is similar to an e.s.r.-active CrIII dinitrosyl (1 8 16.5 cm-I) identified by Beck and Lunsford,'* and the formation of Cr' and CrO is most improbable.l9 Cr" coordinated to three surface oxygens and characterised by N-0 stretching frequencies similar to the CrIII dinitrosyl was proposed to be present on the surface of CO reduced catalyst^.^ This site may account for the shift of the symmetric component to higher wavenumbers upon prolonged reduction [fig. 4(a)] and high- temperature (873 K) reduction [fig. 5(a)], as CrIII is unlikely to survive such conditions. Although the assignment of the 1886/1766 cm-' site to a high oxidation state of chromium would be appropriate, no evidence of oxidation occurring at high temperatures has been presented.6 Alternatively Cr" coordinated to four surface oxygens characterised by a high average stretching frequency, indicating a low electron density, has been p r ~ p o s e d .~ The conditions and the nature of the reduced chromium would favour the formation of this site. A highly coordinatively unsaturated chromium ion bound to the surface would become mobile at high temperatures, enabling the chromium to rearrange itself in a different coordination state, possibly in a vacancy site on the surface. The assignments of absorption maxima to differing Cr sites on the catalyst (table 1) permit an interpretation of fig. 1 (b). The main features of this figure are: (a) a broadening of the symmetric stretching component to higher wavenumbers, (b) a broadening of the asymmetric stretching component and (c) an increase in mononitrosyl absorption with increasing Cr concentration.Hogan20 showed that as the Cr concentration of a catalyst was reduced the efficiency per Cr for ethene polymerisation increased. This was thought to arise from there being a limited number of locations on the silica surface which generate high-activity sites.2' It is suggested that the order in which the sites are formed is first the tetrahedral Cr" and then later the dichromate species, and possibly Cr" coordinated to three surface oxygens. The formation of dichromates would be expected to be favoured over chromate at higher Cr loadings where the Cr to surface hydroxyl ratio is increased and the Cr atoms are closer to one another. The increase in mononitrosyl at higher Cr loadings may be due to the formation of a site (or sites) not energetically or sterically favoured at lower loadings.Recent evidence suggests at least two types of site are active for low- and high-pressure ethene polymerisation, the relative concentrations of which are dependent upon CO reduction time.''2 Both Cr" and C P ' are thought to be catalytically active, extensive reduction being responsible for a change in the dominant active site from CrII' to Cr'1.2,5 This reduction results in a change in the nature of the active site, whereby the active site changes from one active at low temperature to one which is active only at higher temperatures. Both active sites are deactivated by heating in vacuo or in CO at high temperature (873 K). The activity at 343 K of a catalyst treated in vacuo decreases initially then increases slightly with treatment time before falling off further.This maximum was thought to indicate the formation of a new type of site active only at high polymerisation temperatures.88 Modijied Phillips Catalyst : I.R. Study Table 1. Summary of infrared band assignments for NO adsorbed on chromia/silica-titania catalyst (4.2 wt % Ti) structure of oxidation structure of coordination band position/cm-' complex state site number 1807 mononitrosyl 2 chromate a 1856/ 1742 dinitrosyl 2 chromate 2 1868/ 1736 dinitrosyl 2 dichromate 2 1875/ 1755 dinitrosyl 2 chromate 3 1875/ 1755' dinitrosyl 3 dichroma te 3 1886/ 1766 dinitrosyl 2 chromate 4 ~~ ~~~~ a Coordination number thought to be variable.' ' Oxidising conditions favour Crrrr formation, which in the presence of NO is believed to give bands similar to those for one of the Cr" complexes.Supporting evidence for this suggestion has been reported by Ghiotti et af. Extensive reduction [fig. 4(a)] resulted in a shift of the symmetric component to higher wavenumbers and a loss in intensity of both components consistent with the loss of the site giving bands at 1856/1742 cm-' in the presence of NO. The loss of these bands assigned