J. CHEM. SOC. FARADAY TRANS., 1994,90(6), 91 1-917 91 1 Mechanism of Branched Carbon-chain Formation from CO and H, over Oxide Catalysts Part 1.-Adsorbed Species on ZrO, and CeO, during CO Hydrogenation Ken-ichi Maruya," Akihiro Takasawa, Makiko Aikawa, Takashi Hataoka and Kazunari Domen Research Laboratory of Resources Utilization, Tokyo Institute of Technology, 4259 Nagatsuta , Midori-ku, Yokohama 227, Japan Takaharu Onishl Tokyo Polytechnic College ,232-I Oga wanishi, Kodaira City, Tokyo 187, Japan The adsorbed species on ZrO, and CeO, during CO hydrogenation forming branched carbon chains, especially isobutene, (over ZrO,), have been investigated by chemical trapping, in situ IR, and solid-state NMR methods. These indicated that methoxide and formate were present as the surface species on ZrO, .CO hydrogenation on ZrO, with pre-adsorbed methoxide or formate showed that the pre-adsorption of methoxide promotes the forma- tion of the higher hydrocarbons containing mainly isobutene but that formate retards the reaction. Chemical trapping experiments on CeO, after CO hydrogenation at 523 and 673 K led to the formation of methane together with methanol, indicating the presence of methyl, p-methylene, or carbene as a surface species. q2-Formaldehyde was suggested to be the precursor of the methyl, p-methylene, or carbene species, the latter of which in turn gives C, species by insertion of CO. Following the products of CO hydrogenation and chemical trapping on CeO, at 523 K with time indicated that an aldol-condensation-type reaction leads to C, and branchedchain C, products from C, oxyhydrocarbon.We have previously reported that isobutene,' 2-met hylpropanal,2 isobutyl alcohol2 and isoprene3 are formed from hydrogenation of CO under mild reaction conditions over ZrO,, CeO,, In,O,-CeO, and Ln,O,(Ln = lanthanide)-CeO, catalysts, respectively. It has also been shown that branched alkanes are formed from CO hydro- genation over Tho, :La,O, '* and Dy,O, 'catalysts under very severe conditions. All of these products have branched carbon chains and all the catalysts are oxides. These results seem to indicate that the formation of branched carbon-chain compounds from CO hydrogenation is a characteristic of oxide catalysts which are difficult to reduce such as ZrO,, lanthanide and actinide oxide^.^ The mechanism of branched-hydrocarbon formation is not yet clear, even though there have been a few proposal^.^-'^ On the other hand, there have been many reports on the synthesis of iso- butyl alcohol with modified catalysts for methanol synthe- sis.The synthesis of this branched higher alcohol has generally been explained by a conventional aldolic conden- sation mechanism'' and recently Nunan et al. proposed a modified aldol-condensation-type mechanism on the basis of 13Ctracer experiments.', However, the mechanism of forma- tion of C, from C, species is still a matter of controversy. In this paper the species adsorbed on ZrO, and CeO, catalysts during CO hydrogenation are investigated by means of chemical trapping, in situ IR and solid-state NMR methods, and the mechanism of branched carbon-chain formation, especially of C, species, on the oxide catalysts will be dis- cussed.Experimental The catalysts were prepared by the precipitation of hydrox- ides from the aqueous nitrates of Zr and Ce with ca. 3% aqueous ammonia solution and calcination of the precipitates at 773 K for 3 h. CO hydrogenation was carried out typically in a gas-circulation system of total volume 470 cm3 and reactor volume of 55 cm3. Catalysts, except for CeO, used for CO hydrogenation at 523 K, were evacuated at 973 K for 3 h before the reaction. For CO hydrogenation at 523 K, the CeO, catalyst was evacuated at 973 K for 3 h and then treated with H, at 67 kPa and 773 K for 16 h before the reaction.The pre-adsorption of methoxide on ZrO, was carried out by treatment with dimethyl ether at 35 Torr and 643 K. The pre-adsorption of formate on ZrO, was achieved by evacuation at 973 K for 3 h, treatment with water vapour at room temperature and at 15 Torr, CO treatment at 643 K for 30 min, and then evacuation for 1 h at 643 K. After evac- uating the gas-circulating system, except for the reactor, CO hydrogenation over ZrO, with pre-adsorbed methoxide or formate was carried out at 643 K. Chemical trapping was carried out by modification of the method described in the 1iterat~re.I~ During CO hydro-genation the catalyst was rapidly cooled by liquid N, ,slowly warmed to room temperature under evacuation, and treated r'O-I P, 0 .