Heterogeneously catalysed cyclisation reactions of ethyne over single crystal palladium and palladium catalysts

Materials

Chemistry

and Physics,

29 (1991)

105

105-115

Review

crystal palladium

palladium catalysts

Richard M. Lambert and R. Mark Ormerod Department IEW (U.K.)

of Chemistry,

University

of Cambridge,

tijield

Road, Cambridge

CB2

Abstract The unusual cyclisation reactions of ethyne on Pd surfaces are described and discussed with particular reference to analogies with the chemistry of transition metal clusters. Detailed spectroscopic and kinetic characterisation of the adsorbed reactant, intermediate and product species permits the molecular pathway to be constructed in considerable detail. A C,H4 tilted metallopentacycle is found to be a crucial surface intermediate en route to either benzene formation or the formation of heterocycles. The single crystal data enable useful predictions to be made about the behaviour of practical supported metal catalysts and about new areas of catalytic chemistry. These ideas are borne out in practice.

Introduction The view has often been expressed [I] that well-characterised metal single crystal surfaces can serve as useful model systems with which to investigate aspects of the catalytic chemistry of practical, high-area supported metal catalysts. There are a number of objections which might be raised against this, due to the widely differing conditions under which observations are usually made on the two types of specimen; this is particularly pertinent in the case of electron spectroscopic measurements which must necessarily be carried out under conditions of very low pressure. However, in favourable cases electron spectroscopic studies and kinetic studies (including isotope tracing) carried out on model planar single crystal catalysts can yield fundamental and unambiguous information about catalytic reaction mechanisms and the nature of promoter action. An example of the success of this approach is provided by recent work on the selective oxidation of ethene to the epoxide ]2, 31. On the other hand, it has been proposed [4, 51 that organometallic cluster compounds can act as useful model systems by means of which the structure, bonding and reactivity of adsorbates on extended metal surfaces can be rationalised. In this sense, studies on model planar systems may be

0254-0584/91/$3.50

0 1991

-

Elsevier Sequoia, Lausanne

106

regarded as providing a bridge between cluster chemistry and the catalytic chemistry of high area supported metal catalysts - the bridge has of course nothing to do with the relative sizes of the metal ensembles in the three systems, rather it is related to our ability to carry out detailed measurements, especially structural observations, with the three types of system. The low temperature tricyclisation of ethyne to benzene over single crystal Pd(l11) was reported almost simultaneously by Tysoe et al. [6] and by Sesselman et al. [7]; these observations were subsequently confirmed by other workers [ 8,9]. This interesting reaction can be studied over an extremely wide range of pressure - all the way from ultra high vacuum conditions to one atmosphere pressure [ 91. One remarkable feature of the reaction is the way in which benzene evolution into the gas phase occurs at very low temperatures (- 200 K). It also exhibits a number of interesting analogies with well-known organometallic cluster chemistry of ethyne and its derivatives. It therefore provides a unique opportunity to study a most interesting catalytic process over a range of conditions which permit reliable and detailed characterisation of the various surface species involved in the reaction mechanism. This review summarises recent work carried out in our laboratory which was aimed at elucidating the mechanistic chemistry of ethyne cyclisation reactions in as much detail as possible, exploring the analogies with known cluster chemistry and extrapolating the single crystal data to predict the behaviour of practical supported catalysts and at the same time to open up new areas of catalytic chemistry. This has included identification and isolation of the crucial reaction intermediate, determination of the bonding of reactants, products and intermediates at the metal surface and elucidation of the molecular pathway by a combination of kinetic and spectroscopic studies, including isotope tracing.

Results

and discussion

of ethyne tricyclisation over Pd(ll1) Tysoe et of the reaction kinetics; a C, species was detected in the gas phase, the results indicating that this species was the precursor to benzene formation. The chemical identity of this precursor could not however be established by the molecular beam data alone. Angle resolved photoelectron spectroscopy (ARUPS) showed [lo] that at low temperature ( < 200 K) chemisorbed ethyne lies essentially flat on the metal surface. This species is believed to be the precursor to benzene formation; photoelectron spectra show that at higher temperatures it undergoes a rearrangement to yield another species which is characterised by an unsaturated C =C bond and with the molecular axis perpendicular to the metal surface. This second species, thought to be vinylidene, is likely to be the precursor to ethene formation - a reaction which competes with benzene Following

the discovery

al. [ 61 carried out a molecular beam study [lo]

