high-pressure catalytic reactions over single-crystal ... files/148-ss_rpt14_1991.pdf ·...

High-pressure catalytic reactions over single-crystal metal surfaces

Jose A. Rodriguez and D. Wayne Goodman Department of Chemistry, Texas A&M University,

College Station, TX 77843-33255, USA

NORTH-HOLLAND

2 J.A. Rodriguez and D.W. Goodman

Contents

1. Introduction 2. Ammonia synthesis

2.1. Kinetics over Fe single-crystal catalysts 2.2. Kinetics over Re single-crystal catalysts

3. CO methanation 3.1. CO methanation on clean metal single crystals

3.1.1. Monometallic surfaces 3.1.1.1. Ni(100) and Ni( l l l ) 3.1.1.2. Ru(ll0) and Ru(001) 3.1.1.3. W(110) 3.1.1.4. Rh(ll 1) 3.1.1.5. Fe(ll 1) 3.1.1.6. Mo(100) 3.1.1.7. Co(161), Co(1120) and Co(1012) 3.1.1.8. Summary

3.1.2. Bimetallic surfaces 3.1.2.1. Ni/W(l l0) and Ni/W(100) 3.1.2.2. Cu/Ru(001) and A g / R h ( l l l ) 3.1.2.3. Co/W(l l0) and Co/W(100)

3.2. CO methanation on chemically modified surfaces 3.2.1. Electronegative impurities

3.2.1.1. Atomic chlorine, sulfur and phosphorus on Ni(100) 3.2.1.2. Atomic sulfur on Ru(001), Rh( l l l ) and W(ll0)

3.2.2. Electropositive impurities 3.2.2.1. Potassium on Ni(100)

3.2.3. Related theory 3.3. Metal-support interactions and CO methanation

4. CO oxidation 4.1. CO oxidation by 0 2

4.1.1. Rh(100) and Rh( l l l ) surfaces 4.1.2. Ru(001) surface 4.1.3. Pt(100) surface 4.1.4. Pd(ll0) surface 4.1.5. Ir(ll0) and Ir( l l l ) surfaces 4.1.6. Summary

4.2. CO oxidation by NO 4.2.1. Rh(100) and Rh( l l l ) surfaces

5. Ethylene epoxidation 5.1. Kinetics over Ag(ll0) and Ag(l l l ) catalysts 5.2. Chlorine promotion of Ag(110) and Ag(l l l ) surfaces 5.3. Cesium promotion of Ag(l l l )

6. Thiophene hydrodesulfurization 6.1. Kinetics on Mo surfaces 6.2. Kinetics on Re surfaces

7. Alkane hydrogenolysis 7.1. Ethane hydrogenolysis

6 6 6 9

10 10 10 10 12 13 14 14 14 16 16 17 17 18 18 19 19 19 22 23 23 25 26 27 27 27 29 29 32 32 33 35 35 36 36 39 4O 41 42 44 45 46

High-pressure catalytic reactions over single-crystal metal surfaces 3

7.2.

7.3.

7.1.1. Monometallic surfaces 7.1.1.1. Ni(100) and Ni(111) 7.1.1.2. Ru(001) and Ru(1,1,10) 7.1.1.3. I r ( l l l ) and Ir(ll0)-(1 x 2) 7.1.1.4. Pt( l l l ) 7.1.1.5. Re(001) 7.1.1.6. W(100) 7.1.1.7. Summary

7.1.2. Bimetallic surfaces 7.1.2.1. Ni /W(l l0) and Ni/W(100) 7.1.2.2. Ni /P t ( l l l ) 7.1.2.3. R e / P t ( l l l ) and Pt/Re(001) 7.1.2.4. Cu/Ru(001)

Propane hydrogenolysis 7.2.1. I r ( l l l ) and Ir(ll0)-(1 x 2) surfaces Butane hydrogenolysis 7.3.1. I r ( l l l ) and Ir(ll0)-(1 x 2) catalysts 7.3.2. Rh( l l l ) and Rh(100) catalyst~

46 46 47 48 49 5O 5O 5O 51 51 51 53 54 54 54 54 54 56 57 57 57 58 58

7.3.3. Pt(100), Pt(lll) , Pt(332), Pt(557), Pt(10,8,7) and Pt(13,1,1) catalysts 7.4. Neopentane hydrogenolysis

7.4.1. I r ( l l l ) and Ir(ll0)-(1 X 2) catalysts 7.4.2. Pt(ll l) , Pt(100), Pt(10,8,7) and Pt(13,1,1) catalysts

7.5. n-Hexane hydrogenolysis 7.5.1. Pt(100), Pt(l l l) , Pt(332), Pt(10,8,7) Pt(13,1,1), Au-Pt(100) and Au-

Pt(111) catalysts 58 7.6. n-Heptane hydrogenolysis 59

7.6.1. Pt(ll l) , Pt(557), Pt(10,8,7) and Pt(25,10,7) catalysts 59 7.7. Cyclopropane hydrogenolysis 60

7.7.1. Ni(100) and Ni( l l l ) surfaces 60 7.7.2. Mo(100) surface 61 7.7.3. Pt(s)-[6(lll) x (100)] surface 61 7.7.4. I r ( l l l ) and Ir(ll0)-(1 x 2) surfaces 63

7.8. Methylcyclopropane hydrogenolysis 64 7.8.1. I r ( l l l ) and Ir(ll0)-(1 X 2) surfaces 64

7.9. Methylcyclopentane hydrogenolysis 65 7.9.1. Pt(100), Pt(l l l) , Pt(332) and Pt(557) catalysts 65

7.10. Cyclohexane hydrogenolysis 66 7.10.1. Pt(lll) , Pt(557), Pt(25,10,7) and Pt(10,8,7) catalysts 66 7.10.2. Ru(001) and Cu/Ru(001) catalysts 67

8. Isomerization and cyclization of alkanes 69 8.1. Butane isomerization 69

8.1.1. Pt(ll l) , Pt(100), Pt(332), Pt(557), Pt(13,1,1) and Pt(10,8,7) catalysts 69 8.2. Neopentane isomerization 70

8.2.1. Pt(100), Pt(l l l) , Pt(13,1,1) and Pt(10,8,7) catalysts 70 8.3. Methylpentane isomerization and cyclization 72

8.3.1. Pt(ll l) , Pt(ll9), Pt(311) and Pt(557) catalysts 72 8.4. n-Hexane isomerization and cyclization 72

8.4.1. Pt(100), Pt(l l l) , Pt(332), Pt(13,1,1), Pt(10,8,7), Au-Pt(100) and Au- Pt( l l l ) catalysts 72

4 J.A. Rodriguez and D.W.. Goodman

8.5. n-Heptane cyclization 8.5.1. Pt(l l l) , Pt(557), Pt(10,8,7) and Pt(25,10,7) catalysts

8.6. Summary 9. Hydrogenation of olefins

9.1. Ethylene hydrogenation 9.1.1. P t ( l l l ) surface 9.1.2. Ni(100), Ni(ll0) and Ni( l l l ) surfaces

9.2. Propylene hydrogenation 9.2.1. I r ( l l l ) and Ir(ll0)-(1 X 2) surfaces

9.3. 1,3-Butadiene hydrogenation 9.3.1. Pt(100), Pt(ll0), Pt ( l l l ) and Ni05Pt0.5(lll) surfaces

9.4. Cyclohexene hydrogenation 9.4.1. Pt(223) surface

10. Dehydrogenation of cyclohexane 10.1. Kinetics on Pt surfaces 10.2. Kinetics on Ru surfaces 10.3. Kinetics on W surfaces

11. Cyclotrimerization of acetylene 11.1. Kinetics on Pd surfaces

12. Water-gas shift reaction 12.1. Kinetics on Cu(ll0) and Cu( l l l ) catalysts 12.2. Sulfur poisoning of Cu( l l l ) catalysts 12.3. Cesium promotion of Cu( l l l ) and Cu(ll0) catalysts

13. Methanol synthesis 13.1. Kinetics on Pd(ll0) 13.2. Studies on Cu(ll l) , Cu(100), ZnOx/Cu( l l l ) and Cu/ZnO(0001) surfaces

14. Conclusions

Acknowledgements References

73 73 75 77 77 77 79 80 80 81 81 85 85 87 87 91 93 93 93 94 95 98 98 99

100 101 102

102 103

Surface Science Reports 14 (1991) 1-107 North-Holland

........:.:.:.:-:.:~:i:!:i:!:!:i:i:~:!:i:i~i!:i:i:i:!::::::::-:.:..:.......

surface science reports

High-pressure catalytic reactions over single-crystal metal surfaces

Jos6 A. Rodriguez and D. Wayne Goodman Department of Chemistry, Texas A&M University, College Station, TX 77843-33255, USA

Manuscript received in final form 1 July 1991

Studies dealing with high-pressure catalytic reactions over single-crystal surfaces are reviewed. The coupling of an apparatus for the measurement of reaction kinetics at elevated pressures with an ultrahigh vacuum system for surface analysis allows detailed study of structure sensitivity, the effects of promoters and inhibitors on catalytic activity, and, in certain cases, identification of reaction intermediates by post-reaction surface analysis. Examples are provided which demonstrate the relevance of single-crystal studies for modeling the behavior of high-surface-area supported catalysts. Studies of CO methanation and CO oxidation over single-crystal surfaces provide convincing evidence that these reactions are structure insensitive. For structure-sensitive reactions (ammonia synthesis, alkane hydrogenolysis, alkane isomerization, water-gas shift reaction, etc.) model single-crystal studies allow correlations to be established between surface structure and catalytic activity. The effects of both electronegative (S and P) and electropositive (alkali metals) impurities upon the catalytic activity of metal single crystals for ammonia synthesis, CO methanation, alkane hydrogenolysis, ethylene epoxidation and water-gas shift are discussed. The roles of "ensemble" and "ligand" effects in bimetallic catalysts are examined in light of data obtained using surfaces prepared by vapor-depositing one metal onto a crystal face of a dissimilar metal.

0167-5729/91/$37.45 © 1991 - Elsevier Science Publishers B.V. All rights reserved


I. Introduction

One of the important challenges in basic and applied science is to understand how the atomic structure, composition and electronic properties of the surface of a catalyst determine catalytic activity and selectivity. Over the past 20 years many electron spectroscopies have been developed that permit the study, on the molecular level, of surfaces and adsorbates in ultrahigh vacuum (UHV) conditions [1,2]. This has furthered our understanding of a broad range of phenomena that occur on metal surfaces: adsorbate bonding, coverage effects, coadsorption effects and reactions [1-3].

Unfortunately, the conditions under which the analytical techniques of surface science can be applied are highly idealized. The very low pressures required for surface science studies are typically many orders of magnitude below the pressures used in practical catalytic processes. In order to overcome this problem, several laboratories have developed experimental systems which combine a high-pressure reactor system with an UHV analysis chamber [4-9]. The high-pressure reactor allows the kinetics of catalytic reactions to be measured on a given surface, while analysis of the structure and composition of the surface both before and after reaction can be accomplished in the UHV analysis chamber.

In many high-pressure /UHV studies a clean and well-defined, single-crystal plane is used to model a site or set of sites expected to exist on practical high-surface-area catalysts. This approach allows direct comparison of reaction rates measured on single-crystal surfaces with those measured on more realistic supported metal catalysts, and also allows detailed study of structure sensitivity, the effects of promoters and inhibitors on catalytic activity, and, in certain cases, identification of reaction intermediates by post-reaction surface analysis.

In this article, we review existing literature that deals with high-pressure catalytic reactions on metal single-crystal surfaces. We focus our attention on the kinetics and surface chemistry of the catalytic reactions. Examples are provided which demonstrate the relevance of single-crystal studies for modeling the behavior of high-surface-area supported catalysts. For an excellent review on the instrumentation used in this type of experiments we refer the reader to ref. [9].

2. Ammonia synthesis

The success of ammonia synthesis (N 2 + 3H 2 ~ 2NH 3) as a method of nitrogen fixation was a landmark in the development of catalytic processes [10,11]. The search for a catalyst was conducted by Haber, Bosch and Mittasch in the laboratories of Badische Anilin und Soda Fabrik (BASF) between about 1905 and 1910. The discovered catalyst contained iron with some alumina and potassium oxide. Large-scale production of synthetic ammonia using such a catalyst started in Germany in 1914. The industrial catalysts used today are based mainly on iron promoted with A1203, K 2 0 , CaO and MgO [10]. Although the kinetics of the ammonia synthesis reaction has been investigated by numerous workers, the mechanism of the reaction is by no means unequivocally established.

2.1. Kinetics over Fe single-crystal catalysts

The adsorption and dissociation of N 2, H 2 and NH 3 on the (100), (110) and (111) faces of iron have been investigated in detail under UHV conditions. The initial sticking coefficient for dissociative nitrogen chemisorption was found to be very low (10-7-10-6) , with the

High-pressure catalytic reactions ocher single-crystal metal surfaces 7

activity of the surfaces following the sequence: (111)> (100)> (110) [12,13]. On Fe(100), atomic N forms a c(2 × 2) structure [12]. By contrast, the F e ( l l l ) and Fe( l l0 ) planes reconstruct under the influence of N atoms. A series of complicated LEED patterns was observed in these cases, which could be reconciled with slightly distorted (111) planes of FeaH on top of the iron substrate [12,13]. The activation energy for nitrogen desorption ranges between 210 and 240 kJ mol - ~ (depending on the surface orientation). From this, the strength of the F e - N bond is estimated to be 580 kJ mol-~ [12,13], which can be compared with the value of 950 kJ tool-1 reported for the dissociation energy of N 2.

The dissociative adsorption of H 2 on the (100), (110) and (111) faces of iron is non-activated and proceeds with initial sticking coefficients in the range between 0.1 and 0.2 [14,15]. At the limit of zero coverage, the strength of the F e - H bond is of the order of ~ 250 kJ mol - [14,15]. The initial hydrogen adsorption energy varies between 88 and 110 kJ mol-~ on the various crystal planes [14,15]. At low temperatures ( < 200 K), NH 3 adsorbs non-dissociatively on the Fe(100), Fe( l l0 ) and F e ( l l l ) surfaces [16,17]. The heat of adsorption depends on substrate orientation and adsorbate coverage, showing a maximum value of 70 kJ mol - ~. Over Fe, NH 3 decomposition occurs stepwise yielding Nn2.a, NH a and H a [16,17].

On the basis of information obtained by using single crystals under UHV conditions, the individual reaction steps for ammonia synthesis can be formulated as follows [18]:

H2,g ~ 2 H a,

N2,g ~ S2. a ~ 2 Na,

N a + H a ~- NH a,

NH a + H a ~ NH2, a,

NH2, a + H a ~ NH3, a,

NH3. a ~ NH3,g.

(1)

(2)

(3)

(4)

(5)

(6)

In this mechanism the rate-limiting step is the dissociation of adsorbed N 2 [18,19]. A kinetic model constructed using this mechanism and parameters derived from quantum-mechanical calculations and from the UHV studies described above predicts rates of NH 3 synthesis that are in excellent agreement with high-pressure measurements over industrial catalysts [19].

At 798 K and a total pressure of 20 atm of a stoichiometric mixture of hydrogen and nitrogen (PH2/PN2 = 3), ammonia was formed on F e ( l l l ) at a rate of 4.6 × 10 -8 mol cm -2 s-~ with an activation en6rgy of 81.2 kJ mol-1 [20]. The activity ratio of Fe ( l l l ) :Fe (100 ) : Fe( l l0 ) for NH 3 synthesis was found to be 418:25:1 [20]. The same order of reactivity has been found for the dissociative chemisorption of N 2 (see above), the rate-determining step in the ammonia synthesis reaction.

The high catalytic activity of Fe(111) can be attributed to the presence of metal atoms with a coordination number of 7 (C 7 coordination sites, the most highly coordinated surface sites Fe can expose), which are in the second and third layers and are exposed to the reactant gases [21]. A second possibility is that, since the F e ( l l l ) surface is very open, the high catalytic activity is due to the high surface free energy and low work function that are characteristic of a rough surface [21].

The high-pressure kinetic data presented in fig. 1 for ammonia synthesis over Fe ( l l l ) , Fe(211), Fe(100), Fe(210) and Fe( l l0 ) show that the presence of highly coordinated sites is more important than surface roughness for catalytic activity [21]. The Fe(210) plane contains

8 J.A. Rodriguez and D. I~ Goodman

®

Fe(111) Fe(210)

C7 C6

- ~ Fe(lO0) C4

C8

Fe(211) Fe(110)

T = 673 K

12

o 10 x

O4 E ~ 8

z I¢1 • 6 0 E fii F- 4

"12:N2

(111) (211) (100) (210) (110)

SURFACE ORIENTATION Fig. 1. (a) Schematic representat ion of five Fe single-crystal surfaces. The coordination number for each surface atom

is indicated. (b) Rate of NH 3 synthesis over five different faces of iron (from ref, [21]).

n o C 7 sites but it is an open face exposing second- and third-layer atoms, much like the F e ( l l l ) plane. On the other hand, the Fe(211) crystal contains C 7 sites but it is less open than the Fe(210) face. If surface roughness was the important ingredient for an active iron catalyst then one would expect Fe(210) to be the most active surface in fig. 1. The results are contrary to this expectation, showing that the Fe(211) crystal is almost as active as Fe ( l l l ) , while the Fe(210) is much less active. The experimental data sxtppo~" ~ J ~ ' ~ , ~ ' ~ ~h~ C7 sites are the most active centers in iron ammonia synthesis catalysts [21].

The surfaces of industrial catalysts are promoted with K 2 0 [10]. Model studies with adsorbed potassium (without oxygen) on Fe(100) indicate that small K concentrations increase the sticking coefficient for dissociative N 2 adsorption (the rate-limiting step in NH 3 synthesis) by more than two orders of magnitude up to a maximum value of 4 × 10 -5 (at 430 K) [18]. N 2 is more tightly held in the vicinity of an adsorbed K atom. The N 2 adsorption energy is in fact increased from 30 to 45 kJ mol-1, while simultaneously the activation energy for dissociation is lowered by 10 kJ mol-L [18]. Adsorption of K on Fe surfaces causes a very pronounced lowering of the work function; this enhances back-donation of metallic electrons to the antibonding N2(2~-) orbital, strengthening the M - N 2 bond and weakening the N - N bond.

Over actual ammonia synthesis catalysts the promoter effects of K are reduced by coadsorbed oxygen [18]. Under reaction conditions K alone is not thermally stable on Fe


surfaces. Addition of oxygen or aluminum oxide prevents the desorption of K, keeping a relatively large concentration of this promoter on the surface of the catalyst [18,20b]. Studies of ammonia synthesis (T = 673-723 K with a 20 atm stoichiometric mixture of N 2 and H 2) over iron surfaces doped with K and K + O [20] show no promotional effect on Fe(110) and a marked increase in the reaction rate on Fe(111) and Fe(100). No change in the apparent activation energy of the reaction was observed when K was added to the Fe surfaces [20c]. To explain these results, a kinetic model has been proposed in which potassium enhances the rate of ammonia synthesis by lowering the concentration of NH3 on the surface (making more active sites available to chemisorb the reactants) and by increasing the rate of N 2 dissociation [20c].

2.2. Kinetics ouer Re single-crystal catalysts

The synthesis of ammonia from N 2 and H 2 was investigated over model single-crystal rhenium catalysts at 20 atm reactant pressure (PH2/PN2 = 3) and temperatures between 720 and 900 K [22]. The results of fig. 2 indicate that ammonia synthesis is a structure-sensitive

C6 (SOLID) C9 (SOLID) C10 (DOTTED)

O 4

E

_¢

o _e 0 E "l- Z

I,-- ,,¢ n-

1017

1016

1015

1014

T.F.=314 ii iFo Ptota! -- 20 atm H2:N2=3:1

Tcrystal = 870K

T.F.=6 m

°)

(~~)0.03

1013

Fig. 2. Structure sensitivity in the ammonia synthesis over Re single-crystal faces. The turnover frequencies (T.F.) are given as NH 3 molecules c m - 2 s - i. Schematics of the atomic structure of each surface are given above each bar (from

ref. [22]).


reaction over Re surfaces. An apparent activation energy of 81.2 + 4.6 kJ mol -~ was observed, regardless of the catalyst surface structure [22]. Kinetic data, reactant pressure dependence and deuterium isotope effect indicate that the rate-determining step is the dissociative chemisorption of N2, as in the case of Fe catalysts [22].

In fig. 2 the most active crystal faces are those that expose highly coordinated C~ and C 10 sites [21]. Surface roughness on the Re catalyst is only important to the extent that it can expose these highly coordinated surface atoms. An identical result was obtained for Fe substrates (see above). Theoretical work suggests that highly coordinated metal atoms will show the greatest catalytic activity because these atoms can experience the largest electronic charge fluctuations within the solid [21]. On these highly coordinated sites bond breaking and bond formation can occur within the lifetime of the reaction intermediates.

3. CO methanation

The reaction of low concentrations of CO in a mixture with H 2 to form CH 4 was developed as a gas purification process in the 1950's [10]. At the present time, the methanation reaction has a critical role in the production of synthetic natural gas from hydrogen-deficient carbonaceous materials [10,23,24]. In addition, the reaction is an obvious starting point in studies of fuel and chemical synthesis from carbon sources [10,23,24]. In the last ten years the methanation reaction over single-crystal surfaces has been the subject of many investigations. Here we review the results of these investigations. We begin with a discussion of the studies on monometallic (section 3.1.1) and bimetallic (section 3.1.2) single-crystal surfaces. Next the results of studies dealing with the effects of electronegative (section 3.2.1) and electropositive (section 3.2.2) impurities on the kinetics of the methanation reaction are presented. Finally, we show studies concerned with metal -support interactions and the methanation reaction (section 3.3).

3.1. CO methanation on clean metal single crystals

3.1.1. Monometallic surfaces

3.1.1.1. Ni(lO0) and Ni ( l l l ) . The data in the Arrhenius plot of fig. 3a represent steady-state specific methanation r a t e s ( C H 4 molecules / s i t e , s) on both the Ni(111) and Ni(100) surfaces [25]. At a given temperature the rate of production of C H 4 o v e r an initially clean catalyst crystal was constant, with no apparent induction period [25]. The atomic configurations of the Ni(100) and Ni(111) surfaces are shown in figs. 3b and 3c, respectively. The similarity between the data for the close-packed (111) and for the more open (100) crystal plane of Ni is evident in both the specific rates and activation energy (103 kJ/mol) . The single-crystal results are compared in fig. 3a with three sets of data taken from ref. [26] for alumina-supported nickel catalysts. This comparison shows extraordinary similarities in kinetic data taken under nearly identical conditions. Thus, for the H 2 + CO reaction over nickel, there is no significant variation in the specific reaction rates or the activation energy as the catalysts change from small metal particles to bulk single crystals. These data provide convincing evidence that the methanation reaction rate is indeed structure insensitive on nickel catalysts.

Post-reaction analysis of the surface of the Ni crystal catalysts with Auger electron spectroscopy (AES) showed a low level of carbon species and the absence of oxygen [6,25]. The Auger lineshape for the carbonaceous residue was similar to that of nickel carbide,

High-pressure catalytic reactions oL,er single-crystal metal surfaces 11

E t .

0 ® C J= @

Q Temperature (K) 800 700 600 500 450 ( ~

] ~ I I i I [

® I3H=*CO--C"4+H=OI Ni(lO0)©

~ single crystal =" 10 " ~ N i (111) •

0 ¢ 10 "11 ~ c r y s t a l

- 10-2 Ni 'AI20 Q

10 .3

10-4__ I ~ I I , I I ~ I

1.2 1.4 1.6 1.8 2.0 2.2 Reciprocal temperature x 10 3

(lOO)

(111 l

Fig. 3. (a) Comparison of the rate of methane formation (CH 4 molecules/si te-s) over single-crystal nickel catalysts and supported Ni /Al203 . Reaction conditions: 120 Torr, H 2 : C O = 4 (from ref. [25]). (b) Atomic configuration of a

Ni(100) surface. (c) Atomic configuration of a Ni(111) surface.

indicating that the carbon was in a "carbidic" form [6]. Experiments were carried out studying the interaction of CO (24 Torr, 0-1000 s exposures) with Ni(100) at different temperatures (450-800 K) [27]. AlES data showed the deposition of carbon on the surface and the absence of oxygen. Two kinds of carbon were formed on the surface: a carbidic type, which occurs at temperatures < 650 K, and a graphite type at temperatures > 650 K. The carbidic type saturates at 0.5 monolayers, and can be readily removed from the Ni(100) surface by heating the crystal to 600 K in 1 atm of H z with methane formed as the product. In contrast, the graphitic type is a poison. The deposition of an active carbon residue and the absence of oxygen on the nickel surface following heating in pure CO is consistent with a well-known disproportionation reaction, the Boudouard reaction,

2 COg ~ C a qt_ C02,g ' (7)

which has been studied on supported Ni catalysts [28,29] and on Ni films [30]. On Ni(100), the carbon formation data from CO disproportionation indicates a rate equivalent to that observed for methane formation in a H 2 / C O mixture. Therefore, the surface carbon route to product is sufficiently rapid to account for methane production with the assumption that kinetic limitations are not imposed by the hydrogenation of this surface carbon.

A set of experiments was performed [27] in which a Ni(100) surface was precarbided by exposure to CO and then treated with hydrogen in the reaction chamber for various times. This study showed that the rate of carbon removal in hydrogen compared favorably to the carbide formation rate from CO and to the overall methanation rate in H 2 / C O mixtures. Thus in a H e + CO atmosphere the methanation rate is determined by a delicate balance of the carbon formation and removal steps and neither of these is rate determining in the usual sense [6,25,27]. More recent studies [31] using isotopically labeled CO have shown that the CO dissociation step is essentially unidirectional in that the rate of C a and O a recombination is insignificantly slow compared to the C a hydrogenation rate.


800K 700K 600K 500K 450K

~ Ni(lO0) 10 H2/co=4/~

"~ ~r

{ 1 ~ P T ~

Z ~ 10"1

pT,, 1

10 -3

! o I-- 10"4 E---~ h L J I t l

1.2 1.4 1.t6 118 I 2.0 2.2

10

1

10"1

10"2

10"3

10 "4

800K 700K 600K 500K 450K i i i i i = [ / ~

Ru(110) H21CO~t411

X~=120 T~rr /X au(001)

PT¢10 ~

PT,'I Tort

[ I [ I I I L i 1 i 1.2 114 1.6 1.8 2.0 2.2

l IT x 10 3 (K "1) l IT x 10 3 (K "1) Fig. 4. (a) Arrhen ius plot of CH 4 synthesis on a Ni(100) catalyst at total reactant pressures of 1, 10 and 120 Torr; H 2 : CO = 4 (from ref. [25]). (b) Arrhen ius plot of CH 4 synthesis on a Ru(110) catalyst at total reactant pressures of 1, 10 and 120 Torr; H 2 :CO = 4. Data at two tempera tu res for a Ru(001) catalyst at 120 Torr are plotted with a × (from

ref. [25]).

Fig. 4a shows the changes in the methanation reaction rate as the total pressure is increased from 1-120 Torr at a fixed H 2 : CO ratio [25]. At low temperatures the rates fall on the same straight line at all pressures. As the temperatures is increased, a deviation from linearity is seen - the higher the pressure the higher the deviation temperature. Accompany- ing this non-linear rate behavior is an increase in the active carbon level on the surface of the catalyst crystal [25]. It has been proposed [25] that this departure from the linearity of the rate in fig. 4a and the accompanying increase in the surface carbon level is due to a decrease in the surface coverage of hydrogen and thus a decrease in the rate of hydrogenation of surface carbon.

According to the mechanism proposed above for CO methanation, if reaction conditions are altered such that the surface hydrogen concentration decreases (e.g. low H 2 pressure and high temperature), then a correlation between decreasing methane yield and increasing surface carbide should be observed. This correlation holds very well as evidenced by the data in fig. 5. Thus, the proposed reaction mechanism involving the dissociation of CO and the subsequent hydrogenation of the resulting carbon species (C a ) accounts quite satisfactorily for the effect of pressure on the methanation rate, for the variation in the measured surface carbon level as reaction parameters are changed, and for the formation at characteristic temperature and pressure conditions of a catalyst-deactivating graphitic carbon.

3.1.1.2. Ru(llO) and Ru(O01). Fig. 4b displays steady-state specific rates for CO methanation on two faces of ruthenium: the zig-zag, open (110), and the close-packed (001) [25]. While the comparison is limited, it is clear that the H 2 + CO reaction is quite similar in regard to the specific reaction rate and the activation energy for these two crystal planes of ruthenium. Thus, it appears that CO methanation is structure insensitive on ruthenium surfaces [25].


0 . 5 0 - • 1 m H2/CO P,,,, NO. Ratio (Tot r.)

~ 1 4.0 1 \ T = 625K 2 4.0 lo

3 4.0 36 i 4 4.0 61 _ 0 . 4 0 - s ,o so

6 4.0 120 1 3 e . 7 40 soo

8 4,0 1 DO0 9 1.0 10

0 . 3 0 - \ 14 ~o ;o so 2 ~ 11 1.0 200 12 1,0 500

¢~ 13 0.1 110 14 0.4 34

O. 2 0 - - I~ 2 . ! 8--L

0 . 1 0 - 11e 5 t b ~ ~

, - v . 7 I I J 1 10 100

M e t h a n a t i o n Rate (NcH 4 x 10 "2)

Fig. 5. Methane production rate (molecules/Ni-surface-atom.s) at 625 K over a Ni(100) catalyst versus surface carbon coverage (under steady-state conditions). The H2:CO ratio and the total pressure (Torr) for each point are

indicated in the insert (from ref. [25]).

Post-reaction surface analysis of the Ru crystal catalysts with AES showed the presence of carbidic carbon [25]. The hydrogenation of this carbonaceous residue can be followed readily [25,32]. Furthermore, the specific rates of carbide formation on ruthenium surfaces from CO decomposition are equal to the rates of methane formation in CO + H 2 mixtures [25,32]. This experimental evidence suggests that the reaction mechanism for CO methanation on ruthenium surfaces is similar to that mentioned above for methanation over nickel surfaces. In fig. 4 the variation of the reaction rate with pressure is very similar for the Ni(100) and Ru( l l0 ) crystals. In both cases, the non-linear rate behavior is accompanied by an increase in the active carbon level on the surface of the catalyst crystal. It has been proposed [25] that the departure from linearity of the rate in fig. 4 is due to a decrease in the surface coverage of hydrogen, which causes a decrease in the rate of hydrogenation of surface carbon. In fact, since the binding energy of hydrogen on Ru is lower than on Ni [25,33], the deviation from linearity should be expected at lower temperature for ruthenium. This is particularly evident in the 1 Torr data of figs. 4a and 4b.

3.1.1.3. W(llO). The methanation activities for W( l l0 ) [34] and Ni(100) [24] are compared in fig. 6 over a range of temperatures (fig. 6a) and H 2 partial pressures (fig. 6b). The data clearly indicate that W(110) is an active methanation catalyst, with an activity that in some cases can surpass the activity of Ni(100). Plotting the data in an Arrhenius fashion (fig. 6a) yields an apparent activation energy of 56 kJ mol -~ for W(ll0) , as compared to 103 kJ mol - j for Ni(100). The activation energy over W( l l0 ) is in reasonable agreement with the value of 63 kJ mol-1 observed on catalysts prepared by decomposition of W(CO) 6 on alumina [35].

Auger electron spectra of the W( l l0 ) surface after steady-state reaction conditions indicated that the active methanation surface was highly carbidic [34], in contrast to the case of nickel, where the active methanation surface is the metal itself with only a low surface coverage (0.05-0.1 monolayers) of carbidic carbon species present [6,25]. The idea that the active W surface is carbidic is consistent with the significant hydrogenation activity reported for W carbide catalysts [34].


800K 700K 600K 500K 4501<

|

¢B 100 Tort H 2

.-_ " . , 10-1 ~- 10 N i ( 1 0 0 ) o.1 To. CO m ", 500K

-~ ",,, 0.1 Torr CO f j -,

"6 *J 1 '~ W(110) E " ~ \N /Ea.=- 103 kJ mo1-1 10 -2 -

W ( 1 1 0 ,~ I0"I z 00) -r

z 10"3 10.2 Ea=S6 kJ tool - 1 " /

.D

Z : ',\ 10 -4 10 "3! ', - I I I I > Q \~' o 1 10 100 1000

E ~ I - 10 "4 , ~ , , t , , ~ H2 Pressure (Torr)

1.2 1.4 1:6 1.8 210 2:2

lIT x 103 (K "1)

Fig. 6. (a) Methanation activtty for W(110) compared to that for Ni(100), plotted in an Arrhenius fashion (from ref. [34]). (b) H 2 dependence of the methanation activity for W(ll0) compared to that of Ni(100) (from ref. [34]).

3.1.1.4. Rh(111). CO hydrogenation over clean R h ( l l l ) was studied at a temperature of 573 K and partial pressures of 4.5 atm of H 2 and 1.5 atm of CO [36]. Under these reaction conditions the Rh catalyst produced primarily methane (90 wt%) at an initial rate of 0.15 molecules /s i te .s . Small amounts of C 2 and C 3 hydrocarbons were also formed, but no oxygenated hydrocarbons were detected. The rates of formation of all the products were found to be the same on the R h ( l l l ) single crystal and on a polycrystalline Rh foil, suggesting that CO hydrogenation is structure insensitive on these surfaces [36]. The results of AES showed the presence of ~ 1 monolayer of carbon on the surface of the Rh catalysts after 3 h of reaction [36]. The close proximity of the Rh 256 and 302 AES peaks to the C 272 AES peak prevented an analysis of the lineshape of the carbon peak in order to determine the chemical nature of the carbonaceous residue.