-I I I I I I OO 20 40 60 chemical trapping time/h Fig.1 Dependence of the HCO,CH, formation by chemical trap- ping with dimethyl sulfate on the chemical trapping time with a vapour of diluted aqueous HC1 solution at room tem- perature or of dimethyl sulfate at 453 K under circulation of He. The products were collected at liquid-nitrogen tem- perature together with chemical trapping reagents. The iden- tification of products was carried out using GC-MS. IR spectra of the catalysts, which had been treated for 1 h with a vapour of diluted aqueous HCl solution, showed >90% dis- appearance of methoxide, indicating the rapid completion of chemical trapping.However, it takes much longer to com- plete the treatment with dimethyl sulfate, as shown in Fig. 1. The catalysts for the CP MAS NMR measurement, which were rapidly cooled after CO hydrogenation, were evacuated while warming slowly, and then placed in a Pyrex tube in a glove box under Ar. CP MAS NMR spectra were recorded on a JEOL GX 270 at room temperature. Methoxide and carbonate species were assigned on the basis of adsorbed species obtained by the adsorption of l3CH3QH and 13C0, on fresh ZrO,. Formate resonances were assigned according to species obtained by the adsorption of "CO on ZrO, which was treated with a water vapour at room temperature and then evacuated at 573 K for 10 min. Results Adsorbed Species on ZrO, during CO Hydrogenation 13C CP MAS NMR Spectra 13C CP MAS NMR spectra of adsorbed species on ZrO, are shown in Fig. 2.Measurements of the same sample a week later showed a 10% decrease in the peak intensities. In situ IR measurements showed no change of the peak intensities upon evacuation at room temperature. These results are indicative of the stability of the adsorbed species. The chemi- cal shifts of methoxide, formate and carbonate species adsorbed on ZrO, are presented in Table 1. The 13C NMR spectrum of ZrO, after reaction at 523 K shows only two peaks due to methoxide and formate species at 652.7 and 170.0, respectively. However, the spectrum at 643 K is more complex. The methoxide peaks consist of a main peak at 656.5 accompanied by a shoulder of CQ.653. Two peaks are observed for the formate species. A new peak at 6173.1 appears in addition to the peak at 6169.2. II -l""l""r-.'.l~"',"-,I. *..I 250 200 150 100 50 0 -50 6 Fig. 2 CP MAS MNR spectra of ZrO, treated with a mixture of CO and H, at (a)523 K and (b) 643 K.* spinning side band J. CHEM. SOC. FARADAY TRANS., 1994, VOL. 90 Table 1 Chemical shifts of methoxide, formate, and carbonate species adsorbed on ZrO, ~~~ ~~~ ~ species chemical shifts (63 ref. 'CH 30-55.1" this work 52.7' this work 56.5, 53.0 (sh)'Pd this work "CH,O-Zr complex 60.1 13~0,-163.0 this work ~13~00-169.5 this work 170.p this work 193.1, 169.2'' this work " Methanol adsorption. 