to NO liganded to tetrahedral Cr" is not necessarily identified with the loss of the low-temperature active site for several reasons. A comparison of experimental observations with model predictions for polymer characteristics2 and a direct correlation of [Cr(N0)J3+ concentration detected by e.s.r. spectroscopy with low-temperature activity'* suggest that Cr'II is the low-temperature active site.If Cr"' and Cr" coordinated to three surface oxygens, which are thought to exhibit very similar absorption bands,7' l8 are formed, then the inability to discriminate between the two sites using infrared spectroscopy would compromise their correlation with activity. Infrared spectra of NO adsorbed on a standard reduced catalyst following exposure to ethene at 293 and 343 K indicated all sites were affected by ethene pretreatment, while exposure at 413 K showed little change within experimental variation [fig. 4(b)]. Strong ethene chemisorption on inactive and active sites occurs on the reduced catalyst,22 therefore the loss of bond intensity may be partly due to ethene chemisorption preventing coordination of NO to the chromium.The changes in bond intensity are consistent with the tendency of chemisorbed molecules to be desorbed with increasing temperature. Comparison of NO adsorbed on a standard reduced catalyst treated with oxygen [fig. 3(a)] and one pretreated with ethene prior to exposure to NO and oxygen [fig. 4(c)] showed that no 1875 or 1755 cm-' bands were present in the latter case, while bands corresponding to the 1868/ 1736 cm-' site were observed. Although this indicates a change in the nature of the 1868/ 1736 cm-' site it does not mean necessarily that it was the active site. The reason for the change is also unknown. High-temperature evacuation of CO reduced catalysts has been studied before.6, ' 9 23 A decrease in the concentration of the site assigned to tetrahedral Cr" accompanied by a decrease in sorptive capacity for CO and reduced reactivity of Cr" towards reoxidation was interpreted as a loss of polymerisation activity and tetrahedral Cr" assigned as an active site.However, Myers and L u n ~ f o r d ~ ~ found that their catalyst was more active, and upon NO adsorption observed that the concentration of the dinitrosyl previously assigned to tetrahedral Cr" was reduced and the mononitrosyl concentration was increased. They concluded that the active site on Cr" catalysts is a Cr" moiety characterising mononitrosyl formation. We previously reported that after an initial fall in activity an increase in activity occurs before the activity falls off with increasing treatment time.' The maximum was attributed to the formation of a new type of site.Spectra of adsorbed NO [fig. 5 (b)] indicate an increase in mononitrosyl concentrationS . J. Conway, J. W. Falconer and C. H. Rochester 89 and fall in dinitrosyl concentration, particularly the 1856/ 1742 cm-l dinitrosyl, suggesting that the site characterised by mononitrosyl formation is responsible for the increase in activity. Here a correlation between the fall in activity and the loss of the 1856/ 1742 cm-' site is observed; nevertheless this cannot be an unambiguous assign- ment. For ethene polymerisation proceeding by the Langmuir-Hinshelwood mechanism two vacant coordination sites are required. Ghiotti et aL7 identified a mononitrosyl which was capable of coordinating a second NO by an activated process thought to involve a ligand-displacement reaction in which at least one surface oxygen is detached, forming a dinitrosyl characterised by absorption bands at 1887 and 1765 cm-'.The changes with time of NO adsorbed on a catalyst evacuated at high temperature are consistent with the formation of a similar dinitrosyl [fig. 6(a) and (b)]. A comparison of high-temperature reduction and evacuation showed the former treatment to be more effective for catalyst deactivation. This was thought to be due to the formation of thermally stable CO-Cr complexes, allowing the Cr greater surface mobility and subsequent rearrangement into an inactive f0rm.l Consequently, the comparative stability of the site (1868/ 1736 cm-l) assigned to a dichromate species may be due to CO coordinated in such a way as to block vacant coordination positions and prevent rearrangement.Furthermore, dichromates have been suggested to