107

formation when adsorbed hydrogen is available [ 111. More recent studies using near-edge X-ray absorption fine structure observations (NEXAFS) [ 121 indicate that the flat-lying ethyne species is characterised by an H-C-C angle of 117” and a C-C bond length of 1.3 1 A suggesting significant re-hybridisation of the molecule upon adsorption; i.e. the precursor to benzene formation is a flat-lying (T-T bonded ethyne molecule. Such bonding is of course well known in inorganic cluster compounds of ethyne [ 13, 141. Additionally, an X-ray structure determination of the Pd complex, Pd(PPh&(Cz(COOMe), [ 151 indicates that the Pd and both acetylenic carbons are virtually coplanar, with the acetylene bent into a cti configuration (R-C-C angle = 149”) and a C-C bond length of 1.28 A, in good accord with our results for an extended metal surface. As will become apparent later, the bonding geometry of the reactively formed benzene is of importance in determining the unusual kinetics of benzene desorption. Initial studies using ARUPS [IO] indicated that the reactively formed benzene is not co-planar with the metal surface. For normal exit photoemission, the selection rules are such that no emission should be observed from the eZs orbital of a flat-lying benzene molecule (16). In the present case, the eZg emission is visible indicating that the product benzene molecule is formed in a tilted geometry. Further NEXAFS studies [ 171 on the adsorption geometry of benzene on Pd(ll1) at high coverages have indeed confirmed that the molecule is very significantly tilted with respect to the metal surface. Figure 1 shows the observed dependence of the intensity of the C( 1s) --) TT* resonance as a function of the angle between the synchrotron light polarisation and the surface normal. Also shown are the calculated angular dependences for a series of different tilt angles: it can be seen that the data correspond to a tilt angle of about 30” between the molecular plane and the metal surface. As already noted, this finding is of crucial importance to a full understanding of the reaction kinetics.

Molecular plane angle

C K-Edge NEXAFS Benzene/Pd( 111)

20 40 60 Polarization vector angle

80

Fig. 1. T-resonance intensities as a function of the angle between light polarisation and the surface normal. Experimental data (black squares) and calculated curves for four different angles of the molecular plane with respect to the surface.

108

Patterson and Lambert [ 18 ] used deuteriumisotope tracing to demonstrate that the molecular pathway from ethyne to benzene involves no cleavage of C-C or C-H bonds. Figure 2 shows the results of multiplexed temperature programmed reaction spectra obtained following the coadsorption of a mixture of &Ha, CaHD and CzD2 on Pd(lll). It can be seen that all possible deuteriobenzenes are formed, and all appear in the two peaks at - 200 K and - 500 K. Table 1 shows the observed relative yields of the various benzenes (corrected for mass spectrometer fragmentation) and the calculated yields for both associative and dissociative reaction mechanisms. It is at BENZENE FROM ACETYLENE

ON Pd(ll1)

isotopic distribution GJH~+C~HD+C~D~

TEMPERATURE

from

/K

Fig. 2. Temperature programmed C,H,/C2DH/C2D, TABLE

reaction spectra of benzene following mixture at 150 K.

a 2 L dose of a known

1

Relative yields of reactively

formed benzenes from a 48% C2H2, 46% C,D2, 6% CzHD mixture observed

calculated mechanism

T<350

‘AH,

11.0

CAD C,H,Dz CeH,D, CBHZD, GHD, ‘AD,

3.1 32.2 8.6 31.4 3.9 9.8

K

T>350 8.5 4.7 28.7 11.4 30.9 6.1 9.7

K

assoc.

dissoc.