3.1.1.5. Fe(111). The hydrogenation of CO on the (111) face of iron was examined at partial pressures of 4.5 atm of H 2 and 1.5 atm of CO and at a temperature of 573 K [37]. The major product of the reaction was methane (~ 70 wt%), which was formed at an initial rate of 1.35 molecules/sites. The formation of C 2 (~ 20 wt%), C 3 and C 4 products was also observed.

3.1.1.6. Mo(lO0). Fig. 7 shows the effects of temperature and pressure on the rate of CO methanation over Mo(100). On this surface, the hydrogenation of CO produced primarily methane (~ 90 mol%), ethene and propene [38]. An activation energy of ~ 100 kJ mol - ~ was found for the methanation reaction on Mo(100). The observed rate law for methanation (see fig. 7b) is given by

= b j o + 0 . 3 2 1 D + 1.0 r C H 4 '~" C O - - H 2 . ( 8 )

The positive power of the rate dependence on the pressure of CO is unusual since the


u)

"6 o O

E t - O .t..,

t.. 0 I1.

0 r,.

e-

io -3

I I I I I

\ . c

1 3 2 0 To r r

® , , 1.6 1 . 8 2.0

1/'1" x 103(K "1)

O

"6 O E r- O om

E

~. j6 ~

i I i l i i

6 o l m tota l pressure

T • 3 0 0 * C

P¢O ¢ o n s t ~

R'? o /

I I i i I t i o "~ I

Partial Pressure (atm)

Fig. 7. (a) Arrhenius plot for CO methanat ion on Me(100) (from ref. [38]). (b) Rate of formation of methane over Me(100) versus the partial pressure of each reactant. Constant H 2 pressure of 3 atm for determination of CO dependence, and constant CO pressure of 3 atm for determination of H 2 pressure dependence. In all the cases the

total pressure was 6 atm, with nitrogen or argon used as a buffer gas (from ref. [38]).

methanation rate has a negative-order dependence on CO partial pressure over Ni, Ru, Fe and Co catalysts [5,38,39].

Auger electron spectra taken after the hydrogenation reaction indicate that the "active" surface is covered by a submonolayer of a carbidic carbon species. The reaction is poisoned as the carbidic species is converted to graphitic carbon. The rate of poisoning is determined by the ratio of CO to H E in the reaction mixture and by the reaction temperature (lower CO : H 2 ratios and lower temperatures prolong the lifetime of the active catalyst).

The following set of elementary Me(100) [38]:

C O g H C O a ,

H2 ,g ~-- H 2 , a,

CO a + H2,a ~ C H O H a,

1 C H O H a + ~H2,a -'-> CH a + H 2 0 ,

3 fast C H a + ~ H 2 , a - - - - - - > C H 4 , g .

steps was proposed for the methanation reaction on

(9)

(a0)

(11)

(12)

(13)

In terms of this model the rate-determining step is reaction (12) and all the steps preceding it are in quasi-equilibrium. A mathematical analysis of this kinetic model [38] leads to a rate


expression of the form

= b p + t ) . 5 1 0 + 1.0 (14) t ' C H 4 ,~a C O a H 2 '

which is in reasonable agreement with the experimental results of fig. 7b. Reactions (9) to (13) probably take place on top of a carbidic overlayer [38]. This overlayer will deactivate by forming graphite on the surface, which will block the reaction sites.

No differences in either rates or product distributions were observed between CO hydrogenation on Mo(100) and over polycrystalline Mo foils [38]. Thus, the reaction does not appear to be structure sensitive on molybdenum surfaces.

3.1.1.7. Co(O001), Co(1120) and Co(1012). CO hydrogenation over Co(0001) was investigated at temperatures between 490 and 570 K, and partial pressures of 0.66 atm of H 2 and 0.33 atm of CO [40]. Under these reaction conditions methane was the major product, with lesser amounts of heavier hydrocarbons. The apparent activation energies for methane, ethane and propane formation were found to be equal (~ 70 kJ mol-~, T < 570 K), which suggests a similar rate-determining step [40]. The HREELS spectrum, taken after the reaction, showed peaks at 2980, 1420 and 845 cm-1, which can be attributed to vibrations of adsorbed CH x (x = 1,2,3) groups [40].

Studies for CO hydrogenation on Co(0001), Co(1120) and Co(1012) show very similar rates of methane production over these surfaces [40]. The apparent activation energy for CO methanation is identical on Co(0001) and Co(1120) (70 kJ mol-L) [40]. However, the product distribution for heavy hydrocarbons (MW > 16) is different. Co(117.0) yielded a larger chain growth probability than Co(0001). Post-reaction spectroscopy with HREELS and AES revealed that during the reaction longer-chain hydrocarbon fragments grew on Co(1120), while on Co(0001) only CH x (x = 1-3) species were observed [40].

3.1.1.8. Summary. Table 1 shows the apparent activation energies observed for CO methanation on different single-crystal surfaces. In general, the energies are close to the value of 110 kJ mol-1. The exceptions are W(I I0) and Co(001), with apparent activation energies of 56 and 70 kJ mo1-1, respectively. The studies reviewed in this section provide convincing evidence that the methanation reaction is structure insensitive on surfaces of Ni, Ru, Rh, Fe and Mo. On these metals the methanation of CO occurs in the presence of an active carbidic overlayer. The transformation o f this overlayer into graphite leads to a decrease in the catalytic activity of the metal surfaces.

Table 1 Metal single-crystal catalysts; specific activities and activation energies in CO methanation

Catalyst T PH 2 Pco TOF a) Ea Ref. (K) (Torr) (Torr) (site- l s 1) (kJ mol-z)

Ni(100) 450-700 96 24 0.0002- 8 103 [25] Ni(lll) 550-700 96 24 0.04 - 5 103 [25]

440-540 60 20 0.0002- 0.05 111 [99] Ru(110) 500-700 96 24 0.002 - 10 123 [25] Ru(001) 550-650 96 24 0.01 - 0.8 ~ 120 [25] W(ll0) 480-750 100. 0.1 0.002 - 0.5 56 [34] Mo(100) 510-640 1100 220 0.001 - 0.2 100 [38] Co(001) 490-550 500 250 0.01 - 0.07 ~ 70 [40]

~) TOF = turnover frequency = CH 4 molecules/site-s.


3.1.2. Bimetallic surfaces Catalytic properties of metal surfaces can be altered greatly by the addition of a second

transition metal [41]. In many cases, mixed-metal systems are superior over their single-metal counterparts in terms of catalytic activity a n d / o r selectivity [41]. Many fundamental studies have focused on trying to understand the roles of "ensemble" and "ligand" effects in bimetallic catalysts [41,42]. Ensemble effects are defined in terms of the number of surface atoms needed for a catalytic process to occur. Ligand effects refer to those modifications in catalytic activity or selectivity that are the product of electronic interactions between the components of a bimetallic system. In gathering information to address these issues, it has been advantageous to simplify the problem by utilizing models of bimetallic catalysts such as the deposition of metals onto single-crystal substrates. Work on ultrathin metal films supported on well-defined metal surfaces (Cu on Ru(001) [43-46]; Cu, Ni, Pd and Pt on W(l l0 ) and W(100) [47-51]; Fe, Ni and Cu on Mo( l l0 ) [52,53]; and Fe and Cu on Re(001) [53,54]) has shown that a metal atom in a matrix of a dissimilar metal can be significantly perturbed, and that this perturbation can dramatically alter the chemical and electronic properties of both constituents of the mixed-metal system.

The studies reviewed here are part of a continuing effort to identify those electronic and structural properties of bimetallic systems which can be related to their superior catalytic abilities.

3.1.2.1. Ni/W(llO) and Ni/W(IO0). The Ni/W(100) and N i / W ( l l 0 ) systems are particularly interesting because they involve the addition of an active metal for CO methanation (Ni) to relatively inactive tungsten surfaces.

At ~ 100 K, Ni is adsorbed layer by layer on W( l l0 ) and W(100) [49]. The results of low-energy electron diffraction (LEED) indicate that at coverages up to 1 ML the Ni films grow pseudomorphically with respect to the W( l l0 ) and W(100) substrates [49]. This growth pattern leads to Ni monolayer densities on W(l l0 ) and W(100) which are 21% and 38% less than the corresponding monolayer densities for Ni(111) and Ni(100), respectively.

Specific rates of CH 4 production, expressed as turnover frequencies (CH 4 molecules/Ni a tom. s), over Ni-covered W(l l0 ) and W(100) surfaces are shown in fig. 8 [51]. Under the experimental conditions of this figure, the reaction rates on the clean W surfaces were ~ 102 times lower than on the Ni-covered surfaces. Fig. 8a shows that for a total pressure of 120 Torr, Arrhenius behavior is observed over the entire temperature range studied (450-700 K) as the CH 4 production rates vary by almost 3 orders of magnitude. The similarity between N i / W ( l l 0 ) and Ni/W(100) at all the coverages studied is evident in both the turnover frequencies and activation energy, 77 + 4 kJ mol - 1. The activation energy for the Ni-covered W surfaces is lower than the value of 103 kJ mo1-1 reported for Ni(100) and N i ( l l l ) [25]. However, the specific rates for CO methanation on N i / W ( l l 0 ) and Ni/W(100) correlate well with those observed on supported Ni catalysts, Ni single crystals and Ni films [51]. These results are further manifestations of the structure-insensitive behavior of the CO methanation reaction and suggest that the mechanistic steps which control the rate of CO hydrogenation are the same in all these surfaces.

Lowering the total pressure has a significant effect on the rate of methane production (fig. 8b) for Ni supported on W( l l0 ) [51]. A similar effect was observed for Ni(100) (see fig. 4a), and was attributed to a decrease in the concentration of atomic hydrogen on the surface as the pressure was lowered and the temperature was increased (see section 3.1.1) [25]. The departure from Arrhenius behavior occurs at lower temperature for N i / W ( l l 0 ) . This corre-

1 8 J.A. Rodriguez and D. HI.. Goodman

]1 ' I I I I I I I I 1 I ' ~ ' i ' i ' l 1 C O + H 2 ~ C H 4 %~. N I I W ( I 1 0 )

• C O + H 2 C H 4 ' ~ • o.a MI. m

( ~ : I ~ ~ \ o 0.4 . L Hi P : 1 2 0 T o r r ~ • 0 . 5 ML NI

lU~ \ C] 0 . 8 ML NI ~ ~ "l" 1 .0 ML NI

>. >"

Z 2 :

10-' ~ lo-' ° ° 14,1 ~ P = I O

t r I,L

m ~ > o o z z

m., i f f 2 ~ lo -a I- I- u.I uJ Z Z 1 T o r r ~ ~j~_

• .3 ML <{ P = "T" [] .S ML N I / W ( 1 1 0 ) ~ "I" I-" • 1.0 ML N I /W ( 1 1 0 ) ~ [ ] I ' - ~ I ~ U.I IJJ ~; O .4 ML N I / W ( 1 0 0 )

• .7 ML N I / W ( 1 0 0 ) ~ ; 4. .0 ML N I / W ( 1 0 0 )

- , [ ~ I , I , I I 10 "3 i l i l I t [ , I 1 0 1 1 2 1 . 4 1 . 6 1 . 8 2 . 0 2 . 2 2 . 4 . 0 1 . 2 1 . 4 1 . 6 1 . 8 2 . 0 2 . 2 2 . 4

IO00 /T (K "1) 1000/T (K -1)

Fig. 8. (a) Arrhenius plot for CH 4 synthesis over several different Ni coverages on W(ll0) and W(100) at a total reactant pressure of 120 Torr ( H 2 : C O = 4) (from ref. [51]). (b) Arrhenius plot for CH 4 synthesis over several

different Ni coverages on W(110) at total reactant pressures of 1, 10 and 120 Torr (H 2:CO = 4) (from ref. [51]).

lates [51] with the fact that the activation energy for H 2 desorption from Ni/W(110) ( ~ 71 kJ mo l - t [49]) is lower than that from Ni(100) (96 kJ mo l - l [55]).

3.1.2.2. Cu/Ru(O01) and Ag/Rh(111). A bimetallic system that has been extensively studied in supported catalyst research is copper on ruthenium [41,45,46]. The immiscibility of copper in ruthenium circumvents the complication of determining the three-dimensional composition. The adsorption and growth of copper films on the Ru(0001) surface have been examined [43,44,56,57] by work function measurements, LEED, AES, XPS, UPS and TPD. The experimental evidence indicates that for submonolayer depositions at 100 K the Cu grows in a highly dispersed mode, forming two-dimensional islands pseudomorphic to the Ru(001) substrate upon annealing to 300 K. The pseudomorphic growth implies that the copper -copper bond distances are strained approximately 6% beyond the equilibrium bond distances found for bulk copper.

Copper surfaces are inactive catalysts for CO methanation. A study of the rate of CO methanation over Cu/Ru(001) [46] indicates that copper merely serves as an inactive diluent, blocking the active sites of the ruthenium surface on a one-to-one basis. Similar results have been found in analogous studies [58] introducing silver onto a R h ( l l l ) methanation catalyst.

3.1.2.3. Co/W(llO) and Co/W(IO0). Cobalt forms pseudomorphic monolayers on W(l l0 ) and W(100), which are thermally stable to 1300 K [59]. The cobalt overlayers are geometrically strained with respect to bulk cobalt surfaces. The pseudomorphic monolayer of C o / W ( l l 0 ) has an atomic density 21% less than Co(001), while the pseudomorphic monolayer of Co/W(100) has an atomic density 45% less than Co(001) [59]. However, the CO methanation


activity of these Co /W surfaces is very similar and compares favorably with that of alumina- supported Co catalysts [59], a fact that suggests that CO methanation is structure insensitive on cobalt surfaces [59].

AES spectra showed the after-reaction Co/W surfaces to have high coverages of both carbon and oxygen, with carbon lineshapes characteristic of carbidic carbon [59]. The catalytic activity is apparently not correlated with surface carbon level [59].

3.2. CO methanation on chemically modified surfaces

The addition of impurities to a metal catalyst can produce dramatic changes in the activity, selectivity and resistance to poisoning of the catalyst. For example, the selectivity of some transition metals can be altered greatly by the addition of light metals such as potassium, and the activity can be reduced substantially by the addition of electronegative species such as sulfur. Although these effects are well recognized in the catalytic industry, the mechanisms responsible for chemical changes induced by surface additives are poorly understood. An important question concerns the underlying relative importance of ensemble (steric or local) versus electronic (nonlocal or extended) effects. A general answer to this question will improve our ability to design efficient catalysts.

Catalyst deactivation and promotion are extremely difficult questions to address experi- mentally [60]. For example, the interpretation of related data on high-surface-area supported catalysts is severely limited by the uncertainty concerning the structural characterization of the active surface. Specific surface areas cannot always be determined with adequate preci- sion. In addition, the knowledge of the crystallographic orientation, the concentration and the distribution of impurity atoms, as well as their electronic states is generally poor. The use of metal single crystals in catalytic reaction studies essentially eliminates the difficulties mentioned above and allows, to a large extent, the utilization of a homogeneous surface amenable to study using modern surface analytical techniques. In this section we review studies dealing with the effects of electronegative and electropositive surface impurities on the rates of CO methanation over single-crystal catalysts. Although the studies to date are few, the results appear quite promising in addressing fundamental aspects of catalytic poisoning and promotion.

3.2.1. Electronegative impurities Impurities whose electronegativities are greater than those for transition metals generally

poison a variety of catalytic reactions, particularly those involving H z and CO. Of these poisons sulfur is the best known and is technologically the most important [10,60].

3.2.1.1. Atomic chlorine, sulfur and phosphorus on Ni(lO0). The effects of preadsorbed CI, S and P atoms on the adsorption-desorption of H e and CO on Ni(100) have been extensively studied [61-65] using Auger electron spectroscopy, low-energy electron diffraction and temperature-programmed desorption. Fig. 9 shows the variation of the saturation coverage of H and CO on Ni(100) with the coverage of C1, S and P. Both CO and H z adsorption decrease markedly in the presence of surface impurities. The effects of P, however, are much less pronounced than those of CI and S. The similarity in the atomic radii of CI, S and P (0.99, 1.04 and 1.10 A, respectively [66]) suggests a relationship between electronegativity and the poisoning of chemisorptive properties by these surfaces impurities [61-63]. Impurities that are strongly electronegative with respect to nickel, C1 and S, modify the chemisorptive behavior far more strongly than would result from a simple site-locking model. The initial effects of

20 J./I. Rodriguez and D. W.. Goodman

z 0

ffJ

o o

0.6 i

0.5

0.4

03 !

0.2

0.1 ]

01 0

- I I ~ l -" ~ I I 0.1 0.2 0.3 0.4 0.5 0.6 0.7

ADDITIVE COVERAGE (ML)

0.8 • CI

z o_ o.6

~ 0.4 z

0.2

OI I 0.1 0.2 0.3 0.4 0.5

ADDITIVE COVERAGE (ML)

Fig. 9. Variation of the saturation coverage of (a) CO and (b) H 2 on Ni(100) with the precoverage of CI, S and P (from ref. [63]).

these impurities as shown in fig. 9 indicate that a single impurity atom can successfully poison more than just its nearest-neighbor nickel atoms. This type of poisoning supports an interaction that is primarily electronic in nature.

The experimental results [61-63] indicate that the presence of electronegative CI, S and P atoms causes a reduction of the adsorption rate, the adsorption bond strength and the capacity of the Ni(100) surface for CO and H 2 adsorption. In general, these poisoning effects become stronger with increasing electronegativity of the adsorbed impurity.

Kinetic studies [61-63] have been carried out for CO methanation over Ni(100) surfaces covered with chlorine, sulfur and phosphorus impurities. Fig. 10 shows the rate of CO methanation as a function of sulfur and phosphorus coverage over a Ni(100) catalyst at 120 Torr and a Ha~CO pressure ratio equal to 4. In the case of CI overlayers no change in the methanation rate was observed. At the reaction conditions of fig. 10, H 2 reacts with the adsorbed CI forming HCI. This reaction "cleans" the Ni(100) surface and prohibits the study of the effects of CI on the methanation kinetics. The results presented in fig. 10 correlate very


SULFUR ON Ni(100) SULFUR ON Ni(lOO) AT 0.25 AT 0.5 MONOLAYERS MONOLAYERS

(a) (b)

Z 10"1 PHOSPHOROUS i 114

0 10 "2 ~ 1 , I I I l I I I

0 0.1 0.2 0.3 0.4 0.5

ADATOM COVERAGE IN MONOLAYERS

Fig. 10. Plot of the ra te of m e t h a n a t i o n (CH 4 m o l e c u l e s / s i t e . s ) over a su l f ided and phosph ided Ni(100) catalyst at 120 Tor r and a p ressure ra t io of H 2 to C O of 4 ( f rom ref. [63]).

well with the studies on the effects of S and P overlayers upon the adsorption of H 2 and CO on Ni(100) (refs. [61-63] and fig. 9).

Fig. 10 shows a non-linear relation between the sulfur coverage and the methanation rate. A steep drop in catalytic activity is observed at low sulfur coverages, and the poisoning effect maximizes quickly. A similar reduction of methanation activity by sulfur poisoning has been observed for alumina-supported nickel catalysts [67]. The initial attenuation of catalytic activity by sulfur suggests that ten or more equivalent nickel sites are deactivated by one sulfur atom. There are two possible explanations for this result: (1) an electronic effect that extends to the next-nearest-neighbor sites or (2) an ensemble effect, the requirement being that a certain number of surface atoms is necessary for a reaction to occur. If extended electronic effects are significant, then the reaction rate is expected to be a function of the relative electronegativity of the poison. In contrast, if an ensemble of ten nickel atoms is required for the critical step of methanation, then altering the electronegative character of the poison should produce little change in the poisoning of the reaction. Substituting phosphorus for sulfur (both atoms are approximately the same size [66]) results in a marked change in the magnitude of poisoning at low coverages as shown in fig. 10. Phosphorus, because of its less electronegative character, effectively poisons only the four nearest-neighbor metal atom sites. These results support the conclusion that extended electronic effects do play a major role in catalytic deactivation by sulfur.

22 J.A. Rodriguez and D. HI.. Goodman

1 . o ,

0 . 8

uJ I - I~: 0 .6 uJ > m I-- __1~0.4 w n,-

0 0

0 .2

I I I

R U T H E N I U M

I I I

O . 1 O . 2 O . 3

i 1.0

0 . 8

~. 0 . 6

~__j~O.4

n,.

0.2

o

I I I I

R H O D I U M

0 .4 o.s 0 0 .1 0 .2 0 . 3 0 . 4 o .s

S U L F U R C O V E R A G E ( m o n o l a y e r ) S U L F U R C O V E R A G E ( m o n o l a y e r )

Fig. l I . Methanation rate as a function of sulfur coverage on: (a) Ru(001) (from ref. [68]), and (b) R h ( l l l ) (from ref. [58]) catalysts. Pressure 120 Torr, H 2 : CO = 4, reaction temperature 600 K.

3.2.1.2. Atomic sulfur on Ru(O01), Rh(111) and W(llO). Fig. 11 presents the effects of sulfur coverage on the rate of CO methanation on Ru(001) and R h ( l l l ) catalysts [58,68]. As for the case of Ni(100), a precipitous drop in the catalytic activity is observed for low sulfur coverages. The initial changes in the rates suggest that more than ten Rh or Ru atom sites are deactivated by one sulfur atom.

4) t~ r r ¢.. .o tJ ¢0 4) rr"

. 4

4) r r

1 .0 / O 750 K --e--Reaction Temperatures

A 7 0 0 K PH2 = 100 T o r r

0 .8 Q 600 K PCO : 1 Torr

0 .4 --~ " , , , , r~

0 .2 - - i /

x %

0 .0_ , , ~ 0 .0 0.1

Inhibition of Methanation over W (110) by Adsorbed Sulfur

x Similar Data for Ni (100) 600 K PH2 = 96 Torr

Pco = 24 Torr

[ " i i. ~ - ~ . ~ - ~ 0 .2 0 .3 0 .4 0 .5 0 .6

Sulfur C o v e r a g e

Fig. 12. Relative change in methanation rate over W(110) as a function of sulfur coverage (from ref. [69]). For comparison the data for Ni(100) are also shown (from ref. [63]).


Kinetic data for sulfur-covered W(ll0) surfaces indicate that the activation energy of the methanation reaction does not change with sulfur coverage [69]. In this respect tungsten is similar to nickel [61]. Sulfur decreases the rate of CO methanation on W(ll0) [69]. In fig. 12 the relative change in rate is plotted as a function of sulfur coverage. For comparison, the data for the Ni(100) surface [61] are included. While sulfur clearly exhibits long-range effects on nickel, the operation of long-range effects over tungsten occurs only at the lowest coverages. At very low sulfur coverages (~ 0.03 ML), the decrease in activity is quite steep and extrapolates to between 10 and 12 atom sites deactivated per sulfur atom adsorbed [69]. Apparently the adsorption of sulfur occurs initially in a random, disordered fashion, so that there is little overlap of the inhibiting effect of sulfur atoms on open sites. As the sulfur coverage increases, clustering into islands occurs [69]. Thus, the inhibiting effect of additional sulfur atoms is diminished due to overlap with the effect of previously adsorbed sulfur atoms.

3.2.2. Electropositive impurities We have discussed above the role of electronegative impurities in poisoning Ni(100),

Ru(001), Rh(111) and W(110) toward methanation activity. These results have been ascribed, to a large extent, to an electronic effect. In the context of this interpretation it is expected that an electropositive impurity might have the opposite effect, i.e., increases the methanation activity of a metal surface. A study of CO hydrogenation over potassium-covered Ni(100) [63,70] has shown that this is not the case, although certain steps in the reaction mechanism are strongly accelerated by the presence of the electropositive impurity.

3.2.2.1. Potassium on Ni(lO0). Fig. 13 shows kinetic measurements of CO methanation over a Ni(100) catalyst containing well-controlled submonolayer quantities of potassium adatoms [70]. These data indicate a decrease in the steady-state rate of methanation with potassium coverage. A coverage of about 0.22 ML of potassium would be sufficient to terminate the reaction completely. The presence of K did not alter the apparent activation energy associated with the kinetics, as shown in the Arrhenius plot of fig. 13b. However, the potassium did change the steady-state coverage of active carbon on the catalyst. This carbon level changed from 10% of a monolayer on the clean catalyst to 30% of a monolayer for a catalyst covered with 0.1 ML of potassium [70].

As shown in fig. 14, adsorbed potassium caused a marked increase in the steady-state rate and selectivity of Ni(100) for higher hydrocarbon (MW > 16) synthesis [70]. At all the temperatures studied, the overall rate of higher hydrocarbon production was faster on the potassium-doped surfaces, so that potassium may be considered a true promoter with respect to this reaction, Fischer-Tropsch synthesis. The effects of potassium upon the kinetics of CO hydrogenation over Ni(100) (i.e., a decrease in the rate of methane formation and an increase in the rate of higher hydrocarbon production) are similar to those reported for high-surface- area supported Ni catalysts [71,72]. This agreement between bulk, single-crystal Ni and supported Ni indicates that the major mechanism by which potassium additives alter the activity and selectivity of industrial catalysts is not related to the support material, but that it is rather a consequence of direct K-Ni interactions.

Adsorbed potassium causes a marked increase in the rate of CO dissociation on a Ni(100) catalyst [70]. The increase of the initial formation rate of "active" carbon or carbidic carbon via CO disproportionation is illustrated in fig. 15. The relative rates of CO dissociation were determined for the clean and potassium-covered surfaces by observing the growth in the carbon Auger signal with time in a CO reaction mixture, starting from a carbon-free surface. The rates observed in fig. 15 are the observed rates of carbon formation extrapolated to zero


80ILK 700K 6110K 5001(4501( I i I i I I I ' I

® .,.® 1.O - ~ H 2/I[:O=4/1

O0 0.05 0.10 0.]5 0.20 sO d I I I I I I I I I I 1 1.2 1.4 1.8 1.8 2/) 2,2 I~TASSUU ~ ImJ

w r • l ~ a c - ~

Fig. 13. (a) Relative methanation rate as a function of potassium coverage at various reaction conditions: (~) Pco = 1.0 Torr, PH: = 99.9 Torr, T = 600 K; (O) Pco = 24 Torr, Prt 2 = 97.6, T = 538 K; ( zx ) Pco = 24 Torr, PH2 = 96 Torr, T = 600 K; (O) Pco = 24 Torr, PHz = 96 Torr, T = 594 K (from ref. [70]). (b) A comparison of the rate of methane synthesis over a clean single-crystal Ni(100) catalyst with the rate over a potassium-doped catalyst. Total

reactant pressure is 120 Torr, H 2 :CO = 4 (from ref. [70]).

93%

J ~ 22% . . ~

METHANE ETHYLENE ETHANE C3 +

Fig. 14. Comparison of the product distributions (weight percent) observed for clean and K-doped catalysts at T = 500 K, H 2 :CO = 4, and a total pressure of 120 Torr. Potassium coverage is 0.10 ML (from ref. [70]).


Fig. 15. Relative initial

Z _o I-.

0 M. Z o m n . ,¢c u UJ >

o < [

U. o UJ I-- < : n-"

UJ > l - -

,.J

4

3

2

14

I I I I I

0 .02 . 04 .06 .08 . I 0

POTASSIUM COVERAGE (ML)

rate of reactive carbon formation from CO disproportionation as a function of potassium coverage. Pco = 24 Torr, T = 500 K (from ref. [70]).

carbon coverage. The presence of K adatoms leads to a reduction of the activation energy of reactive carbon formation from 96 kJ mo1-1 on clean Ni(100) to 42 kJ mo1-1 on a 10% potassium-covered surface [70].

In spite of increasing the rate of CO dissociation or carbide build-up, potassium decreases the overall rate of methanation. This reduction in methanation activity must be related to a poisoning of either the hydrogen adsorption or the hydrogen addition steps [70]. The enhancement of steady-state carbide coverage caused by potassium favors C-C bond formation and the synthesis of heavy hydrocarbons (MW > 16).

3.2.3. Related theory CO is generally thought to be adsorbed on transition-metal surfaces by the Blyholder

mechanism [73,74], which involves tr-donation of electron density from CO into the unoccupied metal orbitals and rr-back-donation of electron density from occupied metal orbitals into the lowest unoccupied molecular orbitals (2zr*) of the CO molecule. The mechanism is similar to that observed for CO bonding in transition metal compounds [75]. Theoretical results with the constrained space orbital variation (CSOV) method show that, in this type of synergistic bond with o'-donation/~--back-donation, 7r-back-donation is energetically more important in determining the character of the bond than is tr-donation, at least for the case of Cu surfaces [76-78]. Recently, inverse photoemission results have supported the predominant importance of 27r* back-donation in CO chemisorption on Pd and Ru [79]. The thought that the antibonding 27r orbitals of CO are populated upon adsorption is consistent with the results of HREELS [80,81], which show that the C-O stretching frequency and the C-O force constant of CO adsorbed on metals are lower than those of free CO.

The standard picture used to describe the effects of electron-transferring species upon CO chemisorption is an extension of the basic Blyholder model. Electropositive impurities donate charge to the metal. This excess charge is partially accommodated in increased 7r-back-dona-


tion to the 2~-* orbitals of CO. This increases the meta l -CO bond strength, while decreasing the C - O bond strength. Opposite effects are expected for coadsorption with electronegative impurities. In a few coadsorption cases, this general picture has been to some extent substantiated by calculations with different quantum-chemical methods [82-89].

Theoretical work has been undertaken to address directly the predicted magnitude of the near-surface electronic perturbations by impurity atoms. Early work was concentrated on the indirect interactions between adsorbates which occur via the surface conduction electrons [90-92]. These calculations suggested that atomic interactions through several lattice spacings can occur. Recent theoretical studies have expressly addressed the surface electronic perturbations by sulfur [93] as well as by C1, P and Li [94]. The sulfur-induced total charge density vanishes beyond the immediately adjacent substrate atom site. However, the Fermi level density of states, which is not screened, and which governs the ability of the surface to respond to the presence of other species, is substantially reduced by sulfur even at non-adjacent sites. The results for several impurities indicate a correlation between the electronegativity of the impurity and its relative perturbation of the Fermi level density of states [93,94], a result which could be very relevant to the poisoning of CO methanation by S and P as discussed above. Finally, an alternate model, which produces the same final results, involves an electrostatic, through-space (as opposed to through-metal) interaction between the charge distribution of the coadsorbed species [95,96].

Both sets of theories, that is, "through-metal" or "through-space", are consistent with adsorbate perturbations sufficiently large to effect chemically significant changes at next- nearest-neighbor metal sites. This perturbation length is sufficient to adequately explain the observed poisoning of catalytic activity by surface impurities discussed above.

3.3. Metal-support interactions and CO methanation

The early concept of a support or a carrier was that of an inert substance that provided a means of spreading out an expensive catalyst over a large surface area. However, the support may actually modify the activity of the catalyst, depending upon the reaction and reaction conditions [10,97,98]. The oxidic support materials (e.g., La2 0 3, Cr203, ZnO, MgO, TiO 2, ZrO 2) can favorably or adversely influence the performance of a metal in a particular catalytic process [97,98]. Titania (TiO 2) is a typical example of an "interacting" support [97,98].

Surface science methods have been applied to study the CO hydrogenation activity of nickel overlayers on the TiO2(100) surface [99]. Results of ultraviolet photoelectron spectroscopy indicate that there is an electron transfer from TiO2(100) to Ni when Ni is deposited onto a reduced TiO2(100) surface [99]. When the Ni/TiO2(100) surface was used as a methanation catalyst, the CH 4 yield varied as a function of the Ni coverage, as shown in fig. 16a. At a temperature of 190 o C, an average Ni thickness of ~ 5 .& gave optimum activity. An Arrhenius plot of the specific rate of methane formation over the 5 A Ni-covered TiO2(100) surface is included in fig. 16b. The methane yields from the Ni/TiO2(100) catalyst are 3.3-3.7 times that from a pure Ni(111) catalyst. The apparent activation energy for methane production over the Ni/TiO2(100) surface (105.5 ± 2.5 kJ mo1-1) is very close to that seen over Ni(111) (111.8 __ 3.8 kJ mo1-1) [99].

A study of the methanation activity of a Ni(111) surface containing controlled amounts of TiO x (x ~ 1.0-1.5), showed that the nickel catalyst is optimally promoted at a titanium coverage of ~ 0.1 monolayer, displaying an activity and product distribution similar to those seen over 8 A Ni-covered TiO2(100) catalysts [100]. The fact that T i O J N i ( l l l ) and


| I I I [ I i I "~

~- 3 ~ N i / T l O z (1001

/ ,,/CO=6OT,,,/ZOT.,

,

m~ V I I I I l I I //.* 0 8 16 2 4 BULKM

NI THICKNESS ( A )

i0"1

I0 ~ z

uJ > 0

I0 -~

I0' 1.8

T ( C ) 2 6 0 2 4 0 2 2 0 2 0 0 180

i i i 1 r I , ,

H 2 / C O = 6 0 T o t r / 2 O T o r r " ~ ( I 0 0 }

Q N i ( l l l ) -

I I I I 2.0 2.2

1 0 3 / T ( K )

Fig. 16. (a) Methane yield from Ni/TiO2(100) as a function of the average Ni thickness. PH2 = 60 Torr, P¢o = 20 Tort, T = 190 ° C (from ref. [99]). (b) Arrhenius plots of the methane turnover number (CH 4 molecules / s i te -s ) for

N i ( l l l ) and 5 ,~, Ni/TiO2(100), PH = 60 Torr, Pco = 20 Torr (from ref. [99]). 2

Ni/TiO2(100) are so similar in their catalytic behavior, suggests [100] that in high-surface-area Ni/TiO 2 catalysts, TiO x species diffuse from the support material to the nickel. Dispersal of the oxide on the metal and the formation of nickel-titanium bonds modify the catalytic properties of the surface.