'At 523 K. sh = shoulder. At 643 K.CO, adsorption. CO adsorption on H,O-treated ZrO, . Table 2 Amount of foxmate and methoxide species obtained by chemical trapping after CO hydrogenation over ZrO, " amount of adsorbed species/pmol g-T/K formate methoxide CO, ' 523 46 140 250 & 50 643 6 30 110 f30 Chemical Trapping of Adsorbed Species on ZrO, Table 2 shows the products and the amounts of species obtained in chemical trapping experiments. Only methyl formate and methanol were detected. (MeO),CO from the carbonate species was not detected. The total amount of carbon species on the catalyst surface was estimated by the amount of CO,, which was formed upon treatment of the catalyst with 0, at 973 K after CO hydrogenation. The sums of the amounts of methyl formate and methanol are 186 pmol g-' at 523 K and 36 pmol g-' at 643 K, while at these tem- peratures 250 & 50 and 110 _+ 30 pmol g-' CO, is formed. Therefore, the respective sums of amounts of formate and methoxide species are 74 f20% and 33 27%, based on the total amount of surface carbon species. The XPS measure-ments showed no clear difference between peak height of carbon species on the fresh and used ZrO, surfaces.CO Hydrogenation on ZrO, with Pre-adsorbed Methoxide or Formate Fig. 3 shows the product yield with time for CO hydro- genation over ZrO, with pre-adsorbed methoxide. The reac- tion times of 10, 30 and 50 min refer to product collection 81 4 6-4--0 20 40 reaction time/rnin Fig. 3 Dependence of hydrocarbon yield on the reaction time of CO hydrogenation on ZrO, with pre-adsorbed methoxide: C, (O), c3 (A) and c, (0) J.CHEM. SOC. FARADAY TRANS.,1994, VOL. 90 cb, 1.c \ h5-O 0.8 -.5 2 '5. 0.6 se8 0.4 0c> 0.2 0 20 40 reaction time/min Fig. 4 Dependence of the yields of hydrocarbon, (a)CO, (O),and H,O (A) on the reaction time of CO hydrogenation over ZrO, with pre-adsorbed formate w I IO0L 20 40 60 reaction ti me/m in Fig. 5 Dependence of hydrocarbon selectivity (based on carbon) on the reaction time of CO hydrogenation on ZrO, with pre-adsorbed formate: C, (O),C, (A),C, (0)and C, (0) times from 0-20, 20-40, 40-60 min, respectively, at liquid- nitrogen temperature. The product yield within the first 20 min is high and within the next 20 min has already reached the steady state. The hydrocarbon distribution seems to be almost constant over the whole reaction time.Fig. 4 shows the yields of hydrocarbons, H, and CO, formed with time during CO hydrogenation over ZrO, with pre-adsorbed formate. The yield of CO, is very high in the first 20 min, after which it decreases rapidly. Fig. 5 shows the hydrocar- bon distribution with time. C, and C, hydrocarbon selec- tivities decrease with increasing reaction time, while those of C4 and C, hydrocarbons increase. CO Hydrogenation and Chemical Trapping of Adsorbed Species on CeO, Catalysts CO Hydrogenation at 673 K over CeO, The products of CO hydrogenation at 673 K over CeO, are presented in Table 3. The main species are C,, C, and C, hydrocarbons which formed with similar yields.The selec- tivities of isobutene in C4 hydrocarbons and of isoprene in 913 Table 3 Chemical trapping products at 523 and 673 K on CeO, amount of products product selectivity from chemical in CO hydrogenation trapping/pmol (%)" product 523 K 673 K 523 K 673 K hydrocarbon c, 0.47 0.23 26 0.001 0.018 11 28c2 tr 0.008 11 9c3 c4 tr 0.018 20 21 c5 tr 0.011 20 8 C6 + b b 38 8 oxyhydrocarbon bCH,OH 0.36 0.18 56b bCH,CHO C b 1 bC,H,CHO 0.001 b(CH,),CHCHO 0.04 b 26 b 5 b(CH,),CHCH,OH b b 1 b(CH,),CHCOCH c b 7 b(CH ,),CHCOC,H b b b(CH,),CHCOCH(CH,), b 55 2.5 ~~~~~~~~~ a Formation rates of hydrocarbons are 0.55 x lo7 mol g-' h-' at 523 K, (surface area = 11 m2 g-I), and 427 x lo7 mol g-l h-' at 674 K (surface area = 21 m2 g-').Formation rates of oxyhydrocarbons are 2.5 x lo7 at 523 K and almost zero at 673 K. Not detected. Trace amount although the overall selectivity (21%) of C, hydrocarbons at 673 K with CeO, is much lower than that with ZrO, (63.4%).' CO hydrogenation on CeO, with pre-adsorbed methoxide resulted in the increase of methane alone and had almost no effect on the other hydrocarbons, which is different to the case for ZrO, . Chemical Trapping on CeO, after CO Hydrogenation at 673 K The chemical trapping products from CeO, after CO hydro-genation at 673 K are mostly methane and methanol as shown in Table 3. +c6+ hydrocarbons form in much smaller amounts and the product distribution is similar to that of CO hydrogenation. CO Hydrogenation over CeO, at 523 K The catalyst for CO hydrogenation