10.7 4.1 32.0 8.0 31.3 4.0 10.0

1.6 9.4 23.4 31.2 23.4 9.4 1.6

109

once apparent that the data for both peaks agrees with the calculated values for an associative mechanism (i.e. no C-C or C-H cleavage). Closer inspection of Table 1 shows that the results for the low temperature peak are in very good agreement with the associative mechanism, whereas those for the high temperature peak show small systematic deviations from the calculated distribution: the observed yields of those benzene molecules containing an odd number of D atoms are greater than the calculated yields. Control experiments show that these deviations are merely due to H/D scrambling between the product molecules, a process which becomes significant at higher temperatures. The conclusion therefore is that all the benzene in both peaks is formed by an associative mechanism. Deuterium isotope tracing was also used by Patterson and Lambert (19) to demonstrate that the surface intermediate involved in benzene formation has the molecular formula C4H4. This was achieved by using dissociative chemisorption of cti-3,4-dichlorocyclobutene (DCB) as a reagent with which to seed the surface with a C4H4 species. This C4H, species reacted with coadsorbed C,D2 forming benzene with exactly the same kinetics as observed with ethyne alone as a reactant. In these experiments the only products formed were C6H4D2 (from C4H4 + C,D,) and t&H6 (from 3&D,). They indicate that the molecular formula of the reaction of the C, intermediate is indeed C4H,; furthermore they suggest that it is benzene desorption which is the rate determining step for the evolution of benzene into the gas phase. This latter conclusion is strongly confirmed by the observation that coadsorption of C,H6 and C2Dz leads to the evolution of C&H6 (by simple desorption) and C6D~j (formed by reaction, followed by desorption) with identical kinetics (Fig. 3). c&i

/

+ 3c2!& + c&j

1

TEMPERATURE

1x10-’

+ cc303

AMP

/K

Fig. 3. Comparison of desorption kinetics of chemisorbed CsHB (solid curve) and reactively formed CsDB (broken curve).

110

The use of DCB as a reagent for producing high surface concentrations of the relevant C4H4 reaction intermediate was further exploited in order to determine the chemical identity and chemisorption geometry of this species. High resolution electron energy loss spectroscopy (HREELS) carried out as a function of temperature and scattering geometry showed [20] that the C4H4 species is a metallopentacycle rather than say a cyclobutadienyl or cyclobutene species (this was achieved by comparison with infrared reference spectra of known compounds). It was further shown that this metallopentacycle is very significantly tilted with respect to the metal surface (by comparing the HREELS spectra collected at specular geometry with those collected at off-specular geometry). NEXAFS studies have confirmed that the intermediate is a metallopentacycle tilted by about 35” with respect to the surface [ 2 1 I. Examples of similar metallocyclopentadienyl species exist in organoseveral, e.g., metallic chemistry, and Fe&C)&%H& and Rh,(PF,),(C,(CO,Me),), have been successfully isolated [ 22,231. Furthermore several workers have proposed that a o+ r bonded cyclopentadiene is the crucial intermediate in the homogeneously catalysed cyclotrimerisation of ethyne [ 15, 24-271. In addition Whitesides and Ehmann [ 281 have used isotopically labelled but-2-yne (CH3C2CD3) to show that the COAX catalysed cyclotrimerisation reaction does not proceed via a metal-complexed cyclobutadiene intermediate, whilst Nakamura and Hagihara [29] demonstrated that the cyclobutadiene complex, Co(C,H,)(C,Ph,) does not react with dimethylacetylenedicarboxylate. On the basis of the above spectroscopic and kinetic results we may postulate the following reaction mechanism. &H,(g)

-

-150 K

CzHz(aXa-- rr bonded } A

C,H,(a){tiIted

metallocycle}

z

C6H6(a){tilted at high coverage} C,H,(g){high

A-200 K C,H,(g){low

T peak} -500

T desorption peak}

AL! the catalytic chemistry which results in benzene formation takes place at low temperature (,< 200 K) and a key aspect of this mechanism is the role of the tilted benzene which is formed at high surface coverages. Thus in a temperature programmed reaction measurement, when benzene is formed on a crowded surface, steric hindrance prevents this reactively formed benzene from adopting a flat-lying adsorption geometry. The tilted molecule is bonded relatively weakly to the metal surface and therefore undergoes facile desorption at - 200 K giving rise to the low temperature peak. As the temperature rises and the surface becomes depleted of reactant and product species, the remaining benzene molecules can lie flat, bonding more strongly to the surface, and ultimately desorbing at - 500 K in the high temperature peak. The tilted geometry adopted by the C4H4 metallacycle intermediate almost certainly results from ~-interaction between the metallocycle and surface Pd atoms adjacent to the Pd atom to which the metallocycle

