Journal of Molecular Catalysis A: Chemical 131Ž. 1998 39±53

Thermochemistry of the selective dehydrogenation of cyclohexane to benzene on Pt surfaces

Bruce E. Koel a,), David A. Blank b, Emily A. Carter b a Department of Chemistry, UniÕersity of Southern California, Los Angeles, CA 90089-0482, USA b Department of Chemistry and Biochemistry UniÕersity of California, Los Angeles, CA 90095-1569, USA Received 14 August 1997; revised 18 October 1997; accepted 24 October 1997

Abstract

We use a quasiempirical valence bondŽ. QVB schemew E.A. Carter, Chem. Phys. Lett. 169Ž. 1990 218x for calculating the heats of formation of adsorbed species on surfaces to provide reliable estimates of the relative stabilities of several of the surface intermediates and adsorbate-surface bond strengths that are important in the selective dehydrogenation of cyclohex- ane to benzene over Pt surfaces. We estimate heats of adsorption and formation for adsorbed cyclohexylŽ. c-C611 H , a cycloallylic intermediateŽ. c-C69 H , cyclohexadiene Ž. c-C 68 H , cyclohexadienyl Ž. c-C 67 H , phenyl Ž. c-C 65 H , and benzyne Ž.c-C64 H on Pt surfaces. Estimates of these needed formation energies are then combined with the experimentally measured adsorption energies of cyclohexaneŽ. c-C612 H , cyclohexene Ž. c-C 610 H , and benzene Ž. C 66 H , to provide heats of reaction and an equilibrium thermodynamic description of a selective dehydrogenation mechanism that involves the step-wise, sequential removal of one H atom at a time from cyclohexane to form benzene. In addition, several further decomposition products of benzene are considered as precursors to undesirable carbon-forming reactions. In agreement with experimental observations, a cycloallylic speciesŽ. c-C69 H is shown to be an important stable intermediate in cyclohexane dehydrogena- tion that could also be involved in the catalytic rate-limiting step. Carbon±carbon bond cleavage and other possible surface reaction pathways are not considered herein. Addition of measured or estimated values for the activation barriers involved in the reaction on PtŽ. 111 can now give a fairly complete description of the energetics of this prototypical hydrocarbon conversion reaction on PtŽ. 111 surfaces. q 1998 Elsevier Science B.V.

Keywords: Cyclohexane; Benzene; Pt

1. Introduction ates, i.e., cyclohexene and a C69 H species, have been identified and the rate constants for some The dehydrogenation of cyclohexane to ben- of the sequential reaction steps have been mea- zene is a prototypical hydrocarbon reforming suredwx 3±5 . Dehydrogenation of cyclohexane to reactionwx 1,2 that has been studied extensively benzene proceeds efficiently on PtŽ. 111 , but at the molecular level in ultrahigh vacuum very little benzene is desorbed into the gas Ž.UHV . On Pt Ž. 111 , which is the substrate most phase since the Pt surface readily dehydro- thoroughly investigated, some of the intermedi- genates the benzene product. Studies of the cyclohexene intermediate have been carried out. ) Corresponding author. Also, studies are available for cyclohexyl and

1381-1169r98r$19.00 q 1998 Elsevier Science B.V. All rights reserved. PII S1381-1169Ž. 97 00255-0 40 B.E. Koel et al.rJournal of Molecular Catalysis A: Chemical 131() 1998 39±53 cyclohexadiene, which are likely intermediates lution electron energy loss spectroscopy in the reaction. In related work, the influence of Ž.HREELSwx 13,23 and reflection-absorption IR Biwx 6 and Cs wx 7 adatoms on this reaction, as spectroscopyŽ. RAIRSwx 20 have shown that well as many aspects of this reaction on ordered cyclohexane is adsorbed with the molecular wx Pt±Sn alloys 8±10 have also been investi- plane parallel to the surface with C3v symmetry, gated. probably in a chair configuration with three Even with all of the experimental work, sev- axial H atoms projecting toward the surface, eral of the likely intermediates associated with forming hydrogen bonds. Adsorption leads to a this reaction are still unidentified. The primary broad, low frequency C±H stretching band due challenge is to observe and identify metastable to extensive `softening' of the axial C±H bonds. intermediates by surface spectroscopic methods The surface coverage in the monolayer is be- and this is an elusive goal. Another key goal is tween 0.13wx 8 to 0.21 wx 6 monolayerŽ. ML . to determine the adsorption bond strengths and Upon heating, some cyclohexane desorbs, but surface bond energies for the reactant, products, most of itŽ 68%wx 6. converts to adsorbed ben- and intermediates in the reaction in order to zene which decomposes to hydrogen and ad- provide a thermochemical description of the sorbed carbon. Under UHV conditions, no sig- reaction. Temperature programmed desorption nificant gas-phase benzene is produced follow- Ž.TPD and laser induced thermal desorption ing cyclohexane adsorption. From the fraction Ž.LITD have been used for estimating the ad- of cyclohexane desorbed in TPD at 228 K, an sorption energies of cyclohexane, cyclohexene, adsorption energy of 13.9 kcalrmol can be and benzene. Unfortunately, these methods can- estimatedwx 8 . Dehydrogenation begins at 180± not be used for several of the adsorbed hydro- 195 K, corresponding to an activation energy, r carbon intermediates, since the barriers to sub- Ea, of 9.5±13.5 kcal mol over the 5±30% ML sequent surface reactions are less than the corre- coverage rangewx 5 . Benzene is not formed until sponding desorption activation energies and thus 300 Kwx 3±5,23 , in contrast to early reports of decomposition reactions occur rather than des- benzene formation at 200 Kwx 6,18 . Unfortu- orption into the gas phase. Furthermore, ad- nately, cyclohexeneŽ. c-C610 H remains as the sorbed molecules of this size are beyond the only species from cyclohexane dehydrogenation scope of current first principles theoretical that has been identified on PtŽ. 111wx 3 , although methods. Yet, it is essential to obtain equilib- an intermediate with C69 H stoichiometry has rium thermodynamic data on the relative stabili- been inferred to exist over the temperature range ties of surface species in order to better under- 180±270 K using LITD combined with Fourier stand the mechanism of this reaction. In this transform mass spectroscopyŽ. FTMSwx 4 and paper, we utilize a scheme that can be used to bismuth-postdosing thermal desorption spec- estimate the stabilities of surface species in troscopyŽ. BPTDSwx 22 , with the form of this order to provide an energetic foundation for the species favored to be a cycloallylicŽ. c-C69 H mechanism of selective dehydrogenation of cy- structure, rather than a simple loss of three axial clohexane to benzene. Below, we briefly sum- hydrogens, as deduced from HREELS datawx 23 . marize the experimental observations for this No direct evidence exists for cyclohexylŽ c- reaction and various mechanisms that have been C611 H.Ž. , cyclohexadiene c-C 68 H , or cyclo- postulated to explain the experimental data. hexadienylŽ. c-C67 H formation from cyclohex- The adsorption and decomposition of cyclo- ane. hexane on the PtŽ. 111 surface have been studied Some information has recently become avail- using a variety of surface science methodsw 3± able for the reactivity of the chemisorbed cyclo- 8,11±24x . Cyclohexane adsorbs molecularly at hexyl species on PtŽ. 111 . Using electron in- 100 K. Vibrational studies with both high reso- duced dissociationŽ. EID , low energy electron B.E. Koel et al.rJournal of Molecular Catalysis A: Chemical 131() 1998 39±53 41 bombardment of physisorbed layers of cyclo- remaining benzene to decompose to hydrogen hexane was used to cleanly prepare monolayer and adsorbed carbon. coverages of adsorbed cyclohexyl specieswx 8,25 . The adsorption and reaction of both 1,3-

