Langmuir 1992,8, 862-869
862
Reactions of 1-Propanol and Propionaldehyde on Rh(ll1) N. F. Brown and M. A. Barteau* Center for Catalytic Science and Technology, Department of Chemical Engineering, University of Delaware, Newark, Delaware 19716 Received June 24, 1991.In Final Form: December 9, 1991 Temperature programmed desorption (TPD) and high resolution electron energy loss spectroscopy (HREELS) were used to investigate the reactions of 1-propanoland propionaldehyde on the clean Rh(ll1) surface. The principal emphasis of this study was to determine the reaction pathways and intermediates involved in alcohol and aldehyde decomposition. Propoxide intermediates were formed by the dissociation of 1-propanol;these were easily isolated in the HREEL spectrum at 147 K. Decarbonylation occurred at ca. 226 K, resulting in the formation of adsorbed hydrocarbon fragments and adsorbed CO. No aldehyde or acyl intermediates were detected during the course of the decompositionof this primary alcohol. 1-Propanol decomposed on the Rh(ll1) surface to produce CO, Hz,and surface carbon. In contrast, propionaldehyde adsorbed on Rh(ll1) decomposed to form ethane, as well as CO, H2, and surface carbon. A t 145 K, qZ(C,O)-propionaldehydewas observed by HREELS. By 251 K, some decarbonylation had occurred to yield bridge-bonded CO and ethyl species. A portion of the ethyl species formed by low temperature decarbonylation of propionaldehyde was hydrogenated to ethane, while the rest dehydrogenated to form ethylidynes,observedat 301 K by HREELS. Ethylidynespeciesdehydrogenatedto acetylides which ultimately decomposed to carbon and hydrogen. The surprising divergence of reaction pathways for propanol and propionaldehyde is consistent with observations for ethanol and acetaldehyde decomposition on Rh(ll1) and suggests that higher alkoxide decomposition on this surface may involve C-H scission along the alkyl chain, rather than at the a-carbon.
Introduction A great deal of effort has been devoted over the last two decades to studies of alcohol and oxygenate synthesis from CO and H2 with transition metal catalysts. A simple reaction network common to the group VI11 metals has been suggested by several and can be summarized as follows: CO + H2
-
CH,(ad)
-
nCH,(ad)
R(ad)
R(ad) + CO(ad) RCO(ad) + H(ad)
+ H20
RCO(ad)
RCH20H,RCHO
(1) (2) (3)
(4)
In this scheme hydrocarbons are assembled by addition of monomeric CH, units; oxygenate formation involves the termination of hydrocarbon chain growth via CO insertion to form acyl species. These surface acyls are converted to the various oxygenated products, including alcohols, aldehydes, and carboxylic acids. The mechanistic details of the individual steps in this network are much less clear, as are the kinetics of the various competing reactions and their dependence on catalyst properties. For example, although palladium appears to be a selective methanol synthesis catalyst and rhodium the metal of choice for higher oxygenate synthesis: a broad range of activities and selectivities have been reported for supported catalysts containing these metals. Clear distinctions remain, however. Despite its high selectivity for methanol synthesis, palladium has not been shown to exhibit activity for higher alcohol synthesis from CO and H2. The absence of higher alcohol production over palladium catalysts has been attributed to a lack of chain growth activity due to the inability of palladium to dissociate CO under conditions favorable for alcohol (1)Bell, A. T. Catal. Reu.-Sci. Eng. 1981, 23, 203. (2) Biloen, P.; Sachtler, W. M. H. Adu. Catal. 1981, 30, 165. (3) Ponec, V. Catalysis (London) 1981, 5 , 48. (4) Sachtler, W. M. H.; Ichikawa, M. J.Phys. Chem. 1986,90, 4752.