at 523 K was treated with H, at 773 K for 16 h to decrease the induction time, which was, however, still long as shown in Fig.6 and 7. Hydrogen pretreatment leads to very high initial yields of hydrocarbons, which rapidly decrease and reach the steady state after 72 h, as shown in Fig. 6. Initially methane forms at the highest rate, however, higher hydrocarbons become the main pro- ducts in the steady state. Fig. 7 shows the rate of formation of oxyhydrocarbons with time. Methanol and c6 ketone pass through maximum yields. Propanal increases significantly after 2 days and then rapidly decreases.The rate of formation of 2-methylpropanal increases gradually over 5 days. Diiso- propyl ketone, which forms in the highest yield among the oxyhydrocarbons, reaches steady yield within 2 days. Chemical Trapping on CeO, after CO Hydrogenation at 523 K The yield of chemical trapping products is different from that of the products of CO hydrogenation, as shown in Fig. 8. Methanol and propanal reach maximum values after 1 day. The maximum yield of methane occurs a little later, while 2-methylpropanal starts to form after 2 days, and is still increasing after 4 days, C, and C5+ aldehydes were not C, hydrocarbons at 673 K are 66 and 71%, re~pectively,~,~detected. 914 J. CHEM. SOC. FARADAk' TRANS., 1994, VOL.90 n 0,-0.14 0 r 5.I '=-r 37 0.12 D CT, 2 h n0, cn O.lOl\ .-QP 2 +.' iQ .-EE + -D .-al > reaction ti me/days Fig. 8 Dependence of product yield from chemical trapping on the reaction time of CO hydrogenation on CeO, at 523 K: methane (a);methanol (0);C, ,propanal (A) and C,, 2-methylpropanal (0) 0 1 2 3 4 5 reaction time/days methoxide. However, treatment of catalysts with 0, after CO hydrogenation indicates that the sum of the amounts of Fig. 6 Dependence of the rate of hydrocarbon formation on the reaction time of CO hydrogenation over CeO, at 523 K: C, (O), C, formate and methoxide species is 74 and 33% of the amount (01, and c6 (0) of CO, corresponding to the total carbon species on the cata- C, (V), C, (O),C, (0) lyst at 523 and 643 K, respectively.In order to confirm the amounts determined in the chemical trapping experiments, 0.1: the ratio of the amount of formate or methoxide species at 523 K to that at 643 K was compared to the ratio of the IR band intensity at 523 to that at 673 K.', The ratio of inten- c sity of the 0-C-0 asymmetric stretching band due to the I r formate species at 523 (1566) to that at 673 K (1560 cm-') r I cn / was 7.4, while the ratio of the amount of methyl formate h 0, obtained at 523 K to that formed at 643 K is 7.7 from Table -2. Similarly, the ratio of intensity of the C-0 stretchingO 0.1( band due to methoxide at 523 (1144) to that at 673 K (11345--. i c cm-') was 5.4, while the ratio of the amount of methanol at 2 523 K to that at 643 K is 4.7 from Table 2.This indicates C .-that the amounts of methanol and methyl formate deter- c E mined by chemical trapping are near to the actual amounts *b of adsorbed methoxide and formate species, although it is C presumed that the difference between the molar absorption 5 O.O! coefficients of the bands at 523 and 673 K is negligible and iQ 2 that the difference between the peak intensities at 643 and 673 K are within experimental errors. Therefore, the large z r>. difference between the amount of CO, and the sum of methyl 0 formate and methanol suggests the presence of species other than formate and methoxide on the catalyst surface. The species most likely to escape detection by NMR and chemical I trapping measurements is coke, because (i) the catalyst was 0 1 2 3 4 5 dark grey and grey after CO hydrogenation at 643 and 523 reaction time/days K, respectively, (ii) the NOE effect for coke will be small Fig.7 Dependence of the rate of oxyhydrocarbon formation on the because of the lack of hydrogen, and (iii) the reaction with reaction time of CO hydrogenation over CeO, at 523 K: C,, meth-aqueous diluted HCl solution vapour and dimethyl