In TPD experiments, H2 was the dominant reac- cyclohexadiene and 1,4-cyclohexadiene on tion product desorbed, with no cyclohexene and PtŽ. 111 have been investigatedw 4,10,11,15, only small amounts of benzene desorption ob- 24,26x . Cyclohexadiene chemisorption is com- served. The H2 evolution in the TPD spectra pletely irreversible and upon heating it dehydro- was H2 desorption rate-limited, and so no infor- genates to form benzene. Further heating causes mation is available on the activation energy for this adsorbed benzene product to either desorb subsequent C±H bond breaking in adsorbed cy- from the surface or dehydrogenate completely clohexyl. However, on two ordered SnrPtŽ. 111 to form surface carbon. HREELSwx 26 and SFG surface alloys which are less reactive than wx24 indicate that the nature of the surface bond- PtŽ. 111 , cyclohexene desorption is observed in ing is a quadra-s-type with four covalent Pt±C two peaks at 208 and 345 K. The peak at 208 K bonds with sp3 hybridization at those carbons. is cyclohexene desorption rate-limited with a In the SFG studies, a red-shift of the C±H desorption energy of 12.6 kcalrmol, and so we stretching frequency to 2770 cmy1 was ob- can place an upper limit on the activation bar- served and interpreted as a strong interaction wx rier for the reaction of cyclohexyl on PtŽ. 111 to between the surface and the CH2 groups 24 . be 12.6 kcalrmol. Saturation coverage for chemisorbed cyclohexa- Cyclohexene adsorption on PtŽ. 111 has been diene is about 0.18 MLwx 10,26 . LITDrFTMS investigated by several groupsw 3,4,6,9,11,15, experiments show that the activation energy for 22,24x . Generally, it is thought that cyclohexene conversion of cyclohexadienes to benzene is adsorbs molecularly on the PtŽ. 111 surface at very lowŽ at low coverages conversion occurs at ( r wx 100 K. TPD and HREELS show two different 115 K, Ž..Ea 7 kcal mol 4 , but studies using states of molecularly adsorbed cyclohexene in BPTDS indicate that dehydrogenation of both the first adsorbed layer. The desorption energy 1,3- and 1,4-cyclohexadiene occurs at 230±260 of the most tightly bound state desorbing at 281 K with an activation energy of 14"2 kcalrmol K in TPDŽ. at the same temperature as ethylene wx26 . is 17.1 kcalrmolwx 9 . Both of these states are The adsorption and reaction of the product of correlated to di-s-bonded cyclohexene as re- this reaction, benzene, has been extensively vealed by HREELS but in two different confor- studied on PtŽ. 111wx 6,8,16,17,27±32 . Benzene mations. The saturation monolayer coverage is adsorbs molecularly on PtŽ. 111 at low tempera- 0.14wx 9 to 0.23 w 19 x ML. Upon heating, cyclo- tures with the ring plane parallel to the surface hexene begins to dehydrogenate at 150 K at low wx16,17,27,29,30 . The monolayer coverage is wx ( r wx coverages 4 , with Ea 9 kcal mol, or at about 0.16 ML 8,32 . No desorption of benzene s higher coverages at 200±240 K Ž Ea 14.4 occurs at low coverages, but a state is formed at kcalrmol. wx 22 . Greater than 75% w 9,19 x of the higher coverages that desorbs at 480 K. With chemisorbed layer decomposes during TPD. A increasing coverage in the chemisorbed mono-

C69 H reaction intermediate in this reaction has layer, this peak grows and shifts to 450 K and a also been identified. Benzene is formed on the new broad peak forms at 320 Kwx 8 . Desorption ( r surface at temperatures near 300 K Ž Ea 18 activation energies of 30 and 20 kcal mol can kcalrmol. , and Henn et al.wx 22 deduced that been estimated for the high and low temperature the barrier for conversion of the c-C69 H species desorption states, respectively. About 55% of to benzene was 20.8 kcalrmol. Further heating the benzene monolayer dehydrogenates upon wx causes a fraction of the benzene formed to heating 32 , and reaction rate-limited H2 evolu- desorb from the surface at 400±500 K and the tion in peaks at 400, 500, and 525±750 K in 42 B.E. Koel et al.rJournal of Molecular Catalysis A: Chemical 131() 1998 39±53

TPD gives the barriers for benzene decomposi- formed is strongly bound to the surface by 89.7 tion to be 24.7 and 31.1 kcalrmol for losing kcalrmol and is calculated to have a relatively one H and three H atoms, respectivelywx 8 . low barrier to 10.1 kcalrmol for the first subse- While it seems clear that cyclohexeneŽ c- quent C±H bond cleavage. Kang and Anderson