have a good geometrical fit with the silica surface, which may promote thermal stability.21 Since only ca. 10 % of the total Cr is active on a catalyst containing 1 wt YO Cr3, probing of inactive sites is inevitable. Inactive catalysts deactivated by high-temperature reduction still exhibit intense bands [fig. 5(a)]. However, the presence of bands characteristic of the 1868/ 1736 and 1875/ 1755 cm-l sites on an inactive catalyst cannot be interpreted as meaning these sites are always inactive, as Cr species must be considered as families containing sub-species in which the behaviour of a species has to be considered as an average.6, 7 9 25 Welch and McDanie126 favour a deactivation mechanism involving the aggregation of Cr into aggregates of CrO or Cr20,, whereas othersGy7 favour one involving the coordination of surface oxides to the Cr.In both cases a loss of coordinative unsaturation occurs, but a comparison of fig. 5(a) and (6) suggest that differing mechanisms occur with high-temperature evacuation and high-temperature CO treatment. In fig. 5 (b) the significant absorption due to mononitrosyl species suggests that some coordination of surface oxides occurs during high-temperature evacuation. NO adsorbed on different coordinatively unsaturated Cr is characterised by specific absorption bands, showing NO to be sensitive to the nature of the Cr site. Dinitrosyl species typical of these observed on a reduced Cr/silica-titania (4.2 YO Ti) cogel were also observed on a reduced Cr/silica gel.This indicates that the same adsorption sites are present on both catalysts and no observable Ti-0-Cr bonding occurs on the cogel, as adsorption sites bonded to different support atoms would be expected to generate dinitrosyl species with different absorption ~pectra.~' We thank the S.E.R.C. for a Studentship, BP Chemicals for support through the CASE scheme and Dr G. W. Downs for helpful discussions. References 1 S. J. Conway, J. W. Falconer and C. H. Rochester, J. Chem. SOC., Faraday Trans. I , 1989, 85, 71. 2 S. J. Conway, J. W. Falconer and C. H. Rochester, submitted. 3 M. P. McDaniel, Adv. Catal., 1985, 33, 47. 4 M. P. McDaniel and M. B. Welch, J .Catal., 1983, 82, 98. 5 D. L. Myers and J. H. Lunsford, J. Catal., 1985, 92, 260. 6 B. Fubini, G. Ghiotti, L. Stradella, E. Garrone and C. Morterra, J . Catal., 1980, 66, 200. 7 G. Ghiotti, E. Garone, G. Della Gatta, B. Fubini and E. Gianello, J. Catal., 1983, 80, 249. 8 M. A. Sutton, Ph.D. Thesis (Nottingham University, 1981).90 ModiJied Phillips Catalyst : I. R . Study 9 R. E. Nightingale, A. R. E. Downie, D. L. Rotenberg, B. Crawford Jr and R. A. Ogg Jr, J . Phys. Chem., 1954, 58, 1047. 10 D. D. Eley, C. H. Rochester and M. S . Scurrell, Proc. R. SOC. London, Ser. A, 1972, 329, 375. 11 R. P. Rosen, K. Segawa, W. S. Millman and W. K. Hall, J . Catal., 1984, 90, 368. 12 E. L. Kugler, R. J. Kokes and J. W. Gryder, J . Catal., 1975, 36, 142. 13 E. L. Kugler and J. W. Gryder, J. Catal., 1975, 36, 152. 14 H. L. Krauss and R. Hopfl, Proc. 2nd Eur. Symp. Therm. Anal., 1981, p. 175. I5 A. Zecchina, E. Garrone, G. Ghiotti and E. Borello, in Catalysis, Heterogeneous and Homogeneous, ed. B. Delmon and G. Jannes (Elsevier, Amsterdam, 1975), p. 243. 16 E. Garrone, G. Ghiotti, S . Coluccia and A. Zecchina, J. Phys. Chem., 1975, 79, 984. 17 H. L. Krauss and B. Weisser, 2. Anorg. Allg. Chern., 1975, 412, 82. 18 D. D. Beck and J. H. Lunsford, J. Catal., 1981, 68, 121. 19 C. Groeneveld, P. P. M. M. Wittgen, A. M. van Kersbergen, P. L. M. Mestrom, C. E. Nuijten and G. C. A. Schmit, J . Catal., 1979, 59, 153. 20 J. P. Hogan, J . Polym. Sci., 1970, 8, 2637. 21 D. R. Witt, in Reactivity and Mechanism and Structure in Polymer Chemistry, ed. A. D. Jenkins and A. Ledwich (Wiley, New York, 1974), p. 431. 22 R. Merryfield, M. P. McDaniel and G. Parks, J . Catal., 1982, 77, 348. 23 A. Zecchina, E. Garrone, G. Ghiotti, C. Morterra and E. Borrello, J. Phys. Chem., 1975, 79, 966. 24 D. L. Myers and J. H. Lunsford, J . Catal., 1986, 99, 140. 25 H. L. Krauss, B. Rebenstorf, U. Westphal and D. Schneeweis, in Preparation of Catalysts, ed. B. 26 M. B. Welch and M. P. McDaniel, J . Catal., 1983, 82, 110. 27 S . J. Conway, unpublished results. Delom, P. A. Jacobs and G. Poncelet (Elsevier, Amsterdam, 1976), p. 489. Paper 8/01532E; Received 19th April, 1988