sulfate anol (0)C, ,propanal (A) C,, 2-methylpropanal (0);C, , methyl would give no gaseous products. Since a broad peak around isopropyl ketone (0);c6, ethyl isopropyl ketone (0);isopropyl ketone (m) and C,, di- 60 was also observed on fresh ZrO,, it could not be due to carbon species from CO hydrogenation, though the species could not be identified.Thus, only formate and methoxide can be possible intermediates in the formation of isobutene Discussion from CO. On the other hand, the pre-adsorption of methox- Adsorbed Species on ZrO, ide leads to the initial high yield of' hydrocarbons with high C, selectivity as shown in Fig. 3, while the treatment of CeO, In situ IR spectra of CO hydrogenation over ZrO, at 523 with pre-adsorbed formate with a mixture of CO and H, and 673 K showed only the presence of formate and methox- results initially in the formation of a large amount of CO, ide species.14 Solid-state NMR and chemical trapping mea- and only small amount of hydrocarbons, and lower C, selec- surements also show no species other than formate and tivity as shown in Fig.4 and 5. Since the decrease of CO, is J. CHEM. SOC. FARADAY TRANS., 1994, VOL. 90 accompanied by an increase in hydrocarbons and the large amount of CO, is formed from the decomposition of the adsorbed formate species, it is unlikely that surface-adsorbed formate is an intermediate in hydrocarbon formation, espe- cially C, hydrocarbons. Therefore, only methoxide may be an intermediate in the formation of C, hydrocarbons from CO. Reaction Mechanism Selective Ethene Formation on CeO, As described above, the formation of 2-methylpropanal and isobutene seems to be accounted for by the aldol-condensation-type reaction from acetaldehyde. The forma- tion of the isobutyl group in the synthesis of isobutyl alcohol has also been described by the aldol-condensation type reac- Adsorbed Species onCeO,'9, tion.' However, the formation of acetaldehyde, which Since formate is not an intermediate but rather a poison for hydrocarbon formation, chemical trapping by a vapour of aqueous diluted HCl solution was carried out only for trap- ping adsorbed molecules such as alkoxide, acyl, q2-aldehyde, alkyl, p-methylene, and carbene species.The main products from chemical trapping on the CeO, catalyst are methanol and methane after CO hydrogenation at both 523 and 673 K. Methanol is presumably derived from methoxide, as in ZrO, . 30 pmol g-' methoxide is adsorbed on ZrO, at 643 K, i.e. 0.53 pmol m-,, while 0.18 pmol g-' is absorbed on CeO, at 673 K, i.e.8.6 x lo-, pmol m-,, and 0.36 pmol g-' at 523 K, i.e. 33 x lop3pmol m-2. Therefore, twice as much methoxide is adsorbed on ZrO, than on CeO, , even at higher temperatures. The pre-adsorption of methoxide only enhances the formation of methane for CeO,, but the formation of higher hydrocarbons with high C, selectivity for ZrO, . This may indicate that the propaga- tion step from C, to C, species is slower for CeO, than for ZrO, . The formation of methane during chemical trapping on CeO, is the most notable point. The reaction of methoxide with protons gives methanol but not methane. It is known that CeO, is partially reduced by H, and reoxidized by H,O to form H, .l5 The reaction of methoxide with H, ,which is formed by the reaction of the partially reduced CeO, with H,O during chemical trapping, should be negligible, because the coexistence of H,O in a mixture of CO and H, retards CO hydrogenation.16 Therefore, the surface species giving methane during chemical trapping could be a C, species such as methyl, p-methylene or carbene.0.23 pmol g-' of C, species is adsorbed at 673 K (Table 3), i.e. 0.011 pmol md2, as the surface area is 23 m2 g-'. There are 5.4 x lo-'' mol surface Ce ions on the fluorite structure and the unit cell con- tains two Ce ions on the ~urface.'