C610 H.Ž. and a cycloallylic species c-C 69 H are conclude that the other two hydrogens come off isolable, stable intermediates in cyclohexane de- with lower barriers to form benzene and did not composition, and cyclohexylŽ. c-C611 H and cy- calculate aspects of this last step. In summary, clohexadieneŽ. c-C68 H are highly likely inter- they find that the initial removal of a H atom mediates due to the benzene formation in their from cyclohexane is the slowest step in the dehydrogenation reactions, there is little direct dehydrogenation mechanism, followed by the experimental information available concerning formation of the c-C69 H intermediate which the mechanism of conversion of cyclohexane to loses three more H atoms even more easily. benzene on PtŽ. 111 . However, most investiga- This is in contrast to the experimental observa- tors would likely consider a sequential, tion of the c-C69 H intermediate's respectable dehydrogenation mechanism as follows. Cyclo- stability. hexane adsorbs and dissociates to form cyclo- Some time ago, we presented a quasiempiri- hexylŽ. c-C611 H . Step-wise loss of H atoms cal valence bondŽ. QVB scheme for calculating leads to cyclohexeneŽ. c-C610 H , a cycloallylic the heats of formation of adsorbed species on wx speciesŽ. c-C69 H , cyclohexadiene Ž. c-C 68 H , surfaces 35,36 . This approach is a powerful cyclohexadienylŽ. c-C67 H and then to benzene method for roughly estimating the relative sta- Ž.C66 H on Pt surfaces. Benzene desorption is bilities of surface intermediates and adsorbate- then required to form the gas phase product of surface bond strengths. The first application of the selective dehydrogenation reaction. this method considered the mechanism and en- Previous theoretical work assessing mecha- ergetics of ethylene decomposition on PtŽ. 111 nistic aspects of cyclohexane decomposition on wx36 . In this study we estimated the heats of Pt has been limited to semiempirical methods. formation, heats of reaction, and Pt±carbon bond The calculations of Kang and Andersonwx 33 strengths for adsorbed ethylene, ethylidene, eth- were carried out using the semiempirical Atom ylidyne, ethyl, vinyl, acetylene, vinylidene, Superposition-Electron Delocalization Molecu- ethynyl, methyl, methylene, methyne and car- lar OrbitalŽ. ASED-MO method for the binding bide. While the details of the mechanism of of various C6 hydrocarbons on clusters of 15 Pt ethylene to ethylidyne conversion continues to atoms. A point to be kept in mind is that since be debated and investigated, it is important to this method is not a total energy method and is point out the strong support for the validity of in addition only a rough approximation to our method that has been accumulating recently. Hartree±Fock theoryŽ which itself is not able to First, our method relies on an accurate knowl- describe covalent metal±carbon bonds properly edge of the Pt±C bond strength. Our previous wx34. , it cannot be expected to provide reliable work used DŽ.Pt±C s53 kcalrmol, which was relative energetics. With this caveat, we summa- based on an ab initio value from a Pt±CH 3 rize their findings. In their model, cyclohexane complexwx 37,38 . Single crystal microcalorime- is adsorbed with a binding energy of 8.5 try measurements recently made in King's lab kcalrmol. The barrier for C±H bond cleavage wx39 to directly measure the adsorption energies in cyclohexane is calculated to be 29.5 of di-s-bonded ethylene at low temperatures kcalrmol, and the activation barriers for a sec- and ethylidyneŽ. from ethylene adsorption at ond and third hydrogen transfer from the lower 300 K can be usedwx 40 to provide another axial C±H bonds are calculated to be 27.4 and estimate of the Pt±C bond strength of 57.6 r r 17.5 kcal mol. The c-C69 H intermediate thus kcal mol, in fairly good agreement with our B.E. Koel et al.rJournal of Molecular Catalysis A: Chemical 131() 1998 39±53 43 previous estimate. Second, in agreement with face-adsorbate bond energies must both be our equilibrium thermodynamic predictions for known. In general, these adsorbates correspond the most stable C2 hydrocarbon surface species to gas phase molecular fragments that are rela- wx36 , ethylidene was identified using NMR over tively unstableŽ. i.e., radicals or excited states Pt catalysts at high pressureswx 41 . Also, work and hence their heats of formation are not di- from Somorjai's labwx 42 using SFG to study the rectly measurable. Furthermore, most surface- dehydrogenation of ethylene assigned the inter- adsorbate bond energies for such fragments are mediate detected to ethylidene, in agreement also unavailable. We have developed a general with our suggested decomposition mechanism. approach for estimating heats of formation for The success of our method for illuminating adsorbateswx 35 , involving a straightforward, important aspects of the mechanism of ethylene simple procedure:Ž. i determine how many decomposition prompted the present study. In chemical bonds are formed from the adsorbate the next section, we outline and illustrate our to the surface and characterize them according strategy for calculating equilibrium thermody- to either primarily covalent, ionic, or donor± namic energetics of the adsorbates and reactions acceptor character;Ž. ii obtain the experimental necessary to describe selective cyclohexane de- heat of formation for the gaseous species in its hydrogenation to benzene on Pt surfaces. The ground electronic stateŽ or estimate it via a mechanism of benzene decomposition is also Born±Haber cycle of known heats of formation interesting and very important to understand for and bond dissociation energies.Ž. ; iii use elec- eliminating undesirable coking products, how- tronic excitation energies, electron affinities, or ever we will not consider these issues in the ionization potentials to calculate a heat of for- present context except for the likely initial steps mation for a gas phase species that most resem- involving the formation of phenyl and benzyne. bles the local electronic state of the molecule when it is adsorbed on the surface; and finally Ž.iv use measured heats of adsorption Ž where 2. Results and discussion available. or use estimated or calculated bond energiesŽ from, for example, ab initio electronic 2.1. Scheme for calculating heats of formation structure calculations. for s and p covalent or of adsorbates on surfaces donor±acceptor bondswx 43,44 to predict the heat of formation of the adsorbed species by sub- In order to calculate heats of formation of tracting the adsorbate-surface bond energies c-C6 adsorbates of interest for cyclohexane de- from the gas phase heats of formation discussed hydrogenation, the heats of formation of the inŽ. iii . Table 1 provides schematic drawings corresponding gas phase species and the sur- and heats of formation for several c-C6 gas

Table 1 r Structures and heats of formation, D Hf,3008 Ž.kcal mol , for the electronic ground states of gas phase c-C6 species used to calculate the energies of c-C6 adsorbates 44 B.E. Koel et al.rJournal of Molecular Catalysis A: Chemical 131() 1998 39±53

Table 2 tion of the adsorbed species requires knowledge Ž.D 8 Estimated heats of formation Hf,300for C 6 adsorbates on of the Pt±C s-bond energy andŽ. possibly sin- PtŽ. 111 and related gas phase values p D 8 glet±triplet excitation energies to `prepare' - Species Hf,300 References s y wx bonds to form di- -bonds to the surface. In C612 HŽg. cyclohexane 29 45 wx y wx previous work 36 , we employed DŽ.Pt±C s53 CH612Ža. 43 8 wx r c-C611 HŽg. cyclohexyl radical 13 45 kcal mol, an ab initio value for Pt±CH3 com- y c-C611 H Ð Ža. 40 plexeswx 37,38 . We will also employ 53 C H cyclohexene y145wx 610Žg. kcalrmol throughout this paper as our Pt±C CH610Žg. Ž.) 88 y wx C610 HŽa. di-s-bonded 18 9 single bond energy. This value is supported by wx c-C69 HŽg. cycloallylic radical 30 45 measurements of the activation energy for con- c-C H Ž.) 119 69Žg. version of cyclohexyl to cyclohexene on c-C H tri-s-bonded y40 69Ža. r C H 1,3-cyclohexadiene 26wx 45 Sn PtŽ. 111 alloys that place this Ea at about 13 68Žg. wx CH68Žg. Ž.) 204 kcalrmol 9 . Given that PtŽ. 111 is a more C H tetra-s-bonded y8 68Ža. reactive surface than these alloys, we view this C H 1,4-cyclohexadiene 26wx 45 68Žg. value as an upper bound for PtŽ. 111 surfaces. CH68Žg. Ž.) 204 y C68 HŽa. tetra-s-bonded 8 This then provides an upper bound for the Pt±C wx c-C67 HŽg. cyclohexadienyl 57 45 single bond strength, which is obtained from the c-C67 H Žg. Ž.) 235 y inequality requiring the endothermicity to be c-C67 HŽa. penta-s-bonded 30 wx C66 HŽg. benzene 20 45 less than or equal to the activation barrier of 13 y wx CH66Ža. 10 8 kcalrmol: wx c-C65 HŽg. phenyl radical 81 46 c-C H s-bonded 28 65Ža. D 8 qD 8 q y ½5Hf,300Ž.Ž.c-C 6 H 10Ža. Hf,300 H Ža. c-C65 H Ža. s p-bonded 2 wx o,m, p-C64 HŽg. Ž. singlet benzyne 107,122,137 47 wx yD 8 F o,m, p-C64 HŽg. Ž. triplet 145,139,139 48,49 Hf,300Ž.c-C 6 H 11Ža. 13. o,m, p-C64 HŽa. di-s-bonded 39,33,33 o,m, p-C H di-s qp-bonded 9,3,3 s r 64Ža. Now, using D HadsŽ.c-C 6 H 10Žg. 17 kcal mol HP Ž.2 S5245wx Žg. wx9 and D H dissŽ.H s19 kcalrmolwx 50 , then HÐ y10wx 50 ads 2Žg. Ža. we have the following relations: s DHf,3008 Ž.Ž.c-CH 6 10Ža. DHf,3008 c-CH 6 10Žg. phase molecules in their electronic ground states, yD sy y as required in stepŽ. ii , and also gives Ž highly . HcadsŽ.-CH 6 10Žg. 1 17 schematic drawings of adsorbed species on sy18 kcalrmol PtŽ. 111 that show the correlation between the gaseous and adsorbed species and indicate the s y DHf,3008 Ž.HŽa. DHf,3008 Ž.HŽg. DHHads Ž.Žg. bonding to the surface, as required for stepsŽ. iii andŽ. iv . The heats of formation for both gas s52.1y62sy10 kcalrmol, phase and adsorbed species on PtŽ. 111 are sum- where marized in Table 2. s y The heats of formation of the gas phase DHHadsŽ.Žg.ŽDH ŽH g. . species were obtained as follows. Values of s qDHHdiss r2, D Hf,3008 for c-C 6 Hx , x 6±12, and H atom ads Ž.2Žg. were taken directly or derived from Benson's and compilationwx 45 . Values for phenyl radicalŽ c- . s C65 H , and ortho-, meta-, and para-benzyne D Hf,3008 Ž.Ž.c-C 6 H 11Ža. D Hf,3008 c-C 6 H 11Žg. Ž.c-C64 H were taken from more recent data wx46 . The method for obtaining heats of forma- yDŽ.Pt±C s13 kcalrmolyD Ž.Pt±C . B.E. Koel et al.rJournal of Molecular Catalysis A: Chemical 131() 1998 39±53 45