4. CO oxidation

4.1. CO oxidation by 0 2

The catalyzed oxidation of CO by 0 2 (2CO + 02 --> 2CO2) is an important process in the pollution control of combustion products. Currently, the removal of CO as CO 2 from automobile exhaust is accomplished by catalytic converters which employ a supported Pt, Pd and Rh catalyst. The relative simplicity of CO oxidation makes this reaction an ideal model system of a heterogeneous catalytic reaction. Each of the mechanistic steps (adsorption and desorption of the reactants, surface reaction, and desorption of products) has been probed extensively under UHV conditions with the modern techniques of surface science [101].

At UHV pressures, the reaction proceeds through a Langmuir-Hinshelwood mechanism involving adsorbed CO and O. The surface of the catalyst is almost entirely covered by CO, and the reaction rate is determined by the rate of CO desorption. The observed activation energy is close to the binding energy of adsorbed CO. Depending on the relative partial pressures of the reactants, self-sustained kinetic oscillations have been observed for CO oxidation on Pt [102-104] and Pd [105] single-crystal surfaces that show reconstructive phase transitions. Many of the reaction parameters determined in UHV can be applied directly to the kinetics at higher pressures [106]. In this section, we review studies dealing with the catalytic oxidation of CO by 0 2 over Rh, Ru, Pt, Pd and Ir single crystals under high-pressure reaction conditions.

4.1.1. Rh(lO0) and Rh(l l l ) surfaces Fig. 17a compares the CO oxidation rates measured over single crystals of Rh with those

observed on supported Rh/AI20 3 catalysts [106,107]. It is clear that there is an excellent


1 0 0 0

U. 100 0 I -

UJ I - 10

¢C

Z 1 0 m I - <~ ~; 0.1 n" o 14.

t~ 0 . 0 1 o

0.001

0 . 0 0 0 1 I t 1.2 1.6

I |~i~ I I I I I

\ ~ onh I I 1 1 )

• Rh 1100)

I PCO = 8 T o . Rh/AI2031203 ~ P o 2 = 8 Torr

l t I t t

I t Rh/A1203~]

I I I I I t l 2.0 2.4 2.8

1 / T x 1 0 3 (K)

100.0 i • - i . . . . ! i • • i . . . .

_ \ ~) SO.O PO 2 = 8 Ton"

~ • : Rh (100); T=SISK

20.0 ~ 5 0 0 K

~) 10.0

5 . 0

2

° ® ", U ~ . O

1.0 * • - : • • • , I I , . n 1.0 2.0 5.0 10.0 20.0 $0.0 100.0

CO PARTIAL PRESSURE (Tort)

10"

Z o I - : i

B u

10 =

101

lo o Io ° lO =

PCO=8 Tort; T= 500K • : Rh (100) Q f O ~ O

@ J i

1 0 1 l O ~

0 2 PRESSURE (Torr )

Fig. 17. (a) Comparison of the specific rates of the CO+O 2 reaction measured over Rh(100), Rh(l l l ) and Rh/AI203 (from refs. [106,107]). (b) Measured rates of the CO + 02 reaction over Rh(100) and Rh(ll l) at various partial pressures of CO, an O z partial pressure of 8 Torr and temperatures of 515 and 500 K (from ref. [107]). (c) Rates of CO oxidation over Rh(100) and Rh(l l l ) at various partial pressures of 02, a CO partial pressure of 16 Torr

and a temperature of 500 K (from ref. [107]).

a g r e e m e n t be tween the mode l and s u p p o r t e d systems in bo th the specif ic r eac t ion ra tes and a p p a r e n t ac t ivat ion energ ies (109 kJ m o l - t for R h ( l l l ) and Rh(100) versus 125.5 kJ m o l - for R h / A 1 2 0 3 ) . These resul ts indica te tha t the kinet ics of C O oxida t ion on R h is not sensi t ive


to changes in catalyst surface morphology. Under the conditions of fig. 17a, the surfaces are predominantly covered with CO so that the reaction is limited by the adsorption rate of oxygen [106,107]. As the temperature is increased, the reaction rate increases because more vacant sites are available for oxygen adsorption as a result of the higher CO desorption rate. In fig. 17a the CO oxidation rate increases with temperature following an apparent activation energy very similar to that for CO desorption.

Figs. 17b and 17c show the dependence of the rate of CO oxidation over Rh(100) and R h ( l l l ) on the partial pressures of CO and 0 2 [106,107]. The behavior of the reaction can be analyzed by using a kinetic model established from surface science studies of the interactions of CO and 0 2 with Rh [106]. In this model the mechanism of CO oxidation follows the sequence

COg ~- COa, (15)

O2.g ~- 2Oa, (16)

CO a + O a ~ CO2.g. (17)

The corresponding expression for the rate of CO oxidation is

2k%, Po__z rco 2 = kc o l kco j (T ) e c o , (18)

where kco ., ko." and kco~(T) are the rate constants for adsorption and desorption of CO and 0 2. Eq. (18)"is in excellent agreement with the experimental results of figs. 17b and 17c. The reduction in catalytic activity seen in fig. 17c for Po,. > 200 Torr is a consequence of an increase in the coverage of atomic oxygen on the R h ( l l l ) surface [107]. For these cases, it appears that appreciable coverages of oxygen effectively poison the adsorption of CO.

4.1.2. Ru(O01) surface Specific rates of CO oxidation (molecules of CO 2 produced per surface site per second)

over Ru(001) are plotted in Arrhenius form in fig. 18a [108]. The good agreement between the results on the unsupported single crystal and those obtained on the supported catalysts attest to the appropriateness of the model. Linear Arrhenius behavior is observed below ~ 550 K, yielding an activation energy of 81.6 kJ mol-1.

The effects of variations in the partial pressures of CO and 0 2 upon the reaction rate are complex (see fig. 18) [108]. At low CO partial pressures, the rate dependence is approximately positive first order in CO. This changes to negative first order at CO partial pressures above ~ 16 Tort. At low oxygen partial pressures (fig. 18c), the oxidation reaction is positive order in O z. The slope of the curve changes for 0 2 pressures above ~ 4 Torr, becoming zero order in 0 2 . Post-reaction surface analysis indicates that the optimum rate of CO oxidation on Ru(001) is observed when the surface is covered by almost a monolayer of oxygen [108]. This contrasts with the case of Rh catalysts, on which the optimum activity was obtained for surfaces essentially free of oxygen adatoms [107].

4.1.3. Pt(lO0) surface The CO 2 formation rate over Pt(100) as a function of inverse temperature is shown in fig.

19 and compared to data obtained on a P t / S i O 2 catalyst [109]. The structure insensitivity of the reaction is evidenced in the match between the supported, nonoriented catalyst and the single crystal. For Pt(100) the CO oxidation reaction is always positive first order in 0 2


1 0 0 . 0 0 Lk o I--

14.1 I-. 1 0 . 0 0

n- Z 0 F-

1.00

n- O i i

¢N 0 0.10 0

0.01

1.6

l I I I I I I I I I I

8 Torr

® I t I 1 [ I I I I 1

1.8 2.0 2.2 2 .4 2 .6 2.8

1 / T x 1 0 3 (K)

50

Io e-

z 5 o

:E n- O M. 04 1

0 U

I I I I

Ru(O00 I ) PO,2= 8 T o r t

® I I I I

5 10 5 0 100 51

CO PARTIAL PRESSURE (tort)

500

u.

0

er 0

(J

50.0

10.0

5.0

1.0

0.5

0 , '

i i i ' i i !

Ru(00OI)

,co: g2o,, / / @

I | I I I I

0.5 1.O 5.O 10.O 50.0 1OO,O 500 .0

0 2 PARTIAL PRESSURE ( to r t )

Fig. 18. (a) Arrhenius plot of the rate of CO 2 formation over Ru(001) and R u / S i O 2 catalysts (from ref. [107]). (b) Effect of CO partial pressure on the rate of CO oxidation on Ru(001), Po , = 8 Torr, T = 500 K (from ref. [107]). (c)

Effect of 0 2 partial pressure on the rate of CO oxidation on Ru(001), Pco = 16 Torr, T = 500 K (from ref. [107]).

1 0 0 0 L I ~ I L

,,,00 \ \ \ 0 2 = 8 To r r

fl: ,t(loo,--~ _o 1 l--

\ t k \ I ~ t r i l l 1 ) &

0 X \

O.Ol d 0 5~ ~/s,o~ ~,.

1 (3

0 . 0 0 1 I I I I I ! . 4 1 .6 1 .8 2 .0 2 .2 2 , 4 2.11

1 / T x 10 3 (K)

Fig. 19. Specific rates of CO 2 formation as a function of inverse temperature for single-crystal and supported catalysts (from ref. [109]).


1OO

60

U. 0

4( a. 6

z O

,<

a. O 1 ik

U

0 .1 0 .1

i I I I I co ORDER= -0.9 Pt(tO0) • Po z = 1 0 0 tor t , 4 8 0 K ~ _ ~ o Po l=40 tor t , 8 0 0 K

POs: 8 ton', eBO K = "

c o ORDER = O

I I J I 0 . 6 1 6 1 0

C O P A R T I A L P R E S S U R E ( t o r r )

® I

B0 1 0 0

u. O

UJ I - CC

Z

2 I-

I

0 L IN

0 U

lOO

80

lO

6

1

o.8

i i i i i pt(lO0) • PCO:16 ton', 650 K _ ~ (3 Pco:6 ton', BSOK o Pco:12 to~, 67S K ~Y" ~ " A Pco=6 tort, 600K : / m Pco = 3 t~r, 47B K

/

/ ® I l t I I 6 10 60 100 500 1000

0 2 P A R T I A L P R E S S U R E ( t o r r )

Fig. 20. (a) 0 2 and (b) C O par t ia l p ressure d e p e n d e n c e of C O oxida t ion over Pt(100) at several t e m p e r a t u r e s ( f rom

ref. [109]).

pressure up to an 0 2 : CO ratio of 150 (fig. 20a). The reaction order in CO changes from - 0.9 at 650 K to ~ 0 at 450 K (fig. 20b). A variation in activation energy from 138.5 to 54.4 kJ mo1-1 is also observed in this temperature regime [109]. This may indicate a change in reaction mechanism from the CO desorption controlled Langmuir-Hinshelwood mechanism that controls the kinetics at high temperature.

The investigation of kinetic oscillations in the catalytic oxidation of CO over Pt surfaces has attracted considerable attention during the past years. U H V studies indicate that under properly chosen conditions the rate of CO oxidation at a clean and well-defined Pt(100) surface exhibits sustained temporal oscillations, which are reflected in periodic changes of the work function [102]. These oscillations are associated with periodic variations of the surface

32 J.A. Rodriguez and D. IV.. Goodman

structure. The clean Pt(100) surface reconstructs forming a hexagonal arrangement of atoms in the topmost layer. If the adsorbate coverage on this hexagonal phase exceeds a critical value, the reconstruction is removed. The driving force of the kinetic oscillations is basically the change in adsorption properties as the surface switches from one configuration into the other [102].

This mechanism cannot be used to explain the presence of kinetic oscillation on Pt surfaces, which do not reconstruct under high-pressure conditions [110,111]. An alternative model that involves the formation of an oxide in the near-surface region of the catalyst has been proposed [110]. In the two branches of the CO oxidation reaction, the surface is oxidized or reduced, thus changing the surface oxide coverage. This coverage determines in which of the two branches the reaction will occur [110]. The main defect of this model is the fact that Pt oxides are unstable under reaction conditions [109]. Studies of CO oxidation over Pt(100) indicate that no strongly bound, deactivating surface oxygen species is formed [109].

Oscillatory reaction rates have been observed for CO oxidation over P t ( l l l ) , Pt(100) and Pt(13,1,1) catalysts at atmospheric pressure [111]. It appears that silicon impurities were present on the surfaces supporting the oscillations. These impurities play an important role in the oscillatory behavior by either catalyzing the formation of a Pt oxide or by increasing the sticking coefficient of oxygen on the Pt surface [111].

4.1.4. Pd(llO) surface Fig. 19 also displays steady-state specific rates for CO oxidation over Pd(110) and P d / S i O 2

catalysts [109]. For the single crystal the apparent activation energy is 138.6 kJ mol - I at temperatures in the range between 475 and 625 K. At Po:/Pco ratios lower than 12, the CO oxidation reaction on Pd( l l0 ) is negative first order in CO pressure and positive order in 0 2 pressure, in good agreement with supported catalyst data [109]. The reaction becomes negative first order in 0 2 pressure and approximately first order in CO pressure for P%/Pco ratios greater than 12 [109]. Under such oxidizing conditions, a strongly bound oxygen species is formed on the Pd( l l0) catalyst. This strongly bound oxygen species decomposes in post-reaction TPD yielding CO 2, suggesting the formation of a CO x species during the reaction [109]. Deliberate oxidation of the Pd surface prior to the CO + 0 2 reaction resulted in a rate decrease but did not affect the activation energy significantly, indicating that the oxide served merely as a simple site blocker [109].

Reaction rate oscillations have been observed for CO oxidation on Pd( l l0) [105]. The existence region for oscillations was determined for pressures ranging from 10 -3 to 1.0 Torr and depended on the pressures of O z and CO and the sample temperature. A densely packed CO structure and strongly bound oxygen or subsurface oxygen limit the reaction at high and low rates, respectively [105]. The extension of oscillations from UHV conditions to the 1.0 Torr pressure region indicates that the mechanism responsible for the oscillations may be independent of the reactant pressure up to several Torr.

4.1.5. lr(llO) and Ir(lll) surfaces The structure insensitivity of the CO oxidation reaction is evidenced in the results of fig. 19

[109]. There is a match between the rates observed on I r ( l l0) and I r ( l l l ) single crystals and those for a supported Ir catalyst. Fig. 21 shows the reaction rate dependence on CO and 0 2 partial pressures for I r ( l l 1) [109]. Stoichiometric mixtures (Pco/P% = 2) have reaction orders of - 0 . 9 + 0.2 in CO and 0.9 + 0.2 in 0 2. Under highly oxidizing conditions, I r ( l l l ) shows negative-order dependence in 0 2 partial pressure, indicating the presence of a strongly bound


1000

SO0

lOO

50 Z

s U

I I I I I

It(1 1 1) 0 Pco :4 Iorr, SSO K c~ P c o = 6 tort . 4 8 3 K

1 I I I I s lO so lOO soo lOOO

0 2 P A R T I A L P R E S S U R E ( t o r t )

100

0 SO t

< E

Z 0 1 0

• S E 0 U.

1 0.1

: I I I I

• Ir(111) PO 2 =20 torr

I I I I I O.S 1 8 10 50 100

C O P A R T I A L P R E S S U R E ( t o r r )

Fig. 21. D e p e n d e n c e on C O and 0 2 par t ia l p ressures of C O oxida t ion on Ir(111) ( f rom ref. [109]).

oxygen species. This species reduces the overall rate of the reaction and can be detected as CO 2 desorbing at ~ 670 K in post-reaction temperature-programmed desorption [109].

4.1.6. Summary The results discussed above demonstrate the structure insensitivity of the CO oxidation

reaction. The observed activation energies and reaction orders are summarized in table 2. For CO + 0 2 mixtures close to stoichiometric, in general, the reaction rate is independent of total pressure, first order in 02 pressure and negative first order in CO pressure. This behavior is consistent with a Langmuir-Hinshelwood mechanism involving adsorbed CO and O, with the reaction rate governed by the desorption of CO [106]. On Rh, Pd, and Ir catalysts, increases in the 0 2 :CO ratios result in an eventual decrease in the rates of CO oxidation, as a consequence of the formation of an oxide-like species over the surfaces. In contrast, on Ru(001) the oxide is substantially more active than the clean substrate.

34

Table 2 CO oxidation by 0 2

(a) Activation energies

J.A. Rodriguez and D.W. Goodman

Catalyst Pco Po2 T T O F a) Ea Ref. (Torr) (Torr) (K) (kJ m o l - ])

Rh(100), 8 8 450-600 0.1 -1000 108.8 [107] Rh(111) Ru(0001) 16 8 380-550 0.05- 50 81.6 [108] Pt(100) 16 8 500-725 0.1 - 500 137.7 [109] Pd(110) 16 8 475-625 0.05- 400 138.6 [109] Ir(ll0), 16 8 425-625 0.02- 100 92.1 [109] I r ( l l l )

(b) Reaction orders b)

Catalyst T (K) in CO in 0 2 Ref.

Rh(1 l l ) 500 ~ - 1.0 ~ 1.0 [107] Ru(000]) 500 ~ 0.0 [108] Pt(100) 650 -0.9_+0.2 1.0+0.1 [109] Pd(110) 525 - 1.0+0.1 1.0_+0.1 [109] I r ( l l l ) 525 - 0 . 9 + 0 . 2 0.95:0.2 [109]

a) TOF = turnover frequency = CO 2 molecules /metal-surface-s i te , s. b) Pco = 16 Torr, Po2 = 8 Torr.

1 0 3 I I I i

CO + N O - - - ~ CO2 + 1 / 2 N2 PCO = PNO=8 Torr

• : Rh (100) O : Rh (111)

10 2 kl. 0 F"

W I--

Z 0 101 I'-

n" 0 U.

0 (.1 10 0

10-11 i i I I I • 1.4 1.5 1.6 1.7 1.8 1.9 2.0 2.1

1 / T X 1 0 3 (K "1 )

Fig. 22. Arrhenius plot of th~ CO 2 formation rate from the CO + NO reaction on R h ( l l 1) and Rh(100). / )co = PNO = 8 Torr (from ref. [112]).


4.2. CO oxidation by NO

4.2.1. Rh(lO0) and Rh(111) surfaces Supported Rh catalysts are commonly used for controlling pollution from combustion

products such as CO and NO. CO oxidation (and NO reduction) occurs mainly through the overall reaction

CO + NO ~ CO 2 + N 2 . (19)

This catalytic process has been the subject of a large body of research in recent years [106,112,113].

Arrhenius plots for CO oxidation by NO on Rh(100) and R h ( l l l ) are shown in fig. 22 for equal partial pressures of CO and NO (8.0 Torr) [112]. Significantly different activation energies (Rh( l l l ) : 121.4 kJ mol-~; Rh(100): 100.5 kJ mol -~) and specific activities are observed on the twO surfaces, indicating that this reaction is sensitive to the geometric structure of the catalyst surface [112]. The CO + NO reaction exhibits substantially different kinetic behavior between supported R h / A I 2 0 3 and Rh(111) [106].

The dependence of the reaction rate on the partial pressures of the reactants is shown in fig. 23 [112]. The data indicate little if any variation of the rate with changing CO pressures on either surface. Similar NO pressure independence is observed over the two surfaces for NO

~' 2 .0 0 I--

MJ 1.0 I-- <¢ r r

~ 0 .5

I-" .<

r r O 0 .2 1.1.

N

0 0 . 1 (J 1.0

I I I I I I

Rh ( 1 0 0 ) m • I I 515K •

NO+CO Pco = " 81/2 Ton" N2 + C02 j

5 (011K 1)

0 I I I I I I

2 .0 5 . 0 1 0 . 0 2 0 . 0 5 0 . 0 1 0 0 . 0 2 0 0 . 0

NO PRESSURE (Torr)

A

u. 2 . 0 O I - , , . . , ,

uJ I - 1-01 <¢

z o.5 O I-- .<

iv- 0 .2 O u.

O 0.1 ¢-) 1.0

I . i T , , . I I , , i . . . . i

Rh ( 1 0 0 ) • ,

m515K • N 0 + C 0 " - - - ' - 1 / 2 N 2 + C O 2 - P"o =8 Ton.

8 R h ( 1 1 1 ) O 8 o ©

5 0 0 K

I • 1 , , , , I I , I . . . . I

2 . 0 5 . 0 1 0 . 0 2 0 . 0 5 0 . 0 1 0 0 . 0 2 0 0 . 0 CO PRESSURE (Torr)

Fig. 23. NO and CO partial pressure dependences of the CO + NO reaction on Rh(111) and Rh(100) catalysts (from ref. [112]).


pressures below 20 Torr. On Rh(l l l ) , the reaction rate becomes positive order at higher NO partial pressures. This behavior is in excellent agreement with the predictions of a kinetic model based on the following set of elementary processes [106]:

COg ~- COa, (20)

NOg ~- N O a , (21)

NO n --* N, + O a, (22)

NO. + N a --* N 2 + O a ( t%N2) , (23)

2 N a ~ N2,g (/3-N2), (24)

C O a -1- Oa --9 CO2,g" (25)

For Rh(111) the model suggests that the surface is predominantly covered by adsorbed N atoms and NO, and the rate is limited by the surface reaction of N a (~-N 2 formation), which frees adsorption sites on the surface for the other reactants [106,112]. The similarity between the activation energy for reaction (24) (~ 130 kJ mol -~) and the apparent activation energy for the overall reaction (~ 121 kJ tool -~) supports this conclusion [106,112]. The reaction on Rh(100) may be rate limited by 6-N 2 formation due to the greater activation energy for/3-N 2 formation on this surface [112].

5. Ethylene epoxidation

The catalytic oxidation of ethylene to ethylene epoxide (2C2H 4 + 0 2 ~ 2C2H40) is a several billion dollar per year industry [114], providing the necessary intermediate in the synthesis of ethylene glycol, which is used in polyester and antifreeze production. The most common catalyst consists of reduced silver particles dispersed on a-Al203 [115,116]. Cesium is added to the catalyst as a promoter. In addition, chlorinated hydrocarbons are usually present in the reactant feed in order to improve the selectivity of the catalytic process toward ethylene epoxide production. In the presence of oxygen, ethylene can react to yield ethylene epoxide or undergo total combustion producing CO 2 and H20,

kl C2H 4 + O 2 > C2H4O

~ (26)

C O 2 + H 2 0

The heats of reaction, per mole of ethylene, are - A H 1 = 146 kJ tool-1 and -zaH 2 = 1320 kJ mol-1 [10]. The oxidation of ethylene to ethylene epoxide over Ag is the simplest example of a kinetically controlled selective catalytic reaction.

5.1. Kinetics over Ag(l lO) and A g ( l l l ) catalysts

Recent studies [117] show that turnover rates for ethylene oxidation over the (110) and (111) planes of Ag differ by a factor of only 1.8, a result indicative of structure insensitivity [117,118]. The Ag(l l l ) and (110) single-crystal surfaces are excellent kinetic models of high-surface-area supported Ag catalysts, with virtually identical activation energies and reaction orders with respect to ethylene and oxygen pressures.


-6

3 0

I I I I I I

PEt = 20 torr I k - ~ P 0 2 = 150 torr

I I [ I I I t.6 L7 L8 L9 z.o 2., z.2 z.3

T-I/(IO-3K'I)

Fig. 24. Arrhenius plot of the steady-state rates of CO 2 and ethylene epoxide (EtO) production on Ag(ll0) at PEt = 20 Torr and Po2 = 150 Torr (from ref. [8]). Shown for comparison is the rate of EtO production on a

silica-supported Ag catalyst (form ref. [119]).

Fig. 24 shows steady-state rates of C2H40 and CO z production as a function of temperature, in Arrhenius form, for PEt = 20 Torr and Po2 = 150 Torr on a clean Ag(ll0) surface [8]. Shown for comparison is the specific rate of C 2 H 4 0 production over a silica-supported Ag catalyst (from ref. [119]). The apparent activation energy for C2H40 production, while varying with temperature from 93.8 kJ tool -1 near 450 K to 22.1 kJ tool -t near 580 K, is virtually identical on Ag(ll0) and on the high-surface-area catalyst. The activation energies for C2H40 and CO z production are quite similar, a finding that indicates a common rate-limiting step [8]. The specific activity of Ag(ll0) is about 100 times higher than those of supported Ag catalysts [120]. A simple and therefore appealing explanation for the discrepancy in turnover rates is the contamination of the supported silver by impurities [117,118,120].

Fig. 25a illustrates the effect of O z pressure upon the steady-state rates of C2H40 and CO2 production over Ag(ll0) at 490 K and PEt = 4.1 Tort [8]. At low Po2, the reaction orders in Po2 approach unity. When Po2 exceeds about 8PEt the orders in Po2 begin to approach to zero. The selectivity increases with Po2, reaching a maximum value at Po2 > 8PEt" Very similar trends have been observed for supported Ag catalysts [8]. Also shown in fig. 25a is the coverage of atomic oxygen on the Ag(ll0) catalyst after reaction conditions [8]. Oxygen coverages were measured by rapidly (17-45 s) transferring the single crystal at reaction temperature from a reaction chamber to an UHV chamber for surface analysis. It has been argued that this method underestimates the oxygen coverages during the reaction [121]. No adspecies containing atoms other than C, O, H and CI (0Cl < 0.02) were observable by surface analysis. Measurements of the surface concentration of weakly bound species such as CzH4, a and 02, a was not possible because they desorb during transfer of the crystal into the UHV chamber.

38

2,5

|U)

I 2o

~ 1.5 u

o E

v

7 0 k-

0.5

J.A. Rodriguez and D. IF.. Goodman

I I I 1 I 6 0

Acj (110) 490 K CO 2- TON

- PEt =4.1 lorr ~ _ _ _ ~ 50 05 _ _ _ f - ~ - ~ - . . . . . . s-~,o-

- 40 0.4

F-

E eo . . . . 2 3 ° b 0.3

/ / l ~ - ~ 20 o 0.2

~ I0 0.1

I I 1 1 ~ 0 0 25 50 75 IO0 125 ~50

02 PRESSURE/torr

o

31-, ' ~ ' I ~, S SE,o -

0 IY

Lf f TL' 2::2 lO ISiO z

t / "',/ . - __.---*-- ° / ' , . . ¢ . . • O~ "~- - - •

I @ O f / I I I I I 0 5 IO 15 20 25

PEt / to r r

~SE,o

40%

0,3

30

0.2

Do 20 i1:

0,1 I0

0 - - 0

Fig. 25. Effects of (a) 0 2 and (b) C2H 4 partial pressures on the rates of production of ethylene epoxide (EtO) and CO 2 on Ag(ll0). The selectivity for ethylene conversion into EtO is defined as: SEto=TONEto/(TONEto

+ ½TONco 2) (from ref. [8]).

Fig. 25b displays the effect of e thylene pressure upon the rates of C 2 H 4 0 and CO 2 product ion on A g ( l l 0 ) for Po2 = 150 Torr and T = 490 K [8]. At very low PEt, the react ions are near ly first order in PEt" AS PEt increases, the order decreases and finally drops to Zero


as the surface saturates in ethylene. There is not general agreement about the mechanism for ethylene epoxidation on silver.

Arguments concerning the roles of various forms of oxygen on the surface of silver remain the subject of much debate [115-127]. Recent isotopic tracer experiments [124] indicate that ethylene can react with adsorbed atomic oxygen to give the corresponding epoxide, provided that subsurface oxygen is present. Once the adsorbed atomic oxygen has been formed, the rate-limiting step for epoxidation is the insertion of this oxygen into the C-C double bond [124,125]. The rate-limiting step for combustion is probably a C-H bond activation, although C-C bond activation cannot be definitely excluded [125]. Subsurface oxygen seems to be necessary for adsorbed oxygen to react to epoxide [124]. Quantum-chemical calculations (using model clusters of the Ag(ll0) surface and the Hartree-Fock-Slater Xa method) [125] show that the presence of subsurface oxygen reduces the bond energy between silver and adsorbed atomic oxygen, and converts the repulsive interaction between oxygen adatoms and gas phase ethylene into an attractive one, thus making possible the epoxidation reaction. Experiments using Ag(l l l ) as a catalyst [122] have lent support to the view that atomic oxygen reacts with ethylene to form both ethylene epoxide and CO 2. Oxygen dissolved on Ag(l l l ) promotes selective oxidation [122].

5.2. Chlorine promotion of Ag(llO) and Ag(111) surfaces

Kinetic data for Ag(ll0) [128,129] and Ag(l l l ) [121,130] indicate that adsorbed chlorine reduces the activity and increases the selectivity of these surfaces toward ethylene epoxidation. Typical results are presented in fig. 26. The effects of chlorine addition are very similar on Ag(l l l ) and Ag(ll0). Fig. 27 summarizes the results of a series of experiments where the

I c l / I A g ( + l O -= )

0 4 8 12 16 2 0 4 I I I 1 I

C t / A g AES RATIO (+10 -z } / ' . " ' % . ,

O 4 8 12 16 20 2 4 80

~ = I ~ I ' I = I ' I '

2,O ~ TON¢o = u B °/5'~_ 80

~'- .... SEt 0 a

~ - - - ~ ' ' - - . . J r , a - ' ~ ' ° °- i l l , ) _ ~ ~ ,.5 6o ; _= \ . . o / 49o K ~" ~- '= Q . / - ~ " - , , ~SOt,,, O= ~ w u

~ X "%, 4.1tore et I-- ..I 4 0 /

~= hO - a ~ . - 4 0 ~ o - - -

o ',, ~ ,7, "~ _ - - " \

T-ONI~tO "., ell \ 4 9 0 K 563K

0 0 .5 -- ~ ' o ~ ' \ , -- 2 0 ~ 20 - - - 20 torr Et 4.1 lore Et ~- ( ~ -Xo~=~ ,~ ~ ISo 1o. O= lSo to, o=

O i I i I I I I I ""~ 's: . , . 0 0 , I r I ~ I I 1 I 0 0,2 0 ,4 0 .6 0.8 1.0 0 0,2 0 .4 0 .6 0.8 1.0

ec lec ;' 8ct / Uc t

Fig. 26. (a) Var iat ions in the steady-state reaction rates and selectivity with Cl coverage on Ag(111) at 490 K, 4.1 To r r o f C 2 H 4 and 150 Tore 02 ( f rom ref. [130]). (b) Selectivity variat ions on A g ( l ] ] ) and Ag(110) as a funct ion o f C!

coverage ( f rom ref. [130]).


0.9 (1.

Z 0 .7 @-

h i £3 Cr 0.5 0

Z 0 0 3 V-

U.I n"

O9

Z 07

l.IJ 0.5 a n-

O 03

Z 0 I-- Ol

~J tY

8c~ 0 0.I 0.2 0.3 04 05 0.6 _i I I I I I I u |

/oj

° -J

, , ° j

_ . / O ~ o . . . ~ . _ - - 14 o"

" co2 "~'-, \ -I'° eo

4 -I 2 2

O - 0 °

O_____.~ 0 0 0

0 0

1 1 1 I l l 0 O. 1 0.2 0 . 3 0 . 4 0.5 0,6

Oc~

o

b.l

F ig , 27. Inf luence of ch lor ine c o v e r a g e upon the k inet ic p a r a m e t e r s for C O 2 and e t h y l e n e epox ide (EtO) product ion on A g ( l l 0 ) ( f r o m ref. [129]) .

influence of chlorine coverage upon the kinetics of ethylene oxidation on Ag(110) was determined [129]. The fact that the kinetic parameters for full combustion essentially track those for epoxidation suggest that a common adsorbed intermediate is formed in the rate-determining step of both pathways, and that the selectivity is only determined subsequent to this step by the branching ratio of two relatively rapid competing reactions available to this intermediate [129]. Chlorine improves the selectivity by altering the branching ratio to favor ethylene epoxide. It has been proposed [129,130] that this occurs mainly via an ensemble effect, in which the chlorine blocks the Ag sites that are necessary for fragmentation and full combustion of the intermediate. An alternative model [121-123] suggests that the role of C1 as a promoter can be rationalized in terms of the way in which it alters the electron density on O~. By competing for metal electrons and thus reducing the amount of negative charge on O~, adsorbed chlorine favors electrophilic addition of O~ to ethylene (leading to C2H40 formation) and disfavors H-stripping by O a (which would lead to combustion).

5.3. Cesium promotion of Ag(lll)

The effects of Cs precoverage upon the steady-state rates of CO 2 and C2H40 production over Ag( l l l ) are shown in fig. 28 [131]. Under the experimental conditions of these measurements, Cs always tends to decrease the activity for C2H40 and CO 2 production, with


Ics / IAq (+ I0 "2) 0 2 4 6 8

12 ~ I ~ l L 1 J 60%

"T Ira._ i [ 40

30 ,.,° oo

~ 20

" ° f I - 2 TONEto • lO

[ " " o I I I f I o

0 0.1 0 2 0.3 0.4 0.5 0.6

Ocs Fig. 28. Rates of ethylene epoxide (EtO) and CO 2 production as a function of Cs precoverage on Ag(111). The

selectivity to produce EtO is defined as SEt O = T O N E t O/(TONEt O + ½TONco 2) (from ref. [131]).

an increase in the selectivity toward C2H40 when 0cs exceeds 0.15. Somewhat different trends were observed when the reaction was carried out at 500 K with a 1 : 1 mixture of C2H 4 and 0 2 at a total pressure of 10 Torr [132,133]. In this case, the activity and selectivity for C2H40 production were slightly enhanced as Cs was added, until a maximum was reached at 0.25. For Ocs > 0.3, an increase in the coverage of Cs induced a reduction in selectivity and activity.