~ Therefore, the surface concentration of the C, species is estimated to be ca. 0.1% of the surface Ce ion concentration. Similarly at 523 K, the con- centration of C, species is ca.0.043 pmol rn-, and therefore the surface concentration is about 0.4%.The high concentra- tion of C, species is characteristic of CeO, but not the ZrO, catalyst. Chemical trapping on CeO, after hydrogenation at 523 K yields aldehydes such as 2-methylpropanal in addition to methanol and methane (Table 3). Adsorbed molecules which would give such aldehydes on treatment with acid could be q2-aldehyde or acyl species. Treatment of the latter with acidic water usually gives carboxylic acid,'* however, it has been reported that the treatment of an acyl-zirconium complex with hydrochloric acid gives an aldehyde product.lg Therefore, adsorbed acyl species may be the source of alde- hyde upon chemical trapping. The plot of chemical trapping products us.time (Fig. 8) shows that methanol is the first should be the starting aldehyde for aldol condensation is not clearly understood. The following results are notable when considering a possible mechanism for C, aldehyde formation : (i) the high selectivity of ethene during CO hydrogenation over CeO, at 673 K and (ii) the formation of methane during chemical trapping. First, as reported previously,20 CeO, is a good catalyst for the formation of ethene from CO and H,, i.e. it has higher C, than C, selectivity (29% us. 9%) in total hydrocarbons and gives a higher proportion of ethene (96%) in C, hydrocarbons than propenes (84%) in C, hydrocar- bons. Transition metals" and ZrO, catalysts usually form ethene with lower C, hydrocarbon selectivity than that expected from the Schulz-Flory distribution and with higher ethene selectivity in C, hydrocarbons than propenes in C, hydrocarbons.These results seem to suggest that there could be one particular reaction path which forms only ethene in the course of CO hydrogenation over CeO, . Considering the reaction to give ethene alone, coupling of either p-methylene,' or carbene species,, could explain the unusually high ethene selectivity. Secondly, methane formation upon chemical trapping indicates the presence of methyl, p-methylene or carbene species on the catalyst during CO hydrogenation as seen above. This indicates that there is a p-methylene or carbene species on CeO, ,which induce high ethene selectivity.On the other hand, it has been reported that the thermolysis of a trimeric (q2-formaldehyde)zir-conocene compound leads to the expulsion of methylene to form hydrocarbons [eqn. (l)] and 'metal oxide' fragments,, and that the q2-methoxymethylzirconocene,species decom- poses at 523 K to give ethene and the methoxyzirconium complex shown in eqn. (2).,"4 /"\"'$, -(CPZ~O)~ hydrocarbons (1)+ Cp2Zr-0 However, Toreki et al. reported that [{Ta(silox),Cl) ,(pH),] (silox = BuiSiO) and [{Ta(silox),H,},J reacts with CO to form [{Ta(silox),Cl) ,@-H)(p-CHO)] and [{Ta(silox),Cl) ,@-CH,)] respectively, as in the following reaction^.^ HI r. co /:Ap-H),] lTa: ,TaC I(silox),] (3)[{Ta(silo~)~CI}~( [(si10x)~C H H\ ,Hco product, followed by propanal, and then finally 2-[{Ta(~iIox)~H2)~] methylpropanal.This is the order of the species accumulated on the CeO, surface, suggesting chain growth by an aldol- condensation-type reaction as 2-methylpropanal is the end H They also showed that aqueous degradation of p-formyl and product from acetaldehyde in this type of reaction. The p-formaldehyde species produces methanol, and that the p-results of CO hydrogenation at 523 K (Fig. 7) support this methylene species is quenched by H,O to give methane in .