111

is a-bonded. It may very well be that this tilted configuration facilitates the approach of the third adsorbed ethyne molecule which is then incorporated to form the benzene product. The very efficient nature of the overall reaction is thus understandable: the tilted intermediate assists formation of the product; the tilted product molecule is weakly bonded and therefore easily removed from the surface at low temperature. The hypothesis that a high molecular packing density in the reacting adsorbed layer leads to formation of tilted, weakly bonded benzene molecules is very strongly supported by the results of experiments involving twodimensional compression of ethyne overlayers on Pd(ll1) by addition of a chemically inert ‘spectator’ molecule. The early work of Tysoe et al. [6] showed that there was a critical ethyne coverage threshold of - 0.3 monolayers below which benzene formation does not occur - presumably because the reactant molecules are too far apart. This critical coverage is characterised by the appearance of a (&x &)R30” LEED st ructure [ 61. When two species are coadsorbed on the surface, broadly speaking there are two possibilities. They may either form a mixed adsorption layer with the two adsorbates distributed essentially randomly within a single adsorbed phase. Alternatively, if the lateral interactions between the adsorbed species are significantly different, phase separation can occur with the formation of separate islands of each of the two adsorbates. We have found that coadsorption of NO with ethyne leads to such islanding behaviour [30], and this may be exploited in order to test the compression/ tilting hypothesis referred to above. The choice of NO as spectator species was dictated by the following criteria. 1. Pure NO and pure ethyne form quite distinct and different surface structures which are readily distinguishable by their LEED patterns (this is not true of CO for example). One may therefore readily distinguish between the cases of homogeneous mixed adsorption layers and island formation. 2. NO desorbs from Pd(l11) at temperatures above those at which benzene formation occurs. 3. The N(KLL) Auger and N(ls) core level spectra provide convenient spectroscopic monitors of NO coverage in the presence of the organic species (again this would not be true of CO). It is found that when NO and ethyne are coadsorbed on Pd(l11) the effects of compression on the ethyne phase lead to dramatic changes in the threshold coverage required for formation and desorption of benzene at low temperatures. These effects are independent of the order of dosing of the adsorbates, as indeed one would expect if separate islands are being formed. This behaviour is illustrated in Figs. 4 and 5 which show the effect operating in both modes. As can be seen from Pig. 4 a 0.6 L dose of ethyne on clean Pd(ll1) leads to virtually no C&H6 desorption: no benzene is observed in the low temperature peak and only a very small amount in the high temperature peak. This ethyne dose corresponds to a surface coverage close to the critical surface coverage for benzene formation. If however the surface is precovered with a fixed amount of NO it is immediately apparent that low temperature

112

COMPRESSION OF REACTANT OVERLAYER BY SPECTATOR NO

.:

‘-.-w-e--_().65

L

. ...

.-.. ._._ -_...__ _._., --.__-___o~&j

L

COMPRESSION OF REACTANT OVERLAYER BY SPECTATOR NO

- -0.22 L : __.. ‘_--.__---___._..-._.-.__~~, fj L _ ._*...... _~~~~~._.-.~~~.~.~ 2L ..--

L

TEMPERATURE/K Fig. 4. Benzene formation induced by coadsorbed NO. Surface predosed with 0.7 L NO at 300 K followed by varying doses of &Hz at 170 K. Fig. 5. Benzene formation induced by coadsorbed NO. Varying precoverages of C,H2 dosed at 140 K followed by 1.0 L NO.

beruene desorption commences for vgy low doses of adsorbed ethyne. By 0.65 L there is effectively 100% conversion of ethyne to benzene and all the bewwze is &sorbed at low ~~~~~~e - this is ta be compared with the upper trace in Fig. 4 where on the clean Pd( 111) surface essentially no benzene formation occurs at ah for the same initial coverage of ethyne. Figure 5 shows the complementary experiment in which the surface was precovered with varying ~0~~ of C&Hafollowed by a tied 1.0 L dose of NO. Notice in this case that low temperature evolution of benzene into the gas phase is detectable at extremely small doses of ethyne (0.06 L) dramatically illustrating the effects of compression by NO: although the total number of ethyne molecules on the surface is very small, they are compressed into islands in which the local density exceeds the critical value for benzene formation and highly selective ethyne + benzene conversion occurs for ethyne loadings which would not yield any benzene at all in the absence of NO. Figure 6 provides a graphical summary of the results and illustrates how the ethyne coverage threshold for benzene formation is dramatically reduced by coadsorbed NO. These observations confirm nicely the idea that steric crowding leads to reactively formed benzene being generated in a tilted geometry followed by subsequent low temperature evolution into the gas phase. The sensitivity of the benzene-forming reaction to the details of metal surface structure, as deduced from single crystal studies, is borne out by our work (31 ‘f on a variety of supported Pd catalysts. Figure 7 shows results