p r o d ~ c t i o n .The ~ ability of palladium to catalyze CO insertion has been examined by adding hydrocarbons to the syngas feed~tream.~These studies showed that hydrogenation of adsorbed hydrocarbons is more efficient on palladium catalysts than hydroformylation, consistent with the absence of higher alcohol production over palladium. Rhodium differs from palladium in both of these critical characteristics. As noted by Sachtler: it lies between those elements which readily dissociate CO and those which do not. This “compromise”is postulated to be effective for the synthesis of higher oxygenates, since the formation of these products will require both molecular and dissociated CO on the catalyst ~ u r f a c e . l - ~ ~ ~ ~ ~ Secondly, rhodium complexesin solution are the preferred catalysts for hydroformylation, and studies of supported rhodium catalysts have provided clear evidence by means of isotopic labeling for the CO insertion step in higher oxygenate s y n t h e ~ i s . ~ - ~ Implicit in such explanations of selectivity variations in syngas chemistry on the group VI11 metals is the assumption that the origin of these variations is kinetic rather than mechanistic. Wide selectivity variations can still arise from a common network of reactions such as that sketched above if the kinetics of individual steps vary in less-thancompletely coupled ways from catalyst to catalyst. I t may not be necessary to invoke changes in mechanism from one metal to the next just to explain selectivity differences. We have examined the decarbonylation of alcohols and aldehydes on rhodium and palladium surfaces in order to determine the extent to which a common network and a common set of intermediates can describe the reverse of the oxygenate synthesis reaction. Kinetic and spectroscopic studies of oxygenate decarbonylation on the Pd(5) Chuang, S. C.; Tian, Y. H.; Goodwin, J. G., Jr.; Wender, I. J . Catal. 1985, 96, 396. (6) Sachtler, W. M. H. Proc. Ibero-Am. Symp. Catal., 10th 1986,1327. (7) Orita, H.; Naito, S.; Tamaru, K. J. Catal. 1984, 90, 183. (8) Ichikawa, M.; Fukushima, T. J.Chem. Soc., Chem. Commun. 1985, 321. (9)Chuang, S. S. C.; Pien, S.-I. J. Mol. Catal. 1989,55, 12.
0743-7463/92/2408-0862$03.00/00 1992 American Chemical Society
Langmuir, Vol. 8, No. 3, 1992 863
1 -Propanol and Propionaldehyde on Rh(ll1)
(111) surface1&l2have provided evidence for adsorbed intermediates and the pathways connecting them that are in excellent agreement with the above network for oxygenate synthesis. On the clean Pd(ll1) surface, primary alcohols are dissociatively adsorbed at low temperature to form stable alkoxide species. Dehydrogenation of these alkoxides typically occurs to form adsorbed aldehydes, bound to the surface in an v2(C,0) configuration. The chemistry of v2 aldehydes on Pd(ll1) is the same whether they are obtained by adsorption of aldehydes or by dehydrogenation of the corresponding alcohols. These species dehydrogenate further to form acyl intermediates on the Pd(ll1) surface; although the C1homologue, formyl, has not been isolated, higher acyls are stable and easily identified by high resolution electron energy loss spectroscopy (HREELS).'O The higher acyls typically decompose between 250 and 300 K, depending on the structure of the hydrocarbon backbone, to liberate CO, hydrogen atoms, and hydrocarbons one carbon shorter than the parent molecule. Thus we have been able to demonstrate the sequence of intermediates between higher alcohols and CO + H + hydrocarbon species on the metal surface and to isolate and to characterize spectroscopically each of the intermediates in the sequence1@l2
- -
+
+
ROH RO v2-R'CH0 R'CO R' CO H Note that from right to left this is precisely the sequence proposed for higher alcohol synthesis by hydroformylation of surface hydrocarbons in CO + Hz The overall chemistry of alcohols on the Pd(ll1) surface is in good agreement with that observed on the Ni(ll1) and Pt(ll1) surfaces. Gates et al.13 have shown, for example, that ethanol decomposes on Ni(ll1) as on Pd(111)to yield CO, H2, and CH4, along with some acetaldehyde. Careful kinetic studies utilizing isotopic labeling demonstrated that the reaction sequence involves first 0-H scission to form alkoxides, followed by dehydrogenation of the a carbon which is the rate-determining step for the production of both acetaldehyde and methane. Thus, nickel appears to be somewhat less active than palladium for decomposition of adsorbed aldehydes and their subsequent reaction products. Work by Sexton