The thermal stability of Cs on Ag is greatly enhanced under reaction conditions due to the formation of a surface cesium "oxide" ( ~ CsO3) which exists in islands that show a (2v~- × 2x/3-)R30 o LEED pattern [131]. The formation of this "oxide" is very rapid and can also be accomplished with small doses of 0 2 gas fed to C s / A g ( l l l ) in U H V conditions. It decomposes near 600 K in UHV, yielding nearly simultaneous desorption of 0 2 and Cs gas.

It has been proposed that the principal role of Cs is in the secondary chemistry, where it acts to supress the isomerization (and hence the further oxidation) of ethylene epoxide [122b].

6. Thiophene hydrodesulfurization

Organosulfur compounds are common impurities in crude oil. Due to a variety of reasons [10], it is necessary to reduce the sulfur content of petroleum fractions. The process used for this purpose is catalytic t reatment with hydrogen (hydrodesulfurization) to convert the various sulfur compounds present in the oil to hydrogen sulfide [10,134]. The most frequently used catalyst is a mixture of cobalt and molybdenum on a y-alumina support, which is sulfided before use. The ratio of M o / C o is always considerably greater than 1. The hydrodesulfuriza-


tion (HDS) of thiophene is often used to test the activity of C o - M o / A I 2 0 3 catalysts [134]. The main reaction pathway is

thiophene + 3 H 2 ---, H2S + butene (mixed isomers). (27)

In recent years the adsorption and decomposition of thiophene on metal surfaces have been extensively studied in UHV [135-141]. The kinetics and mechanism of thiophene hydrodesulfurization at atmospheric pressures have been investigated over single-crystal surfaces of Mo and Re [142-146].

6.1. Kinetics on Mo surfaces

The catalytic hydrodesulfurization of thiophene over Mo(100) can be readily performed [142,144]. An Arrhenius plot of the appearance rates of each of the products is shown in fig. 29a. The curves for the butenes all have similar temperature dependence. Arrhenius behavior is only observed for the rate of production of butadiene (apparent activation energy 60.3 + 8 kJ mol-~ [144]). The reaction product distribution on Mo(100) is compared in fig. 29b with that observed over a MoS 2 catalyst [144]. It is clear that the two distributions are almost identical, suggesting that over both surfaces the reaction proceeds via the same mechanism.

Fig. 30 illustrates the dependence of the reaction rates for butane, the butenes and butadiene on the hydrogen and thiophene partial pressures [144]. The dependence among the three butene isomers shows very little variation. The rate dependence on PHz is approximately 1, 1/2 and 0 for butane, the butenes and butadiene, respectively. The rate of butadiene production has a first-order dependence on thiophene pressure, while the hydrogenated products show only very weak dependence.

TPD and HREELS experiments investigating the surface chemistry of thiophene on Mo(100) [138] indicate that the mechanism for molecular decomposition changes as a function of adsorbate coverage. At low coverages, the molecule adsorbs parallel to the surface. C-S bonds are broken, and subsequent dehydrogenation of the remaining hydrocarbon fragments

-1.0

-1.5

.-2.0

g, - - -2.5

-3.0

®

i

60

I - O r~ ~ ,o n .J

I- o

~ 2o u.l .J o

® ~] Mo(IO0) PTh = 2.5t

[~ MoS2 PTh =20t

1.4 1.6 1.8 2.0 ~ ~ ~ ~

1/1" (x lO "3) PRODUCT

Fig. 29. (a) Arrhenius plot of the rate of product appearance during thiophene HDS over Mo(100). PH2 = 780 Torr, Pth iophene = 2.5 Tor t (from ref. [144]). (b) Comparison of the product distributions from thiophene HDS over Mo(100)

and powdered MoS 2 (from ref. [144]).


-1.0

-1.5

A IJ. I - -2.0 o l O i

-2.5

-3.0

®

/ v "

/ ÷ Lx = 0.58

o /' ,=1.10

/ ' , / o ,,/,,i-~=-o.la

0.0

-1.0

-2.C

-3.(

4.C

J

i 2'.0 2'.2 214 2'.6 2'.8 3.0

l o g PH2

, J

I - • . / =0:9.,

~ X ~ ~,~ x ~ r , = 0 13 ~ ~ x "

I I I - Io o'.o log P T h

Fig. 30. (a) Hydrogen and (b) thiophene partial pressure dependence of the rates of product appearance during thiophene HDS on Mo(100) [144]. T = 613 K, Pthiovhene = 2.7 Torr (a), PH2 = 780 Torr (b).

takes place in the 100-340 K temperature range [138]. At saturation coverages, the thiophene molecules chemisorb in a perpendicular configuration. Sequential dehydrogenation starts with C-H bond breaking in one of the a positions at 230 K. The loss of a fl-hydrogen follows at about 350 K, when a stable intermediate forms which does not react any further until heated above 550 K. Full dehydrogenation is achieved at ~ 700 K, leaving only carbon and sulfur on the surface [138]. Mo(110) exhibits reactivity similar to Mo(100), suggesting that the decomposition of thiophene over molybdenum is not sensitive to the surface structure [137].

The active catalytic surface of a Mo(100) single crystal during thiophene HDS is deficient of hydrogen and covered with adsorbed carbon from thiophene decomposition [143,144]. Deposition of atomic sulfur onto the catalyst surface does not occur during desulfurization of thiophene. On Mo(100), the rate of hydrogenation of adsorbed sulfur to HzS is much lower than the rate of thiophene hydrodesulfurization [145]. These results are consistent with a mechanism in which the desulfurization occurs via either an intramolecular step (in which the sulfur atom is hydrogenated by atoms originating from the thiophene molecule) or a direct hydrogenolysis of C-S bonds to form H2S [144,145]. The thiophene HDS reaction appears to proceed through initial desulfurization to yield butadiene, followed by hydrogenation reactions producing butenes and butane [144].

Fig. 31 shows the thiophene HDS activities for three low-Miller-index planes of molybdenum ( P T h = 3.0 Torr, PH2 = 780 Torr, T--613 K) [146]. Although the surfaces have very different roughness exposing atoms with different coordination number, the rate of thiophene HDS is very similar on all of them. Under reaction conditions the Mo single crystals are covered with a layer of carbon [146]. This overlayer blocks the influence of the metal surface structure on the catalyst activity.

44

2.0 o

x 1.6

E ~ 1.2

0 E o.e

E 0.4 p-

~ o.o ffl

J.A. Rodriguez and D.W. Goodman

(111) (110) (100)

MOLYBDENUM SURFACE

®

Mo(110)

®

Mo(100) Mo(111)

Fig. 31. (a) Thiophene HDS activities for three low-Miller-index planes of molybdenum ( e t h i o p h e n e = 3.0 Torr, PH2 = 780 Torr, T = 613 K). (b) Schematic models of the surfaces studied. The coordination numbers of the surface

atoms are shown (from ref. [146]).

6.2. Kinetics on Re surfaces

The rates of thiophene HDS over four faces of rhenium are displayed in fig. 32 [146]. In contrast to the case of Mo surfaces, on which the reaction is structure insensitive, the Re single crystals show a six-fold increase in activity on going from the hexagonal close-packed Re(0001) surface to the more corrugated Re(1010) surface. The interaction between thiophene and rhenium is not moderated by an overlayer of carbon as was found with molybdenum [146]. The trend of increasing HDS activity of the Re surfaces, (0001) < (1121) < (1120) < 1010), does not follow the increase in surface openness, (0001) < (1010) < (1120) < 1121). This fact indicates that surface openness is not the determining factor in catalyst activity [146]. It appears that the Re(1010) and Re( l l20) surfaces expose a particular configuration of active sites that is not present in a smooth surface such as Re(0001) or in a very open surface such as Re(ll?. l) .

High-pressure catalytic reactions ol,er single-crystal metal surfaces 45

~ 8

x

~ B

'~ o ffJ

"1-

/ / / /

I / ' / / / / / / /

, 1 1 1 ,

, / / ,

/ / / / ,

I / / / ~

/ / i / / /

(OOOl) (1121) '- - (1120) (1010) RHENIUM SURFACE

®

Re(O001) Ro(I01O) ®

Re(1120) Re(1121)

Fig. 32. (a) Thiophene HDS activities for four low-miller-index planes of rhenium (Pthiophcnc = 3.0 Torr, PH2 = 780 Torr, T = 613 K). (b) Drawings of the four rhenium surfaces studied. The coordination numbers of the surface atoms

are shown (from ref. [146]).

7. Alkane hydrogenolysis

Hydrogenolysis of an alkane is the cleavage of a C-C bond accompanied by hydrogenation to form two hydrocarbon molecules from one. The hydrogenolysis of paraffins is of great importance in petroleum processing, in which it is sometimes a desired reaction, as in commercial hydrocracking, and sometimes not desired, as in catalytic reforming [10]. The most active metallic catalysts are those in Group VIII [10,147]. The distribution of primary products from hydrogenolysis varies substantially with the nature of the metal. In addition, the specific activity and selectivity of the catalysts have been found to be sensitive to the average metallic particle size.

The hydrogenolysis reactions usually require a relatively large ensemble or group of active sites on the surface of the catalyst [41,42]. Alkane hydrogenolysis can be selectively sup- pressed by removing or poisoning a fraction of the active sites. This approach has been applied in the design of industrial reforming catalysts [10,41,142]. For example, the selectivity of Ni for dehydrogenation versus hydrogenolysis can be greatly increased by adding an inactive metal (Cu) to form an alloy.

Metal single crystals offer unique possibilities in the study of the effects of surface structure and composition upon alkane hydrogenolysis. In the present section we review


works dealing with the hydrogenolysis of small hydrocarbons over well-defined monometallic and bimetallic surfaces. As a result of this type of studies, a fundamental understanding of the electronic and structural properties responsible for the catalytic behavior of metal systems in alkane hydrogenolysis is beginning to evolve.

7.1. Ethane hydrogenolysis

Ethane hydrogenolysis (C2H 6 + H 2 ~ 2CH 4) is the most extensively studied cracking reaction [10,147]. The activity for ethane hydrogenolysis depends markedly on the size of metal catalyst particles and upon the nature of the metal [147]. This reaction therefore has been described as structure sensitive.

7.1.1. Monometallic surfaces

7.1.1.1. Ni(lO0) and Ni(lll) . Fig. 33 shows the specific reaction rate for methane formation from ethane over Ni(100) and Ni(l l 1) catalysts [148]. The values shown are for a PH2/Pc2H6 ratio of 100 and a total pressure of 100 Torr. At a given temperature the rate of methane production over an initially clean crystal was extremely constant with no apparent induction period [148]. The carbon level during reaction remained constant at a submonolayer coverage [148]. In fig. 33 it is evident that the activity of the (111) surface toward ethane hydrogenolysis is considerably less than that observed for the (100) surface. The activation energy associated with the (111) data (192.6 kJ mol-~) compares much more favorably with results for Ni /SiO 2 [149] than that for the (100) data (100.5 kJ mol-1). The high-surface-area catalyst is expected to contain metal crystallites exposing predominantly (111) faces [148,149].

The differences in the spacing between high-coordination bonding sites on the Ni(100) and Ni(111) surfaces could account for the differences in activity [148]. For the (100) surface the spacing between the four-fold hollow sites is ~ 2.5 ,~. The C - C bond length of an adsorbed intermediate is expected to be from 1.3 to 1.5 ,~ (the range of C-C bond lengths between

U~

E O ¢0

o

,m

, m

X o

0

10 -1

1 0 "2

E 10 .3 ,,d"

,,1.. ro

, ~ 1 0 "4 n'-

(b) (a)

I 1.6 11.8 21,0 2.2

1000/T (K 1) Fig. 33. Methane production from ethane hydrogenolysis over (a) Ni(100) and (b) Ni( l l 1) at a total pressure of 100

Torr and PH2/P,~thanc = 100 (from ref. [148]). The atomic configurations of Ni(100) and N i ( l l l ) are shown in fig. 3.

High-pressure catalytic reactions ot,er single-crystal metal surfaces 47

acetylene and ethane). Upon adsorption of ethane on Ni(100), the C - C bond cannot remain intact when the carbon atoms bond to the preferred high-coordination sites of the surface. The situation is different, however, for Ni ( l l l ) . In this case, the 1.4 ,~ spacing between the high-coordination sites is ideally suited for maintaining the C - C bond intact while bonding each carbon to a three-fold hollow site. An additional contribution to the differences in hydrogenolysis activity between Ni(100) and N i ( l l l ) can come from differences in the electronic properties of these surfaces [148]. If back-bonding from the metal to the unfilled (r* levels of ethane is an important first step toward C - C bond scission in ethane, then the (100) surface should be more active than the (111) surface, given that the appropriate Ni orbitals for such back-bonding are more available in the case of the (100) surface [148,150].

Adsorbed sulfur induces a large decrease ( ~ 90% at 0 s = 0.1) in the rate of ethane hydrogenolysis on Ni(100) [151]. The experimental data suggest that electronic effects, rather than ensemble requirements, dominate the poisoning mechanism.

7.1.1.Z Ru(O01) and Ru(1,1,10). The temperature dependence of the rate of CH 4 formation from ethane hydrogenolysis over Ru(001) [152] and Ru(1,1,10) [153] is plotted in Arrhenius form in fig. 34. The Ru(1,1,10) surface has (001) terraces which are five atoms wide and separated by (110) steps in a zig-zag form close to a kink structure [153,154]. The apparent activation energies in fig. 34 (Ru(001): ~ 84 kJ mol - i at T = 500-570 K; Ru(1,1,10): 92 kJ mol-J at T = 475-522 K) are quite smaller than the value reported for silica-supported Ru catalysts (134 kJ mo l - l [155]).

Fig. 35 shows the rate of ethane hydrogenolysis over Ru(001) as a function of H z partial pressure [152]. In agreement with studies on R u / S i O 2 [155] and Ru(1,1,10) [153], the reaction is negative order with respect to hydrogen for partial pressures of H 2 above 40 Torr. The reaction order with respect to ethane was observed to be +0.85 on Ru(001) [152], +0.55 on

1°' I ] ,.(oool) I C2H6 + Hz ) 2CH4

e= 20

n," 100 Z O

m Z 10-,

IE

10-2 1.5 1.6 1.7 1.8 1.9 2.0 2.1 2.2

1 / T X 10 ~ ( K -1)

u)

7'E1016 o

¢n Q)

"5 o 4}

o

E 1015

~ 1 0 1 4

u..I n,.'

. . . , •

Ru(1,1,10)

° ~ _

o

1.9 2.0 2.1 1000IT (K "1)

Fig. 34. (a) Arrhenius plots of the rate of e thane hydrogenolysis on a Ru(001) catalyst at total pressures of 21 and 100 Torr. /)ethane = 1 Torr (from ref. [152]). (b) Arrhenius plots for e thane hydrogenolysis on Ru(1,1,10) (Petha,c = 5.6

Torr, PH2 = 39.8 Torr) and on R u / S i O 2 (dashed line, from ref. [153]).

48 J.A. Rodriguez and D. IV. Goodman

E 1.o o

0.5

O E -r"

¢J O.1 w I -

(I:

Fig. 35. Effect

10

[C~Hs + H2 ..(o00% 2CH41

PCzHs = 1 TORR

K

I I I I I [ I I I I I I 50 100 500

H2 PRESSURE, torr

of H 2 pressure on the rate of e thane hydrogenolysis on Ru(001) (from ref. [152]).

Ru(1,1,10) [153] and +0.8 on R u / S i O 2 [155]. The order of the hydrogenolysis reaction with respect to the two reactants reflects the ability of hydrogen to compete more favorably for adsorption sites on the catalysts [152]. In fig. 35 the change to positive order at low H 2 pressures indicates the conditions at which the surface concentration of hydrogen falls below the critical opt imum value. This surface depletion of hydrogen causes a departure of the reaction rate from linear Arrhenius behavior at high reaction temperatures (see fig. 34a).

Sulfur adatoms largely reduce the rate of ethane hydrogenolysis over Ru(001) ( ~ 50% at 0 s = 0.1 and ~ 70% at 0 s = 0.5; T = 500-575 K, Ptotal = 100 Torr, H 2 " C 2 H 6 = 99)[152]. A study of sulfur poisoning of Ru(1,1,10) [153] revealed that atoms in step sites of this surface are much more active for ethane hydrogenolysis than atoms in the (001) terraces. The low coordination number of the Ru atoms in the step sites makes them more reactive in C - C bond breaking.

7.1.1.3. Ir( l l l ) and Ir(110)-(1 × 2). Specific rates of ethane hydrogenolysis over I r ( l l l ) and Ir(l l0)-(1 x 2) are shown in fig. 36 [156]. The clean I r ( l l0 ) surface reconstructs into a (1 × 2) superstructure [157], which contains a large fraction (25%) of low coordination number edge atoms (C7). The (1 x 2) reconstruction is expected [156] to be stable under reaction conditions. For temperatures below ~ 550 K, the (111) surface is more active than the (110)-(1 × 2), whereas above 550 K the opposite is true. In the linear Arrhenius regime (T < 550 K) the rate constants, k =A exp(-Ea/RT) , on the two surfaces differ considerably [156]: E a = 147 kJ mol - l and A = 1 x 1013 molecules /s i te - s for I r ( l l l ) , and E a --- 205 kJ mo1-1 and A = 6 × 10 TM molecules /s i te • s for Ir(l l0)-(1 × 2).

Post-reaction surface characterization by AES indicated the presence of a submonolayer carbonaceous residue (0 c ~-0.2) on the Ir single crystals [156]. On both surfaces ethane hydrogenolysis was approximately first order in Petha,e, while the order in PH2 varied from - 1 . 5 over I r ( l l l ) to - 2 . 5 over Ir(l l0)-(1 × 2) [156]. The dependence of the rate of ethane hydrogenolysis with respect to the partial pressures of the reactants was analyzed using a mechanistic scheme in which the rate-limiting step involves an irreversible C - C bond cleavage in a partially dehydrogenated hydrocarbon fragment,

C2H6,g ~ C2Hx, ~ + (6 - x ) H ~ , (28)

H2,g ~ 2 H a, (29)


t~ 102

(9

~ 1 0 -2

O '¢C 10"4 ILl n-"

T, K 600 500 400 600 500 400

I I i I I I I I I I I I 1

10::

I t I I I 10 -4 . I I ~ \ I I

1,6 2.0 2.4 1.6 2.0 2.4

1000/T

~ /SiO2

oMETHANE

X

'½

I x

I r ( 1 1 1 ) I r ( 1 1 0 ) - ( 1 x 2 )

Fig. 36. Specific rates of methane production from ethane hydrogenolysis (left) on 1r(111) and (right) on lr(110)-(l × 2). Pcmane = 1 Torr, PH2 = 100 Torr (from ref. [156]).

C2Hx a slow , 'CHy,a + CHz, a, (30) fast

CHz, a (or C H y , a ) + n H a , C H 4 , g . (31)

According to this model the intermediates preceding C - C bond breaking are C2H 4 on Ir(111) and C2H 2 on Ir(ll0)-(1 × 2). This difference can be explained by arguments based on structural considerations. Under the (1 × 2) reconstruction the I r ( l l0) surface is very open, exposing sites with different coordination number (see fig. 36). It is quite conceivable that on this surface an extensively dehydrogenated hydrocarbon fragment may be formed, whereas on the essentially flat I r ( l l l ) surface its formation is not energetically favorable [156].

7.1.1.4. Pt(ll l) . Values of 153.2 + 13 kJ mol -~ [158] and 144 + 3 kJ mol -I [159,160] have been reported for the apparent activation energy of ethane hydrogenolysis on P t ( l l l ) . Both


values are considerably smaller than the activation energy of ~ 226 kJ mol-1 observed on alumina- and silica-supported platinum catalysts [147,161]. Under reaction conditions the P t ( l l l ) surface is covered by a submonolayer of a carbonaceous residue [158,159]. A comparison of deuterium exchange rates with hydrogenolysis rates for ethane on P t ( l l l ) indicates that C - H bond scission is rapid relative to C - C bond scission [159], suggesting that C - C bond breaking dominates the kinetics of ethane hydrogenolysis over this surface. The reaction proceeds with orders in C2H 6 and H 2 pressures of 1 and -1 .8 , respectively [158]. An analysis of the dependence of the hydrogenolysis rate on the partial pressures of the reactants indicates that the surface hydrocarbon fragment preceding C - C bond breaking is C2H 4 or C2H 3. It has been proposed that ethylidine ( - C - C H 3) is the precursor to C - C bond cleavage [158,159].

7.1.1.5. Re(O01). The rate of ethane hydrogenolysis on Re(001) was examined at PH2 = 100 T o r r , eethane = 10 Torr and T = 573-623 K [160]. The reaction exhibited linear Arrhenius behavior, with an apparent activation energy of 75.3 + 4.2 kJ m o l - ~. At 573 K the initial rate of methane formation was 0.55 molecu le s / s i t e - s . The rate could be increased up to 1.8 molecules / s i te • s by roughening the surface with argon ion sputtering and no annealing. This fact indicates that ethane hydrogenolysis is structure sensitive on Re surfaces. Post-reaction surface analysis revealed a carbonaceous layer on the Re(001) catalyst [160]. The accumulation of carbonaceous deposits was fairly insensitive to temperature and hydrogen partial pressure.

7.1.1.6. W(IO0). Methane was produced catalytically when a W(100) crystal was heated to temperatures between 513 and 613 K in a mixture of H 2 (100 Torr) and ethane (1 Torr) [162]. After a period of several minutes during which little or no methane was formed, constant turnover frequencies for the reaction of ethane were observed. Post-reaction surface analysis indicated that the true catalyst was a monolayer of carbide which was formed on the W(100) surface on exposure to the reactants. When the crystal surface was carburized before reaction, no induction period was detected in ethane hydrogenolysis.

The hydrogenolysis reaction was approximately first order in ethane and half order in H 2, displaying an apparent activation energy of 113 kJ mol -~ [162]. This activation energy is identical to that reported on evaporated tungsten films [163].

7.1.1.7. Summary. Table 3 shows specific activities and apparent activation energies for ethane hydrogenolysis on several metal single-crystal catalysts. The activation energies vary between 75 and 210 kJ m o l - 1. In general, surfaces presenting activation energies below 100 kJ mol-~ show the highest catalytic activities. A comparison of the turnover frequencies (TOF) at P . : = 100 Torr, eethane = 1 Torr and T = 575 K gives the following order of catalytic activity: Ir( l l0)-(1 × 2) > I r ( l l l ) > Ru(001) > Ni(100) > Re(001) > N i ( l l l ) > W(100) > P t ( l l l ) . In this series the T O F varies from ~ 0.001 C H 4 molecules /s i te • s on P t ( l l l ) [158] to ~ 2 CH 4 molecu les / s i t e , s on Ir(l l0)-(1 X 2) [156].

The studies reviewed in this section show unambiguously that ethane hydrogenolysis can be classified as a structure-sensitive reaction. For a given metal, open planes show higher catalytic activity than the close-packed face. The reaction occurs in the presence of a carbonaceous layer on the surface of the catalyst. The fraction of the crystal surface covered by carbon under reaction conditions depends on the nature of the metal: 0 c < 0.2 for Ni and Ir, O c = 0.5 for Pt, and O c = 1 for W(100).

High-pressure catalytic reactions over shlgle-crystal metal surfaces 51

Table 3 Metal single-crystal catalysts, specific activities and activation energies in ethane hydrogenolysis

Catalyst T PH2 Pcthanc TOF a) Ea Ref. (K) (Tort) (Torr) (site- i s- I ) (kJ mol- l)

Ni(100) 450-600 99 0.99 0.0005- 0.3 100.5 [148] Ni ( l l 1) 520-600 99 0.99 0.0005- 0.1 192.6 [148] Ru(001) 500-600 20 1 0.2 - 3 ~ 84 [152]

500-600 99 1 0.04 - 1 ~ 84 [152] Ru(1,1,10) 475-510 39.8 5.6 - 92 [153] I r ( l l l ) 450-550 100 1 0.0004- 0.3 145.2 [156]

575-650 100 1 0.9 - 8 82.5 [156] It(110)-(1 × 2) 475-575 100 1 0.0003- 2 206.7 [156]

575-700 100 1 2 -75 95.4 [156] Pt(111) 550-640 100 1 0.0003- 0.02 153.2 [158] b)

570-625 100 10 0.02 - 0.25 144 [159] Re(001) 573-623 100 10 0.55 - 2 75.3 [160] W(100) 553-613 100 1 0.004 - 0.05 113 [162]

a) T O F = turnover frequency = CH 4 molecu les / s i t e . s . b) D2 pressure.

7.1.2. Bimetallic surfaces

7.1.2.1. Ni/W(llO) and Ni/W(IO0). Fig. 37a shows the turnover frequencies for methane formation from ethane hydrogenolysis over different N i / W ( l l 0 ) surfaces [51]. The reaction rate decreases with increasing Ni coverage. Arrhenius behavior is observed in all the cases, with an apparent activation energy independent of Ni coverage ( ~ 91 kJ mol-~ [51]). Data for ethane hydrogenolysis on Ni/W(100) catalysts are shown in fig. 37b [51]. In contrast to the results on W(ll0) , the turnover frequencies on Ni/W(100) are independent of Ni coverage. As discussed above, kinetic studies show much higher rates for ethane hydrogenolysis on Ni(100) than on Ni ( l l l ) . This difference can be attributed to the spacing of the high-coordination sites on the crystals [148]. The large spacing on Ni(100) ( ~ 2.5 A) facilitates the breaking of the C - C bond in adsorbed ethane. This suggests that, if the separation between the Ni atoms is sufficiently large, the rate of ethane hydrogenolysis will be large and independent of Ni structure. It appears that this is the case for Ni/W(100) [51], where the distance between high-coordination sites is ~ 2.7 ,~ and the rates are independent of Ni coverage.

Also shown in fig. 37b are data from a Ni(100) catalyst [148] and the rates from N i / W ( l l 0 ) in the limit of zero Ni coverage. For all these systems, the reaction rates and activation energies are essentially identical. This suggests [51] that the reaction mechanism on Ni/W(110) is similar to the mechanism on Ni(100) and Ni/W(100) and that the proportion of active metal atoms is changing as a function of Ni coverage. A plausible model is that only isolated a n d / o r low coordination (island edge) atoms exhibit high activity in N i /W ( l l 0 ) . As zero coverage is approached, the proportion of low-coordination Ni atoms increases, and the hydrogenolysis rate gets closer to that of the "open" Ni/W(100) surface.

Z1.2.2. Ni /Pt(111) Specific rates of ethane hydrogenolysis as a function of Ni coverage on P t ( l l l ) are shown

in fig. 38 [158]. The reaction rate increases monotonically as Ni is added to the P t ( l l l ) surface, until a maximum is reached at ~ 0y i = 1.3. The catalytic activity of a P t ( l l l ) surface

52 J.A. Rodriguez and D. W,, Goodman

IOI

>,,. ¢..) l o -1 Z UJ ::) o uJ

u..

. - 162 uJ > O Z a:

I-

tu 16 a Z ,< 3: I- uJ =E

C 2 H 6 ÷ H 2 ~ C H 4

iL N I / W ( 1 1 0 1

• 0 . 1 M L NI O 0 . 4 M L NI • 1 . 2 M L NI

lo-4 1.5 1.7 1.9

I O 0 0 / T (K- l )

® J(/)

>. O Z

O u.I rr u. rl- uJ > O Z

I - UJ Z .< "I- I - uJ :£

10-1

lo-2

2 .1 1 .6 1 .7 1 .8 1 .9 2 . 0 2 .1

I I I I I

C 2 H 6 + H 2 - CH 4

• N I ( 1 0 0 ) . ~,~ O N I / W ( 1 1 0 ) , O N I = O

• 0 . 4 M L N I / W ( I 0 0 ) O 0 . 7 M L N I / W ( 1 0 0 ) "1" 0 . 8 M L N I / W ( 1 0 0 ) • 1 .1 M L N I / W ( 1 0 0 )

1 0 - 3 , I I t I i

I O 0 0 / T (K "1)

Fig. 37. (a) Arrhenius plots of the rates of e thane hydrogenolysis on several Ni /W(110) surfaces. Total pressure 100 Torr, PH,/P,:than,: = 100 (from ref. [51]). (b) Comparison of the methane production from ethane hydrogenolysis over several Ni coverages on W(100), Ni on W( l l0 ) (in the limit of zero Ni coverage), and Ni(100). Ptotal = 100 Torr,

PH2/Petha,c = 99 (from ref. [51]).

70

60

o ~ .u) 50 • E

~ m 40 = E z ~ 0 _ o 30

< I ,U:~ 20 m o

10

0 0,o

C2H6 -I- H2 ~ 2 CH4

' / ~ ; " e ~ T = 580 K / PC2H6 = 1 Torr PH2 -- 100 Torr

. . . . , i t i i i I , , , , i . . . . I i L i i

0.5 1.0 1.5 2.0 2.5 3.0

eNi Fig. 38. Rate of e thane hydrogenolysis as a function of surface Ni coverage on P t ( l l l ) . T = 580 K, eethane = 1 Torr and PH2 = 100 Torr. The rate was normalized using the atom density of the P t ( l l l ) substrate (1.505 × 10 t5 atoms

cm-2) , assuming that each Ni adatom blocks a Pt surface atom (from ref. [158]).


o IM 3.5

~. 3.0

IZ: 2.5 +

--I 2.0 < I-- o I~ 1.5 o ul --I 1.0 o IE

T O.5 o uJ

~ 0 . 0

@ Re/Pt(111)

1 i 1.0 2.0 3.0

RHENIUM COVERAGE (ML)

o

¢n LU I--

n" +

..I <~ I-- o

o

o _u

el-

3.5 I

3.01

2.5

2.0

( ~ Pt/Re(0001)

• EPITAXIAL o ALLOYED

o

o / \ 1.5 / \

1 . 0 ~ \ ~

1 0.0 1.0 2.0 3.0

PLATINUM COVERAGE (ML) Fig. 39. Ethane hydrogenolysis activity of (a) Re/Pt(111) and (b) Pt/Re(001). Pethane = 10 Torr, PH2 = 100 Torr and

T = 573 K (from ref. [160]).

with the equivalent of a monolayer of Ni atoms is ~ 30 times higher than that seen for clean P t ( l l l ) . An analysis of the increase in the rate of hydrogenolysis with Ni coverage suggests an ensemble requirement of two or three Ni atoms for this reaction [158]. For Ni coverages in the range between 1.2 and 3 ML, the catalytic activity of the N i / P t ( l l l ) surface is lower than that reported for Ni(100) under similar experimental conditions ( ~ 0.15 CH 4 molecules /s i te • s [148]) and close to that observed for N i ( l l l ) ( ~ 0.05 c n 4 molecules /s i te - s [148]).

Independent of Ni coverage, the rate of hydrogenolysis was approximately first order in ethane pressure [158]. In contrast, the order with respect to hydrogen pressure varied as follows [158]: 0Ni = 0, n = - -1 .8_ 0.3; 0Ni = 0.4, n = -2 .1 + 0.3; 0Ni = 1, n = --2.4 + 0.3; and 0Ni = 1.3, n = - 2 . 5 + 0.3. The hydrogenolysis reaction was always negative order in hydrogen partial pressure, and the dependence increased with Ni coverage. The dependence of the rate of ethane hydrogenolysis on the partial pressures of the reactants indicates [158] that the hydrocarbon fragments preceding C - C bond cleavage are C2H 4 or C2H 3 (perhaps ethylidine) on clean P t ( l l l ) and C2H 2 on P t ( l l l ) surfaces with Ni coverages of 1 and 1.3 ML. Thus, the extent of dehydrogenation of the ethane molecule on the surfaces increases with Ni coverage. This is consistent with results that show that the presence of Ni adatoms on P t ( l l l ) enhances the capacity of this surface to dissociate ethane ( C - H bond breaking) [158].

7.1.2.3. Re /P t ( l l l ) and Pt /Re(O01) Specific rates of ethane hydrogenolysis on Re/Pt(111) and Pt /Re(001) catalysts are shown

in fig. 39 [160]. The addition of 0.6-1 ML of Re to P t ( l l l ) results in surfaces with a hydrogenolysis activity much larger than that of Pt(111) or Re(001). High catalytic activity is also observed for Pt < JRe(001) . A bimetallic surface of Re2Pt stoichiometry was found to be the most active catalyst for ethane hydrogenolysis (approximately one order of magnitude more reactive than Re(001)). These results show that an electronic interaction exists between the two metals resulting in a hydrogenolysis activity larger than that displayed by either monometallic component alone [160].


1.0 ~ [C2H6.l. H2RU(ooo,),2CH4]_

i ~ o H C 6 99 BCOPPER 0.8 H2 / =1-1 = L" SULFUR III • •O

l~ 0.6 W

~ 0 . 4 UJ

0.~

0.0 t . t 0.00 0.15 0.50 0)75 1.00 1.25 1.50

SURFACE IMPURITY COVERAGE (MONOLAYER) Fig. 40. Relative rate of ethane hydrogenolysis over Ru(001) as a function of or sulfur coverage. PH~ = 99 Torr,

e e t h a n e = 1 Torr, T = 550 K (from ref. [152]).

7.1.2.4. Cu/Ru(O01). Fig. 40 illustrates the effect of Cu coverage upon the rate of ethane hydrogenolysis on Ru(001) [152]. For comparison data for sulfur poisoning is also plotted. The initial slope indicates that Cu merely serves as an inactive diluent [152], blocking active Ru sites on a one-to-one basis with three-dimensional cluster growth occurring at roughly a third of a monolayer. The contrast in reaction rate attenuation by Cu as compared with S can be attributed to differences in relative electronegativities [152], where the more electronegative S alters the surface activity more profoundly than by a simple site blocking mechanism.