And so, using only experimentally measured ab initio calculationswx 51 place this planar quin- heats of adsorption and activation barriers, along tet state in butadiene at (185 kcalrmol, con- with well-known gaseous heats of formation in sistent with our estimate. . This is how D Hf,3008 the above equationsŽ. vide infra , we obtain for the high spin excited state radicals C6 H x ), xs7±10, shown in Table 2, were calculated. y18y10y Ä413yDŽ.Pt±C F13 kcalrmol Finally, values for D Hf,3008 of gaseous, triplet and therefore DŽ.Pt±C F54 kcalrmol. Thus, benzynes were calculated as follows. The exper- the experimental bound on the activation energy imental values of D Hf,3008 of the ground singlet for dehydrogenation of adsorbed cyclohexyl states were added to appropriate single±triplet r provides us with an upper bound of 54 kcal mol splittings. Only the D EST for ortho-benzyne has on the Pt±C single bond strength, in excellent been measuredwx 48 . However, correlation-con- agreement with our chosen ab initio value of 53 sistent multireference CI calculationswx 52 have kcalrmol. Furthermore, the microcalorimetry- predicted all three, and the error in the calcula- derived value of 57.6 kcalrmolwx 40 mentioned tion compared to the ortho-benzyne experimen- previously is within 5 kcalrmol of the value tal result is only 2 out of 38 kcalrmol. Thus, that we are usingŽ an increase of 4 kcalrmol for we used the other ab initio values of 17 and 2 r the adsorption energy of cyclohexene at zero kcal mol for D EST for meta- and para-ben- coverage is required for this value to be consis- zyne, respectively. tent with the above arguments as well. . The heats of formation of the adsorbates Once this value is fixed, we can now derive, were obtained from the procedure outlined above based on the heat of adsorption of cyclohexene, for estimating these values, except for adsorbed the appropriate singlet±triplet excitation energy cyclohexane, cyclohexene, and benzene, where

Ž.D EST to employ in order to prepare C±C the experimental heats of adsorption were used. p-bonds to form di-s-bonds to the metal. Again, We consider adsorbed species forming the max- the method prescribes starting with the gaseous imum number of s-covalent bonds to the sur- species in its ground stateŽ. here cyclohexene , face, with no p-bonding interactions. Since cy- exciting up to a `bond-prepared' stateŽ via, e.g., clohexene and cyclohexadiene are known exper- a singlet-to-triplet excitation. , and then forming imentally to adopt h 2- and h 4-configurations the appropriate number of metal±carbon bonds upon adsorption, we believe this is the preferred to obtain heats of adsorptionrformation. Since geometry for these other species as well. How- sy we know D Hf,3008 Ž.cyclohexeneŽg. 1 ever, reaction energetics involving p-bonding r sy kcal mol and we know its D Hads 17 configurations could be easily calculatedŽ if de- kcalrmol, using DŽ.Pt±C s53 kcalrmol one sired. simply by using the corresponding appro- s r can derive D EST 89 kcal mol for the appro- priate gas phase heats of formation. We did priate singlet±triplet splitting to bond-prepare a include p bonding in addition to s-bonding for cyclic p bond. This is exactly the same value phenyl and the benzyne isomers, because of the as calculated previously for ethyleneŽ Ref.wx 36 uncertainty about the nature of the surface bond- and E.A. Carter, unpublished. . We will use ing and the experimental evidence that phenyl s r D EST 89 kcal mol per p bond to estimate on CuŽ. 111 is inclined only about 308 above the the energies of all high spin radicals that are surface planewx 53 . We used the p-bond strength required. For example, in the case of cyclohexa- of benzeneŽ. 30 kcalrmol to the surface as our diene, we will estimate that the high spin ex- rough estimate ofŽ and presumably an upper cited state is 2=89s178 kcalrmol above the limit to. this stabilization. We include, for com- ground state, since two p bonds are being pleteness, values for phenyl and the benzynes excitedŽ it is bound to be at least this high, without the p-bonding stabilization, and one based on Pauli repulsion arguments, and recent sees that this leads to quite a large valueŽ 28 46 B.E. Koel et al.rJournal of Molecular Catalysis A: Chemical 131() 1998 39±53 kcalrmol. for the endothermicity of the conver- heat of adsorption on PtŽ.Ž 111 30 kcalrmol,wx 8 . sion of benzene to phenyl. However, this value from D Hf,3008 Ž.Ž.c-C 6 H 6Žg. ; viii adsorbed hydro- is close to a measured barrier of 24.7 kcalrmol gen atom Ž.y10 kcalrmol by subtracting the for the initial dehydrogenation of benzenewx 8 , experimental Pt±H bond energyŽ 62 kcalrmol, and so either this species describes well the derived from the experimental heat of adsorp- chemical nature of phenyl on the surface, with tionŽ. 19 kcalrmol of hydrogen of Pt Ž. 111wx 50 . the endothermicity of reaction accounting for from D Hf,3008 Ž.Ž.HŽg. ; ix adsorbed phenyl radical the barrier, or it represents an intermediate Ž28 kcalrmol for s-bonded, y2 kcalrmol for formed in the conversion of benzene to a more sqp bonded. by subtracting the chemisorp- stable adsorbed sqp-bonded phenyl. Finally, tion s bond energy of 53 kcalrmol and the we did not include any estimate of H bonding in chemisorption p bond of 30 kcalrmol from species other than cyclohexane, since in general D Hf,3008 Ž.Ž.c-C 6 H 5Žg. ; and x adsorbed benzynes no vibrational signatures of such bondingŽ i.e., Ž39 and 33 kcalrmol for di-s only bonding of softening of the C±H vibrations. have been ortho and metarpara; 9 and 3 kcalrmol for observed spectroscopicallyŽ there is one uncon- di-sqp bonding of ortho and metarpara. by firmed report of such interactions for cyclohexa- subtracting the s chemisorption bond energy of dienewx 24. . 106 kcalrmol and the p chemisorption bond D 8 r We derived Hf,300 Žvalues again shown in energy of 30 kcal mol from D Hf,3008 Žtriplet- y parentheses.Ž. for: i adsorbed cyclohexane Ž43 benzynesŽg.. . kcalrmol. by subtracting the measured heat of It is critical to re-emphasize the main as- adsorption on PtŽ.Ž 111 13.9 kcalrmol,wx 8. from sumptions made in calculating these heats of

