Infrared Study of CO Adsorption on Magnesium Oxide BY EUGENIO GUGLIELMINOTTI, SALVATORE COLUCCIA, EDOARDO GARRONE, Istituto di Chimica Fisica dell'Universit8 di Torino, Corso M. d'Azeglio 48, 10125 Torino, Italy LUIGI CERRUTI AND ADRIANO ZECCHINA * Received 11th April, 1978 The adsorption of CO at room temperature on well outgassed specimens of MgO gives rise to a large number of bands in the 2200-1000 cm-' range, which can be divided into two main groups. The bands of the former group are destroyed by oxygen at room temperature : some react instan- taneously and are associated with a marked pink colour of the sample ; the others are less reactive as they require prolonged contact time in order to be completely oxidized at room temperature (r.t.). The bands of the latter group, far from being destroyed by oxygen, grow when the oxygen-sensitive species are depleted.The oxygen-sensitive species are thought to be negatively charged polymeric CO structures (CO clusters) of the type (CO)",-, where x = 2 or 4 and n is > 2. The simplest CO clusters (dimers) can be transformed into larger polymers by further CO addition. Under the correct conditions the reverse process can also be carried out. The oxygen-insensitive species have a car- bonate-like structure and are present on the surface in fairly constant ratios with respect to the former group species. A chemisorption mechanism leading both to oxidized (carbonate-like) and to reduced (negative CO polymers) species is proposed. The active centres for CO chemisorption consist of groups of ions (both positive and negative) in strongly uncoordinated situations.1.r. studies on the CO/MgO system have been carried out by Little et a2.l and by Jiru et aL2 Both studies deal with samples outgassed at moderately low temperatures, i.e., when still covered by a fair number of OH groups. Under these conditions, CO is adsorbed as carbonate-like species only if 0, is present. In contrast, other studies carried out by u.v.-vis. reflectance spectroscopy on completely dehydrated samples 9 have shown that CO chemisorption gives rise to peculiar polymeric species character- ized by electronic transitions in the visible and near ultraviolet. The aim of this investigation is to obtain further information on the structures revealed by reflectance spectroscopy.It will be shown that on the basis of vibrational spectra alone the presence of polymeric CO species is confirmed, in agreement with the preliminary results discussed in ref. (5). EXPERIMENTAL The samples were prepared by decomposing magnesium hydroxide pellets under vacuum N m-2) in the i.r. cell at 523 K. Two hydroxide samples of different purity were used. The purest one was the same as that used in the reflectance study,4 whereas the other was prepared following Lunsford and Jayne from Carlo Erba Reagent Grade ACS MgO (Fe impurity 50 p.p.m.). In both cases, specimens were obtained with indistinguishable pro- perties (B.E.T. specific surface area - 250 m' g-'), which do not sinter appreciably in vacuo even under the severe thermal treatments at 1073 K required for the complete surface dehydration.Pellets of % 30mgcm-2 were used. The i.r. spectra were measured on a Beckman I.R. 12 spectrometer using a standard in situ silica cell. High purity (Matheson) CO and O2 were used. On 1073 K outgassed samples the CO coverage, estimated following the procedure described in ref. (6), was 0.01-0.015. 96GUG LIELMI N OTTI, COLU C CIA, GARRONE, CERRUTI , Z E C C HI N A 97 RESULTS EFFECTS OF PRETREATMENT TEMPERATURE The MgO sample formed by decomposition of a magnesium hydroxide pellet was outgassed at 773, 923, 1073 and 1173 K for 4 h. After each outgassing step, the pellet was cooled to r.t. and contacted with 13.3 kN m-2 CO. The sample outgassed at 773 K shows no (or very limited) activity, in agreement with the 1iterature.l For higher pretreatment temperatures, the same number of bands with constant relative intensities are observed but the overall activity increases up we 1073 K, then declines at 1173 K.We ascribe the increase in activity to the increasing surface dehydration and the decrease to sintering of the particles. For the above reasons, only the spectra taken after outgassing at 1073 K will be reported and discussed. co ADSORPTION ON FULLY DEHYDRATED SAMPLE Due both to the exceptional complexity of the spectra and to the fact that the correlations among the bands are only established on the basis of the whole set of spectra reported, we are unable to make the usual detailed description of the results and we anticipate, for the sake of clarity, the correlation table 1 of the bands.The following discussion will, hopefully, justify the correlations proposed. TABLE 1 .