113

REACTIONS OF ETHYNE OVER Pd/AlzOa

BENZENE YIELD AS A FUNCTION OF CzHz PRECOVERAGE

540K ethyne ;//_.~__A--~

A 2 .%i -a p -”

0.2

0.1

Acetylene

coverage

0.3

0.4

: :, L

. .

.._......

0

(monolayers)

1200 Time/s

Fig. 6. Comparison of benzene yield as a function of ethyne precoverage; exposure and (B) on clean Pd(ll1). T= 140 K. Fig. 7. Time variation in benzene, buta-1,3-diene and ethyne production from He to ethyne at 540 K: Pd/Al,OB catalyst.

benzene __ . . . . . . . . .

2400

(A) after 1.0 L NO

on switching gas feed

obtained with a Pd/A1203 catalyst. It can be seen that in addition to benzene, butadiene and butene were also detected and all three products exhibited the same time profile (the fall in activity to a relatively low steady state level is due to carbon deposition). Most interestingly, the synthesis of the C, species, butadiene and butene, and the absence of any C3 or C5 products is a strong indication that the cyclisation of ethyne does indeed proceed through a C4 intermediate - in excellent accord with the single crystal spectroscopic and kinetic data. The present results provide real evidence that such a C, species is indeed present at the surface of a practical working catalyst under reaction conditions. We were also able to show that higher metal loadings (larger particle sizes, lower metal areas) resulted in a higher steady-state level of berzzene fomulc km. This is in excellent agreement with the single crystal measurements which show that the cyclisation reaction occurs preferentially on the (111) plane of Pd; such planes will predominate on large metal particles but not on small ones. In this connection, it is of considerable interest to note that Osella et al. [32] have observed closely similar homogeneous catalytic chemistry with [Co,(CO),(HC,H)]. At high ethyne concentrations they find that cyclotrimerisation of ethyne to benzene becomes the dominant catalytic process, and they too observe co-production of butadiene and butene. The similarities with our work are very striking. They strongly suggest that, in this case at least, there is a close relationship between the bonding modes, intermediate species and molecular mechanisms characteristic of the cluster compound and the extended metal surface. Finally, the presence and chemical importance of the C4H, intermediate are further demonstrated by the results of scavenging experiments [33] in which coadsorbed oxygen may be used to react in a competitive fashion

114 OXIDATION OF C4H4 ON Pdtlll)

PARTIAL

C2H2 AND

C4H4 2CzH2

%-

%

C4H40

C4H40 !n

age .A,‘.

<,‘.

: .;:

.~I

.

.:’ ., ,_ -

.

.’

‘.;_,‘.“.-

200

,,-.“‘-,.-,.;

..,-, . ..‘..,,..,~ .,., _..,a

_

300

,‘....

..,

.‘:;

400

‘..;_

;,;*

500

TEMPERATURE/K

.,; y,,_“;:

b

200

600

i

Fig. 8. Partial oxidation of ethyne to furan on Pd(l11);

300

460

500

660

760

TEMPERATURE/K J

a, b refer to different ethyne precoverages.

Fig. 9. Partial oxidation of C,H4 intermediate to furan. TPR spectra obtained from oxygenprecovered Pd(ll1) after DCB adsorption at 270 K.

with the C,H, intermediate: in the presence of adsorbed &Ha both C6H6 and C,H,O are formed. Figure 8 shows the partial oxidation of ethyne to furan and Fig. 9 shows that furan is formed with essentially the same kinetics when the C4H, species is generated on the surface by dissociative chemisorption of DCB rather than directly from ethyne itself. In this case, the high surface concentration of the reaction intermediate can lead to a very high selectivity to fur-an formation - up to 80%. It is probably fair to say that the notion of partially oxidising ethyne to furan over a transition metal surface would never have arisen in the absence of our detailed understanding of the catalytic chemistry of this interesting system.