D Hf,3008 Ž.Ž.c-C 6 H 12Žg. ; ii adsorbed cyclohexyl, adsorption. First we point out that one must y r c-C611 H Ž.40 kcal mol by subtracting the have an understanding of the type and number Pt±C single s bond energyŽ 53 kcalrmol of adsorbate-surface bonds involved. If one is wx D 8 37,38.Ž.Ž. from Hf,300c-C 6 H 11Žg. ; iii ad- unsure, the error bars associated with D Hads sorbed cyclohexene Ž.y18 kcalrmol by sub- could be large. In the cases presented here, tracting the measured heat of adsorption on spectroscopic evidence from HREELS estab- r wx PtŽ.Ž 111 17.1 kcal mol, 9.Ž from D Hf,3008 c- lished di-s and quadra-s bonding for cyclo- y C610 HŽg..Ž. ; iv adsorbed c-C69 H Ž40 hexene and cyclohexadiene. This suggests we kcalrmol. by subtracting the chemisorption are on firm ground assuming that all unsaturated r D 8 bond energyŽ.ŽŽ 159 kcal mol from Hf,300 c- C atoms in these c-C6 compounds will form a q r CH69Žg..ŽD EST 89 kcal mol .. , assuming that number of s-bonds equal to the number of H c-C69 H forms three covalent s bonds to the atoms lost relative to cyclohexane. Benzene is surface, each worth 53 kcalrmol;Ž. v adsorbed clearly an exception, since the aromaticity of y r cyclohexadiene, c-C68 H Ža. Ž.8 kcal mol by benzene favors p-bonding to the metal. Again, subtracting the chemisorption bond energyŽ 212 spectroscopic evidence supports p-bonding for r q kcal mol.ŽŽ from D Hf,3008 c-C 6 H 8Žg. .2D EST benzene and phenyl, while phenylÐhaving lost r Ž..178 kcal mol assuming that c-C68 H forms an H atom relative to benzeneÐcan use its C four covalent s bonds to the surface, each radical electron to form an additional s-bond to r y worth 53 kcal mol;Ž. vi adsorbed c-C67 H Ž 30 the metal. kcalrmol. by subtracting the chemisorption Secondly, we assume a high spin excited r bond energyŽ.ŽŽ 265 kcal mol from D Hf,3008 c- state is required for bonding. Note this is the q wx CH67Žg...2D EST, assuming that c-C 6 H 7 typical valence bond model of bonding 49 that forms five covalent s bonds to the surface, has been highly successful for gas phase each worth 53 kcalrmol;Ž. vii adsorbed benzene molecules. In addition, confirmation of the qual- Ž.y10 kcalrmol by subtracting the measured ity of our estimate of the energy for such high B.E. Koel et al.rJournal of Molecular Catalysis A: Chemical 131() 1998 39±53 47

Table 3 r Estimated heats of elementary reactions of C6 hydrocarbons on PtŽ. kcal mol Reaction Type D H rxn ™ 1 q y CH612Ža. h -c-C611 H Ža.ŽHa. C±H cleavage 7 ™ CH612Ža. C612 HŽg. desorption 14 1 q ™ h -c-C611 H Ža.ŽH a. C612 HŽa. hydrogenation 7 12™ q h -c-C611 H Ža. h -C610 H Ža.ŽHa. C±H cleavage 12 21q ™ y h -c-C610 H Ža.ŽH a. h -c-C611 HŽa. hydrogenation 12 23™ q y h -c-C610 H Ža. h -C69 H Ža.ŽHa. C±H cleavage 32 ™ CH610Ža. C610 HŽg. desorption 17 32q ™ h -C69 H Ža.ŽH a. h -c-C610 HŽa. hydrogenation 32 34™ q h -C69 H Ža. h -C68 H Ža.ŽHa. C±H cleavage 22 43q ™ y h -C68 H Ža.ŽH a. h -c-C69 HŽa. hydrogenation 22 45™ q y h -C68 H Ža. h -C67 H Ža.ŽHa. C±H cleavage 32 ™ CH68Ža. C68 HŽg. desorption 34 54q ™ h -C67 H Ža.ŽH a. h -C68 HŽa. hydrogenation 32 5 ™ q h -C67 H Ža. CH66Ža.ŽHa. C±H cleavage 10 q ™ 5 y CH66Ža.ŽH a. h -C67 HŽa. hydrogenation 10 ™ 6 q y CH66Ža. h -C65 H Ža.ŽHa. C±H cleavage 2 ™ CH66Ža. C66 HŽg. desorption 30 6 q ™ h -C65 H Ža.ŽH a. C66 HŽa. hydrogenation 2 66™ q y h -C65 H Ža. h -C64 H Ža.ŽHa. C±H cleavage 1Ž.Ž for o-; 5 for m or p- . spin states comes from very recent first princi- activation barriers provide upper bounds to the ples density functional theory calculations on endothermicities and thus allow us to check the butadiene by Petterson et al.wx 51 whose predic- consistency of our model. In all cases consid- tions for the singlet±quintet splitting in planar ered herein and in our past work, this has butadieneŽ. 185 kcalrmol is roughly twice the proven to be trueŽ namely, our endothermicities vertical singlet±triplet splitting in ethyleneŽ 89 have been nearly met or exceeded by the corre- kcalrmol. , which is the value we proposed to sponding, measured activation barriers. . use. We therefore believe that such an estimate With our assumptions now clearly defined, of the excited states' energy is valid. One could we then use these calculated heats of formation imagine refining the model by noting that the for adsorbed species to calculate several heats difference in energy between the first principles of reactions of adsorbed hydrocarbons on excitation energy and our `additive' excitation PtŽ.Ž 111 Table 3 . . In Section 2.2, we employ energy is `excess' Pauli repulsion energy due to these heats of formation and heats of reaction, four spins being in contact rather than two along with experimental activation barriers, to separated groups of two spins. Thus 185-2Ž. 89 obtain a rough potential energy surface for the s7 kcalrmol might be considered an estimate selective dehydrogenation of cyclohexane to of this additional repulsion not accounted for in benzeneŽ. and beyond on clean Pt Ž. 111 . our purely additive model. We retained the purely additive model here for simplicity. 2.2. Calculated energetics for the dehydrogena- Lastly, the model does not account for lateral tion of cyclohexane to benzene on Pt() 111 interactions between adsorbates and therefore is only applicable to low coverage regimes. This Fig. 1 displays a mechanism for the reaction means that when we look at coverage-dependent of cyclohexane on PtŽ. 111 using the energetics binding energies and activation barriers, we must derived above for the step-wise, sequential re- be careful to compare only to low coverage moval of H atoms from cyclohexane. The en- values. For endothermic surface reactions, their ergy zero is the heat of formation of the ele- 48 B.E. Koel et al.rJournal of Molecular Catalysis A: Chemical 131() 1998 39±53

Fig. 1. Thermochemistry for a step-wise, sequential mechanism for dehydrogenation of cyclohexane to benzene on Pt.