-ADSORBED SPECIES number of observed frequencies/cm-1 species modes features A 1 2200 weak and reversible B 2 2064, 13 18 transient species, oxygen sensitive 1480, 1275 C 4 desorbed first 1197, 1066 grow slowly upon ads., very sensitive to oxygen D1 2108, 1358 D1 grows upon C depletion D2 and D3 grow upon ads., 2097,1365 D{D2 2 resistant to desorption D3 2084, 1392 all oxygen sensitive 1582, 1160 1548, 1160 2 1574, 1160 as D1 species F see text oxygen insensitive, not studied in detail In fig. 1 the i.r. spectrum of adsorbed CO is shown at increasing coverages: the dotted curve was obtained by contacting the surface with 650 N m-2 CO for 15 min, while the broken one refers to the same sample contacted for the same time with 2.6 kN m-2 CO.The solid curve is the spectrum of the sample after 15 h at 13.3 kN rn-2. The last spectrum does not yet correspond to the maximum coverage, as small but definite intensity increases are observed for longer contact times. This shows that CO chemisorption is a slow and complex process, strongly pressure and time dependent. High CO pressures considerably increase the rate of adsorption; I 498 CO ADSORPTION ON MgO however the same situation can be reached with lower pressures and longer contact times. Fig. 1 shows that : (1) a weak band at 2200 cm-l is formed immediately upon CO contact with time-independent intensity. This unusual behaviour enables it to be assigned to a single species A.(2) The bands located at 2064 and 1318 cm-' FIG. 1.-Spectra of CO adsorbed at increasing coverages. . . ., 650 N m-' for 15 min; - - -, 2.6 kN m-2 for 15 min ; -, background ; 13.3 kN m-2 for 15 h (nearly maximum coverage). 1 OOy 3/cm-1 FIG. 2.-CO desorption up to 373 K. -, nearly maximum coverage (as last curve in fig. 1) ; , . ., sample outgassed 5 min at beam temperature ; - - -, sample outgassed 5 min at 373 K. The bands due to C species are shadowed.GUGLIELMINOTTI, COLUCCIA, GARRONE, CERRUTI, ZECCHINA 99 initially grow in intensity and then disappear at high coverages. As this behaviour is unique we ascribe them to a single species B. (3) The bands at 2097, 2084, 1392 and 1365-72cm-' (D species) appear immediately (dotted curve), whereas those at 1480,1275, 1197 and 1066 cm-' (C species) require longer contact times and/or higher pressures.(4) The apparent maxima of the bands of the C species shift with the coverage to higher frequencies (Av = + 8 cm-I). Both this fact and their large half width suggest that a family of species with very similar structure absorb at the quoted frequencies. co DESORPTION AND READSORPTION In fig. 2 and 3 the desorption of CO from the same sample is illustrated. Due to complexity of the spectrum, two figures are needed for a complete description of the experiment. In fig. 2 the modifications caused by outgassing in the 310-373 K range are shown, whereas in fig. 3 the remaining part of the experiment is illustrated (tem- perature range 423-473 K). 0 Vlcm-' FIG. 3.-CO desorption in the range 373-473 K.