7.2. Propane hydrogenolysis

7.2.1. Ir(111) and Ir(110)-(1 × 2) surfaces Compared to ethane, considerably less work has been carried out for the hydrogenolysis of

propane over metal catalysts [156]. Examination of the results in fig. 41 shows that the specific activities and apparent activation energies for propane hydrogenolysis are very similar on the I r ( l l l ) and Ir(110)-(1 × 2) surfaces [156]. This suggests that the reaction mechanism on both surfaces is the same. It has been proposed [156] that the precursor for C - C bond breaking is an adsorbed C 3 H 6 species bound to the surface via two single carbon-metal bonds, which may involve more than one metal surface atom.

7.3. Butane hydrogenolysis

7.3.1. l r ( l l l ) and Ir(llO)-(1 × 2) catalysts Specific rates of hydrogenolysis of n-butane over I r ( l l l ) and Ir(ll0)-(1 x 2) are shown in

fig. 42. On supported Ir catalysts the selectivity for n-butane hydrogenolysis is extremely dependent on the average metallic particle size [164]. Fig. 43 shows the selectivity for ethane production as a function of the mean particle diameter [156] for both the supported catalysts and the two single-crystal surfaces. The observed enhancement in the ethane selectivity with decreasing particle size is a consequence [156] of an increase in the amount of Ir atoms with low coordination number in the surface of the catalysts (C 7 sites in Ir(ll0)-(1 × 2)).


4) , i

(n

4) "6 E ui

n"

Z O m I-- o < UJ re

600 500 400 10"; i i i I ; i i

10 .4

I I t I !

1.6 2.0 2.4

10 -2

T, K

102

1

10 .2

10 -4

1000/T

600 500 400 I I t J I I Ir1110)-(1x2)

I 1.6 2.0 2.4

Fig. 41. Specific reaction rates for propane hydrogenolysis on I r ( l l 1) a n d I r ( l l 0 ) - ( 1 × 2). O: methane, [2: ethane. PHz = 100 Torr, Pr,ror, a,e = 1.0 Torr (from ref. [156]).

The data in figs. 42 and 43 imply that different reaction mechanisms occur on I r ( l l l ) and Ir(ll0)-(1 × 2). The dependence of n-butane hydrogenolysis on the partial pressures of the reactants supports the following reaction mechanism on Ir(ll0)-(1 x 2) [156]:

T, K 102 600 500 400 02 600 500 400

1 ] i I t I i i 1 1 i i i i i o~ ~ 1 1 1 ) _~lr(110)-(1x2) !

<~ ~ ' ~ / - I r / : 02

fl" 10 -2 10"2 t

o: lO4 F lO' I Zl PROPANE ~,

I t I I 1.6 2.0 2.4 1.6 2.0 2.4 1000/T

Fig. 42. Specific reaction rates for n-butane hydrogenolysis on It(111) and Ir(110)-(1 × 2). O: methane, Q: ethane, Lx: propane. PH2 = 100 Torr, Pbutane = 1.0 Torr (from ref. [156]).


lO0

80

4O

I-- 20 W

0 0

I I I I

• I r / A I 2 0 3

I I I r / S i 0 2

t i l i 100 200

lr, A Fig. 43. Selectivity for ethane production (mol% of total products) from the reaction of n-butane with hydrogen on Ir catalysts plotted as a function of the mean Ir particle size (from ref. [156]). The abscissae for the two single crystals

were determined by a calculated "effective particle size" (see ref. [156]).

n-C4H10,g ~ C4H8, a + 2 Ha, (32)

C4H8, a ~ 2 C2H4,a, (33)

C2H4, a + 2 H a ~ C2H6,g. (34)

The stoichiometry of the C4H 8 fragment is consistent with a 1,4-diasorbed species or, more specifically a metallacycle pentane that involves the C7 atoms of the surface [156]. The symmetrical decomposition of this metallacycle is responsible for the very high selectivity found over Ir(l l0)-(1 x 2) for ethane production. The formation of a metallacycle pentane is sterically forbidden on the I r ( l l l ) surface [156]. For this close-packed surface (only C 9 atoms, no C 7 atoms, see fig. 36) a significant repulsion is expected between the a-hydrogens of the metallacycle and the adjacent Ir atoms.

7.3.2. Rh(lll) and Rh(l lO) catalysts The specific turnover frequencies for the production of methane, ethane and propane from

n-butane hydrogenolysis on Rh( l l0 ) and R h ( l l l ) are plotted in fig. 44 [165]. For Rh( l l0) , the order of abundance of the products is methane > ethane > propane. The high ethane selectivity seen over I r ( l l0 ) [156] is not observed on the Rh( l l0 ) surface, presumably because the Rh surface does not exhibit the (1 x 2) missing-row reconstruction that is stable on I r ( l l0 ) surfaces under reaction conditions [165]. This precludes the presence of C 7 sites on Rh( l l0) , preventing the formation of the metallocyclopentane species that is responsible for the high selectivity toward ethane in n-butane hydrogenolysis.

On Rh( l l0 ) the apparent activation energy for all three products was 135 + 6 kJ mo1-1 at temperatures below 500 K (see fig. 44a). Above 500 K, the product distribution shifted to reflect more complete hydrogenolysis (i.e. decrease in ethane and propane selectivity). This


I°° I . R h ( 1 1 0 )

i °1 ~ 0.01

0.0011

10 i

'-,, l o R h ( 1 1 1 )

o.o,1 "% l a 1:9 2'o £1 2Jz ~.6 ¢.7 " f .a 1'.9 ' z'.0 £1 2.z z.3

10 "~ K / T 105 K/T

Fig. 44. Specific reaction rates for the hydrogenolysis of n-butane over (a) Rh(110) and (b) Rh(111). 13: methane, O: ethane, zx : propane. Pbutane = 10 Torr, PH2 ~ 200 Torr (from ref. [165]).

behavior is qualitatively similar on Rh(111), with the changes in selectivity and "rollover" occurring at a lower temperature (~ 475 K). The "rollovers" in fig. 44 can be ascribed to a decrease in the surface concentration of hydrogen [165].

7.3.3. Pt(lO0), Pt(lll), Pt(332), Pt(557), Pt(10,8, 7) and Pt(13,1,1) catalysts The hydrogenolysis of n-butane was investigated on the (100), (111) and (13,1,1) faces of

platinum at 573 K with PHJPbutane = 10 and e t o t a l = 220 Torr [166]. Under reaction conditions the flat (100) and stepped (13,1,1) surfaces contain high concentrations of square (100) microfacets. The initial reaction rates of n-butane hydrogenolysis were very similar on the three surfaces investigated: on Pt( l l l ) 0.0071 C4Hl0 molecules/Pt-atom • s, on Pt(100) 0.0087 C4Hl0 molecules/Pt-atom-s, and on Pt(13,1,1) 0.0098 Canl0 molecules/Pt-atom-s. The (111) surface produced a statistical hydrogenolysis distribution (40% methane, 30% ethane and 30% propane) whereas the (100) and (13,1,1) faces displayed much higher selectivities for scission of the internal C-C bond (50%-60% ethane production).

Product accumulation curves determined as a function of reaction time for isobutane hydrogenolysis over six different Pt surfaces are compared in fig. 45 [166]. The hydrogenolysis reaction displays a significant structure sensitivity. At 573 K, the apparent activation energy on Pt(100) (146.5 kJ mol -~) was substantially larger than on the kinked Pt(10,8,7) surface (~67 kJ mo1-1) [166]. Over Pt(ll l) , Pt(332), Pt(557) and Pt(10,8,7) hydrogenolysis of isobutane produced mainly methane (55%-70%) and propane (25%-35%). The (100) and (13,1,1) surfaces yielded more ethane (40%-50%) and less methane and propane than the other platinum crystal faces.

7.4. Neopentane hydrogenolysis

7.4.1. Ir(l l l) and Ir(llO)-(1 x 2) catalysts Studies for neopentane hydrogenolysis on I r ( l l l ) and Ir(ll0)-(1 × 2) at T= 450-600 K,

Pn2 = 100 Torr and Pc~i~,2 = 1 Torr indicated essentially identical catalytic activities (10-4-10 C 5 H 1 2 molecules/site -s) and 1 activation energies (E a = 151 kJ mol- for T < 500 K; Ea = 104

58 J.A, Rodriguez and D.W. Goodman

60

E O

K. a oJ 40 "5

o E p:

20

r~ o n- n

I ! I ! L

- , . c , - c 3 - 573K ~ , ~ _

-- H2/HC =10 ~ \1~'(10,87] P~ot = 220 T o r r /

j ~ j l . ~ (3321

20 40 60 80 100

REACTION TIME, minutes

®

Pt 13,1,1) Pt (100) Pt (332)

Pt (10,8,7) Pt (111) Pt (557)

Fig. 45. (a) Product accumulation curves for isobutane hydrogenolysis over platinum single-crystal surfaces. T = 573 K, PH~ = 200 Torr, Pisobutane = 20 Torr (from ref. [166]). (b) Idealized atomic surface structures for the flat (100) and

(1 If), stepped (332), (557) and (13,l,1) and kinked (10,8,7) platinum single-crystal surfaces (from refs. [16&178]).

kJ mol- ' for T > 500 K) for these surfaces [156]. At T < 500 K the major products of the reaction were methane and isobutane, while methane clearly dominated (> 60%) at T > 500 K. An adsorbed species with CsH 7 or CsH s stoichiometry was the precursor to the initial C-C bond cleavage [156].

7.4.2. Pt(lll), Pt(lO0), Pt(lO,8,7) and Pt(13,1,1) catalysts The hydrogenolysis of neopentane over several faces of platinum was investigated at 573 K,

PH: = 200 Torr and Pc~n,2 = 20 Torr [166]. The catalytic activity of the surfaces followed the trend: Pt( l l l ) < Pt(100) = Pt(13,1,1) < Pt(10,8,7). Methane and isobutane were the major products. Demethylation represented 75%-90% of the reaction over Pt( l l l ) and Pt(10,8,7), but only 65%-75% on Pt(100) and Pt(13,1,1). An apparent activation energy of ~ 220 kJ mo1-1 was observed for neopentane hydrogenolysis ~wer Pt(10,8,7) at temperatures in the range between 540 and 590 K [166].

7.5. n-Hexane hydrogenolysis

7.5.1. Pt(lO0), Pt(lI1), Pt(332), Pt(lO,8,7), Pt(13,1,1), Au-Pt(lO0) and Au-Pt ( l l l ) catalysts Fig. 46 shows the activity for n-hexane hydrogenolysis of a series of five platinum

single-crystal surfaces with variable terrace, step and kink structure [167]. The hydrogenolysis rates display little dependence on surface structure, although hydrogenolysis product distributions were influenced markedly by terrace structure [167]. Surfaces with (100) terraces (Pt(100) and Pt(13,1,1), see fig. 45b) displayed a clear preference for scission of internal C-C bonds leading primarily to the formation of ethane, propane and butane. In contrast, surfaces with (111) terraces (Pt(111), Pt(332) and Pt(10,8,7), see fig. 45b) showed a high selectivity for terminal hydrogenolysis. Multiple C-C bond breaking processes were favored at high reaction


Temperoture (K) 630 570 520

1 0 "! , i I

/ V V ~ C1-C5 alkones "~m Hz/HC =I0

ft. -

.g ,, p, taoo) ~ \ - o Pt ( . l ) ~

• P tu3 ,u ) "~ " a: • pt (10,8,7) \

1(53 • pt (332) N ~

J t I I I

1.5 1.7 1.9 1030/T (K "t)

Fig. 46. Arrhenius plots for n-hexane hydrogenolysis catalyzed over platinum single-crystal surfaces. PH, = 200 Torr, PC6HI 4 = 20 Torr (from ref. [167]). The atomic configurations of the Pt surfaces are shown in fig. 45b.

temperatures (> 600 K) and appeared to occur most easily on the (111), (332) and (10,8,7) platinum surfaces [167]. This phenomenon correlates with a reduction in the surface concentration of chemisorbed hydrogen at high temperature. The Pt surfaces that showed larger changes in selectivity were those with the lower heat of adsorption for hydrogen (those presenting (111) microfacets) [167].

The hydrogenolysis of n-hexane (Pn,/Pnexan~ = 10, Ptotal = 220 Torr and T = 570 K) has been studied on A u - P t ( l l l ) and Au-Pt(100) surfaces [168,169]. The catalysts were prepared by vapor-depositing Au onto the (111) or (100) crystal faces of Pt to form either an epitaxial overlayer or an alloy system. In both cases, the Au effects on P t ( l l l ) were identical: a large reduction in the hydrogenolysis activity of the surface with increasing Au coverage. The corresponding large inhibition of n-hexane hydrogenolysis over Au-Pt(100) was not observed, indicating a less severe ensemble size constraint for the reaction on the more open, four-fold (100) surface [169].

Z 6. n-Heptane hydrogenolysis

7.6.1. Pt(ll l) , Pt(5S7), Pt(lO,8,7) and Pt(25,10,7) catalysts The hydrogenolysis of n-heptane at T = 533-603 K, PH: = 480 Torr and P, C7H,6 = 15 Torr

was studied over the fiat (111), stepped (557), and kinked (10,8,7) and (25,10,7) faces of platinum [178]. The hydrogenolysis activity increased in the order (557)< (10,8,7)< (111)< (25,10,7), with the Pt(25,10,7) surface being ~ 4.5 times more active than the Pt(557) surface. In general, the reaction rates decreased with increasing reaction time [178]. This decrease was shown to be the result of the deposition of irreversibly adsorbed carbonaceous species. The apparent activation energy for hydrogenolysis on an initially clean Pt(10,8,7) surface was 58.6 + 8.5 kJ mol -~ and on the P t ( l l l ) it was 77.9 + 8.5 kJ mo1-1 [178]. Preoxidation of the single-crystal catalysts led to an increase in the rate of n-heptane hydrogenolysis [178].

60 J.A. Rodriguez and D. IV. Goodman

1.0

v UJ I,,- .~IE

Z _o • I--

X E

~ 10_2

(,) •

0

10"~.. 117 1!" 110 2~0 ,.11 212 , I , 2.8 2.4 2.3

TEMPERATURE (1/K X 10 3)

Fig. 47. Product formation from the reaction of cyclopropane with hydrogen over Ni(111) and Ni(100) surfaces. Total pressure 100 Torr, He: cyclopropane = 100 (from ref. [170]).

7. 7. Cyclopropane hydrogenolysis

7.7.1. Ni(lO0) and Ni(111) surfaces Fig. 47 shows the results of the reaction of cyclopropane (1 Torr) with hydrogen (100 Torr)

over Ni(100) and Ni(ll 1) catalysts [170]. As seen for supported Ni catalysts, under hydrogen- rich conditions the only observed products of the reaction were methane, ethane and propane [170]. From the data in fig. 47, we can conclude that cyclopropane hydrogenolysis to ethane and methane is a structure-sensitive reaction on Ni surfaces. In complete analogy with the results for ethane hydrogenolysis (see above), the activity of the (100) surface is decidedly higher than that for the (111) surface.

The results for the single crystals are qualitatively similar to results for supported Ni catalysts [170]. Lower temperatures tend to produce a simple ring-opening product (propane), whereas higher temperatures tend to promote complete C-C bond cleavage, i.e., larger amounts of methane. The apparent activation energies associated with total hydrogenolysis on Ni(l l l ) and Ni(100), 117.2 and 71.2 kJ mol -] respectively [170], are consistent with values typically found for transition metal catalysts (92-126 kJ mo1-1 [171]). Likewise, the corresponding activation energies for the ring-opening reaction over Ni(l l l ) and Ni(100), 58.6 and 62.8 kJ tool-~ respectively, are very close to those reported for a variety of high-surface-area catalysts (46-67 kJ mol-t [172-175]).

Fig. 48a shows the effect that progressive sulfiding of a Ni(l l l ) catalyst has on the cyclopropane/hydrogen reaction. The consequences of the presence of S adatoms are three-fold: (1) a reduction in the C-C bond breaking ability (i.e., a decrease in ethane and methane production), (2) an increase in the relative propane + propylene yield, and (3) the appearance of significant amounts of ethylene product. Qualitatively the results for the (111) and (100) surfaces are similar in many aspects (see fig. 48). A notable difference is the absence of ethylene product for the sulfided Ni(100) surface.


1

i lO.1 d

lO -=

u. ~4~ 1O'S

:E 10 "4

10 4

~ 10"1

|

~)W 10"; :E

• = ° Nl(111),T 550 K Ni(100),T=45 K

o'., o'.= o" o!, 0.5 o o~, o'.= o.~ o'., ~ .= SULFUR COVERAGE (ML) SULFUR COVERAGE (ML)

Fig. 48. Product distribution as a function of S coverage for the reaction of cyclopropane with hydrogen over Ni(111) at 550 K and Ni(100) at 450 K. Ptot~l = 100 Torr, H2: cyclopropane = 100 (from ref. [170]).

7. 7.2. Mo(lO0) surface The catalytic hydrogenolysis of cyclopropane has been investigated over initially clean

Mo(100) surfaces and over surfaces chemically modified by adsorbed carbon, oxygen and sulfur [176]. Initially clean Mo(100) surfaces and surfaces covered by 1 ML of atomic carbon or oxygen were active for single hydrogenolysis to propane and double hydrogenolysis to ethane and methane. A Mo(100) surface with 0.8 ML of sulfur was completely inactive. The reaction was first order in H~ and zero order in cyclopropane.

The accumulation of ethane at 323, 423 and 523 K from an initially clean Mo(100) surface is shown in fig. 49 [176]. Deactivation is observed at temperatures greater than 373 K. Two possible causes of deactivation are carbon deposition and sulfur segregation [176]. After reaction the measured carbon coverage of initially clean Mo surfaces was ~ 0.75 ML. Arrhenius plots for ethane formation on initially clean, C- and O-modified Mo(100) surfaces are reported in fig. 50 [176]. The activation energy is the same for all the surfaces: 37 + 4 kJ mol. However, the activity of initially clean Mo(100) is approximately five times greater than the activity of the modified surfaces. The product distribution was similar for all the active surfaces ( ~ 17% CH 4, 17% C2H 6 and 56% C3H 8) and independent of temperature [176]. The observed selectivities are in good agreement with data reported for catalysts containing Mo in a low oxidation state [176].

It has been proposed that the active sites of Mo(100) for cyclopropane hydrogenolysis are the open four-fold hollows of the surface [176]. On these sites, it is likely that cyclopropane forms a 1,3-diasorbed C3H 6 species. This intermediate can be hydrogenated to yield propane, or it can decompose forming hydrocarbon fragments that upon hydrogenation produce ethane and methane [176].

Z Z3. Pt(s)-[6(111) x (100)] surface The hydrogenolysis of cyclopropane at T = 350-410 K, PH2 = 675 Torr and PC3H6 = 135

Torr was studied over a platinum stepped single crystal, Pt(s)-[5(l l l )× (100)]. This surface


3000

2000 c o

..c:

Z O I - - 1 0 o 0

/ . • 323 K • 423 K • 523 K

I I l i l i I S I I I I I I I I I I I I I I I I I I I I I 1 1 I I I

0.0 40.0 80.0 120,0 time, rain

Fig. 49, Time evolution of ethane formation on initially clean Mo(100) at 323, 423 and 523 K (from ref. [176]).

consists of terraces of (111) orientation and six atomic rows in width, with steps of (100) orientation and one atom in height [4]. The rate of propane formation as a function of temperature is summarized in fig. 51. The apparent activation energy calculated from this plot is 51.1 + 4.2 kJ mol-1 [4]. Activation energy values reported in the literature for this reaction

O.O

" ~ - 1 . 0 t- O c"

- 2 .0

0 L -

0

O -3 .0 ._J

~4.0

Ternperoture (K) 500 400 300

I I I

A Clean Mo o Oxygen a Carbon

&

0 0

l i l i . l l t i , I i i i i t i . , l l , , i l l l l t , l l l

1.75 2.25 2.75 3.25 I / T × I OOO ( I / K )

Fig. 50. Ar rhenius plots for ethane format ion f rom cyclopropane hydrogeno|ysis on clean, C- and O-modi f ied Mo(100). PH2 = 715 Torr, Pc3H6 = 40 Torr (from ref. [176]).


u3 I - .E ~ E

Z E 0 o

,<o)

10 .4 , , ~ '

I Ea-51 1 kJ/mol

10 -s

PH 2 = 6 7 5 torr X

10 -6 I i i i i 2.4 2 .6 2 .8 3.0

IO00/T, K "1 Fig. 51. Arrhenius plot for the hydrogenolysis of cyclopropane on Pt(s)-[6(11 l)x(100)]. PC3H6 = 135 Torr, PH2 = 675

Tort (from ref. [4]).

on platinum catalysts range from 34 to 51 kJ tool-~ [4]. The hydrogenolysis rates observed on Pt(s)-[6(111) x (100)] are in reasonable agreement with rates reported in the literature for alumina-supported Pt catalysts [4].

7. 7.4. I r ( l l l ) and Ir(llO)-(1 x 2) surfaces Specific rates for the reaction of cyclopropane with H 2 over I r ( l l l ) and Ir(ll0)-(1 x 2) are

shown in fig. 52, plotted in an Arrhenius form [177]. The results of this figure indicate a slightly structure-sensitive reaction: the specific activity of the Ir(ll0)-(1 x 2) catalyst is greater than that of the I r ( l l l ) catalyst, the ratio varying between a factor of 2 and 10 [177]. The rates observed for propane production over I r ( l l l ) agree well with values found over silica-supported Ir catalysts [177]. On Ir(111), c-C3H 6 hydrogenolysis produced methane, ethane, ethylene, propane and propylene (in fig. 52 total C 3 product refers to propane and propylene). In contrast, the only reaction products observed on the Ir(ll0)-(1 x 2) surface were methane, ethane and propane. At T < 475 K the major product of the reaction was propane, while at T > 500 K methane production dominated. For T = 375-450 K the apparent activation energy associated with the production of propane (c-C3H 6 + H 2 --* C3H 8, E a = 38 kJ tool -1) was much lower than that for the production of methane and ethane ( c - C 3 H 6 -I- 2H z ~ C H 4 + C 2 H 6 , E a = 104 kJ mo1-1 [177]). These observations can be interpreted [177] by a reaction mechanism involving the relatively facile formation of a metallacycle butane (the step that "opens" the ring), with the rate-limiting step being C - C bond cleavage in this intermediate (i.e., subsequent hydrogenation and desorption are rapid with respect to C - C bond breaking in the metallacycle species).

Fig. 53 illustrates the effects of PH_2 a n d Pc_C3H6 upon the reaction of cyclopropane and hydrogen over I r ( l l l ) at 573 K [177]. A dramatic shift in the selectivity is observed for PH2/Pc.C3H6 < 50. Post-reaction surface characterization by AES indicated that the carbon


i0 z

L i

tn ' I ¢.J a~ o E

hi p- <~ rr"

Z IO-Z 1-- 0 ILl ¢r"

10-4

T,K

6O0 50O 40O I I i I I

x ~ C y c l o p r o p a n e

0 M e t h a n e ~.X Ethooe %

[] Ethylene ~ A Propone ~ X

P,~: z.o ~ \ ~ PHz : I00 Torr I t I I I I

1.6 2.0 2.4 IO00 /T

T,K 700 600 500 400

io ~

,

[12 g

rt*

10-4

I ~ I I I I

_ C y c l o p r o p a n e

? e,hooe O E i h o n e

P r o p o n e ~

~H~ ~ ~00 TOrr @

I I I I I I 1 .6 2 . 0 2 . 4

1 0 0 0 / T

Fig. 52. Specific reaction rates for the reaction of cyclopropane and hydrogen on (a) I r ( l l l ) and (b) lr(110)-(1 ×2) surfaces. PH2 = 100 Torr, PC3Ho = 2.0 Torr (from ref. [177]).

coverage increased from a value of 0 c = 0.46 + 0.1 to values of 0 c = 0.8-1.0_+ 0.1, as PH/P,._C3H, was reduced below a value of 50 at a temperature of 575 K [177]. Fig. 54 compares the changes in selectivity for C 3 and (C I + C 2) production with the variation in the surface carbon coverage measured after reaction. It has been proposed [177] that the selectivity shifts in figs. 53 and 54 are a consequence of changes in either the structure or the binding energy of the adsorbed hydrocarbon fragment that precedes the formation of ethane and methane on Ir ( l l l ) . For Ir(ll0)-(1 × 2) the selectivities of c - C 3 H 6 hydrogenolysis also showed a strong dependence on PH/Pc.c~H, ratio [177]. The onset of the selectivity shifts observed on I r ( l l l ) at PH/Pc.c3H<50 (see fig. 54) occurred on Ir (110) - (1×2) at PH/Pc_c~H. < 10.

7.8. Methylcyclopropane hydrogenolysis

7.8.1. Ir(lll) and Ir(llO)-(1 × 2) surfaces The hydrogenolysis of methylcyclopropane was investigated on I r ( l l l ) and Ir(ll0)-(1 × 2)

at PH2 = 20--500 Torr, Pc-C3H~CH3 = 0.4--10 Torr and T = 375-700 K [177]. On both catalysts the major product of the reaction was n-butane. Other reaction products observed were methane, ethane, propane, isobutane and a mixture of 1- and 2-butenes. The major hydrogenat ion / i somer izat ion channel over I r ( l l l ) and Ir(ll0)-(1 × 2) is given by

c - C 3 H s C H 3 + H 2 ~ n - C 4 H l 0 , (35)


i

I ( .3

E uS t--

z 0

i,i

1(2

I0

IO IOO IOOO I PH2 , Torr PHC, Torr

10

Fig. 53. Specific rates for the reaction of cyclopropane and hydrogen on lr(111) at T = 575 K. (a) Effect of H 2 partial pressure , ec3H6 ~ 2.0 Torr. (b) Effect of cyclopropane partial pressure, PH2 = 100 Torr (from ref. [177]).

whereas the major hydrogenolysis channel is

c - C 3 H s C H 3 + H a ~ C H a + C 3 H 8. (36)

on Ir(l l l) , and

c - C 3 H s C H 3 + H 2 ~ C H 4 + C 2 H 6 + C 3 H 8 (37)

on Ir(ll0)-(1 × 2). The absence of ethane in the major hydrogenolysis channel over I r ( l l l ) can be explained purely on a stereochemical basis [177]. The presence of ethane from hydrogenolysis of methylcyclopropane can only be observed by the fragmentation of a metallacycle butane that has been formed by the activation of the C~-C 2 bond. The methyl group provides a steric limitation to the activation of the C r C 2 bond on a flat I r ( l l l ) surface. On the other hand, a metallacycle butane involving the rupture of the C~-C2 bond can be formed with the low-coordination-number C 7 atoms that are present on Ir(ll0)-(1 x 2) (see fig. 36).

7.9. Methylcyclopentane hydrogenolysis

7.9.1. Pt(lO0), Pt(lll), Pt(332) and Pt(557) catalysts The reaction between methylcyclopentane (20 Torr) and hydrogen (600 Torr) was investi-

gated at T= 540-650 K over four platinum single-crystal surfaces with flat (100) and (111), and stepped (332) and (557) orientations [179]. The main reaction products were 2- and


>- P-

(_9 LtJ ._J LtJ o3

E o

0.1

0.01

1.2

0.8

ec 0.4

C ¢clopropane, fr(lll) 1 I I I I l l I I 1 I I I I I I I

'( h ,!

I

I I

OCl+C 2

l I llllll I I I IIIIII

I I l l l l I I I I I I I I I I

I /

, [ ~ I I I I I I ~ , I , l i l t

\\ ' i

I I I I I 1 t l

1 ] I lllll i I I I i l l l l

(b)

PHz = I00 Torr PHC = 2.0 Torr 1 /

I I I I l l l l l I I I I I I I l l 0.0 , , ,IE,,l , , t1,,,,l

I I0 I0 I00 I000

PHC, Torr PH2 , Torr

Fig. 54. (a) Relative selectivity for hydrogenation/ isomerizat ion (C 3) and hydrogenolysis (C I + C 2) from the reaction of cyclopropane and hydrogen on I r ( l l l ) at 575 K. The reaction conditions are identical with those given in fig. 53. (b) The fractional surface coverage of the carbonaceous residue at the corresponding reaction conditions (from ref.

[177]).

3-methylpentane; very little n-hexane, cyclopentane or methane was formed. The rate of formation of methylpentanes was found to be nearly two times faster on the (100) surface than on the (111) surface. The presence of steps on the platinum surface had no significant effect on the rates of ring opening since these line defects were quickly covered with carbonaceous deposits [179]. Apparent activation energies were determined for the P t ( l l l ) and Pt(100) surfaces [179], and the following values were determined in kJ mol- l : 37-46 for 2- and 3-methylpentane, 59-75 for methane and 105-109 for hydrogenolysis to C2-C 5 products.

7.10. Cyclohexane hydrogenolysis

7.10.1. Pt(lll), Pt(557), Pt(25,10,7) and Pt(10,8,7) catalysts Fig. 55 shows results for cyclohexane hydrogenolysis on the flat (111), stepped (557), and

kinked (25,10,7) and (10,8,7) faces of platinum [180]. The hydrogenolysis rates were highest on Pt(111). In general, the reaction rates decreased with increasing reaction time due to the irreversible adsorption of a tightly bound carbonaceous deposit [180]. Fig. 56 illustrates the


O E

o K

¢:: o x

J= I c

25

2C

15

10

0 0

I ~ I I 1 I I O + H 2 "-~ N V / . . . , ~ ,, )

15 torr Cyclohexone -,oo / / . . (~o, 8,7)

40 80 120 Time (minutes)

60

12

N'~E I°

u

%6

° 4

~z

0 0 1 6 0

I I 1 1 I I I

Q - " [ < cs ~ . . , ~ ( l l l ) 15 torr Cyclohexone

I 0 0 torr H y d r o g e n

573 K . , f

(557)

1 I I I I I I 40 80 IZ0

T i m e ( m i n u t e s )

Fig. 55. (a) n-Hexane and (b) light-alkane production from cyclohexane hydrogenolysis as a function of surface structure and time (from ref. [180]). The atomic configurations of the Pt surfaces are shown in fig. 45b.

dependence of initial rates on H 2 pressure over the P t ( l l l ) and Pt (10,8 ,7)surfaces [180]. Hydrogenolysis leading to n-hexane is positive order in hydrogen on both surfaces at low hydrogen pressure, but becomes negative order at pressures greater than 400 Torr. Ligh(al~

L J " "

kane product ion is slightly negative o rder in hydrogen pressure. • !

ZlO.2. Ru(O01) and Cu /Ru(O01) catalysts The hydrogenolysis of cyclohexane was investigated on Ru(001) at PH2 = 20 -100 Torr,

Pc-c6H12 = 1 -4 Torr and T = 570-650 K [181]. In all the cases investigated, initially c l e a n


3".

E

w #: "6 I

c

i I -- I ' I I P t (111) ( ~

1,5 Torr Cyclohexonl 573 K

~ , . e _ n . hlxon!

. ~" < C6 "~,~ l I I • I

~00 300 5o0 700 900 Hydro(len Presl~ure (Tort)

6" N

o T .E E

E

e,-

- 6

o~

~T

T,

I I ' b I - -

PI (10, 8,7) 15 Torr Cyclohexone

,p, 8 5 7 3 K

6 ~ e

i r~--O ~ x o n e

"G 2

0 , , 1,, _xio . . . .

100 3 o 0 5 0 0 7 0 0 9 0 0 Hydrogen Pressure (Torr)

F

® o

4 ~ E G~

CII

o ~

Fig. 56. Effect of hydrogen pressure upon dehydrogenation and hydrogenolysis rates of cyclohexane over (a) Pt(111) and (b) Pt(10,8,7) (from ref. [180]).

Ru(001) showed the highest activity for producing C j - C 6 hydrocarbons. The hydrogenolysis activity of the catalyst decreased with time up to a constant (steady-state) value. At this point the Ru(001) surface was covered by ~ 1 ML of carbon and the major product of the c-C6H12 + H 2 reaction was benzene [181]. The decrease in hydrogenolysis rates with carbon coverage indicates, at the very least, that C - C bond scission, generally recognized as the rate-determining step in hydrogenolysis reactions, is slower in the presence of carbon than on the clean Ru(001) surface [181]. From the data in fig. 57, it appears that cyclohexane hydrogenolysis on Ru(001) is negative order in cyclohexane and positive order in hydrogen partial pressures [181]. This type of dependence is believed to reflect the response of the surface carbon level to changes in cyclohexane and hydrogen pressures (positive w.r.t. cyclohexane and sharply negative w.r.t, hydrogen) [181].

High-pressure catalytic reactions ouer single-crystal metal surfaces 69

k.

i,.

I i

uJ I-- <{ n-

(n (n > .

z i l l 0 g 0 >.

tM Z < 1,< IM

0

0

H z P A R T I A L P R E S S U R E ( to r r }

$ 10 20 50 1 0 0 1.O0 • ' ' ' i I ' ! ' ' " ' 1

0 . ( a ) •

0 . 2 0

0 . 1 0 -

0.05 ( b )

0.02

0.O1 t ' ' ' 1 I I * ~ , , ! o.s 1.o 2.0 s.o 1 o.o

C6H12 P A R T I A L P R E S S U R E ( torr )

2 0 0

~0.0

Fig. 57. Cyclohexane hydrogenolysis rates over Ru(001) at 650 K as a function of reactant partial pressures; (a) PH2 = 100 Torr, and (b) Pc-CoH,2 = 1 Torr (from ref. [181]).