and co-workers14has shown that Pt(ll1)is more active. These investigators reported that higher alcohols decomposed to CO + H2 + hydrocarbons one unit shorter on this surface. However,volatile hydrocarbon products were not observed; the surface-bound hydrocarbons were identified by their characteristic fingerprints in the TPD spectrum for hydrogen, as can also be done on Pd(l1l).l1 Ironically, while the oxygenate decarbonylation network on the platinum-group metals is consistent with that proposed for oxygenate synthesis on neighboring members of group VI11 such as rhodium, oxygenate decarbonylation on the Rh(ll1) surface appears anomalous in several respects. Ethanol and acetaldehyde have been observed to behave very differently on Rh(lll)l5 than on Pd(l1l).l&l2Even though decomposition of acetaldehyde on Rh(ll1) and Pd(ll1) results in the observation of volatile methane, the pathways and intermediates involved in the two cases are not consistent. It has been observed that acetaldehyde dehydrogenates to form an acetyl +
-+
(10)Davis, J. L.;Barteau, M. A. J. Am. Chem. SOC.1989,111, 1782. (11)Davis, J. L.;Barteau, M. A. Surf. Sci. 1987,187,387. (12)Davis, J. L.;Brateau, M. A. Surf. Sci. 1990,235, 235. (13)Gates, S.M.;Russell, J. N., Jr.; Yates, J. T., Jr. Surf. Sci. 1986, 171,111. (14)Sexton, B. A.;Rendulic, K. D.; Hughes, A. E. Surf. Sci. 1982,121, 181. (15)Houtman, C.J.; Barteau, M. A. J. Catal. 1991,130,528.
intermediate on P d ( l l l ) ,and this species is likely formed on Rh(ll1) as well. From observation of a kinetic isotope effect it has been proposed that the acetyl dehydrogenates to form ketene on Pd(ll1) which then undergoes decarbonylation to yield surface bonded CO and methylene species.1° These methylene species are then hydrogenated to form methane. On R h ( l l l ) , no kinetic isotope effect was observed, and since codosing of deuterium and acetaldehyde resulted in the clean formation of monodeuterated methane, it was proposed that the acetyl intermediate undergoes decarbonylation before dehydrogenation, thereby releasing methyl species.15 These methyl species can then hydrogenate to form methane. On both metals, ethanol dissociates to yield an ethoxide intermediate; these then decompose on Pd(ll1) via the acetaldehyde route, ultimately to release methane. On Rh(ll1) neither acetaldehyde intermediates nor methane was formed from ethan01.l~Thus cleavage of a 0-CH bond must occur before a-CH bond scission in the ethoxide. Such a step would form an oxametallacycle, the dehydrogenation and decarbonylation of which would yield methylene and/or methylidyne intermediates; these must ultimately decompose to deposit carbon on the surface of the metal. Hydrogenation of ethanol-derived CH,, n < 3, intermediates was not observed on Rh(ll1). The focus of the present study was to determine the reaction pathways and intermediates in the decomposition of 1-propanol and propionaldehyde on Rh(ll1) through the use of HREELS. The reaction products were monitored with temperature programmed desorption (TPD). The results of this study demonstrate that 1-propanol follows the trends observed by Houtman15for ethanol on Rh(ll1) rather than the trend observed for higher alcohols on Pd(ll1) by Davis.l*l2 To the extent that observations for oxygenate decarbonylation can be applied to oxygenate synthesis, they suggestthat reaction pathways and intermediates on supported rhodium catalysts may be more complex than the simple network outlined by eqs 1-4 above.
Experimental Section The experiments were performed in a two-level stainless steel vacuum chamber, which has been described previously.16 U1trahigh vacuum conditionswere achieved through the use of both an ion pump and a diffusion pump to create a base pressure of 2 X 10-10 Torr. The top level was equipped with four grid optics (PhysicalElectronics)for LEED and Auger electron spectroscopy (AES)experiments, as well as a quadrupole mass spectrometer (UTI 1OOC)multiplexed with an IBM XT for TPD experiments. The lower level contained a HREELS spectrometer (McAllister TechnicalServices). HREEL spectra were recorded at an electron beam energy of 5 eV which produced an elastic peak with a maximum intensity of 2 x lo5 counts/s and a full width at half maximum (fwhm) of 70 cm-l when the beam was reflected from a clean Rh(ll1) surface. The Rh(ll1) crystal was polished using standard metallographic techniques and was spot-welded on three 0.5-mm tantalum wires that were used to support as well as to heat the crystal. By passing current (