- - -, after outgassing 5 min at 373 K (as last curve in fig. 2) ; - - * -, sample outgassed 15 min at 423 K ; -, sample outgassed 30 min at 473 K. The solid curve in fig. 2 is the starting spectrum, and actually coincides with the last spectrum in fig. 1. Outgassing for 5 min at the beam temperature (b.t.) (fig. 2 dotted curve) decreases the intensity of the bands at 1480, 1275, 1197 and 1066 cm-1 and increases that of the bands at 2108, 1574,1548,1358 and 1160 cm-I. Moreover, the 2200 cm-I band completely disappears. The relative intensities of the former group of bands are not basically modified, so confirming that they are due to the same type of surface species C . The lability of C species upon outgassing is surprising, as fig. 1 has shown that their formation is definitely activated.This fact indicates that the depletion pathway is different from the formation one. In fact, new species are created on the surface.100 CO ADSORPTION ON MgO The classification of the bands growing upon outgassing at b.t. is not easy. Some bands are already present in the initial spectrum (1574, 1548 and 1160 cm-l), whereas the others are nearly new (2108 and 1358cm-l). Thus, different surface species cause these absorptions. In particular, the bands at 2108 and 1358 cm-l (clearly coupled together) are both ascribed to species D1, whereas the other bands belong to another family (E species). Evacuation at 373 K (fig. 2 dashed curve) leads to a further dramatic decrease in the C species, while the D1 and E bands reach their maximum intensity.The residual components of the C species are now shifted to higher frequency, so confirming the composite nature of these bands. By outgassing in the 310-373 K range the intensity of a large number of bands does not change at all. They are ascribed to other less reactive F species. 7 S/cm-l FIG. 4.-CO readsorption after outgassing at 423 K. - * - a, spectrum of a different sample after outgassing 15 min at 423 K ; . . ., 13.3 k N m-2 CO at beam temperature. Successive evacuation at 423 K (fig. 3, dotted curve) causes the disappearance of a couple of weak bands at 2084 and 1392 cm-l characteristic of the D3 species, while the bands of the F species now change in intensity in a complex way. The exposure of the sample to 13.3 kN m-2 CO after each outgassing step de- scribed so far restores the full coverage spectrum. This is shown in fig.4 after desorption at 423 K, the dotted curve being the final spectrum. The C species are regenerated at the expense of D1 and E species in a fast (i.e. non-activated) process, in contrast to the slow formation of C species on the unreacted sample. Finally by outgassing at 473 K the D1, D2 and E species completely disappear, while the F species evolve to a final simplified situation. CO/O2 INTERACTION In fig. 5 and 6 the reactivity of CO surface species towards gaseous oxygen is illustrated. This experiment has been carried out on a lighter sample: as a con- sequence the bands are less intense than those shown in fig. 1-4. The solid curve inGUGLIELMINOTTI, COLUCCIA, GARRONE, CERRUTI, ZECCHINA 101 fig.5 refers to full CO coverage: the broken one has been taken immediately after contact with 2.6 kN r r 2 CO. The same curve has been reported in fig. 6 for com- parison: the other two were taken after 1 and 20 h, respectively. The following important features are observed : (a) the C species are extremely sensitive to oxygen and an immediate reaction takes place leading to F species. In the meantime the I I I I I I I 2 9 It300 .woo 1200 10 ' I 11 ij/cm- 0 FIG. 5.-CO/02 interaction ; first stage. -, nearly total coverage ; - - -, immediately after contact with 2.6 kN m-2 02. 100 FIG. 6.-CO/O2 interaction ; further stages. - - -, as the last curve in fig. 5. - - -, after 1 h; . . ., after 20 h.102 co ADSORPTION ON MgO sample colour turns white (fig.5). (b) The D and E species react at a much lower rate with formation of species of the F family (fig. 6). This experiment clearly shows that the C , D and E species are oxygen sensitive, whereas the F species are not. The transient B species is also oxygen sensitive. In fact if a sample showing the broken or dotted spectra of fig. 1 is contacted with 02, a slow reaction takes place leading to the complete destruction of the bands at 2064- 1318 cm-l and to the formation of F species. Similar experiments carried out with N,O as oxidizing agent (not reported here for sake of brevity) gave the following results : (a) the C species were oxidized leading to species of the F family; (b) the D and E species did not react at r.t.In conclusion, species and/or families of