ments in their standard states, i.e., D Hf,3008 course, the limitations of calculating coadsorp- s s s Ž.H 2Žg. D Hf,3008 Ž.PtŽs. D Hf,3008 ŽCŽgraphite. .0. tion effects on adsorbate bonding due to high Mass balance requires that we consider the heat coverages of reactants, products, or intermedi- of formation of the coadsorbed H atoms re- ates. quired by stoichiometry. This mass balance is The appreciable structure-dependent differ- required even if the reaction is carried out in ences in hydrocarbon chemistry on Pt comes vacuum under conditions where the hydrogen principally from structure-dependent changes in recombines and desorbs at equilibrium, since activation barriers for reactions, and rather small we assume that the nascent products of the barrier changes can cause large alterations in the reaction include coadsorbed hydrogen, rather branching ratios among the many reaction paths than H2Žg. . available for hydrocarbon conversion reactions. We believe that this diagram does not dra- Thus, while the thermochemistry contained in matically change upon changing the Pt crystal Table 3 determines the reaction energetics and face, and so these energetics have some general- the concentrations of these species at equilib- ity for all reactions on Pt, i.e., the Pt±C s bond rium, the concentrations of these species mea- strengths are not a strong function of the surface sured in experiments are often determined by structure. For example, it is well known that the kinetics of the reactions rather than the DŽPt±H . is a constant for Pt Ž.Ž. 111 , 110 , and thermodynamics. Thus, activation energy barri- Ž.100 . The shortcoming of such a diagram is, of ers for these reactions need to be considered to B.E. Koel et al.rJournal of Molecular Catalysis A: Chemical 131() 1998 39±53 49

Fig. 2. Reaction energetics for a step-wise, sequential mechanism for dehydrogenation of cyclohexane to benzene on PtŽ. 111 . understand the species formed, and an overall Consider first the two possible reaction path- understanding of the reaction energetics requires ways for cyclohexane adsorbed on PtŽ. 111 : de- inclusion of the various activation barriers in the hydrogenation and desorption. We find that the reaction mechanism. The activation energy bar- most energetically favorable reaction stepŽ see riers shown in Fig. 2 are specific to reactions on Table 3. is C±H bond cleavage to form an PtŽ. 111 , and were either taken directly from adsorbed cyclohexyl speciesŽ 7 kcalrmol experimental observations on PtŽ. 111 or were exothermic. while desorption of molecular cy- deduced from a combination of the experimen- clohexane is an unfavorable reaction stepŽ endo- tal activation energies and our calculated ther- thermic by 14 kcalrmol.Ž . Experimentally on mochemistry. In extracting predictions of un- PtŽ. 111 at saturation coverage . , (32% of the known barriers, we used these two kinetic prin- adsorbed cyclohexane is observed to desorb s r wx ciples:Ž. i if the first elementary step is rate- molecularly Ž Ed 13.9 kcal mol 8. , and 68% limiting, then every subsequent barrier en route is observed to decompose via some pathway to s r to a stable intermediate must be lower than the form cyclohexene Ž Ea 9.5±13.5 kcal mol for first barrier andŽ. ii if a later elementary step is 5±30% MLwx 5. . Thus, the barrier for dissocia- rate-limiting, then the barriers for all previous tion from the molecularly adsorbed state is 9.5± steps en route to a stable intermediateŽ or final 13.5 kcalrmol, while the barrier for desorption product. must be lower than the observed acti- is 13.9 kcalrmol, and hence adsorbed cyclohex- obs vation barrier, Ea . Below we discuss the heats ane is more likely to dissociate than to desorb, of and barriers to reaction for various steps in especially at low coverages. the dehydrogenation of cyclohexane, comparing Large barriers may exist for C±C bond cleav- our predictions to experiment and to previously age compared with those for C±H bond cleav- - postulated mechanisms where possible. age, since Cx , where x 6, adsorbates are not 50 B.E. Koel et al.rJournal of Molecular Catalysis A: Chemical 131() 1998 39±53 observed until much higher temperatures, if at Ž.endothermic by 17 kcalrmol . Experimentally all. Indeed, one might expect a higher barrier ŽŽ.on Pt 111 at saturation coverage . , -25% of for C±C bond cleavage compared to C±H bond the adsorbed cyclohexene is observed to desorb s r wx ) cleavage because of the less favorable geometry molecularly Ž Ed 17.1 kcal mol 9. and s that the carbon atoms must adopt in transition 75% is observed to decompose Ž Ea 9±14.4 r wx 3 states involving C±C bond cleavage compared kcal mol 4,22. through a proposed h -C69 H Ža. to those for C±H bond breaking. However, we intermediate. The endothermicity we calculate should note that the identification of various to decompose this intermediate is 22 kcalrmol,

Cx- 6 adsorbates by unambiguously distinguish- and this is consistent with an estimated reaction r ing these from possible C6 hydrocarbon species barrier of 18±20.8 kcal mol from experimental is difficult at best using most surface analytical measurementswx 4,22 . This reaction forms ben- spectroscopies and perhaps impossible if mix- zenewx 4,9,22 , presumably through a cyclohexa- tures of hydrocarbon adsorbates are formed; diene intermediate. It is important to note that thus, the validity of this lack of observation is the predicted barrier to hydrogenation of cyclo- still an open question. However, these pathways hexene to cyclohexyl is nearly zero, and this are not considered in this diagram since there is process is clearly dictated by the concentration no evidence at this time that such pathways of hydrogen on the surface. It might even play a involving C±C bond cleavage lead to apprecia- role in the interesting `two-peak' desorption ble benzene production on PtŽ. 111 . spectra of cyclohexenewx 9 . Also, the huge bar- 3 Cyclohexyl dehydrogenates at relatively low rier to hydrogenating the h -C69 HŽa. intermedi- F r wx r temperatures Ž Ea 12.6 kcal mol 8. on ateŽ endothermic by 32 kcal mol with a barrier PtŽ. 111 . Using the experimental activation ener- predicted to be 41 kcalrmol. eliminates hydro- gies given above and our calculated heats of genation pathways to produce cyclohexane and reaction, we predictŽ. Fig. 2 that the barrier for cyclohexene from any intermediate past c-C69 H , formation of adsorbed cyclohexane from coad- including cyclohexadiene and benzene, in sorbed cyclohexyl and hydrogen is 17 kcalrmol agreement with experimental findings. and that the endothermicity of reaction to form Consider the three possible reaction pathways adsorbed cyclohexene is small, only (12 for cyclohexadiene on PtŽ. 111 : hydrogenation, kcalrmol. Thus, the barrier to dehydrogenation dehydrogenation, and desorption. We find that of cyclohexyl is essentially the endothermicity the most energetically favorable reaction step of reaction. Confirmation of our prediction for Ž.see Table 2 is C±H bond cleavage to form a 5 the barrier for formation of adsorbed cyclohex- h -C67 HŽa. cyclohexadienyl speciesŽ 32 ane from coadsorbed cyclohexyl and hydrogen kcalrmol exothermic. , followed by hydrogena- 3 comes from TPD experiments showing tion to adsorbed h -c-C69 HŽa. Ž downhill by 22 reaction-rate limited cyclohexane production at kcalrmol. , while desorption of molecular cy- ( r 272 K Ž.Ea 16.6 kcal mol from hydrogena- clohexadiene is the least favorable reaction step tion of cyclohexene. Ž.endothermic by 34 kcalrmol . Experimentally, Consider the three possible reaction pathways on PtŽ. 111 no adsorbed cyclohexadiene is ob- for cyclohexene on PtŽ. 111 : hydrogenation, de- served to desorb molecularly, no cyclohexadi- hydrogenation, and desorption. We find that the ene is hydrogenated to cyclohexene, and pre- most energetically favorable reaction stepŽ see sumably complete conversion to chemisorbed 3 Table 2. is dehydrogenation to h -C69 HŽa. Ž ex- benzene occurs. This is consistent with Fig. 2. r 4 othermic by 32 kcal mol. , followed by hydro- The activation barrier to react h -C68 HŽa. , pre- genation to adsorbed cyclohexylŽ downhill by sumably to form benzene via a c-C67 HŽa. inter- 12 kcalrmol. , while desorption of molecular mediate, has been measured to be 7±13.5wx 4 cyclohexene is the least favorable reaction step and 14"2 kcalrmolwx 26 on PtŽ. 111 . The en- B.E. Koel et al.rJournal of Molecular Catalysis A: Chemical 131() 1998 39±53 51 dothermicity we calculate for dehydrogenation chemisorbed with the ring plane nearly parallel r of c-C67 HŽa. to benzene is 10 kcal mol, while to the surface, with p bonding from the ring that for hydrogenation is 32 kcalrmol. The 10 contributing strongly to the chemisorption bond, kcalrmol dehydrogenation endothermicity rep- and endothermic by 28±29 kcalrmol for the resents most of the barrier to forming benzene limiting case of s-bonded and di-s-bonded from c-C67 HŽa. since benzene is formed from species, respectively, with no p bonding stabi- cyclohexane and cyclohexene at about 300 K lization at all. Importantly, we don't predict wx F r 3±5,23 , i.e., Ž.Ea 18 kcal mol , and mea- huge endothermicities that disagree with the surements on PtŽ. 111 show benzene desorption known barriers. Experimentally, on PtŽ. 111 F r wx at 310 K Ž Ea 18.9 kcal mol 10. from cyclo- about 55% of the benzene monolayer dehydro- hexadiene dehydrogenation. A betterŽ. lower genates upon heatingwx 32 with barriers of 24.7 value of the upper limit for this barrier comes and 31.1 kcalrmol for losing one H and three H from studies on Pt±Sn alloys, where benzene atoms, respectivelywx 8 . Our predicted endother- was desorbed in TPD in a peak at 235 K micities for forming s-bonded phenyl and di- s r wx Ž Ea 13.6 kcal mol 10. following cyclohexa- s-bonded benzyne without any p bonding sta- diene adsorption on the 63-SnrPtŽ. 111 alloy. bilization are remarkably close to the activation This represents an upper limit on the reaction barriers cited above. Obviously, p bonding con- barrier on PtŽ. 111 , because this alloy is less tributions to the chemisorption bond will lower reactive than PtŽ. 111 , and it is within a few the energies of these species. We conclude that kcalrmol of the value for the calculated en- either this s-bonded species describes well the dothermicity. Again, we predict little or no bar- chemical nature of phenyl on the PtŽ. 111 sur- rier for hydrogenation of cyclohexadiene to c- face, with the endothermicity of this reaction