Studies for Cu/Ru(001) show a reduction in the rate of cyclohexane hydrogenolysis with increasing Cu coverage [181]. Several mechanisms have been proposed which may account for the altered properties of the Ru(001) catalyst upon addition of Cu [181]. In the simplest one, Cu may be acting much like C in that the intermediate preceding hydrogenolysis is stabilized toward C-C bond scission.

8. I s o m e r i z a t i o n a n d c y c l i z a t i o n o f a l k a n e s

Catalytic reforming is one of the basic petroleum refining processes, yielding a large variety of liquid fuels [10,11,134]. In reforming, paraffins are reconstructed without changing their carbon numbers. The reactions, which include isomerization of n-alkanes to isoalkanes and dehydrocyclization of n-alkanes to alicyclics and aromatics, lead to a marked improvement in fuel quality as measured by the octane number [11,134]. Typical industrial reforming catalysts are generated by dispersing crystallites of Pt or Pt alloys on alumina. The isomerization of light alkanes on Pt single-crystal surfaces has been intensively investigated in recent years from the viewpoint of catalysis [166-169,178,182-184]. It has become evident that the catalytic chemistry of the platinum/hydrocarbon interface is interesting, unique and important [183].

8.1. Butane isomerization

8.1.1. Pt(lll), Pt(lO0), Pt(332), Pt(557), Pt(13,1,1) and Pt(lO,8, 7) catalysts The interaction between H 2 (200 Torr) and n-butane (20 Torr) was investigated on the


E 8O0 B 0

E_ ~ 6oo

0 E

J 4o(

~ 2O( 13_

! I I

) - - -'~ ,~t 573 K

I I

~ t{iii}

0 20 40 60 80 ~00 Reaction Time (minutes)

Fig. 58. Product accumulation curves determined as a function of reaction time for isobutane isomerization over Pt single-crystal surfaces (from ref. [166]). PH = 200 Torr, Pisobutane = 20 Torr, and T = 573 K. The atomic configura- 2

tions of the platinum surfaces are shown in fig. 45b.

(100), (111) and (13,1,1) faces of platinum at 573 K [166]. Isomerization, dehydrogenation and hydrogenolysis of n-butane were observed. The reaction rates for the processes followed the order: dehydrogenation > isomerization > hydrogenolysis. The fractional isomerization selectivities (Sisom = Risom//(Risom 4-Rhydrogeno lys i s ) ) varied from 0.65 on P t ( l l l ) to ~ 0.8 on Pt(100) and Pt(13,1,1). The isomerization reaction displayed substantial structure sensitivity: the catalytic activities of the (100) and (13,1,1) platinum surfaces were at least three times higher than that for P t ( l l l ) [166]. AES showed a carbon coverage of 1.7-2.7 ML on the Pt surfaces after a 90-150 min reaction time [166].

Product accumulation curves for isobutane isomerization over the flat (100)and (111), stepped (332), (557) and (13,1,1), and kinked (10,8,7) faces of platinum are shown in fig. 58 [166]. Under the experimental conditions of this figure, isobutane dehydrogenation produced equilibrium concentrations of isobutene within minutes over all six platinum surfaces [166]. The initial rate of dehydrogenation was at least 2 -6 times faster than those for isomerization and hydrogenolysis. In general, the fractional isomerization selectivities (Sisom) were larger than 0.85. In fig. 58 the order of activity for isomerization follows the sequence Pt(100)'--- Pt(13,1,1) > Pt(557) > Pt(10,8,7) ~- Pt(332) > P t ( l l l ) , with the difference in initial rates between the least and most active crystal face being approximately a factor of 6. High concentrations of (100) microfacets were required for high catalytic activity in butane isomerization. Arrhenius plots for isobutane isomerization on the (100) and (10,8,7) platinum surfaces are compared in fig. 59 [166]. The isomerization activity of Pt(100) displayed a maximum at about 600 K and then decreased with increasing temperature.

8.2. Neopentane isomerization

8.2.1. Pt(lO0), Pt(111), Pt(13,1,1) and Pt(10,8,7) catalysts Fig. 60 displays the activity of four different faces of Pt for neopentane isomerization [166].

In contrast to the C 4 alkane reactions, surface structure had only a small influence on the

High-pressure catalytic reactions ocer single-crystal metal surfaces 71

I.O

u o)

E O

E O.i ¢u

-6 E ¢u

O t~

o.o[

0

r r '

0.002 1.5

I 1 t

~ . ~ Convers=on Pt(lO0). PI' (10,8,7)

~ ~ x ~ / ' x / P t (100)

~ Pt(lO,8,7l

Hz/H C : tO ~tot : 220 Tort

I t I . . . . I 1.7 1.9

ICOO/T (K -I) Fig. 59. Arrhenius plots for the initial rates of isobutane isomerization over the flat (100) and kinked (10,8,7) platinum

single-crystal surfaces (from ref. [166]).

initial rates of neopentane isomerization. The fractional isomerization selectivities (Sisom) varied between 0.85 and 0.90. The selectivity in isomerization for n-pentane production displayed notable structure sensitivity [166]. While the Pt(111) and Pt(10,8,7) catalysts yielded only 2%-3% n-pentane in isomerization, the (100) and (13,1,1) surfaces produced 12%-14% n-pentane. For T--540-570 K the initial rates of neopentane isomerization displayed an apparent activation energy of 217.6 kJ mol-1 on Pt(10,8,7) [166]. Above 580 K the initial reaction rates were insensitive to changes of temperature [166]. The rates of isomerization

1000

E

o

0

E 400 I - : r,3

O 200 rr n

-~- l i J

+ -,. ) - ' +

T = 5 7 3 K 8 , 7 ) ~ pH2 = 2 0 0 TORR Pt (10,

PHC=20 TORR j P t (13,1,1)

20 40 60 80

• R E A C T I O N T I M E , m i n u t e s

Fig. 60. Product accumulation curves measured as a function of reaction time for neopentane isomerization over Pt single-crystal surfaces (from ref. [166]).

72 J.A. Rodriguez and D. 145. Goodman

decreased with time due to the deposition of carbon on the Pt catalysts. The rate of deactivation of the surfaces increased with increasing reaction temperature. Deactivation was accompanied by the deposition of about one monolayer of strongly chemisorbed carbonaceous species [166].

8.3. Methylpentane isomerization and cyclization

8.3.1. Pt(lll), Pt(ll9), Pt(311) and Pt(557) catalysts The isomerization of 2- and 3-methylpentane was investigated over Pt(l l l) , Pt(ll9),

Pt(311) and Pt(557) at PH2 = 755 Torr, PHC = 5 Torr and T= 623 K [182,184]. On all the surfaces, the major products of 2-methylpentane isomerization were methylcyclopentane and 3-methylpentane, with minor amounts of n-hexane and benzene. Similar results were observed for polycrystalline Pt and alumina-supported Pt catalysts [182]. Isomerization of 3-methylpentane on Pt(557) yielded mainly methylcyclopentane and 2-methylpentane with no benzene in the products. For the examined platinum catalysts, the flat (111) surface showed the highest activity for hydrogenolysis and the lowest for isomerization.

8.4. n-Hexane isomerization and cyclization

8.4.1. Pt(lO0), Pt(lll), Pt(332), Pt(13,1,1), Pt(lO,8,7), Au-Pt(lO0) and Au-Pt( l l l ) catalysts The reaction of H z (200 Torr) and n-hexane (20 Torr) was studied at 573 K on flat (100)

and (111), stepped (332) and (13,1,1) and kinked (10,8,7) surfaces of Pt [167]. Only the aromatization reaction displayed significant structure sensitivity, which was characterized by about a factor of 4 difference in initial rates between the most active (10,8,7) and least active (100) platinum surfaces [167]. In general, the selectivity for cyclization (benzene+ methylcyclopentane production) varied between 50% and 60%, while that for isomerization (2- and 3-methylpentane production) was in the range between 20% and 30%. Skeletal rearrangement was dominated by cyclic mechanisms involving both 1,5- and 1,6-ring closure [167]. Polymerization of the adsorbed species competed with skeletal rearrangement and led to the growth of a disordered carbonaceous deposit (one or more monolayers) on the platinum surfaces [167]. The primary role of this deposit was that of a non-selective poison.

Arrhenius plots for n-hexane isomerization, C s cyclization and aromatization catalyzed over five platinum surfaces are shown in figs. 61 and 62 [167]. Surfaces with a high concentration of (111) microfacets (Pt(10,8,7), Pt(332) and Pt(l l l) , see fig. 45b) were always more active in aromatization than those rich in (100) terraces (Pt(13,1,1) and Pt(100), see fig. 45b). All other reactions displayed little structure sensitivity at all reaction temperatures.

The catalytic conversion of n-hexane was studied over Au-Pt(100) and Au-P t ( l l l ) alloys [168,169]. In fig. 63 the initial rates (initial rates per surface atom, i.e., gold and platinum) of isomerization, C 5 cyclization, aromatization and hydrogenolysis are shown as a function of surface alloy composition. The data in the figure show the structure sensitivity of alloy catalysis for hydrocarbon conversion [169]. The effects of Au concentration upon the formation of 2- and 3-methylpentane and the rate of hydrogenolysis are different for each surface. In particular, the Au-P t ( l l l ) system shows an increase in the rate of isomerization that is not observed on Au-Pt(100). The total activity of the Au-Pt(100) alloy decreased linearly with increasing surface atom fraction of gold [169]. In contrast, the Au-Pt ( l l l ) alloy showed an enhancement in the total activity as the surface atom fraction of gold was increased from 0 to 0.4, and then fell off with further increase in the surface atom fraction of gold [169].



I I l

E 0 .6 _ A N

~. Hz/HC =lO ~ o 10"3 Plot 220 Torr ~,'kk~ u, k

. t I I , I I o . . . . . .

t w 3 ~ 1 1

c 4xl(~ ~ o ,'~

O - - t D

I ( ~ 3 _ _

• Pt (10,8,7) • Pt (332)

2 z ~ o 4 I I I I I 1.5 1.7 t . 9

IO00/T (K'I)

Fig. 61. Arrhenius plots for n-hexane isomerization to 2-methylpentane (below) and 3-methylpentane (above) catalyzed over several Pt surfaces (from ref. [167]). The atomic configurations of the platinum surfaces are shown in

fig. 45b.

The results of the A u - P t ( l l l ) surface can be interpreted in terms of differences in ensemble requirements [168,169]. It appears that hydrogenolysis and aromatization require large ensembles of platinum atoms. The dilution of the Pt surface with Au reduces the number of these large ensembles, inhibiting these reactions. The blocking of these reaction pathways provides more surface species for reactions that require only small ensembles of Pt atoms. The result is an enhancement in the rate of isomerization per Pt atom with increasing surface atom fraction of Au [168,169]. For Au-Pt(100), there is no enhancement in the rate of n-hexane isomerization. The reaction mechanism that occurs for the (111) alloy surfaces was not available on the (100) alloy surfaces, probably due to differences in the morphology a n d / o r in the electronic properties of these bimetallic surfaces [168,169].

8.5. n-Heptane cyclization

8.5.1. Pt(111), Pt(557), Pt(lO,8,7) and Pt(25,10,7) catalysts Fig. 64 shows toluene accumulation curves from n-heptane cyclization on four different

faces of platinum [178]. The initial rate of toluene production increased in the order (111) = (25,10,7)< (557)= (10,8,7). Fig. 65 compares the initial rates measured for toluene production and n-heptane hydrogenolysis [178]. The selectivity of toluene production versus

14 J.A. Rodriguez and D. W. Goodman


iTi Ptar * 220 Torr

a Ii3 I I I I I I .5 1.6 1.7 1.8 1.9

1000/T (K-I 1

Temperature ( K I 630 570 520

I 0 Pt(III) 0 Pt (10.8.7)

b Pt(lo0)

1.5 1.6 1.7 1.8 1.9

IOCdT (K-II

Fig. 62. Arrhenius plots for (a) n-hexane C,-cyclization and (b) n-hexane aromatization catalyzed over five Pt surfaces

(from ref. [167]).

‘&I+ H2 570K

H2IHC = 10 P,,,=220 TORR

n b l <cg

0, a AU-PT (100)

1 20

1.5

1.0

0.5

0, b AU-P~(lll) I

. .

A .’ .-*-i-;-., ’ .I

‘\

4

\ i

,,! ‘\ i - I 81 . l

\m 4 \ i

Fig. 63. Plot of the initial turnover frequencies per surface atom (platinum and gold) versus the surface atom fraction of gold (from ref. [1691X


2 4 I I I I I CH 3

20_ /~/~/~.._~[~ +4H z f ( 5 5 7 ) _

= , 2 - _ j j y -

00 30 60 9 0 120 150 180 Reoefion Time (minutes}

Fig. 64. Toluene formation from n-heptane over four Pt single-crystal surfaces (from ref. [178]).

hydrogenolysis increased by an order of magnitude in the sequence (25,10,7) < (111) < (10,8,7) < (557). The apparent activation energy for toluene formation was 126.6 + 16.74 kJ mol-1 on the clean Pt(10,8,7) surface and 140.6 + 8.4 kJ mo1-1 on the clean P t ( l l l ) surface [178]. In general, the reaction rates decreased with increasing reaction time. This decrease was shown to be the result of the deposition of irreversibly adsorbed carbonaceous species [178].

8.6. Summary

The results discussed above show significant differences in isomerization activity for platinum single-crystal surfaces with different atomic structures. Fig. 66 summarizes initial rates as a function of crystallographic orientation for the isomerization of n-butane, isobutane, neopentane and n-hexane at 573 K [166,167]. The butanes display the maximum variation in reaction rate with surface structure. High concentrations of (100) terraces were required for high catalytic activity in butane isomerization [166]. The effect of steps and kinks was to increase the isomerization rates, although this effect was small as compared to the changes introduced by terrace structure. Neopentane and n-hexane isomerization display little dependence on platinum surface structure. The absence of structure sensitivity in the isomerization of n-hexane has been explained in terms of a reaction mechanism that is dominated by C 5 cyclic intermediates [166,167]. These intermediates cannot be formed during the isomerization of light alkanes.

The results in fig. 66 indicate that the rate of isomerizati0n depends markedly upon the structure of the reacting hydrocarbon [166]. The order of isomerization activities follows the sequence neopentane > isobutane > n-butane > n-hexane. This trend suggests that isomerization rates increase with increasing degree of substitution and decreasing molecular weight [166].

The isomerization activity of carbon-covered platinum surfaces appears to be lower than that of clean platinum. Hydrocarbon overlayers that are pre-adsorbed on Pt act as poisons


!. E

.-= z o

'~" I i - I

I I I I

i ' i n , l - 6

. I 1 I I ( i t l l (ssT) (~8 ,7 ) (?_5,10,7)

Crystol OPientotion

6

o ~ 4

O O

O O O t:~ o -~ ~ 2

o

" L '1 l I

._. ~ I I I (III) (557) 110,8,7) (25,10o7}

Crysto l Orientot ion

Fig. 65. The dependence of (a) initial rates and (b) selectivities on atomic surface structure at 15 Torr n-heptane, 480 Torr hydrogen and 573 K (from ref. [178]).

[183]. During actual catalytic conditions more than 50% of the catalyst surface was covered by carbonaceous deposits within the first few minutes of reaction [183,185]. These residues grew continuously, passivating the metal and inhibiting further hydrocarbon conversion.

A significant difference in deactivation behavior between the light-alkane and n-hexane reaction studies was in the amount of carbon deposited [166,185]. At any given temperature, less carbon was deposited on the Pt surfaces during light-alkane isomerization. In these cases, the defect sites (steps and kinks) were not readily covered by hydrocarbon residues and were therefore available for catalysis. The presence of kinks and steps had an appreciable influence on the rate of isomerization of isobutane on Pt surfaces [185]. The same effect was not observed for methylcyclopentane or n-hexane conversion, because these heavier alkanes deposited carbonaceous fragments at the surface irregularities and passivated them within the first minutes of reaction, so the catalysis occurred only on the terraces of the Pt surfaces [185].


o ¢11 m

E o

0

t ,

0 E @

o t r

o

¢ :

0.1

0.131

0.001 (13.1,1) (lO0) (111) (10,8,7) (557)

Crys to l O r ~ e n t o t t o n

I i I i I I 1 - A I k o n e I s o m e r i z o t i o n S t ruc tu re Senst l iwty

573K, PH2 = 200 Tort, PMC = 20Tort

- I'1

~ ~ , ~ H e x o n e

I i I t I l (332}

Fig. 66. Structure sensitivities of alkane isomerization reactions catalyzed over platinum single-crystal surfaces (from ref. [166]). The atomic configurations of the Pt surfaces are displayed in fig. 45b.

9. Hydrogenation of olefins

A great variety of hydrogenation reactions involving olefins are carried out industrially. These range from large-scale continuous catalytic operations in petroleum refineries to small-scale batch operations in the pharmaceutical and fine chemical industry [10,11]. The addition of hydrogen across a C=C bond is one of the easiest and most studied of metal-catalyzed reactions [11,183]. The most active metallic catalysts are those in Group VIII. Studies using single-crystal catalysts have been primarily focused on the hydrogenation of ethylene and light olefins (C3-C6) [177,186-192].

9.1. Ethylene hydrogenation

9.1.1. Pt(lll) surface The hydrogenation of ethylene to ethane using both hydrogen and deuterium was investi-

gated on the flat P t ( l l l ) surface at temperatures between 300 and 373 K [186]. Arrhenius plots are shown in fig. 67, yielding an apparent activation energy of 45.2 + 0.5 kJ mol- 1. This value agrees well with results for supported Pt catalysts that show activation energies in the


1 I I I Pt( l l l ) -

Elhyiene hydrogenotion PCzH4 = I0 Tort

40 ~ _ PHz°r OZ" 20T°rr

"~ PNz = 80Tort

~'~ 9 Kcol/mole ("L'-O,4)

Da,

1 I 1 I I

2.7 2.9 31 3.3 3.5 I /T xlO 3 (K -I)

Fig. 67. Arrhenius plots for ethylene hydrogenation with H 2 and D 2 over P t ( l l l ) . PCeH4 = 10 Torr, PH2 or Po, = 20 Torr, and PN_~ = 80 Torr (from ref. [186]).

40 -r

U

3o o

Pt ( l l l ) Ethylene hydrogenotion"

P(::~4 = 10 Torr

PDZ = 20 Torr _

PN2 = 80 Torr

T=333K

10

O ~ 0 I 2 3 4 5 6

Number of Deuterium Atoms/C2H 6 molecule

Fig. 68. Deuterium atom distribution in the resulting ethane from the hydrogenation of ethylene with deuterium over Pt(1 ] ]) (from ref. [186]).


range between 38 and 45 kJ mo1-1 [186]. On P t ( l l l ) , ethylene hydrogenation exhibits a normal isotope effect: the reaction with hydrogen is about 1.3 times faster than with deuterium [186]. Reaction orders of 1.31 + 0.05 with respect to hydrogen (or deuterium) and - 0 . 6 0 + 0.5 with respect to ethylene were observed. All these kinetic data compare favorably with results reported for high-surface-area Pt catalysts [186].

Fig. 68 shows a typical product distribution for the reaction of ethylene and D 2 [186]. The deuterium atom distribution in the reaction products peaks at 1-2 deuterium atoms per ethane molecule produced, similar to what has been reported for supported Pt catalysts [186].

AES spectra taken after reaction conditions revealed a metal surface covered by hydrocarbon species [186]. No reaction self-poisoning was detected during the hydrogenation experiments. Furthermore, when reactions were started on surfaces covered by hydrocarbon fragments, either by retaining the carbonaceous deposits from previous reactions or by predosing the surface to ethylene under UHV conditions, the rates were identical to those seen when starting with a clean P t ( l l l ) surface [186]. Evidence from LEED and TDS suggests that the carbonaceous species found on the catalyst surface after reaction are, in fact, ethylidyne moieties, as obtained by dosing ethylene onto a clean P t ( l l l ) surface in UHV [186]. The hydrogenation rate of the ethylidyne adlayer was much slower than the rates of ethylene hydrogenation [186]. It has been proposed that ethylene hydrogenation does not take place on the clean metallic surface but rather on top of a layer of carbonaceous fragments composed mainly by ethylidyne adsorbates [186]. According to this mechanism H e dissociates on the metal, while ethylene is weakly adsorbed on top of the carbonaceous deposits. The main role of the ethylidyne species is to transfer H atoms from the metal surface into the adsorbed ethylene molecules [186],

H2,g ~ 2 Ha,

Pt3-CCH3, a + H a ~ Ptn=CHCH3,a,

2 Ptn=CHCH3, a + C2H4, a --~ 2 Pt3----CCH3, a + C2H6,g.

(38)

(39)

(40)

The ability of ethylidyne to exchange hydrogen (or deuterium) atoms with P t ( l l l ) surfaces covered with hydrogen (or deuterium) is well known [186]. The fact that ethylene hydrogenation over P t ( l l l ) is very similar to that on supported Pt catalysts is a consequence of the presence of the ethylidyne adlayer, which masks the surface structure making the reaction surface insensitive [186].

9.1.2. Ni(lO0), Ni(llO) and Ni(l l l) surfaces Ethylene hydrogenation was performed over the three low-Miller-index faces of nickel,

(100), (110) and (111), at 298 K under a total pressure of 50 Torr with H z : C 2 H 4 ratios of 10 and 100 [192]. The reaction rate was first order in H 2 pressure and independent of ethylene pressure. At PH/Pc H = 10, the measured steady-state rates were respectively 0.35 and

2 2 4 19 I 2 0.10 × 10 molecules h - cm- for the (111) and (110) faces. On the (100) face practically no

reaction was observed under the same conditions. The trend in catalytic activity of the single-crystal surfaces, (100)<< (110)< (111), agrees well with results that show a higher catalytic activity for Ni crystallites with (111) orientation than for randomly oriented crystallites [192].

Post-reaction surface analysis with AES showed the presence of carbon on the Ni single-crystal catalysts [192]. C2H 4 is dissociatively adsorbed on the (100) face forming acetylenic residues, which completely cover the surface and are not displaced when in contact of hydrogen. Consequently, Ni(100) is practically inactive for ethylene hydrogenation. Over


the ( l i d face, the acetylenic residues are rearranged on the surface in a way that leaves free Ni regions in which the catalytic hydrogenation takes place [192].

Effective poisoning of catalytic activity at sulfur coverages of less than 0.15 ML has been observed for ethylene hydrogenation on nickel [151]. Electronic effects, rather than ensemble requirements, dominate the catalytic poisoning mechanism [151].

9.2. Propylene hydrogenation

9.2.1. Ir( l l l ) and lr(llO)-(1 X 2) surfaces Steady-state rates of the reaction of propylene with hydrogen on the Ir(111) and Ir(110)-(1

x 2) surfaces are shown in fig. 69, plotted in Arrhenius form [177]. Two reaction channels were observed: hydrogenation to propane (E a--- - 0 . 8 kJ mol - l on I r ( l l l ) and - 1 5 . 1 kJ mol - 1 on Ir(ll0)-(1 x 2); T = 450-600 K) and hydrogenolysis to methane and a mixture of ethane and ethylene (Ea = 99.6 kJ mol -~ on I r ( l l l ) and 110.1 kJ mol -~ on Ir(ll0)-(1 x 2); T = 400-500 K). On both Ir surfaces the major reaction product was propane, with Ir(ll0)-(1 x 2) showing a greater activity for hydrogenation. Measurements of the adsorption and desorption kinetics of hydrogen on the clean (111) and (110)-(1 x 2) surfaces show the presence of a higher binding energy adstate on the (110)-(1 x 2) surface [177,193]. Thus, it might be expected that the hydrogen adatom concentration will be greater on the Ir(ll0)-(1 x 2) surface during reaction conditions, and this could explain the greater rate of hydrogenation seen on the Ir(ll0)-(1 x 2) catalyst [177].

AES spectra acquired after the hydrogenation reaction revealed a carbonaceous residue on the surface of the Ir catalysts [177]. The coverage of carbon varied between 0.4 (T = 400 K) and 1.0 ML (T = 600 K). The departure from Arrhenius behavior seen in fig. 69 for propylene

T,K T,K 6 0 0 5 0 0 4 0 0 6 0 0 5 0 0 4 0 0

IOZ l i i i ~ I 0 z i i m i m

I r { l l l ) PHC : 2.0 PHZ = IOOTorr-

T~ T,,

i11

0 E

LU" I0-~

13: z _o I-- L.) <I" [] Elhane la k w rr" 10 -4 I I Ethylene \ Q'N

zx Propone

I I I I I ~ ' ~ ' / l I 1.6 2 . 0 2 . 4

I O 0 0 / T

' I

(:~ I0 -2 OC

Z 0 I-- (o

10 -4

Ir[llO)(I x 2)

[] E1hone \ "~R [] Ethylene \ "-~ zs Propone k " x

P.c o 2.o

PH2 = I00 Torr ~ ' ~ v

I I I I I 1.6 2.0 2.4

I O 0 0 / T

Fig. 69. Specific reaction rates for the reaction of propylene and hydrogen (a) on Ir(111) and (b) Ir(110)-(1 x 2). The partial pressure of propylene was 2 Torr, whereas that of hydrogen was 100 Torr (from ref. [177]).


(O } Propylene

T n, ~ I r ( l l l ) J

"-~ o_Methane ~ T=575K / I o Ethane 1 -'

0 ~ Ethylene / E u"

n,-

7 0.1 o } -

< uJ n,-

10-2 i ] i i i i i i i 1 . 2 . . . . . . . . . I I I I I ' ' , t , i I I , ,

0.8

Oc

0.4

PHz = IO0 Torr

0 I I I I I I I I I J I I I I I I I l O.01 0.1 I

PHC , T0 r r

Fig. 70. (a) Specific reaction rates for the reaction of propylene and hydrogen on the I t ( ! ! ] ) surface at a temperature of 575 K. The partial pressure of hydrogen was maintained at ]00 To r r as the partial pressure of propylene was varied. (b) The fractional coverage of the carbonaceous residue at the corresponding reaction conditions (from ref.

[177]).

hydrogenolysis at 480 K was associated with a large increase in the coverage of the carbonaceous residue and a reduction in the coverage of hydrogen adatoms [177]. Fig. 70 illustrates the dramatic effects of the PHJPc3H~ ratio upon the coverage of the carbonaceous residue and the selectivity of the C 3 H 6 -I- H 2 reaction at a temperature of 575 K [177]. Similar experiments carried out at 450 K showed an essentially constant value of O c - 0.5, with a rate for propane production 10-100 times larger than those for production of methane, ethane and ethylene [177].

9.3. 1,3-Butadiene hydrogenation

9.3.1. Pt(lO0), Pt(110), Pt(111) and Nio.sPto.5(111) surfaces The hydrogenation of 1,3-butadiene was investigated over the (100), (110) and (111) faces

of platinum at T = 300-550 K, PH2 = 100-500 Torr and Pc,H~ = 5 - 3 0 Torr [187-189]. Fig. 71


\ ~ - J P~ = 660,10~ pa

0 ~ i - - ' - - , ~ " ~ Jl. 10 20 30 40

TIME, minutes

100

% @ Pt (1~/1)

PH= "660,102pa / B u t a n e

PC.H~ = 27~102 Pa

T= 413K >

O n l .J

('I" . . .- 0 50 100%

C4H6 CONVERSION Fig. 71. ] ,3-Butadiene hydrogenat ion reaction on Pt( ] 11); (a) composi t ion of the gas phase as a funct ion of t ime; (b) selectivity as a funct ion of conversion. " B - I " : ] -butene, "B-2-cis": 2-cis-butene, "B-2-trans": 2-trans-butene. T = 4]3

K, PH: = 495 Tor r and PCaH~ = 20.3 Tor r ( f rom ref. [189]).

shows a typical product distribution for butadiene hydrogenation on P t ( l l l ) as a function of reaction time. In all the cases examined, butane and 1-butene were the major initial products, with minor formation of 2-cis- and 2-trans-butene. Fig. 72 shows initial rates of butadiene consumption on P t ( l l l ) and Pt(100) as a function of inverse temperature. At 330 < T < 480 K, the apparent activation energies were: 138 kJ mo1-1 on Pt(100) [188], 164 kJ mol -~ on Pt ( l l0) [187] and 159 kJ mol-1 on P t ( l l l ) [189]. The initial turnover numbers increased with the roughness of the surface, ( l I D < (100) < (110), with a ratio of 1 to 4 [189].

Auger spectra taken after reaction conditions showed large amounts of carbonaceous deposits on the surface of the Pt catalysts [188,189]. After any reaction at temperatures below 480 K 0 c varied between 1 and 1.5 carbon atoms per platinum surface atom. At these conditions the carbonaceous adlayers were probably formed by butadiene and C4H 7 entities which directly participated in the reaction [189]. Butadiene is stable under vacuum without


E o

o E loc o

o

c o o,.

E

c o

E 4 1.5

®

Pt (100) PH= = 5 0 0 T o r r ~ ~ePc4Ha = 20 Torr

2

101

®

pt(m) PH 2.,~ 330,102 Pe Pc4HeO27,102pa

\

C4He t l o n

2 2.5 3 1.5 2 2.5

IO00/T, K "1 IO00/T, K "1

Fig. 72. Initial rates of 1,3-butadiene hydrogenation (a) on Pt(100) (from ref. [188]) and (b) on P t ( l l l ) (from ref. [1891).

loss of hydrogen until 390 K; it appears that under an excess of hydrogen and at temperatures below 480 K the molecules do not dehydrogenate on the Pt surfaces [189]. Above 480 K, a continuous decrease in the rate of hydrogenation was observed when the temperature was increased (see fig. 72). This correlated with an increase in the carbon coverage measured after reaction: 2.5 to 3 ML or even more [189]. This deactivation phenomenon was probably the result of an irreversible decomposition of butadiene.

The order of the hydrogenation reaction with respect to 1,3-butadiene was zero on all the surfaces, while the order in hydrogen pressure varied from 0.5 for Pt(100) to 1 for Pt(110) and P t ( l l l ) [189]. On Pt(100) and P t ( l l l ) , the selectivity did not depend on PH2 but changed considerably when PC4H6 varied [188,189]:

The following reaction scheme has been proposed for the hydrogenation of 1,3-butadiene on Pt surfaces [187,188]:

H2,g --~ 2 Ha, (41)

C4H6,g ~ C4H6,a, (42)

C4H6, a + H a ~ C4H7,a, (43)

C4H7, a q- H a ~ CallS,a, (44)

C4H8, a ~ C4H8,g, (45)

C4H8, a + H 2 --* C4Hlo,g. " (46)

The rate-determining step is probably reaction (44), which corresponds to hydrogenation of the C 4 H 7 intermediate. This intermediate exists in two different forms: one leading to


40

m (loo)

PH~ 500Torr PC4H6 = 20Torr

= •

20

W

E 0 Q

0 E

o

E

i 0.25 0~5

i pt(~11) 100 PHz = 660"102Pa

PC4H6: 27,102pa '7 T .413K

a0 o

60

% 40[ l

20

0 ~ 0 0 .25 0.5

OS OS

Fig. 73. Rates of 1,3-butadiene hydrogenation as a function of S coverage (from refs. [188,189]).

1-butene, the other to 2-butenes (cis and trans). 1-Butene can either desorb or be hydrogenated into butane [188,189].

Sulfur adatoms poison the hydrogenation of 1,3-butadiene on Pt(100) [188], Pt( l l0) [187] and P t ( l l l ) [189]. Fig. 73 displays typical deactivation curves. At high coverage, 0 s = 0.5, the residual activity could be due to defects in the surface or in the S adlayer. The electronic effect usually observed in sulfur poisoning does not exist for butadiene hydrogenation on Pt surfaces [187]. For 0 s = 0.2-0.4, sulfur adatoms did not change the apparent activation energy of the reaction [187-189]. On Pt(100) [188] and Pt(111) [189], sulfur preferentially poisons the sites of 1-butene hydrogenation into butane, decreasing the selectivity toward butane production. For the sulfur-modified surfaces butadiene hydrogenation was first order in 1-I 2 partial pressure [188,189].

The hydrogenation of 1,3-butadiene was examined over the (111) face of a Pt0.sNio. 5 alloy [191!. Although the bulk concentration of the alloy was 50 at% in each component, the surface was largely enriched with platinum. On the Pt0.sNi0.5(lll) sample a quasi-complete platinum surface layer (less than 2-4 at% nickel) was present. This platinum outer layer showed novel catalytic properties compared to pure Pt [191]. Fig. 74 compares results for 1,3-butadiene hydrogenation on P t ( l l l ) and Pt0.sNi0. 5 (111). The main features are a higher activity and a better selectivity toward butenes on the alloy sample as compared to pure platinum. The initial activity, measured from the decrease of the butadiene partial pressure, is about 3-4 times larger on the Pt-Ni catalyst.

The different behavior of the alloy sample compared to pure platinum seems surprising, since on the alloy sample an almost complete platinum outer layer is formed. However, either the Ni atoms in the sublayers may influence the surface metallic atoms or the underlying metal can cause an epitaxial strain in the outer layer, which modifies the physical and chemical properties of the platinum surface atoms [191]. The observation of a strongly contrasted (1 × 1) LEED pattern suggests that the surface atoms of the alloy are in registry


1 0 0 . . . . . .

a -- P t(111)

1 I I I I 1 I I o 1 5 t t h )

- . . . . . . . . . . . . . - - . . . . . . . . . . . - - - - -

b - - P t50 N i50 (111)

1 1.3 bu t ad iene

1 5 t ( h )

Fig. 74. Conversion versus time for 1,3-butadiene hydrogenat ion on: (a) Pt(111) and (b) Pt0.sNi0.s(ll 1). PH, = 660 Torr, P¢4H6 = 4.5 Tor r and T = 300 K (from ref. [191]).

with the bulk lattice, which is contracted by 4.3% for the alloy compared to pure platinum [1911.