species which show different behaviour on outgassing also have different reactivity towards oxidizing agents. DISCUSSION STRUCTURE OF ADSORBED SPECIES A SPECIES The presence of a single band indicates that the diatomic CO structure is retained. The reversibility of this species and the positive shift with respect to the gaseous molecule stretch (Av = +47 cm-l) are in agreement with a CO molecule weakly adsorbed on cationic sites. Similar positive shifts have been found for CO inter- acting with exposed Mg2+ ions in magnesium spinel and exchanged zeolites (MgX and MgY).' B SPECIES These are characterized by bands at 2064 and 1318 cm-'. As these frequency values are very near to those characteristic of the D species and a similar reactivity toward oxygen and N20 is also observed, we are led to assume that B transient entities and the D species have similar structures, which will be discussed together in the following section.c SPECIES These are characterized by four bands in the 1500-800 cm-I range and hence must have a more than diatomic nature. The highest mode of the C species is at N 1475 cm-I . In compounds containing carbon4arbon or carbon-oxygen bonds, bands are found : in the 2100-1900 cm-I range in the case of triple or double cumulated bonds (C=C, C=C=O), in the 1850-1600 cm-' range in the case of double bonds (C=O in ketones and esters, C=C in olefins), and in the 1200-900 cm-l range in the case of single bonds (C-0 in alcohols and esters, etc., C-C in hydrocarbons), so that none of these structures can account for the experimental results.A C-0 bond order of between one and two is suggested by the above considera- tion, so that we are forced to consider less common structures, like anionic, resonance stabilized, CO clusters with the general formula : (C0);- where n 2 2 and x = 2, 4 . . .. This conclusion based on pure i.r. data completely confirms the reflectance results.3* Cyclic, resonance stabilized carbanions (a)-(d) are well known : 2-GUGLIELMINOTTI, COLUCCIA, GARRONE, CERRUTI, ZECCHINA 103 other ring sizes (e.g. 3, 7 membered) can also occur. The i.r. spectra of the solid alkali derivatives of all these compounds are indeed characterized by two i.r. bands in the 1500-800 cm-l range,9 which is half the number of observed bands of the C species.However, due to the reduced symmetry caused by the strong interaction with the surface, the two Raman modes in this range could also become i.r. active. For simplicity we refer to the most common six-membered twice-charged ring (c). One reason for this choice is that the electronic transition of the rhodizonate com- pounds lo is close to the 21 500 cm-l absorption shown by C species in the visible region.3 We propose that a substantial interaction occurs with exposed Mg ions, e.g., as in structure (e), where three Mg ions in a triangular array are considered. The n (el reason for such a choice is given later. The interaction between carbon atoms and Mg ions will induce a partial sp3 character in the carbon atoms and cause a slight distortion of the rhodizonate structure from planarity, so increasing the number of i.r.active modes to four without much loss of aromaticity of the molecule. Model (e) is tentative, as many others can be conceived, e.g., with x = 4 or interacting with only two Mg ions. Moreover, as already observed,l resonance stabilized anions with open structure can also have the low bond order characteristic of C species. Open, resonance stabilized (C0):- clusters can be represented as in structure (f) 1 df) where n >, 2 and x = 2 or 4. In our opinion more than diatomic clusters with open structure have to be ruled out because : (a) compounds with similar structure are unknown in the homogeneous phase and (b) the presence of only four i.r. active modes in the 1500-800 cm-l range restricts the n value to 3 or 4, as for higher degrees of polymerization the i.r.spectrum should be more complicated. However, a low degree of polymerization (i.e., a conjugated system of limited extension) does not explain the 21 500 cm-1 band observed by reflectance spectroscopy. D SPECIES Three types of D