C69 HŽa. , and this species should be formed as accounting for nearly all of the benzene soon as hydrogen becomes available from dehy- dehydrogenation barrier, or it represents an in- drogenation of cyclohexadiene. However, fur- termediate formed in the conversion of benzene q ther hydrogenation past c-C69 HŽa. will not occur to a more stable adsorbed s p-bonded phenyl. due to the large barrierŽ. 41 kcalrmol for hy- An obvious feature of the potential energy drogenation to cyclohexene. diagram of Fig. 1 is that at equilibrium in the 5 The final step in the reaction involves the presence of hydrogen, adsorbed h -C67 H is desorption of benzene. It is likely that the more predicted to be the most stable c-C6 surface 3 exothermicŽ. and nearly barrierless hydrogena- species. An adsorbed h -C69 H species is also tion reaction does not compete in UHV because predicted to be a stable species present in appre- of the lack of adsorbed hydrogen under reaction ciable concentrations due to the relatively large conditions. At temperatures G300 K, all ad- barriers for forward and reverse reactions. This sorbed hydrogen immediately desorbs as H2 so prediction has important implications to hydro- that negligible quantities of HŽa. exists at carbon conversion and reforming catalysis where steady-state under ultrahigh vacuum reaction these species should play important roles in the conditions. However, even if hydrogen is avail- catalytic reaction mechanisms. Figs. 1 and 2 are 5 able, the conversion of benzene to h -C67 H Ža. also consistent with important kinetic and mech- on clean PtŽ. 111 is a dead end path due to the anistic aspects of cyclohexane to benzene con- large barriers encountered for the backŽ further version catalysis at high pressureswx 1 :Ž. i kinet- hydrogenation. reactions. ics consistent with a simple sequence of irre- We also show a likely path for an undesirable versible steps; andŽ. ii the rate determining step reaction, that of benzene decomposition. Our being either benzene desorption or the dehydro- predictions are that the first steps are nearly genation of some intermediate depending on the thermoneutral if phenyl and benzyne are reaction conditions. We predict that dehydro- 52 B.E. Koel et al.rJournal of Molecular Catalysis A: Chemical 131() 1998 39±53 genation of the cycloallylic intermediate is en- a role in sequential dehydrogenation process. dothermic by 22 kcalrmol and that this is the These heats of formation were then used to rate determining step when the kinetics are not predict the thermochemistry along the pathway rate-limited by benzene desorption. to benzene. Barriers to each step were taken Two final comments on this energetic and from or deduced from experimental data. mechanistic analysis can be made. First, new Throughout our analysis we have made no at- direct measurements of hydrocarbon heats of tempt to include coverage effects on adsorption adsorption or improvements from chemical energies and reaction barriers, in order to sim- physics experiments or theoretical calculations plify the approach. Our calculations indicate of excitation energies for reactive species and of that in the sequential dehydrogenation mecha- 3 metal±carbon bond strengths will greatly im- nism, a metastable h -C69 H allylic intermediate prove our ability to set thermodynamic limits on is formed and that the rate limiting step in- the stable species formed on surfaces. Secondly, volved in benzene production is either the dehy- the heats of formation and reaction shown in drogenation of this intermediate to chemisorbed Tables 2 and 3 and the type of scheme shown in cyclohexadiene or desorption of the benzene Fig. 1 should be useful in general for under- product. Finally, we calculate phenyl and ben- standing the reactions of cyclohexane, cyclo- zyne energetics for benzene decomposition reac- hexene, and cyclohexadiene on other metal sur- tions that lead to carbon deposition and we find faces, simply by incorporating changes in the the energetics of those reactions to be consistent calculated heats of adsorption for the hydrocar- with the experimentally measured barriers. bon species, as well as any experimental infor- mation about the adsorption energies and barrier heights on other surfaces. In particular, the esti- Acknowledgements mates of heats of formation and reaction ener- getics for the adsorbed hydrocarbon fragments BEK acknowledges the U.S. Department of can be obtained by using the appropriate Energy, Office of Basic Energy Sciences, metal±carbon and metal±hydrogen bond ener- Chemical Sciences Division for support of this gies, dependent on the specific transition metal, research. EAC is grateful for support for this since we have already tabulated the heats of work from a Union Carbide Innovation Recog- formation of the gas phase species in the appro- nition Award, a Camille and Henry Dreyfus priate electronic state. Thus, this approach and Foundation Teacher Scholar Award, and a Al- the resulting values for adsorbate heats of for- fred P. Sloan Foundation Research Award. It is mation are expected to lend new insight into the a pleasure to acknowledge valuable discussions mechanisms of hydrocarbon reactivity on a vari- with Prof. G.B. Ellison. ety of metal surfaces. References