9.4. Cyclohexene hydrogenation

9.4.1. Pt(223) surface Product accumulation curves for the hydrogenation of cyclohexene on Pt(223) are shown in

fig. 75 [190]. The atomic structure of the catalyst consists of terraces of (111) orientation that are five atoms wide, separated by steps of (100) orientation one atom in height. At temperatures below ~ 370 K, self-poisoning did not occur and a large fraction of the adsorbed species that accumulated on the surface was reversibly desorbed at ~ 390 K as molecular benzene. At temperatures higher than 400 K, self-poisoning became rapid and largely irreversible with the formation of extensively dissociated species with multiple metal- carbon bonds [190]. This strongly chemisorbed overlayer decomposed on further heating under vacuum with the evolution of hydrogen and only minor traces of benzene [190].

The turnover frequencies and reaction probabilities at 423 K for the hydrogenation and dehydrogenation of cyclohexene in excess hydrogen are summarized in fig. 76 [190]. The Pt(223) crystal predominantly produces benzene at low pressures (~ 10-7 Torr) and cyclohexane at high pressures (~ 102 Torr). The catalytic activity observed at low pressures is characteristic of a clean metal surface. At high pressures the initially clean surface rapidly

86 J.A, Rodriguez and D.W. Goodman

l H I I I [

O " Z - ~ - - ~ O Pl(223) TCC) J

20 / Hz/HC = IO - ~ I 50 - - /

4 25

o IO 20 30 Reecfion Time'(minutes)

Fig. 75. Product accumulation versus reaction time for cyclohe×ene hydrogenation over Pt(223) at several temperatures (from ref. [190]).

becomes covered with a near-monolayer of reversibly and irreversibly chemisorbed species. The catalytic behavior is then characteristic of an extensively precovered Pt surface [190].

In fig. 77 the initial turnover frequencies obtained for the hydrogenation of cyclohexene on Pt(223) as a function of temperature are compared with those observed on supported Pt catalysts [190,194]. The two sets of data agree well, supporting the view that the reaction at

s~

1.0

I 1 I I I Pt (223)

16 2

t6 4

16 6

t I I i I 1 1 t I ~ i p

o---G- /

I~ l I I I I I I 10 -8 10-6 10"4 13 -2 1.0 ~2

* U

O)

t~

E z

0 t -

I.O

10 z

~5 o

10"4 c~

U

I I I I I O " e I0 8 t0 -6 IO -4 I0 -2 I.O IO 2

Cyglohexene Pressure ( Tor t )

Fig. 76. Correlat ion of cyclohexene reaction rates and reaction probabilities over a ten orders of magni tude pressure range. The reactions were per formed at 423 K over Pt(223) with PH2/Pc6Htlj = 10 (from ref. [190]).


I o

~ 2o .~_

0 (Iv

~ 8

. D

E ~ 4 z

g r -

l- '-

1 i 1 i I I I

_ o

- - Plot -" 77 Tor t , ~ "~1

_ Pt/Si o i

i t I I I I I | 2.2 2.6 3. 0 5.4

IO00/T (K "~) Fig. 77. Arrhenius plot for cyclohexene hydrogenation over Pt(223) (from ref. [190]). For comparison results for a

high-surface-area Pt /SiO 2 catalyst are included (from ref. [194]).

high pressures is structure insensitive [190,194]. The apparent activation energy is 20.9 + 2 kJ mol-~ on the Pt(223) surface.

1 0 . D e h y d r o g e n a t i o n o f c y c l o h e x a n e

The aromatization of cyclohexane and alkylcyclohexanes is an important reaction in petroleum refining processes [134]. The dehydrogenation of cyclohexane has been extensively studied as a model reforming reaction [183]. In industrial bifunctional catalysts aromatization of cyclohexane occurs almost exclusively on the metal component [134]. At temperatures between 460 and 590 K, platinum is more active than palladium, which appears to be more active than osmium, ruthenium, and nickel in decreasing order of activity (Pt > Pd > Os > Ru >_ Ni) [183]. Continuous deactivation is usually observed due to the formation of a carbonaceous layer on the surface of the catalyst [183]. This layer is formed by reversibly adsorbed benzene and by strongly (sometimes irreversibly) chemisorbed carbon deposits.

10.1. Kinetics on Pt surfaces

The adsorption and dehydrogenation of cyclohexane on Pt( l l l ) have been extensively studied [195-202]. A sticking coefficient of near unity was found for c-C6Hl2 on clean Pt ( l l l ) at 100 K [201b]. The heat of chemisorption of the molecule is ~ 58 kJ mol -~ [201]. At ~ 235 K a fraction of the adsorbed c-C6Ht2 molecules desorb, whereas the rest begin to dehydrogenate and finally form benzene by 350 K [201,202].

The interaction between hydrogen (100 Torr) and cyclohexane (15 Torr) was studied on the flat (111), stepped (557), and kinked (25,10,7) and (10,8,7) faces of platinum [180]. Benzene, cyclohexane, n-hexane and alkane fragments with carbon number less than 6 were observed as reaction products. Typical results obtained with the (10,8,7) surface, depicted in fig. 78,


60 1 I I I i t I

- 0 ' ' ~ Pr°ducts /x 5/~00 15 torr Cyclohexone . ~ I00 tort Hydrogen . , ~ .,, _ _

" /

0 I I I I l I I 0 40 80 120 160

Time (minutes)

Fig. 78. Product accumulation curves for the reaction of cyclohexane and hydrogen on the Pt(10,8,7) surface (from ref. [180]).

illustrate the selectivity and deactivation behavior of the four types of reaction products. Over all the investigated surfaces, the major dehydrogenation product was benzene [180]. Fig. 79 shows the production of cyclohexene and benzene as a function of surface structure. The initial rate of benzene production increased in the order (111)< (557)< (10,8,7)< (25,10,7). In fig. 79 the decrease of the dehydrogenation rates was due to reversible product inhibition by benzene, as well as the irreversible adsorption of tightly bound carbonaceous deposits [180]. Product poisoning was found to be structure sensitive, increasing in the sequence (25,10,7) < (10,8,7) < (557) < (111). There was a clear correlation between the activity of the surfaces and their sensitivity to product poisoning [180].

The apparent activation energy associated with the initial rate of benzene production was 71.2_+8.4 kJ mo1-1 on Pt(111) and Pt(10,8,7) [180]. Dehydrogenation to benzene was approximately first order in cyclohexane pressure [180]. Fig. 56 illustrates the dependence of the initial rates of dehydrogenation and hydrogenolysis on hydrogen pressure over the (111) and (10,8,7) surfaces [180]. On both surfaces, at hydrogen pressures less than 400 Torr, benzene production is positive order in hydrogen. Above 400 Torr the reaction becomes zero order in hydrogen pressure on the (10,8,7) surface, but remains positive order up to at least 745 Torr on the (111) surface.

The dehydrogenation of cyclohexane (Pc-c6H,2 = 15 Torr, PH2 = 99.75 Torr, T = 573 K) has been studied over Au-Pt(111) surfaces prepared by vapor-depositing Au onto the (111) face of Pt to form either an epitaxial overlayer or an alloy system [203]. Cyclohexane was converted into benzene, cyclohexene, and very small amounts of n-hexane and smaller hydrocarbons. The latter two classes of reactions were several orders of magnitude slower than dehydrogenation. The influence of Au on the amount of benzene and cyclohexene formed in 123 min reaction time is shown in fig. 80. There is an enhancement of the benzene production by gold, which reaches a maximum at a surface gold content of about 50%. The enhancement was observed with both the epitaxial and the alloy surfaces, although it was generally larger with the surface alloys. For cyclohexene production the maximum enhancement was found at a


10

'E 8

6

%

, j 4 C oa N C

= 2 m

0 0

J I I I I I I

O - © + _ 15 torr Cyelohexone ~ . . . I ~ ' ~ (25,10,71 _

IO0 tort Hydrogen 573 K S -

/ j...wb.'~'~~ (557)

1 I 1 I t I I [ 40 80 120 160

Time (minutes)

= .

I I I I I

0 - . . - 0 + 15 Tort Cyclohexone ~ ( 1 1 1 ) 100 Torr Hydrogen ~ _

8, s 7 3 K - ~ _(tO, B,t~ Oo " " y J " E 6- 71 _ ~ e 4 ~ 1 5 5 7 1

t "

_o

¢j

0 t l . I I I . I 0 40 80 120

Time (minutes)

Fig. 79. (a) Benzene and (b) cyclohexene production from the c-C6H12 +H 2 reaction on several platinum surfaces (from ref. [180]).

surface Au concentration of about 90 at%, which differs considerably from the composition that gave maximum benzene production. The catalytic activity that is observed at high Au coverages is due to contributions from the crystal edges and imperfections that are not covered by gold [203].

Au influences cyclohexane dehydrogenation by restricting the number of Pt atoms available for interaction with adsorbed benzene [203]. When Pt( l l l ) is alloyed with small amounts of Au, deep dehydrogenation of benzene no longer takes place and the product poisoning is reduced. The resulting activity increase is larger than the reduction of the Pt surface area. At an intermediate gold surface concentration, the activity reaches a maximum and declines thereafter as the ensembles of Pt atoms become too small to catalyze cyclohexane dehydrogenation down to benzene [203]. At this point, the cyclohexene-like intermediate that is formed during the dehydrogenation of cyclohexane is no longer quantitatively converted to benzene but it desorbs as cyclohexene. Consequently, the maximum of cyclohexene production is reached at higher gold concentrations than that of benzene formation [203].


C 0

CJ

c

"o

E o

0

E <

3 0 -

7 CD

X

2o 0

?_

A u - P t [111)

,:,-? /

/ o i • I

o I I I o

I

@ - @

-o- A l l o y s

-o- E p i t o x i a l

0 ~ l i - I

0 1 ~ ' Gold C o v e r a g e ( M o n o [ a y e r s ) - o -

I

6 o'.s Gold S u r f a c e C o n c e n t r a t i o n -o-

A u - Pt ( I l l ) c o Q @

*d a ~ a 200- (D

~- ~ - - ~ A l l o y s

o ~ E p i t o x i a l

~-~ ~10o - /~ - E o ~ . . . . . . . . _ ~ - ' - ' j ~ - - - - - - - . - ~ _

! o 0 4 - i "~ ~ t l E

< 0 1 2 3 Gold C o v e r a g e ( m o n o l a y e r s )

t - 0 e.s Gold S u r f a c e C o n c e n t r a t i o n - c -

Fig. 80. Influence of Au surface content on the amount of benzene and cyclohexene produced from cyclohexane dehydrogenation over Au-P t (1 l 1) in 123 min reaction time (from ref. [203]).

High-pressure catalytic reactions oL'er single-crystal metal surfaces 91

10.2. Kinetics on Ru surfaces

Fig. 81 shows rates of cyclohexane dehydrogenation and hydrogenolysis on Ru(001) as a function of reaction time and carbon coverage present on the catalyst surface after reaction [181]. The rate of benzene production over initially clean Ru(001) was quite low but rose with time to reach a constant value when the surface was saturated with carbon. During the induction time, the formation of lower carbon number alkanes from the hydrogenolysis of cyclohexane was much higher than the rates obtained at steady state. The carbon-modified Ru(001) surface likely stabilizes cyclohexane (and possibly other intermediates leading to the product benzene) with respect to carbon-carbon bond scission [181].

Specific rates of cyciohexane dehydrogenation to benzene obtained under steady-state reaction conditions are plotted in Arrhenius form in fig. 82a. The apparent activation energy measured, ~ 75 kJ mol - l , is very similar to values of 63-88 kJ mol- I obtained in Pt single-crystal and supported-catalyst studies [181]. The response of the reaction to changes in the partial pressure of the reactants is shown in fig. 82b. The rate of benzene formation is zero order in cyclohexane pressure and approximately first order in hydrogen pressure, in good agreement with results on supported catalysts [181].

Fig. 83 shows the effect of addition of Cu to Ru(001) upon the rate of cyclohexane dehydrogenation [181]. The overall rate of the reaction is seen to increase by almost an order of magnitude at a Cu coverage of 0.75 ML. Above this coverage, the rate fails to an activity approximately equal to that of Cu-free Ru(001). The observed non-zero rates at the higher Cu coverages are believed to be caused by three-dimensional clustering of the Cu overlayers [181]. The rate enhancement observed for submonolayer Cu deposits may be related to an enhanced activity of the strained Cu film for this reaction due to its altered geometric and

2 0 I l I I 1 0 0 2 I i i I

: " " " . j : ( 4 )

= ='Z" , s 7s ,,,

==-'= (~== = 0 6

' ! L / ~- ~ 5o 0.4

I - " ' u - o 3 Z 0 . 2 Z:E O ~ 0

>.~ ~ o. :DO ~ - - 2 5 i _ ' ~ 0

~ 0.8

I " ~ (3) "~ ',' • (5 ) IZ; 0 . 4

~J Z O.S 0 . 2

=o , ® o o ® - 0 . 0 I I I I I

0 5 1 0 1 5 2 0 2 5 0 . 2 0 . 4 0 . 6 0 . 8 1 . 0

REACTION TIME (minutes) RELATIVE CARBON COVERAGE

Fig. 81. (a) Amount of (1) benzene and (2) lower carbon number alkanes formed from the c-C6H 12 + H 2 reaction on Ru(001) as a function of time; (3) relative quantities of carbonaceous material present following reaction. (b) Relative rates of (4) cyclohexane dehydrogenation and (5) hydrogenolysis as a function of relative carbon coverage present on

the surface following reaction at T = 650 K, H 2 :C6H12 = 100, Ptotal = 101 Torr (from ref. [181]).


O,OSO

0

0 ,0=0

Z _o I,- E 0.01o

m £ w Z

~ 0 . o o | Z ~ : IO

0 .0o2 I.s

C6Ht 2 R u ( O 0 1 ) C6H6. t " 3 H 2

| I

® I I

1,6 1.7

1 / T x 10 = (K -z )

1.8

UJ I -

Z o_

:E

0 u .

uJ 2:

z uJ m

5 2 0 0 0 . 0 5 0

0 . 0 2 0

0 . 0 1 0

0 . 0 0 5

H 2 P A R T I A L P R E S S U R E ( t o r r )

10 2 0 5 0 1 0 0 . . . . i I " I . . . . I

11) = "I

(21

® 0 . 0 0 2 . . . . I s , . I . . . . I

O.S 1.0 2 .0 S.O 10 .0

C 6 H 1 2 P A R T I A L P R E S S U R E ( t o r r )

2 0 . 0

Fig. 82. (a) Arrhenius plot of the rate of cyclohexane dehydrogenation to benzene over Ru(001) under steady-state reaction conditions, H 2:C6Ht2 = 100 and Ptotal = 101 Torr. (b) Cyclohexane dehydrogenation rates over Ru(001) at

650 K as a function of reactant partial pressure: (1) PH2 = 100 Torr and (2) Pco~,2 = 1 Torr (from ref. [181]).

electronic properties [181]. Alternatively, a mechanism whereby the two metals cooperatively catalyze different steps of the reaction may account for the activity promotion [181]. Dissocia- tion of H 2 on bulk Cu is difficult due to kinetic constraints. In the combined Cu/Ru(001) system, Ru may function as an atomic hydrogen source/sink via spillover to/from neighboring Cu [181]. In addition, Cu may serve to weaken the chemisorption bond of benzene, limiting self-poisoning by the adsorbed product [181].

Z 10.©

o_ I-

Z *.0 IC 0 L

w z

z w m

0 4.0

E ~ 2.0 I-

w a-

Fig. 83. Relative rate of

| I I 1

o.o o.s i .o I .S 2.o 2.s

C O P P E R C O V E R A G E ( m o n o l a y e r s )

cyclohexane dehydrogenation as a function of surface Cu coverage on Ru(001) at T = 650 K, H 2 : C 6 H 12 = 25 and Ptotal = 104 Tort (from ref. [181]).


10.3. Kinetics on W surfaces

The dehydrogenation of cyclohexane to benzene was studied over W(l l0) at PH2----- 100 Torr, Pc_c,H,2 = 1 Torr and T = 6 3 0 K [204]. A turnover frequency of 0.005 benzene molecules/si te .s was observed. The dehydrogenation activity of the W(l l0) surface was increased by adsorbing SiO x species [204]. A W(l l0) catalyst covered by a monolayer of silicon oxidized 35% showed a turnover frequency of 0.019 benzene molecules /s i te . s , i.e., about four times the activity of clean W(110).

11. Cyclotrimerization of acetylene

11.1. Kinetics on Pd surfaces

The conversion of acetylene to benzene on Pd surfaces is a reaction which proceeds readily at conditions ranging from UHV (10 -9 Torr) to atmospheric pressures (103 Torr) [205-209]. The cyclization of acetylene to benzene occurs on Pd( l l l ) , Pd(100) and Pd(l l0) at temperatures below 250 K [205]. The results of fig. 84 show a structure-sensitive reaction. In UHV the (111) face is the most active followed by the (110) and then the (100) surface [205]. At high pressure the (111) and (100) surfaces have equal catalytic activity whereas the (110) face is one-fourth as active [205].

At high-pressure conditions, an apparent activation energy of 8.4 kJ mol - ' was observed for the trimerization reaction over Pd(ll0), Pd ( l l l ) and Pd(100) (see fig. 85a). The reaction rate was first order in acetylene partial pressure (see fig. 85b). Analysis of the catalysts by AES after reaction showed partially carbon-covered surfaces [205]. It appears that acetylene cyclization takes place on top of the bare metal sites. The irreversibly bound carbon limits the rate of reaction by blocking the active sites [205]. Pd(ll0), the face with lowest catalytic activity at high-pressure conditions, is the most open surface and leads to the highest decomposition rate of both acetylene and benzene, showing a low percentage of bare Pd atoms [205].

"10

CC,,- 0

(111)

3C2H 2 Pd C6H6

~ U H V

[ High pressure

(110) (100)

Pd surface Fig. 84. Relative amount of benzene formed from acetylene cyclotrimerization on Pd single crystals in UHV and at

atmospheric pressures (from ref. [205]).

94 J.A. Rodriguez and D. W. Goodman

o

c~-5

- 3

i ]

Acetylene 260 Torr Nitrogen 1040 T0rr

E a = 2 kcal/mol

( ~ 170°C 25°0

1.5 2.0 2.5 3.0 3.5

lfTx 103(K -1)

O O

X

O t - O

O"

> 0 t'-

I---

7 - i T I

~ ~ 20PSI total pressure/(111)

, /

/ /® F 1 I

100 200 300

Acetylene pressure (Torr) Fig. 85. (a) Apparent activation energy for acetylene cyclotrimerization over Pd. The same value was obtained on all t h r e e low-Miller-index planes of palladium [205]. (b) Turnover frequency for acetylene cyclization as a function of

acetylene partial pressure (from ref. [205]).

The mechanism for acetylene cyclotrimerization on metal surfaces under high pressure is not unequivocally established. A possible reaction scheme involves the concerted cyclization of three acetylene molecules [210]. Another reaction path includes the formation of a surface C4H 4 species as a key intermediate [209,211,212]. This is followed by the insertion of another acetylene to form either an "open" C 6 H 6 species and then benzene or benzene through a Diels-Alder-type addition [211,212].

12. Water-gas shift reaction

The water-gas shift reaction (CO + H20- -~ H 2 + CO 2) is widely used industrially in various hydrogen production or enrichment processes [10,213]. Many materials are able to catalyze this reaction [10,213]. Originally, the most commonly employed industrial catalysts were based on iron oxides and operated at high temperatures (570-820 K) [10,213]. A


substantial improvement came about by the development of Cu /ZnO-based catalysts, which operate at relatively low temperatures (470-530 K) and allow higher thermodynamic conver- sions [10,213]. In these catalysts, copper is the active species, and the principal role of the zinc oxide is to act as a support for the copper [10,214-218]. It is not clear why ZnO is a superior support [213-218]. In this section we review the results of studies in which the kinetics and mechanism of the water-gas shift reaction have been investigated using the modern techniques of surface science and copper single crystals.

12.1. Kinetics on Cu(llO) and Cu( l l l ) catalysts

Fig. 86 illustrates the effect of temperatures on the rate of the water-gas shift reaction over Cu( l l0) [215] and C u ( l l l ) [214] surfaces. For the same geometric area, Cu( l l0) is 2.5-7.0 times more active than C u ( l l l ) between 550 and 650 K [215]. On a "per copper surface atom" basis, this difference is 1.63 larger due to the higher surface atom density of Cu( l l l ) . The slopes in fig. 86 give apparent activation energies of 41.9 and 71.2 kJ tool - t for the water-gas shift reaction on Cu(110) and Cu(111). These results indicate that the reaction is structure sensitive on copper surfaces, a fact that is consistent with data for high-surface-area supported catalysts, which show an increase in the catalytic activity as the Cu particle size decreases [218]. High-area-supported and unsupported catalysts show apparent activation

T E M P E R A T U R E / K

0.4

0.2

o.o CJ

.~ -0.2 e~ qJ • ~ -0.4

-0.6

-0.8

-1 o0 i

~., - 1 . 2

r~ -1.4 0

-1.6 1 , 4

700 650 600 550 I I I i

Cu (110) e

Ea= 17kcal/mofe " ~

,-... iH,o+co... :co \ I ,0Tor H,O I , , \ I 26 Torr CO I \

0 H2 ' O ~ Z~ C02

I I I I 1.5 1.6 1.7 1.8 1.9

1000 K / TEMP.

Fig. 86. Variation of the wa te r -gas shift reaction rates on Cu( l l0 ) and Cu( l l 1) with temperature, in Arrhenius form. PH2O = 10 Torr and / ' co = 26 Torr (from refs. [214,215]).


energies in the range of 54 to 67 kJ moi -~ [87], well within the range between Cu(110) and C u ( l l l ) (42-71 kJ tool- l ) .

A study of the influence of CO and H 2 0 partial pressures on the reaction rate over Cu(110) [215] and Cu(111) [214], revealed that on both surfaces the rate is nearly independent of Pco (order = zero) and strongly positive order in H 2 0 (0.5-1.0). Analysis of the surface of the catalysts with AES and XPS after reaction conditions showed essentially oxygen-free copper surfaces [214,215]. To further prove that Cu surfaces are fully reduced under water-gas shift conditions at low conversion, experiments were carried out using a heavily preoxidized C u ( l l l ) crystal [214]. After a few minutes reaction time, the surface displayed a fully metallic Cu(2p) XPS spectrum, and gave no oxygen signal in AES or XPS [214]. This suggests that metallic Cu is the active ingredient for high-surface-area C u / Z n O or Cu-based catalysts [214].

Experimental evidence supports the following mechanism for the water-gas shift reaction on Cu( l l0) and C u ( l l l ) [215]:

H2Og ~-- H2Oa, (47)

H 2 0 a --+ OH a + H a, (48)

OH a ~ O a + H a, (49)

2 H a ~ H2,g, (50)

COg ~ CO a, (51)

CO~ + O a ~ CO2,g. (52)

In this mechanism the rate-determining step involves O - H bond cleavage in H 2 0 a (reaction (48)). The enhanced catalytic activity of Cu( l l0) compared to C u ( l l l ) has been attributed [215] to a lower barrier for O - H bond cleavage on the Cu( l l0) surface. Reaction (49) may not necessarily be an elementary step, but a consequence of the process [215]

H2Og ~ H2Oa, (47)

H 2 0 a ~ OH a + Ha, (48)

2 OH a ~ Oa + H2Oa, (53)

H2O a ~ H2Og , (47)

net OH a ~ O a + H a. (49)

Reaction (53) provides an easy pathway for converting OH a to O a, which proceeds rapidly even at 290 K on Cu( l l0) [219,220].

Fig. 87 shows a potential energy diagram for the water-gas shift reaction on Cu(ll0). The diagram was constructed [215] using kinetic and thermochemical data for reactions (47) to (52). The activation energy for the rate determining step ( H 2 0 a ~ OH a + H a) is ~ 84 kJ mol -~. A value of 113 kJ tool - l has been estimated for this step on C u ( l l l ) [214]. This difference is probably due to the fact that the (110) plane is much more open, offering Cu surface atoms which are much more coordinatively unsaturated and hence more active for breaking O - H bonds.


0 u'l

I I i I I I

.-~+C

,.4- o / d

:.-.-.-~ + +

I o"

C.,

" 0 o : : ~ + +

d

0

~. , .o o .,1,- 0

l , , I , l , , , , I f l l r l i i , I I , t l t l r r r , l

r--.,

= 0

S

^

6 , , .

UJ o

z

r r

o ? g ° _ ~

0 -

~ E

0

._=

(el0w/re0~t) / AE)&I3N3 -IVI_LN=I.LOcl


12.2. Sulfur poisoning of Cu(l ll) catalysts

The poisoning of C u / Z n O catalysts by sulfur is one of the most serious problems in the water-gas shift process [213,221]. Fig. 88 shows the effect of pre-adsorbed sulfur atoms upon the rate of the water-gas shift reaction over C u ( l l l ) [221]. Sulfur addition causes a linear decrease in the rate of the reaction, with the rate going to zero at saturation sulfur coverage (0 s-- 0.39). The decay in the rate with 0 s can be easily understood in terms of a simple site-blocking model, where sulfur adatoms sterically prevent the dissociation of water [221]. The rate decay of fig. 88 is well fit by the expression 1 - 2 . 6 0 s. According to a statistical analysis [221], this linear function indicates that each sulfur atom blocks about 2.6 Cu atoms, and that the ensemble required for HzO dissociation is rather small (1 or 2 Cu atoms). This is not unexpected, given the small size of the H eO molecule and its dissociation products (OH~ + H,) .

12.3. Cesium promotion of Cu(lll) and Cu(llO) catalysts

Cesium has been shown to promote the water-gas shift reaction over C u / Z n O catalysts [222,223]. Kinetic data for the reaction on Cs-promoted Cu( l l0) surfaces [215] are displayed in fig. 89. The optimum Cs coverage to promote the reaction is 0cs = 0.25 ML. At this coverage, the reaction rate is five times faster than the clean surface rate at any reaction temperature [215]. The coverage dependence and optimal coverage for cesium promotion observed over C u ( l l l ) are similar to those shown in fig. 89 for Cu(l l0) [224].

An apparent activation energy of 46 kJ mol - ] was determined for the water-gas shift reaction on an optimally promoted C s / C u ( l l 0 ) surface (0cs = 0.27) [215]. This activation energy is very close to the value of 41.9 kJ tool-I found for the clean surface. A tendency toward a slightly higher apparent activation energy was also observed in experiments with Cu( l l 1), where the optimally promoted Cs-covered surface had an activation energy of 83.7 kJ mol - l versus 71.2 kJ mol - t for the clean Cu(111) surface [224].

A study of the pressure dependence of the water-gas shift reaction over optimally promoted C s / C u ( l l 0 ) showed that on this surface the reaction orders in Pco and PH20 were very different from those on clean Cu(l l0) [215]. This fact rules out the possibility that the same elementary step is rate determining with Cs present or absent.

U) .=_ ¢0 Q)

0

0 E 0 qm,

u: 0 I--

0,1 0.2 0.3 0.4

°'° i~ ~ ~ C. l l . ) 10 • ~ PH2 O = 10 tOrr

8 ' \ . Pco =26 to r t

2 (") =H2 ' ~

(o) =co= , \ 0 o.o o11 o'.2 0'.3 o14 0.5

es

0 ~==

I1. 0 I--

0.0 0.1 0.2 0.3 0.4

Cu(111)

PH20 = 4 torr PCO = 26 torr

(.) = H2 , ~

(o) =co2 0.0 0.1 0.2 0.3 0.4 0.5

S/Cu AES RATIO

Fig. 88. Rate of wate r -gas shift reaction over Cu(l 11) as a function of sulfur coverage (from ref. [221]).


Cs/Cu AES Ratio

0 0.02 0.04 0.06 0.06 0.10 1.6

0.6

o l I I I I P 0 0.1 0.2 0.3 0.4 0.5 0.6

1.2

o

[- .<

C e s i u m Coverage (OCs)

Fig. 89. Water-gas shift reaction rate as function of Cs coverage (Oct) on Cs/Cu(ll0) at 493, 523, 553 and 573 K. PH2 o = 10 Tort and Pco = 26 Torr (from ref. [215]).

In order to understand the role of Cs promoters in the water-gas shift reaction, detailed studies were carried out investigating the effects of Cs upon the surface chemistry of H20 , CO and CO 2 on Cu( l l0) [220,225,226]. Post-reaction surface analysis of the Cs-promoted Cu( l l0) catalysts shows that the dominant form of Cs is a surface cesium-carbonate complex (Cs. CO 3) [227]. This same species can be produced by dosing CO 2 to C s / C u ( l l 0 ) under ultrahigh vacuum (UHV) conditions [225]. It decomposes in UHV at 450-600 K (for 0cs < 0.25) via Cs. C 0 3 , a --~ Cs" O a + CO2,g [225]. The equilibrium,

CO2,g d- Cs" O a ~ Cs" CO3,a, (54)

occurs rapidly under water-gas shift reaction conditions [226,227]. The C s - O a species is readily reducible with CO gas via COg + Cs- O~ ~ CO2,g + Cs~ [215,227]. Thus, a mechanism involving the species Cs. O, (produced by the reaction of H 2 0 and Cs [220,226,227]) rather than simply O a may be operative on the Cs-promoted catalysts. The species Cs • O a should be thought of as a cesium-stabilized oxygen adatom, with a dominant bond to Cu and some degree of bonding to a neighboring Cs adatom [220,226].

1 3 . M e t h a n o l s y n t h e s i s

Methanol is synthesized from carbon monoxide and hydrogen by the reaction CO + 2H 2 --* CH3OH. The selective synthesis of methanol is a process of major industrial importance because of the use of methanol as a chemical intermediate, its potential use as a starting material for fuel production, and many other applications [10,228,229]. The most commonly employed industrial catalysts are based on Cu supported on ZnO, or a mixture of ZnO and other oxides such as Cr20 3 [10,228,229]. It has been demonstrated that under certain


conditions Pd is also an active methanol synthesis catalyst, and that the nature of the support has great influence on both the rate and selectivity [230]. Here we review studies that deal with methanol synthesis on Pd(l l0) , Cu( l l l ) , Cu(100), Z n O x / C u ( l l l ) and Cu/ZnO(0001) single crystals.

13.1. Kinetics on Pd(llO)

Fig. 90a illustrates the effect of temperature on the rate of methanol production over a Pd( l l0) catalyst. The rate shows approximately Arrhenius behavior with an apparent activation energy of 77_+ 8 kJ tool -~ and a pre-exponential factor of 8 × 10 4 S - ! over the temperature range from 493 to 553 K [231]. The activation energy is in good agreement with values reported for Pd on "non-interacting" supports such as SiO 2 and basic supports such as ZnO and La20 a [231,233,234]. In addition, the pre-exponential factor determined for Pd(110) agrees well with reported values for P d / S i O 2 [231,234]. The activation energy for Pd( l l0) is, however, significantly different from values obtained on acidic AI20 3 and TiO 2 supports [231,233].

Fig. 90b shows the variation of the reaction rate with pressure at 553 K. Methanol production was roughly first order (1.2 _+ 0.2) in total pressure. The specific rates seen on Pd(l l0) , extrapolated to pressures typical of most methanol synthesis works, are in good agreement with rates observed for Pd dispersed on SiO 2 [231,233,234]. Thus, Pd metal is an active methanol synthesis catalyst, and no specific support interaction is required. Highly interacting supports exhibit rates a n d / o r selectivities substantially different from those of Pd( l l0) [231]. The rates on Pd( l l0) are higher than those seen over Pd supported on zeolite

10 - 2

IJ. 10 - 3 0 I - -

1 0 - 2

®

P, p-

10 "3

10-4 101

®

10 - 4 ~ , . . . . . . . . . . . . . . 0.18 0.19 0.20 0.21 102 103

Pressure (KPa) 1000/T

Fig. 90. (a) Turnover frequency (CH3OH molecules p roduced /Pd-sur face-a tom-s) versus inverse temperature at a total pressure of 244 kPa and a 3.9:1 H 2 / C O mixture (from ref. [231]). (b) Turnover frequency plotted as a function

of total pressure at 553 K, for a 3.9:1 H 2 / C O mixture (from ref. [231]).


and acidic supports (e.g., PdNaY and A1203) [231,233,235], but much lower than those observed on highly active LaaOa-supported Pd [231,233,234].

After reaction on Pd(110), submonolayer quantities of a carbonaceous residue (generally between 0.05 to 0.25 carbon monolayers) were detected by Auger electron spectroscopy on the surface of the catalyst [231]. Significant adsorption of hydrogen into the bulk, determined by post-reaction temperature-programmed desorption, was not observed [231].

13.2. Studies on Cu(lll), Cu(lO0), ZnOx/ Cu(lll) and Cu /ZnO(O001) surfaces

The synthesis of methanol from CO and H 2 was attempted on C u ( l l l ) and over model catalysts consisting of well-defined Cu overIayers on ZnO(0001) [232]. For temperatures between 500 and 600 K, H e + CO total pressures up to 1500 Torr, and reaction times of 40 min, no production of methanol was observed. The same result was obtained when ZnOx films on Cu(111) were used as catalysts [232]. On the basis of these data, an upper limit of < 2 x 10 -3 CH3OH molecu les / s i t e . s was obtained for the catalytic activity of Cu(111), Z n O J C u ( 1 1 1 ) and Cu/ZnO(0001) surfaces [232]. This limit is consistent with the rates expected for high-surface-area C u / Z n O catalysts (extrapolated from a somewhat higher pressure regime) [232].