Acknowledgement RMO holds a University of Cambridge Oppenheimer Research Fellowship.

References 1 2 3 4

G. R. R. R.

A. Somorjai, A&. Catal., 26 (1977) 2. B. Grant and R. M. Lambert, J. Catal., 92 (1985) 364. B. Grant and R. M. Lambert, Langmuir, I (1985) 29. Ugo, Cata&sis Reviews, 11 (1975) 225.

115 5 E. L. Muetterties, T. N. Rhodin, E. Band, C. F. Brucker and W. R. Pretzer, Chem. Rev., 79 (1979) 91. 6 W. T. Tysoe, G. L. Nyberg and R. M. Lambert, f. Cherry. Sot., Chem. commune, (1983) 623. 7 W. Sesselman, B. Woratschek, G. Erti, J. Kiippers, H. Haberland, SI.& Sci., 130 (1983) 245. 8 T. M. Gentle and E. L. Muetterties, J. Phgs. Chem., 87 (1983) 2469. 9 T. G. Rucker, M. A. Logan, T. M. Gentle, E. L. Muetterties and G. A. Somorjai, J. Phys. them., 90 (1986) 2703. 10 W. T. Tysoe, G. L. Nyberg and R. M. Lambert, Surface Science, 135 (1983) 128. 11 W. T. Tysoe, G. L. Nyberg and R. M. Lambert, J. Phys. Chem., 90 (1986) 3188. 12 H. Hofmann, F. Zaera, R. M. Ormerod, R. M. Lambert, D. K. Saldin, J. M. Yao, L. P. Wang, D. W. Bennet and W. T. Tysoe, in preparation, 13 Y. Iwashita, F. Tamura and A. Nakamura, Irwrg. Chem., 8 (1969) 1179. 14 A. Meyer and M. Bigorgne, 0rganomet., 3 (1984) 1112. 15 P. M. Maitlis, Pure CLTXZ Appl. Chem., 33 (1973) 489 and references therein. 16 G. L. Nyberg and N. V. Richardson, Surf: Sci., 85 (1979) 335. 17 H. HoiTmann, F. Zaera, R. M. Ormerod, R. M. Lambert, L. P. Wang and W. T. Tysoe, Surf: Sci., 232 (1990) 259. 18 C. H. Patterson and R. M. Lambert, J. P&s. Chem., 92 (1988) 1266. 19 C. H. Patterson and R. M. Lambert, J. Am. Gem. Sot., 110 (1988) 6871. 20 C. H. Patterson, J. M. Mundenar, P. Y. Timbrell, A. J. Gellman and R. M. Lambert, St& Sci., 208 (1989) 93. 21 H. Hofmann, F. Zaera, R. M. Ormerod, R. M. Lambert, D. K. Saldin, J. M. Yao, L. P. Wang, D. W. Bennet and W. T. l’ysoe, in preparation. 22 W. Hubel, &em. Ber., 95 (1962) 1155. 23 M. A. Bennett, R. N. Johnson and T. W. Tumey, Inorg. Chem., 15 (1976) 107. 24 M. A. Bennett and P. B. Donaldson, Inorg. Chem., 17 (1978) 1995. 25 B. L. Booth, R. N. Has&dine and I. Perkins, J. Churn. Sot., L)alton ?hm.s., (1981) 2593. 26 M. H. Chisholm, K. Folting, J. C. Huffman and I. P. Rothwell, J. Am. Ch.em. Sot., 104 (1981) 4389. 27 J. P. Collman, 1. W. Kang, W. F. Little and M. F. Sullivan, Znorg. Chem., 7 (1968) 1298. 28 G. M. Whitesides and W. J. Ehmann, J. Am. Ch.em. Sot., 91 (1969) 3800. 29 A. Nakamura and N. Hagihara, Bull. Charm. Sot. Japan, 34 (1961) 452. 30 R. M. Ormerod and R. M. Lambert, Surf. Sci., 225 (1990) L20. 31 R. M. Ormerod and R. M. Lambert, J. Chem. Sot., Chem. COE=WFZU~. (1990) 142 1. 32 D. Oseha, S. Aime, D. Boccardo, M. Castiglioni and L. Milone, Znorg. China. Acta, MO (1985) 97. 33 R. M. Ormerod and R. M. Lambert, Catal. Lett., 6 (1990) 12 1.