3. Summary wx1 J.H. Sinfelt, in: J.R. Anderson, M. BoudartŽ. Eds. , Catalysis, Science and Technology, Vol. 1, Chap. 5, Springer-Verlag, We have applied a quasiempirical valence Berlin, 1981. wx bond approach for calculating the heats of for- 2 S.M. Davis, G.A. Somorjai, in: D.A. King, D.P. Woodruff Ž.Eds. , The Chemical Physics of Solid Surfaces and Hetero- mation of adsorbed species to understand mech- geneous Catalysis, Vol. 1, Chap. 7, Elsevier, Amsterdam, anistic aspects of the selective dehydrogenation 1984. wx of cyclohexane to benzene over PtŽ. 111 sur- 3 D.P. Land, C.L. Pettiette-Hall, R.T. McIver Jr., J.C. Hem- minger, J. Am. Chem. Soc. 111Ž. 1989 5970. faces. We have calculated the heats of forma- wx4 C.L. Pettiette-Hall, D.P. Land, R.T. McIver Jr., J.C. Hem- tion of all the relevant surface species that play minger, J. Am. Chem. Soc. 113Ž. 1991 2755. B.E. Koel et al.rJournal of Molecular Catalysis A: Chemical 131() 1998 39±53 53

wx5 D.H. Parker, C.L. Pettiette-Hall, Y.Z. Li, R.T. McIver Jr., wx30 M. Abon, J.C. Bertolini, J. Billy, J. Massardier, B. Tardy, J.C. Hemminger, J. Phys. Chem. 96Ž. 1992 1888. Surf. Sci. 162Ž. 1985 395. wx6 J.A. Rodriguez, C.T. Campbell, J. Phys. Chem. 93Ž. 1989 wx31 E.L. Garfunkel, M.H. Farias, G.A. Somorjai, J. Am. Chem. 826. Soc. 107Ž. 1985 349. wx7 J.M. Davidsen, F.C. Henn, G.K. Rowe, C.T. Campbell, J. wx32 J.M. Campbell, S. Seimanides, C.T. Campbell, J. Phys. Phys. Chem. 95Ž. 1991 6632. Chem. 93Ž. 1989 815. wx8 C. Xu, Y.-L. Tsai, B.E. Koel, J. Phys. Chem. 98Ž. 1994 585. wx33 D.B. Kang, A.B. Anderson, J. Am. Chem. Soc. 107Ž. 1985 wx9 C. Xu, B.E. Koel, Surf. Sci. 304Ž. 1994 249. 7858. wx10 J. Peck, B.E. Koel, J. Am. Chem. Soc. 118Ž. 1996 2708. wx34 E.A. Carter, W.A. Goddard III, J. Phys. Chem. 88Ž. 1984 wx11 J.L. Gland, K. Baron, G.A. Somorjai, J. Catal. 36Ž. 1975 1485. 305. wx35 E.A. Carter, Chem. Phys. Lett. 169Ž. 1990 218. wx12 L.E. Firment, G.A. Somorjai, J. Chem. Phys. 66Ž. 1977 wx36 E.A. Carter, B.E. Koel, Surf. Sci. 226Ž. 1990 339. 2901. wx37 J.J. Low, W.A. Goddard III, J. Am. Chem. Soc. 108Ž. 1986 wx13 J.E. Demuth, H. Ibach, S. Lehwald, Phys. Rev. Lett. 40 6115. Ž.1978 1044. wx38 J.J. Low, W.A. Goddard III, Organometallics 5Ž. 1986 609. wx14 S. Lehwald, H. Ibach, Surf. Sci. 89Ž. 1979 425. wx39 Y. Yeo, A. Stuck, C.E. Wartnaby, D.A. King, Chem. Phys. wx15 M.C. Tsai, C.M. Friend, E.L. Mutterties, J. Am. Chem. Soc. Lett. 259Ž. 1996 28. 104Ž. 1982 2539. wx40 H. Gross, B.E. Koel, C.T. Campbell, D.A. King, to be wx16 J. Stohr,È F. Sette, A.L. Johnson, Phys. Rev. Lett. 53Ž. 1984 submitted. 1684. wx41 J.M. Griffiths, A.T. Bell, J.A. Reimer, J. Phys. Chem. 97 wx17 A.P. Hitchcock, D.C. Newbury, I. Ishii, J. Stohr,È J.A. Hors- Ž.1993 9161. ley, R.D. Redwing, A.L. Johnson, F.J. Sette, Chem. Phys. 85 wx42 P. Cremer, C. Stanners, J.W. Niemantsverdriet, Y.R. Shen, Ž.1986 4849. G.A. Somorjai, Surf. Sci. 328Ž. 1995 111. wx18 C.T. Campbell, J.A. Rodriguez, F.C. Henn, J.M. Campbell, wx43 E.A. Carter, W.A. Goddard III, J. Am. Chem. Soc., 108 P.J. Dalton, S.G. Seimanides, J. Chem. Phys. 88Ž. 1988 Ž.1986 2180, 4746. 6585. wx44 E.A. Carter, W.A. Goddard III, Organometallics 7Ž. 1988 wx19 C.T. Campbell, J.M. Campbell, P.J. Dalton, F.C. Henn, J.A. 675. Rodriguez, S.G. Seimanides, J. Phys. Chem. 93Ž. 1989 806. wx45 S.W. Benson, Thermochemical Kinetics, Wiley, New York, wx20 M.A. Chesters, P. Gardner, Spectrochim. Acta A 46Ž. 1990 1976. 1011. wx46 G.E. Davico, V.M. Bierbaum, C.H. DePuy, G.B. Ellison, wx21 M.E. Bussell, F.C. Henn, C.T. Campbell, J. Phys. Chem. 96 R.R. Squires, J. Am. Chem. Soc. 117Ž. 1995 2590. Ž.1992 5978. wx47 P.G. Wenthold, R.R. Squires, J. Am. Chem. Soc. 116Ž. 1994 wx22 F.C. Henn, A.L. Diaz, M.E. Bussell, M.B. Hugenschmidt, 6401. M.E. Domagala, C.T. Campbell, J. Phys. Chem. 96Ž. 1992 wx48 D.G. Leopold, A.E.S. Miller, W.C. Lineberger, J. Am. Chem. 5965. Soc. 108Ž. 1986 1379. wx23 D.P. Land, W. Erley, H. Ibach, Surf. Sci. 289Ž. 1993 237. wx49 L.B. Harding, W.A. Goddard III, Annu. Rev. Phys. Chem. wx24 X. Su, R.Y. Shen, G.A. Somorjai, Chem. Phys. Lett. 280 29Ž. 1978 363. Ž.1997 302. wx50 B. Poelsma, G. Mechtersheimer, G. Comsa, Surf. Sci. 111 wx25 C. Xu, B.E. Koel, Surf. Sci. 292Ž. 1993 L803. Ž.1981 519. wx26 M.B. Hugenschmidt, A.L. Diaz, C.T. Campbell, J. Phys. wx51 L. Triguero, L.G.M. Pettersson, B. Minaev, H. Agren, J. Chem. 96Ž. 1992 5974. Chem. Phys., submitted. wx27 S. Lehwald, H. Ibach, J.E. Demuth, Surf. Sci. 78Ž. 1978 577. wx52 S.G. Wierschke, J.J. Nash, R.R. Squires, J. Am. Chem. Soc. wx28 M.C. Tsai, E.L. Mutterties, J. Am. Chem. Soc. 104Ž. 1982 115Ž. 1993 11958. 2534. wx53 M.X. Yang, M. Xi, H. Yuan, B.E. Bent, P. Stevens, J.M. wx29 J. Somers, M.E. Bridge, D.R. Lloyd, T. McCabe, Surf. Sci. White, Surf. Sci. 341Ž. 1995 9. 181Ž. 1984 L167.