~~hypothesis, although the formation of ketone cannot be so quantitative yield. Also, Casey et ~ 1 and Gladysz et easily explained. have shown that o-formyl and hydroxycarbene species react [(sil~x)~Ta~~~Ta(silox)~(4) '0'1 IH with protons to give methanol.26 These reactions, and the in situ IR measurements during CO hydrogenation at 473 K over ZrO, showing the presence of formyl, dioxymethylene and methoxide specie~,'~*~~ and the formation of formalde- hyde with no induction time2' suggest that similar reactions occur for ZrO, and CeO, catalysts as shown in Scheme 1.r! A Scheme 1 Proposed mechanism of selective ethene formation Formation of Isobutenefrom C,Species In the above discussion an aldol-condensation type reaction for the formation of 2-methylpropanal from acetaldehyde and the thermolysis of q2-formaldehyde to form carbene and then ethene have been proposed. The next question concerns the C, precursor of acetaldehyde, since the latter is a starting compound in an aldol-condensation-type reaction to form isobutene, for which selectively formed ethene is not the pre- cursor.29 The formation of acetaldehyde or ethanol on oxide cata- lysts, or those for isobutyl alcohol formation, has been explained by the insertion of CO into the carbon-metal bond of the q2-formaldehyde species7p8 and the reaction of formal- dehyde with adsorbed formyl species.12 However, the carbon atoms of the carbonyl group of q-and/or pformaldehyde, J. CHEM.SOC. FARADAY TRANS., 1994, VOL. 90 y3 CH2 M-H y3 co -A c=o -M -0-M --M -0-M __ -L+-I\ -0-M I I It I1 Scheme 3 Formation of acyl species from q2-formaldehyde for alkene hydrogenation,20 supporting the above suggestion. The formation of acyl species leads to the formation of 2- methylpropanal and then isobutene.The formaldehyde for the aldol-condensation-type reaction could result from a large amount of adsorbed methoxide species. The selective formation of isobutene is due to the 2-methylpropanal being an end product in the aldol-condensation-type reaction. Formation of Diisopropyl Ketone CO hydrogenation over CeO, at 523 K results in the forma- tion of ketones, although they are hardly detected by chemi- cal trapping. Diisopropyl ketone can be formed in two ways: (i) the ketonisation reaction of 2-methylpropanal or (ii) the aldol-condensation-type reaction of acetone, which could be formed oia ketonisation of acetaldehyde. As methyl isopropyl ketone and ethyl isopropyl ketone are formed and diiso- propyl ketone reaches a steady state faster than 2-methylpropanal ketones are likely to be formed by the aldol condensation-type reaction of acetone with formaldehyde as shown in Scheme 4.Diisopropyl ketone is the end product from the reaction of acetone with formaldehyde. This is the reason for the high selectivity of diisopropyl ketone com- pared to other ketones. CO 0:: PCH3C-H A2CH3-g-CH3 CH20 H2 CH3-CHZ-C-CH340 H2O I I1CHo CH30---$>CH3-CH-C-CH3 H!0 ketone and adsorbed formyl species are ele~trophilic.~~*~~.~' Therefore, the above indication that p-methylene or carbene species are present on the CeO, catalyst could naturally suggest that these species or their precursors may participate as a key intermediate in the formation of acetaldehyde, and, therefore, 2-methylpropanal and isobutene.Where p-methylene or carbene are assumed to be precur- sors of acetaldehyde, carbonylation followed by reaction with hydrides gives acyl species as shown in Scheme 2. H3C. EH2 -'k,-,O -M-CH3-M-I M-H c=o or or L-4-0-y-uu ?\ I I Scheme 2 Formation of acyl species from p-methylene or carbene On the other hand, the reaction of q2-formaldehyde with hydrides gives adsorbed methyl species.31 The presence of p-methylene or carbene species on CeO, suggests that the mechanism outlined in Scheme 2 is pre- dominant for this catalyst. The experimental results that CO hydrogenation over CeO, results in the selective formation of ethene at 673 K and the formation of aldehyde and ketones at 523 K indicate that the thermolysis of p-methylene or carbene species has a higher activation energy than carbon- ylation or the aldol-condensation-type reaction. On the other hand, no detection of ethene on ZrO, may suggest the forma- tion of acyl species through the path-way shown in Scheme 3.When hydrogenation of p-methylene or carbene to form methyl species is fast, the pathways in Scheme 2 and 3 are hard to differentiate. ZrO, is much more active than CeO, CH30CH3-CH-C-CH2-CH3I II CH20 v H2$ H20 t CH,O CH3 I1 I CH3-AH-C-CH-CH3 Scheme 4 Formation pathway for ketones Conclusions This study shows that: (i) The formation of branched carbon- chain compounds from CO hydrogenation over oxide cata- lysts may be attributed to the aldol-condensation-type reaction of aldehydes and ketones with formaldehyde.2-Methylpropanal is the end product from the reaction of acetaldehyde with formaldehyde and diisopropyl kentone is that from acetone with formaldehyde. 