The synthesis of methanol on Cu(100) was studied in a temperature range of 500-550 K at pressures between 300 and 760 Torr using a gas mixture of CO 2 : CO : H 2 = 1 : 2 : 12 [236]. At 550 K the reaction was first order in total pressure. Turnover frequencies for methanol production as a function of reaction temperature are shown in fig. 91 [236]. The apparent

10-=

"2,

0 ' ~ Z

a ~ IU

Wi 010"

0 10"

10

TEMPERATURE, K 550 525 500

CuO/ZnO CuOlZnOIAl,O~= ~,,~ "~'C~

CuO/ZnO ORaney C~Zn

"~-II V 5% Cu/SI0=

~ n 2.48% Cu/SI0= A 0/Cu(100) "~

O~ ~O~

~ 0 ~ Cu(100) • 0/Cu(100) ~ - -0~ ~ , ~

I I I 1.800 1.900 2.000

1 O00/'r Fig. 91. Temperature dependence of methanol turnover frequencies for Cu(100) (e, o) [236], oxidized Cu(100), (&, zx) [236], Cu/SiO 2 ( l l , v ) [237,238], CuO/ZnO (zx, O) [232], C u O / Z n O / A l 2 0 3 (o) [232], and Raney Cu-Zn (~) [232]. Filled symbols represent data obtained under the actual experimental conditions Pa-= 750 Torr,

CO2 :CO:H 2 = 1:2:12. Open symbols represent data extrapolated to 14.8 atm (from ref. [236]).


activation energy determined from the slope of the Arrhenius plot was 73.4 kJ tool-~ [236]. This E a is within the range of values (40-110 kJ mo1-1) reported in the literature for methanol synthesis over supported Cu catalysts.

In fig. 91 the catalytic activity of Cu(100) is approximately one order of magnitude lower than that of silica-supported Cu catalysts. It appears that Cu-oxide support interactions play an important role in creating active Cu ~+ sites for methanol synthesis [236]. The extremely low TOF's seen on Cu(100) are probably a consequence of the very low level of ionic copper species present in the surface under reaction conditions. Post-reaction Auger spectra showed a very small amount of oxygen on the Cu(100) surface [236].

14. Conclusions

The combination of a high-pressure reactor with an ultrahigh vacuum surface analytical chamber in the study of catalytic reactions can provide information about the molecular events that control the mechanism of reactions at the gas/solid interface. Using these techniques, the concepts of structure sensitivity and structure insensitivity can be investigated. For structure-sensitive reactions, the activity of a particular site or set of sites can be examined in detail, exploring the effects of atomic structure and electronic properties on the catalytic behavior of a surface. For structure-insensitive reactions (CO methanation, CO oxidation, etc.) excellent agreement has been obtained between studies on single-crystal surfaces and studies on high-surface-area supported catalysts, demonstrating the relevance of kinetics measured on well-ordered single-crystal surfaces for modeling the behavior of practical catalysts.

Single-crystal surfaces offer a unique possibility to investigate the roles of "ensemble" and "ligand" effects on the performance of bimetallic catalysts. The works discussed above show that in some cases well-defined bimetallic surfaces (generated by vapor-depositing one metal onto a crystal face of a dissimilar metal) can have catalytic properties that are very different from those displayed by either monometallic component alone. Model studies on single-crystal surfaces are also helpful in developing a better understanding of the effects of surface additives (alkali metals in ammonia synthesis, CO methanation, water-gas shift, ethylene epoxidation, methanol synthesis, etc.) and poisons (sulfur in CO methanation, alkane hydrogenolysis, ethylene hydrogenation, water-gas shift, etc.) on the catalytic activity and selectivity of metals. The influence of additives and poisons on the surface chemistry of adsorbed reactants, products and intermediates can be studied using UHV techniques, and this information can be related to the effects of the impurities upon the catalytic behavior of the metal surface. Of particular interest is the possibility that these types of studies will help to clarify the relative importance of electronic and geometric contributions in determining additive and poisoning effects.

The types of studies reviewed here, in conjunction with accurate studies on high-surface- area supported catalysts, hold great promise for contributing to an overall understanding of surface-catalyzed reactions.

Acknowledgements

We acknowledge with pleasure the partial support of this work by the Department of Energy, Office of Basic Energy Sciences, Division of Chemical Sciences and the Robert A. Welch Foundation. The authors would like to thank Prof. C.T. Campbell for useful comments during the course of the preparation of this review article.


References

[1] G.A. Somorjai, Chemistry in Two Dimensions: Surfaces (Cornell Univ. Press, Ithaca, NY, 1981); A. Zangwill, Physics at Surfaces (Cambridge Univ. Press, New York, 1988).

[2] G. Ertl and J. Kiippers, Low Energy Electrons and Surface Chemistry (VCH, Weinheim, 1985); D.P. Woodruff and T.A. Delchar, Modern Techniques of Surface Science (Cambridge Univ. Press, New York, 1986).

[3] M.R. Albert and J.T. Yates, The Surface Scientist's Guide to Organometallic Chemistry (American Chemical Society, Washington, DC, 1987).

[4] D.R. Kahn, E.E. Petersen and G.A. Somorjai, J. Catal. 34 (1974) 294. [5] B.A. Sexton and G.A. Somorjai, J. Catal. 46 (1977) 167. [6] D.W. Goodman, R.D. Kelley, T.E. Madey and J.T. Yates, J. Catal. 63 (1980) 226. [7] H.P. Bonzel and H.J. Krebs, Surf. Sci. 91 (1980) 499. [8] C.T. Campbell and M.T. Paffett, Surf. Sci. 139 (1984) 396. [9] C.T. Campbell, Adv. Catal. 36 (1989) 1.

[10] C.N. Satterfield, Heterogeneous Catalysis in Practice (McGraw-Hill, New York, 1980). [11] G.C. Bond, Heterogeneous Catalysis: Principles and Applications (Oxford Univ. Press, New York, 1987). [12] F. Bozso, G. Ertl, M. Grunze and M. Weiss, J. Catal. 49 (1977) 18. [13] F. Bozso, G. Ertl and M. Weiss, J. Catal. 50 (1977) 519. [14] F. Bozso, G. Ertl, M. Grunze and M. Weiss, Appl. Surf. Sci. 1 (1977) 103. [15] R. Imbihl, R. J. Behm, K. Christmann, G. Ertl and T. Matsushima, Surf. Sci. 117 (1982) 257. [16] M. Grunze, F. Bozso, G. Ertl and M. Weiss, Appl. Surf. Sci. 1 (1978) 241. [17] M. Weiss, G. Ertl and F. Nitschke, Appl. Surf. Sci. 2 (1979) 614. [18] G. Ertl, Catal. Rev. Sci. Eng. 21 (1980) 201. [19] P. Stoltze and J.K. Norskov, J. Catal. 110 (1988) 1. [20] (a) N.D. Spencer, R.C. Schoonmaker and G.A. Somorjai, J. Catal. 74 (1982) 129;

(b) S.R. Bare, D.R. Strongin and G.A. Somorjai, J. Phys. Chem. 90 (1986) 4726; (c) D.R. Strongin and G.A. Somorjai, J. Catal. 109 (1988) 51.

[21] D.R. Strongin, J. Carraza, S.R. Bare and G.A. Somorjai, J. Catal. 103 (1987) 213. [22] M. Asscher, J. Carraza, M.M. Khan, K.B. Lewis and G.A. Somorjai, J. Catal. 98 (1986) 277. [23] M. Vannice, Catal. Rev. 14 (1976) 153. [24] R.D. Kelley and D.W. Goodman, in: Fundamental Studies in Heterogeneous Catalysis, Eds. D.A. King and

D.P. Woodruff, Vol. 4 of The Chemical Physics of Solid Surfaces and Heterogeneous Catalysis (Elsevier, Amsterdam, 1982).

[25] R.D. Kelley and D.W. Goodman, Surf. Sci. 123 (1982) L743. [26] M.A. Vannice, J. Catal. 44 (1976) 152. [27] D.W. Goodman, R.D. Kelley, T.E. Madey and J.M. White, J. Catal. 64 (1980) 479. [28] P.R. Wentrcek, B.J. Wood and H. Wise, J. Catal. 43 (1976) 363. [29] J.A. Rabo, A.P. Risch and M.L. Poutsma, J. Catal. 53 (1978) 295. [30] M. Araki and V. Ponec, J. Catal. 44 (1979) 439. [31] D.W. Goodman and J.T. Yates, J. Catal. 82 (1983) 255. [32] D.W. Goodman and J.M. White, Surf. Sci. 90 (1979) 201. [33] D.W. Goodman, T.E. Madey, M. Ono and J.T. Yates, J. Catal. 50 (1977) 279. [34] T.J. Udovic, R.D. Kelley and T.E. Madey, Surf. Sci. 150 (1985) L71. [35] A. Cichowlas, E.P. Yesodharam and A. Brenner, Appl. Catal. 11 (1984) 353. [36] D.G. Castner, R.L. Blackadar and G.A. Somorjai, J. Catal. 66 (1980) 257. [37] D.J. Dwyer, K. Yoshida and G.A. Somorjai, Advanced in Chemistry, Vol. 178 (American Chemical Society,

Washington, DC, 1979) p. 65. [38] M. Logan, A. Gellman and G.A. Somorjai, J. Catal. 94 (1985) 60. [39] M.A. Vannice, J. Catal, 37 (1975) 449. [40] J.C. Geerlings, M.C. Zonnevylle and C.P.M. De Groot, Catal. Lett. 5 (1990) 309; Surf. Sci. 241 (1991) 302, 315. [41] J.H. Sinfelt, Bimetallic Catalysts (Wiley, New York, 1983);

J.A. Rodriguez and D.W. Goodman, J. Phys. Chem. 95 (1991) 4196; J.A. Rodriguez, R.A. Campbell and D.W. Goodman, J. Phys. Chem. 95 (1991) 5716.

[42] W.M.H. Sachtler, Faraday Disc. Chem. Soc. 72 (1981) 7. [43] J.E. Houston, C.H.F. Peden, D.S. Blair and D.W. Goodman, Surf. Sci. 167 (1986) 427;

J.A. Rodriguez, R.A. Campbell and D.W. Goodman, J. Phys. Chem. 94 (1990) 6936; 95 (1991) 2477.


[44] J.E. Houston, C.H.F. Peden, P.J. Feibelman and D.R. Hamann, Phys. Rev. Lett. 56 (1986) 375. [45] C.H.F. Peden and D.W. Goodman, J. Catal. 104 (1987) 347. [46] C.H.F. Peden and D.W. Goodman, Ind. Eng. Chem. Fundam. 25 (1986) 58. [47] R.W. Judd, M.A. Reichelt and R.M. Lambert, Surf. Sci. 198 (1988) 26. [48] I. Hamedeh and R. Gomer, Surf. Sci. 154 (1985) 168. [49] P.J. Berlowitz and D.W. Goodman, Surf. Sci. 187 (1987) 463; Langmuir 4 (1988) 1901. [50] R.A. Campbell, J.A. Rodriguez and D.W. Goodman, Surf. Sci. 240 (1990) 71. [51] C.M. Greenlief, P.J. Berlowitz, D.W. Goodman and J.M. White, J. Phys. Chem. 91 (1987) 6669. [52] R.A. Campbell, J.A. Rodriguez and D.W. Goodman, Surf. Sci., in press. [53] J.W. He, W.L Shea, X. Jiang and D.W. Goodman, J. Vac. Sci. Technol. A 8 (1990) 2435. [54] J.W. He and D.W. Goodman, J. Phys. Chem. 94 (1990) 1496;

J.A. Rodriguez, R.A. Campbell and D.W. Goodman, Surf. Sci. 244 (1991) 211. [55] K. Christmann, O. Schober, G. Ertl and M. Neumann, J. Chem. Phys. 60 (1974) 4528. [56] K. Christmann, G. Ertl and H. Shimizu, J. Catal. 61 (1980) 397. [57] C. Park, E. Bauer and H. Poppa, Surf. Sci. 187 (1987) 86. [58] D.W. Goodman, in: Proc. IUCCP Conf. on Heterogeneous Catalysis (Texas A&M Univ., 1984). [59] B.G. Johnson, C.H. Bartholomew and D.W. Goodman, J. Catal. 128 (1991) 231. [60] B. Imelik, C. Naccache, G. Couduier, H. Praliud, P. Meriaudeau, P. Gallezot, G.A. Martin and J.C. Vedrine,

Eds., Metal-Support and Metal-Additive Effects in Catalysts (Elsevier, Amsterdam, 1982). [61] D.W. Goodman and M. Kiskinova, Surf. Sci. 105 (1981) L265. [62] M. Kiskinova and D.W. Goodman, Surf. Sci. 108 (1981) 64. [63] D.W. Goodman, Appl. Surf. Sci. 19 (1984) 1. [64] S. Johnson and R.J. Madix, Surf. Sci. 108 (1981) 77. [65] E.L. Hardegree, P. Ho and J.M. White, Surf. Sci. 165 (1988) 488. [66] C. Kittel, Introduction to Solid State Physics, 6th ed. (Wiley, New York, 1986) p. 76. [67] J.R. Rostrup-Nielsen and K. Pedersen, J. Catal. 59 (1979) 395. [68] C.H.F. Peden and D.W. Goodman, in: Proc. Symp. on the Surface Science of Catalysis, Philadelphia, 1984, Eds.

M.L. Devinny and J.L. Gland, ACS Symposium Series (American Chemical Society, Washington, DC). [69] P.D. Szuromi, R.D. Kelley and T.E. Madey, J. Phys. Chem. 90 (1986) 6499. [70] C.T. Campbell and D.W. Goodman, Surf. Sci. 123 (1982) 413. [71] G.A. Mills and F.W. Steffgen, Catal. Rev. 8 (1973) 159. [72] P. Schoubye, J. Catal. 14 (1969) 238. [73] G. Blyholder, J. Phys. Chem. 68 (1964) 2772. [74] G. Blyholder, J. Phys. Chem. 79 (1975) 756. [75] F.A. Cotton and G. Wilkinson, Advanced Inorganic Chemistry, 4th ed. (Wiley, New York, 1980) pp. 82-86. [76] P.S. Bagus, K. Hermann and C.W. Bauschlicher, J. Chem. Phys. 81 (1984) 1966. [77] P.S. Bagus and K. Hermann, Phys. Rev. B 33 (1986) 2987. [78] K. Hermann, P.S. Bagus and C.J. Nelin, Phys. Rev. B 35 (1987) 9467. [79] J. Rogozik and V. Dose, Surf. Sci. 176 (1986) L847. [80] K.L. D'Amico, F.R. McFeely and E.I. Solomon, J. Am. Chem. Soc. 105 (1983) 6380. [81] H. lbach and D.L. Mills, Electron Energy Loss Spectroscopy and Surface Vibrations (Academic Press, New

York, 1982). [82] J.A. Rodriguez and C.T. Campbell, J. Phys. Chem. 91 (1987) 2161. [83] J. Benziger and R.J. Madix, Surf. Sci. 94 (1980) 119. [84] N.K. Ray and A.B. Anderson, Surf. Sci. 125 (1983) 803. [85] R.C. Baetzold, Phys. Rev. B 30 (1984) 6870. [86] M.C. Zonnevylle and R. Hoffman, Langmuir 3 (1987) 452. [87] A.J. Freeman, C. Fu and E. Wimmer, J. Vac. Sci. Technol. A 4 (1986) 1265. [88] J.M. MacLaren, J.B. Pendry, R.W. Joyner and P. Meehan, Surf. Sci. 175 (1986) 263. [89] J.M. MacLaren, D.D. Vvedensky, J.B. Pendry and R.W. Joyner, J. Chem. Soc. Faraday Trans. I, 83 (1987) 1945. [90] T.B. Grimley, Proc. Phys. Soc. 90 (1967) 751. [91] T.B. Grimley and S.M. Walker, Surf. Sci. 14 (1969) 395. [92] T.E. Einstein and J.R. Schrieffer, Phys. Rev. B 7 (1973) 3629. [93] P. Feibelman and D. Hamman, Phys. Rev. Lett. 52 (1984) 61. [94] P. Feibelman and D. Hamman, Surf. Sci. 149 (1985) 48. [95] J.K. N0rskov, S. Holloway and N.D. Lang, Surf. Sci. 137 (1984) 65. [96] N.D. Lang, S. Holloway and J.K. Norskov, Surf. Sci. 150 (1985) 24.


[97] S.J. Tauster, Acc. Chem. Res. 20 (1987) 389. [98] G.L. Hailer and D.E. Resasco, Adv. Catal. 36 (1989) 173. [99] C.C. Kao, S.C. Tsai and Y.W. Chung, J. Catal. 73 (1982) 136.

[100] Y.W. Chung, G. Xiong and C.C. Kao, J. Catal. 85 (1984) 237. [101] T. Engel and G. Ertl, in: Fundamental Studies in Heterogeneous Catalysis, Eds. D.A. King and D.P. Woodruff,

Vol. 4 of The Chemical Physics of Solid Surfaces and Heterogeneous Catalysis (Elsevier, Amsterdam, 1982). [102] R. Imbihl, M.P. Cox and G. Ertl, J. Chem. Phys. 84 (1986) 3519;

G. Ertl, Adv. Catal. 37 (1990) 213. [103] M. Eiswirth, K. Krischer and G. Ertl, Surf. Sci. 202 (1988) 565. [104] M. Eiswirth, P. M611er and G. Ertl, Surf. Sci. 208 (1989) 13. [105] M. Ehsasi, C. Seidel, H. Ruppender, W. Drachsel, J.H. Block and K. Christmann, Surf. Sci. 210 (1989) L198. [106] S.H. Oh, G.B. Fischer, J.E. Carpenter and D.W. Goodman, J. Catal. 100 (1986) 360. [107] D.W. Goodman and C.H.F, Peden, J. Phys. Chem. 90 (1986) 4839. [108] C.H.F. Peden and D.W. Goodman, J. Phys. Chem. 90 (1986) 1360. [109] P.J. Berlowitz, C.H.F. Peden and D.W. Goodman, J. Phys. Chem. 92 (1988) 5213. [110] B.C. Sales, J.E. Turner and M.B. Maple, Surf. Sci. 114 (1982) 381. [111] R.C. Yeates, J.E. Turner, A.J. Gellman and G.A. Somorjai, Surf. Sci. 149 (1985) 175. [112] C.H.F. Peden, D.W. Goodman, D.S. Blair, P.J. Berlowitz, G.B. Fisher and S.H. Oh, J. Phys. Chem. 92 (1988)

1563. [113] K.C. Taylor, in: Catalysis: Science and Technology, Vol. 5, Eds. J.R. Anderson and M. Boudart (Springer,

Berlin, 1984). [114] J.C. Zomerdijk and M.W. Hall, Catal. Rev. Sci. Eng. 23 (1981) 163. [115] W.M.H. Sachtler, C. Backx and R.H. van Santen, Catal. Rev. Sci. Eng. 23 (1981) 127. [116] R.A. van Santen and H.P.C.E. Kuipers, Adv. Catal. 35 (1987) 265. [117] C.T. Campbell, J. Catal. 94 (1985) 436. [118] D.J. Sajkowski and M. Boudart, Catal. Rev, Sci. Eng. 29 (1987) 325. [119] H.R. Dettwiler, A. Baiker and W. Richarz, Helv. Chim. Acta 62 (1979) 1689. [120] C.T. Campbell, J. Vac. Sci. Technol. A 2 (1984) 1024. [121] S.A. Tan, R.B. Grant and R.M. Lambert, J. Catal. 100 (1986) 383. [122] (a) R.B. Grant and R.M. Lambert, J. Chem. Soc. Chem. Commun. (1983) 662;

(b) R.B. Grant, C.A.J. Harbach, R.M. Lambert and S.A. Tan, J. Chem. Soc. Faraday Trans. I, 83 (1987) 2035. [123] R.B. Grant and R.M. Lambert, J. Catal. 92 (1985) 364. [124] R.A. van Santen and C.P.M. de Groot, J. Catal. 98 (1986) 530. [125] P.J. van den Hoek, E.J. Baerends and R.A. van Santen, J. Phys. Chem. 93 (1989) 6469. [126] J.A. Rodriguez and C.T. Campbell, Surf. Sci. 206 (1988) 426. [127] E.A. Carter and W.A. Goddard, J. Catal. 112 (1988) 80; Surf. Sci. 209 (1989) 243. [128] C.T. Campbell and M.T. Paffett, Appl. Surf. Sci. 19 (1984) 28. [129] C.T. Campbell and B.E. Koel, J. Catal. 92 (1985) 272. [130] C.T. Campbell, J. Catal. 99 (1986) 28. [131] C.T. Campbell, J. Phys. Chem. 89 (1985) 5789. [132] R.B. Grant and R.M. Lambert, J, Catal. 93 (1985) 92. [133] R.B. Grant and R.M. Lambert, Langmuir 1 (1985) 29. [134] B.C. Gates, J.R. Katzer and G.C,A. Shuit, Chemistry of Catalytic Processes (McGraw-Hill, New York, 1979). [135] F. Zaera, E.B. Kollin and J.L. Gland, Langmuir 3 (1987) 555. [136] G.R. Schoofs, R.E. Preston and J.B. Benziger, Langmuir 1 (1985) 313. [137] J.T. Roberts and C.M. Friend, Surf. Sci. 186 (1987) 201. [138] F. Zaera, E.B. Kollin and J.L. Gland, Surf. Sci. 184 (1987) 75. [139] J.P. Fulmer, F. Zaera and W.T. Tysoe, J. Phys. Chem. 92 (1988) 4147. [140] B.A. Sexton, Surf. Sci. 163 (1985) 99. [141] J.A. Rodriguez, Surf. Sci. 234 (1990) 421. [142] A.J. Gellman, M.H. Farias and G.A. Somorjai, J. Catal. 88 (1984) 546. [143] M.E. Bussell and G.A. Somorjai, J. Catal. 106 (1987) 93. [144] A.J. Gellman, D. Neiman and G.A. Somorjai, J. Catal. 107 (1987) 92. [145] A.J. Gellman, M.E. Bussell and G.A. Somorjai, J. Catal. 107 (1987) 103. [146] M.E. Bussell, A.J. Gellman and G.A. Somorjai, J, Catal. 110 (1988) 423. [147] J.E. Sinfelt, Adv. Catal. 23 (1980) 91. [148] D.W. Goodman, Surf. Sci. 123 (1982) L679.


[149] G.A. Martin, J. Catal. 60 (1979) 345, 452. [150] M.C. Desjonqueres and F. Cyrot-Lackmann, J. Chem. Phys. 64 (1976) 3707. [151] A.G. Sault and D.W. Goodman, Adv. Chem. Phys. 76 (1989) 153. [152] C.H.F. Peden and D.W. Goodman, ACS Symp. Ser. 288 (1985) 185. [153] C. Egawa and Y. Iwasawa, Surf. Sci. 198 (1988) L329. [154] E. Shincho, C. Egawa, S. Naito and K. Tamaru, Surf. Sci. 149 (1985) 1. [155] J.H. Sinfelt and D.J.C. Yeates, J. Catal. 8 (1967) 82;

J.H. Sinfelt, Catal. Rev. 3 (1969) 175. [156] J.R. Engstrom, D.W. Goodman and W.H. Weinberg, J. Am. Chem. Soc. 110 (1988) 8305. [157] C.M. Chan, M.A. Van Hove, W.H. Weinberg and D.E. Williams, Solid State Commun. 30 (1979) 47; Surf. Sci.

91 (1980) 430. [158] J.A. Rodriguez and D.W. Goodman, J. Phys. Chem. 94 (1990) 5342. [159] F. Zaera and G.A. Somorjai, J. Phys. Chem. 89 (1985) 3211. [160] D.J. Godbey, F. Garin and G.A. Somorjai, J. Catal. 117 (1989) 144. [161] G. Leclerq, L. Leclerq and R. Maurel, J. Catal. 44 (1976) 68. [162] M.L Wax, R.D. Kelley and T.E. Madey, J. Catal. 98 (1986) 487. [163] J.R. Anderson and B.G. Baker, Proc. R. Soc. London A 271 (1963) 402. [164] K. Foger and J.R. Anderson, J. Catal. 59 (1979) 325. [165] A.K. Datye, B.F. Hegarty and D.W. Goodman, Faraday Disc. Chem. Soc. 87 (1989) 18. [166] S.M. Davis, F. Zaera and G.A. Somorjai, J. Am. Chem. Soc. 104 (1982) 7453. [167] S.M. Davis, F. Zaera and G.A. Somorjai, J. Catal. 85 (1984) 206. [168] J.W.A. Sachtler and G.A. Somorjai, J. Catal. 81 (1983) 77. [169] R.C. Yeates and G.A. Somorjai, J. Catal. 103 (1987) 208. [170] D.W. Goodman, J. Vac. Sci. Technol. A 2 (1984) 873. [171] C.M. Machiels and R.B. Anderson, J. Catal. 58 (1979) 253. [172] R.A. Dalla Betta, J.A. Casumano and J.H. Sinfelt, J. Catal. 19 (1970) 343. [173] T.S. Sridhar and D.M. Ruthven, J. Catal. 24 (1972) 253. [174] A. Verma and D.M. Ruthven, J. Catal. 46 (1977) 160. [175] A. Verma and D.M. Ruthven, J. Catal. 19 (1970) 401. [176] D.S. Kellog, M.S. Touvelle and P.C. Stair, J. Catal. 120 (1989) 192. [177] J.R. Engstrom, D.W. Goodman and W.H. Weinberg, J. Phys. Chem. 94 (1990) 396. [178] W.D. Gillespie, R.K. Herz, E.E. Petersen and G.A. Somorjai, J. Catal. 70 (1981) 147. [179] F. Zaera, D. Godbey and G.A. Somorjai, J. Catal. 101 (1986) 73. [180] R.K. Herz, W.D. Gillespie, E.E. Petersen and G.A. Somorjai, J. Catal. 67 (1981) 371. [181] C.H.F. Peden and D.W. Goodman, J. Catal. 104 (1987) 347. [182] F. Garin, S. Aeiyach, P. Legare and G. Maire, J. Catal. 77 (1982) 323. [183] S.M. Davis and G.A. Somorjai, in: Fundamental Studies in Heterogeneous Catalysis, Eds. D.A. King and D.P.

Woodruff, Vol. 4 of The Chemical Physics of Solid Surfaces and Heterogeneous Catalysis (Elsevier, Amster- dam, 1982) ch. 7.

[184] A. Dauscher, F. Garin and G. Maire, J. Catal. 105 (1987) 233. [185] F. Zaera and G.A. Somorjai, Langmuir 2 (1986) 686. [186] F. Zaera and G.A. Somorjai, J. Am. Chem. Soc. 106 (1984) 2288. [187] J. Oudar, S. Pinol, C.M. Pradier and Y. Berthier, J. Catal. 107 (1987) 434, 445. [188] C.M. Pradier, E. Margot, Y. Berthier and J. Oudar, Appl. Catal. 31 (1987) 243. [189] C.M. Pradier, E. Margot, Y. Berthier and J. Oudar, Appl. Catal. 43 (1988) 177. [190] S.M. Davis and G.A. Somorjai, J. Catal. 65 (1980) 78. [191] J. Massardier and J.C. Bertolini, J, Catal. 90 (1984) 358. [192] G. Dalmai-Imelik and J. Massardier, in: Proc. 6th Int. Congr. on Catalysis, London, 1976, Eds. G.C. Bond, P.B.

Wells and F.C. Tomkins (The Chemical Society, London, 1977) p. 90. [193] J.R. Engstrom, W. Tsai and W.H. Weinberg, J. Chem. Phys. 87 (1987) 3104. [194] E. Segal, R.J. Madon and M. Boudart, J. Catal. 52 (1978) 45. [195] C.M. Tsai, C.M. Friend and E.L. Muetterties, J. Am. Chem. Soc. 104 (1982) 2539. [196] A.P. Hitchcock, D.C. Newbury, I. Ishii, J. Stohr, J.A. Horsley, R.D. Redwing, A.L. Johnson and F. Sette, J.

Chem. Phys. 85 (1986) 4849. [197] J. Stohr, F. Sette and A.L. Johnson, Phys. Rev. Lett. 53 (1984) 1684. [198] L.E. Firment and G.A. Somorjai, J. Chem. Phys. 66 (1977) 2901. [199] J.E. Demuth, H. Ibach and S. Lehwald, Phys. Rev. Lett. 40 (1978) 1044.


[200] D.B. Kang and A.B. Anderson, J. Am. Chem. Soc. 107 (1985) 7858. [201] (a) J.A. Rodriguez and C.T. Campbell, J. Phys. Chem. 93 (1989) 826;

(b) C.T. Campbell, to be published. [202] C.T. Campbell, J.A. Rodriguez, F.C. Henn, J.M. Campbell, P.J. Dalton and S. Seimanides, J. Chem. Phys. 88

(1988) 6585. [203] J.W.A. Sachtler and G.A. Somorjai, J. Catal. 89 (1984) 35. [204] A. Sault and D.W. Goodman, J. Catal. 126 (1990) 57. [205] T.G. Rucker, M.A. Logan, T.M. Gentle, E.L Muetterties and G.A. Somorjai, J. Phys. Chem. 90 (1986) 2703. [206] W.T. Tysoe, G.L. Nyberg and R.M. Lambert, Surf. Sci. 135 (1983) 128. [207] W. Sesselman, B. Woratschek, G. Ertl, J. Kiippers and H. Haberland, Surf. Sci. 130 (1983) 245. [208] B. Marchon, Surf. Sci. 162 (1985) 382. [209] C.H. Patterson, J.M. Mundenar, P.Y. Timbrell, A.J. Gellman and R.M. Lambert, Surf. Sci. 208 (1989) 93;

C.H. Patterson and R.M. Lambert, J. Phys. Chem. 92 (1988) 1266. [210] G.N. Schrauzer, P. Glockner and S. Eichler, Angew. Chem. Int. Ed. Engl. 3 (1964) 185. [211] K.P.C. Vollhardt, Acc. Chem. Res. 10 (1977) 1. [212] P.M. Maitlis, Acc. Chem. Res. 9 (1976) 93. [213] D.S. Newsome, Catal. Rev. Sci. Eng. 21 (1980) 275. [214] C.T. Campbell and K. Daube, J. Catal. 104 (1987) 109. [215] J. Nakamura, J.M. Campbell and C.T. Campbell, J. Chem. Soc. Faraday Trans. 86 (1990) 2725. [216] G.C. Chinchen, M.S. Spencer, K.C. Waugh and D.A. Whan, J. Chem. Soc. Faraday Trans. 1, 83 (1987) 2193. [217] S. Kinnaird, G. Webb and G.C. Chinchen, J. Chem. Soc. Faraday Trans. I, 83 (1987) 3399. [218] E.G.M. Kuijpers, R.B. Tjepkema, W. Van der Wal, C.M. Mesters, S.F. Spronck and J.W. Geus, Appl. Catal. 25

(1986) 139. [219] K. Bange, D.E. Grider, T.E. Madey and J.K. Sass, Surf. Sci. 137 (1984) 38. [220] W.D. Clendening, J.A. Rodriguez, J.M. Campbell and C.T. Campbell, Surf. Sci. 216 (1989) 429. [221] C.T. Campbell and B.E. Koel, Surf. Sci. 183 (1987) 100. [222] D.C. Bybell, P.P. Deutsch, R.G. Herman, P.B. Himelfarb, J.G. Nunan, C.W. Young, C.E. Bogdan, G.W.

Simmons and K. Klier, Prep. Petrol. Div. Am. Chem. Soc. 31 (1986) 282. [223] K. Klier, C.W. Young and J.G. Nunan, Ind. Eng. Chem. Fundam. 25 (1986) 36. [224] C.T. Campbell and B.E. Koel, Surf. Sci. 186 (1987) 393. [225] J.A. Rodriguez, W.D. Clendening and C.T. Campbell, J. Phys. Chem. 93 (1989) 5238. [226] J.A. Rodriguez, W.D. Clendening, J.M. Campbell, W. Min and C.T. Campbell, J. Vac. Sci. Technol. A 7 (1989)

2118. [227] J. Nakamura, J.M. Campbell and C.T. Campbell, in preparation. [228] H.H. Kung, Catal. Rev. Sci. Eng. 22 (1980) 235. [229] K. Klier, Adv. Catal. 31 (1982) 243. [230] J.H. Lunsford, Chem. Ind. 22 (1985) 95. [231] P.J. Berlowitz and D.W. Goodman, J. Catal. 108 (1987) 364. [232] C.T. Campbell, K.A. Daube and J.M. White, Surf. Sci. 182 (1987) 458. [233] Y.A. Ryndin, R.F. Hicks, A.T. Bell and Y.I. Yermakov, J. Catal. 70 (1981) 287. [234] R.F. Hicks and A.T. Bell, J. Catal. 91 (1985) 104. [235] F. Fajula, R.G. Anthony and J.H. Lunsford, J. Catal. 73 (1982) 237. [236] J. Szanyi and D. W. Goodman, Catal. Lett., submitted. [237] R. Burch and R.J. Chappel, Appl. Catal. 45 (1988) 131. [238] C.A. Leon y Leon and M.A. Vannice, Appl. Catal. 69 (1991) 291.

high-pressure catalytic reactions over single-crystal ... files/148-ss_rpt14_1991.pdf ·...

Documents