species have been found each characterized by two bands at : 2108 and 1358 cm-l (D1), 2097 and 1365 cm-l (Dz) and 2084 and 1392 cm-l (D3). Two bands in the 2150-1300cm-1 region indicate that again more than diatomic structures are involved and undoubtedly suggest that ketenic groups are present in the adsorbed species.12 Moreover, as the D3 species can be transformed into C species by adding CO from the gas phase, they can be considered as less complicated versions of the C species.104 CO ADSORPTION ON MgO The two requirements are completely fulfilled by the dimeric CO structures :13 0 I! C II C -0’ ‘Mg’ (s) In fact they are the smallest (more than diatomic) charged fragment of the C structures containing a ketenic chromophore.Moreover, the two observed bands are typical of in-phase and out-of-phase vibrations of cumulated double bonds. The C-D3 transformation process can consequently be represented as in scheme (1) Of course reaction (l), leading to resonance stabilized structure, can occur only if suitable arrangements of the counter Mg2+ ions are present (at least a pair of Mg2+ ions in a suitable position being required). If this favourable arrangement is missing, structure (g) cannot show any tendency to incorporate CO to give a cyclic species, This could explain the presence of D1 and D2 species, which as a consequence are assumed to differ from D3 species only by the number and geometry of surrounding ions.Finally, owing to the spectroscopic similarity between D and B species, we con- clude that this transient species also has a D-like ketenic structure. However, the reason for its transient nature cannot be easily understood in the light of this dis- cussion. E SPECIES These absorb in two different regions, namely 1585 and 1160 cm-l. In this case more than diatomic species are also necessarily involved. We think that, as in the case of D species, we are dealing with three components (El, E2, E3). In fact the absorption at high frequency is clearly the superposition of three bands which are affected by outgassing in different ways.For instance the El component does not change during the desorption process in the 373-423 K range, unlike the others. To make a reasonable hypothesis of their structure the following have to be taken into account : (i) the frequencies of the observed bands imply bond orders of < 2; (ii) the E2 and E3 species are less complicated versions of the C species as they can be transformed into each other by addition or removal of CO. These requirements are met by the dimeric fragments(C0);‘ in a steric arrangement different from the one of D species, e.g., involving three magnesium ions :GUGLIELMINOTTI, COLUCCIA, GARRONE, CERRUTI, ZECCHINA 105 Similar bands have been found by Buchner13b in the spectra of the reaction products of CO with alkali metals in liquid ammonia.Moreover, the structure is very close to the cis-hyponitrite NzOi- found by some of us on NO/Mg0,14 which shows two infrared absorptions in a similar position. The E2, E3-C transformation process can be consequently illustrated by scheme (2) It is noticeable that schemes (1) and (2) are very similar : in fact they represent parallel ways for formation and destruction of C species. Species D1, E2 and E3 are present in fairly fixed ratios at all outgassing stages : hence a sort of equilibrium between the two types of species can be hypothesized following scheme (3) During the last outgassing stage at 473 K both types of species disappear simul- taneously leaving on the surface only F (carbonate-like) species.This is further evidence that all these species have similar stability and similar destruction patterns during the outgassing procedure. F SPECIES The band position l5 and stability towards oxygen demonstrate that F species have a carbonate-like structure. Moreover, as in the previous cases, we are dealing with families of similar species (organic and bidentate-type carbonates) rather than single well-defined structures. Owing to the complexity of the spectrum, we shall not attempt a full assignment. The organic carbonates (bands at 1774, 1740, 1710 and 