2-Methylpropanal undergoes hydrogenation and dehydration yielding iso-butene. (ii) Acetaldehyde and acetone are produced by hydro- genation and the hydration, respectively, of an acyl intermediate. The acyl species is assumed to be formed by the carbonylation of methyl or carbene species which are thermal decomposition products of q2-formaldehyde. References 1 T. Maehashi, K. Maruya, K. Domen and T. Onishi, Chem. Lett., 1984,747.J. CHEM. SOC. FARADAY TRANS., 1994, VOL. 90 917 2 3 4 5 K. Maruya, T. Arai, K. Domen and T. Onishi, in Catalytic Science and Technology, ed. S. Yoshida, N. Takezawa and T. Ono, Kodansha, Tokyo, 1991, vol. 1, p. 457. K. Maruya, K. Ito, K. Kushihashi, Y.Kishida, K. Domen and T. Onishi, Catal. Lett., 1992, 14, 123. H. Pichler and K. H. Ziesecke, Brennst. Chem., 1949,30,13. R. Kieffer, J. Varela and A. Deluzarche, J. Chem. SOC., Chem. 17 18 19 T. Arai, K. Maruya, K. Domen and T. Onishi, J. Catal., 1993, 141, 533. P. Pino, F. Piacenti, and M. Bianchi, Organic Synthesis via Metal Carbonyl, ed. I. Wender and P. Pino, Wiley-Interscience, New York, 1997, p. 233; J. Falbe, ‘New Syntheses with Carbon Monoxide,’ Springer, New York, 1980. C. A. Bertelo and J.Schwartz, J. Am. Chem. SOC.,1975,97,228. 6 7 8 Commun., 1983,763. K. Maruya, A. Inaba, T. Maehashi, K. Domen and T. Onishi, J. Chem. SOC., Chem. Commun., 1985,487. T. J. Mazanec, J. Catal., 1986,98, 115. S. C. Tseng, N. B. Jackson and J. G. Ekerdt, J. Catal., 1988, 109, 284. 20 21 22 23 T. Arai, K. Maruya, K. Domen and T. Onishi, J. Chem. SOC., Chem. Commun., 1987,1757; Bull. Chem. SOC. Jpn., 1989,62,349. W. A. Hermann, Ado. Organomet. Chem, 1982,20,159. E. 0.Fisher, Adv. Organomet. Chem, 1976,14, 1. K. Kropp, V. Skibbe, G. Erker and C. Kruger, J. Am. Chem. SOC.,1983,105,3353. 9 10 R. Kieffer, G. Cherry, J. Varela and R. Touroude, J. Chim. Phys., Phys. Chim. Biol., 1987,84,901. H. Idriss, R. Kieffer, P. Chumette and D. Durand, Znd. Eng. Chem. Rex, 199 1,30,1130.24 25 G. Erker, C. Kruger and R. Schlund, 2.Natuqorsch., Teil B, 1987,42,1009. R. Toreki, R.E. LaPointe and P. T. Wolczanski, J. Am. Chem. SOC.,1987,109,7558. 11 12 13 14 J. P. Hindermann, G. J. Hutchings and A. Kienemann, Catal. Rev. Sci. Eng., 1993,35, 1. J. G. Nunan, C. E. Bodgan, R. G. Herman and K. Klier, Catal. Lett., 1989, 2, 49; J. G. Nunan, C. E. Bodgan, K. Klier, K. J. Smith, C-W. Young and R. G. Herman, J. Catal., 1992,116, 195. A. Delzarche, R. Kieffer and A. Muth, Tetrahedron Lett., 1977, 38, 3357; A. Deluzarche, J. P. Hindermann and R. Kieffer, Tetrahedron Lett., 1978, 39, 2787; J. Saussay, J. C. Lavalley, T. Rais, A. Chakor-Alami, J. P. Hindemann and A. Kienemann, J. Mol. Catal., 1984, 26, 159. H. Abe, K. Maruya, K. Domen and T. Onishi, Chem. Lett., 1984, 1875. 26 27 28 29 30 31 32 C. P. Casey and S. M. Neumann, J. Am. Chem. SOC., 1978, 100, 2544. J. A. Gladysz and W. Tam, J. Am. Chem. SOC., 1978,100,2545. T. Onishi, H. Abe, K. Maruya and K. Domen, J. Chem. SOC., Chem. Commun., 1985,103. K. Maruya, T. Fujisawa, A. Takasawa, K. Domen and T. Onishi, Bull. Chem. SOC. Jpn., 1989,62, 11. D. M. Roddick and J. E. Bercaw, Chem. Ber, 1989,122, 1579. G. Erker, U. Do$ J. L. Atwood and W. E. Hunter, J. Am. Chem. SOC., 1986,108,2251. T. Arai, K. Maruya, K. Domen and T. Onishi, Bull. Chem. SOC. Jpn., 1989,62, 349. 15 K. Otsuka, M. Hatano and A. Morikawa, J. Catal., 1983, 79, 493. 16 K. Kushihashi, K. Maruya, K. Domen and T. Onishi, J. Chem. SOC., Chem. Commun., 1992,259. Paper 3/056006;Received 16th September, 1993