1270-1220 cm-l) in any case disappear at 473 K, whereas some bidentate car- bonates (1665-1650, 1325 cm-l) are left on the surface as the only residual species. We stress here that during the adsorption process the overall intensity of the F species parallels the overall intensity of species C , D and E.This implies that upon CO adsorption both F species (on one hand) and C, D and E species (on the other hand) are simultaneously formed in fairly constant ratios. ADSORPTION MECHANISM The large number of different species and the slow formation rate indicate that CO adsorption is a very complicated process leading to both reduced and oxidized species at the same time, whose relative concentrations are fairly constant. A suitable overall mechanism can be schematized as follows :4 (n +x/2)CO + xO& -+ x/2CO$- + (CO):- (4)106 CO ADSORPTION ON MgO where x = 2 or 4 and the involved oxygen ions are considered to be in low coordina- tion state (and hence have a reduced Madelung potential and an enhanced instability).Scheme (4) indicates that a disproportionation reaction of CO is taking place. Thus, we neglect the possible contribution to the reaction from electron transfer processes from the solid, which have been evidenced by other studies.16 Although the CO coverage is not too far from the reported concentration of electron releasing centres, the constant presence of both oxidized and reduced species strongly favours a disproportionation scheme like (4). Moreover the CO reaction on CaO and SrOY1' leading to much higher coverages, further supports this idea. Therefore, we assume that each reduced species is formed along with an oxidized partner. An adsorption/desorption scheme taking into account most of the results is reaction (5) : -co t +"+ F,C slow s1 .Cq F,B 1 .CQ 16 I I It tI Three different sites (Sly S2, S , ) are assumed, arbitrarily, as no definite evidence is available about the independence of the three mechanisms.The transient nature of the B species is taken into account, as well as different pathways for C formation and depletion. The possible equilibrium between D1 and E is indicated by dashed arrows. In the above reaction scheme, both reduced and oxidized species are envisaged as being depleted together by reversal of reaction (4), with complete restoration of the surface. This is probably an over-simplification. It has been shown (fig. 3) that the last desorption step leaves on the surface small amounts of bidentate carbonates without any reduced partner.In our opinion, this suggests that the reduced dimeric species E and D can be, to some extent, also depleted in a different way, which leads to the formation of carbon atoms (not visible in i.r.), according to eqn (6) : (C0)2 - -+. co,,, + c + 0::. (6) We observed a moderate darkening of the pellets after several adsorption/ desorption cycles and Lunsford and Jayne revealed the presence of traces of C 0 2 in the gas phase during desorption. In conclusion, a Boudouard reaction seems to occur to a small extent at 473 K in agreement with old observations.18 The CO disproportionation shown in scheme (4) would be the first step towards Boudouard reaction. ADSORBING SITES Scheme (4) has the merit of pointing out that basic oxygen ions on the surface act as initiators of the reaction, but it does not take into account the contribution ofGUGLIELMINOTTI, COLUCCIA, GARRONE, CERRUTI, ZECCHINA 107 the positive surface ions to the stabilization of negative CO polymers.It can thus generate the incorrect idea that the reaction centres consist oiily of single uncoordin- ated oxygen ions instead of uncoordinated oxygen ions surrounded by a suitable environment of Mg ions as the proposed assignment of species E, C and D suggests. It has been also shown that the negative CO clusters and the carbonate-like species are formed together, so that both kinds of species should lie in the region. same surface - \-co Fm. 7.Tentative model for the mechanism and site of reaction. A possible model of the site which seems to explain much of experimental data is given in fig. 7.The initial attack of the CO molecule occurs on a coordinatively unsaturated 02- ion, e.g., in a corner or kink position [fig. 7(a)]. The formed car- bonate-like species can be located in such a way that three Mg ions become strongly uncoordinated and available to stabilize a hexagonal dinegative ring [fig. 7(b)], formed most likely through the transient intermediate B. It is noteworthy that if the C-C distance in the rhodizonate ion is used,g the ring of this model compound exactly fits the oxygen vacancy created in the first adsorption step.108 co ADSORPTION ON MgO Desorption at moderate temperatures leads to ring destruction and to the forma- tion of E and D species : a plausible location of these fragments is illustrated fig.7(c) and (d). CO elimination from both dimeric fragments and carbonate-like species restores the initial situation. The Boudouard-like reaction is shown in fig. 7(e) and (f). The dimeric species release one CO molecule leaving one carbon atom on the surface, according to scheme (6) [part (e)] : finally the surface carbonates are desorbed as C 0 2 CO attack on the oxygen ion located at the crystal edge is probably also occurring. However, due to the less favourable arrangement of the ions, the stabilization of large CO clusters is more difficult and the process thus stops at low degrees of polymer- ization. In agreement with this hypothesis, we observed that some D species (D1, DJ and the El one do not show any tendency to incorporate CO to give larger clusters.[part ff >I - CONCLUSIONS The i.r. evidence gives further support to the assignment of U.V. bands of CO adsorbed on MgO to dimeric and polymeric structures. Unexpected species are revealed by vibrational spectroscopy, as well as species like carbonates and molecular CO, which are not seen by U.V. spectroscopy. A detailed assignment has been attempted (except for the carbonates), although this assignment is difficult due to the peculiar reactions occurring at the surface, which do not have an exact correspondence in ordinary chemistry. The tentative nature of the assignments must be acknowledged. An overall disproportionation reaction scheme is shown to hold, and several steps in the reaction network have been understood. The overall trend is to yield polymeric reduced species, and this coincides with what was already found for NO chemisorption on the same samples.4* l4 In the case of NO, the degree of polymer- ization is necessarily smaller and only dimers are found.A further interesting coincidence is the existence in both cases of a depletion pathway which does not coincide with the reversal of the formation reaction, leading in the case of CO to carbon atoms and to N,O in the case of NO. R. St. C. Smart, T. Slager, L. H. Little and R. G. Greenler, J. Phys. Chem., 1973, 77, 1019. H. Kolbel, M. Ralek and P. Jiru, Z. Naturforsch., 1970, 25a, 670. A. Zecchina and F. S. Stone, J.C.S. Chem. Comm., 1974, 582. F. S. Stone and A. Zecchina, Proc. Sixth Znt. Congr. Catalysis (Chem. SOC., London, 1976), p. 162. A. Zecchina, Proc. Sixth Znt. Congr. Catalysis Discussion (Chem. SOC., London, 1976), p. 179. J. H. Lunsford and J. P. Jayne, J. Chem. Phys., 1966,44,1492. ’ (a) C. Morterra, G. Ghiotti, F. Boccuzzi and S. Coluccia, J. Catalysis, in press; (b) C. L. Angel1 and P. C. Schaffer, J. Phys. Chem., 1966,70,1413. R. West, H. Y. Niu, D. L. Powell and M. V. Evans, J. Amer. Chem. SOC., 1960,82,6204. R. West in Non-benzenoid Aromatics, ed. J. P. Snyder (Academic Press, New York, 1969), vol. 1 , p. 31.. lo B. Eistert and G. Bock, Angew. Chem., 1958,70,595. l1 A. Zecchina and F. S. Stone, J.C.S. Faraduy Z, 1978, 74, 2278. l2 J. H. Wotiz and W. D. Celmer, J. Amer. Chem. SOC., 1952, 74, 1860. l3 (a) W. Buchner, Helv. Chim. Acta, 1963, 46, 2111 ; (6) 1966, 49, 907. l4 L. Cerruti, E. Modone, E. Guglielminotti and E. Borello, J.C.S. Faraday I, 1974, 70, 729. l5 L. H. Little, Infrared Spectra of Adsorbed Species (Academic Press, London, 1966), p. 74. l6 A. J. Tench and R. L. Nelson, Trans. Faraday Soc., 1967, 63,2254. l7 A. Zecchina et al., unpublished data. l8 J. Cleminson and H. V. A. Briscoe, J. Chem. Soc., 1926, 2148. (PAPER 8/685)