Importance of non-additive effects on chemisorption and catalysis

Journal of Molecular

Catalyszs, 40 (1987) 37 - 48

37

IMPORTANCE OF NON-ADDITIVE EFFECTS ON CHEMISORPTION AND CATALYSIS PART II. ACTIVATION OF HYDROGEN BY COPPER IN ITS GROUND AND EXCITED ATOMIC STATES M E RUIZ Znstztuto Mexzcano de1 Petrhleo, (Mexzco) 0

Apdo

Postal 14-805, Mexzco Czty 07730 D F

NOVARO

Znstztuto de Fiszca, UN A M , Apdo

Postal 20-364,

Mexzco Czty 01000 D F (Mexzco)

and J GARCIA-PRIETO 80 St George St, Lash Mzller Chemzcal Ont M5S 1Al (Canada) (Received May 12,1986,

Laboratory,

Unzuerszty

of Toronto,

Toronto,

accepted October 20,1986)

Summary

In Part I of this series, a multibody analysis of the adsorption energy was proposed as a measure of the delocahzation of adsorbate-metal catalyst mteractions. Furthermore, the idea of descnbmg the catalyst as a third body promoting the breaking of chemical bonds in dissociative adsorption processes was presented and exemplified m the Li + Hz reaction. Now we choose another metal atom copper, m order to take advantage of recent studies on the Cu + H, reaction, where the copper atom was taken to be m its ground (‘S) state and m the excited states Cu(?P) and Cu(‘D). We shall here show that detailed knowledge of the three-body non-additive contributions to the copper-hydrogen mteraction energy once more provides us with a valuable guide to understandmg why and how the diverse states of Cu behave so differently m the capture and cleavage of the H2 molecule

Introduction

The importance of understandmg the basic prmclples which rule the activation of small molecules by mdividual metal atoms m relation to real catalytic processes need not be emphasized. Theoretical [l] and experimental [2] studies have been addressed to this problem, proposmg that the mdividual metal atom-small molecule mteractions may be correlated with data concernmg the actual surface adsorption on a heterogeneous catalyst. With regard to homogeneous processes, we also find m the literature attempts 0304-5102/87/$3

50

@ Elsevler Sequola/Prmted in The Netherlands

38

to correlate the usual expenmental [3] and theoretlcal [4] studies of catalytic activation of molecules by organometalhc complexes to specific charactenstics of the metal atom electronic states and their derived potential energy surfaces. In [ 11, the results of several papers on the Cu* + Hz photoactivated reaction [5] were summarized m an attempt to show such correlations for the case of metallic copper catalyst. However, one question which remams open m that approach is how the mformation about collective nonlocal effects - which are not present m a smgle Cu atom plus a smgle HZ molecule model of the active site - may alter the obtamed overall mteraction picture In a previous paper in this senes [6], we dealt with the problem of how to estimate the delocalized nature of the adsorbate-catalyst mteractions wlthm a model with only a small number of metal atoms formmg a finite small-srze cluster plus a single adsorbate molecule. Therem, moreover, rt was shown that m the specific case of a single llthmm atom with a H, molecule forming a LIHZ system subJect to a senes of geometrical changes, the threebody non-additive (z e , mtrmslcally delocahzed m the sense defmed m the followmg section) energy term was related to the metal catalytic effect We thus propose to vlsuahze the metal catalyst as a thud body which promotes cleavage of the bond m the hydrogen molecule. Here we shall agam take this approach for the case of the dissociative adsorption of the H2 molecule with another model of the catalyst, a smgle copper atom, considermg either its ground (2S) state or any of the two lowest-lymg excited states ( 2D and ‘P). The idea is to see whether, as m the case of LiH2, the three-body non-addltrve term 1s a valuable guide to estimate if the activation of the H-H bond can take place or not. Concernmg the very different behavlour of the Cu (2S), (2P) and ( 2D) states towards Hz capture and activation for both thermal and photochemical reactions [ 51, we may ask d this approach is also capable of ratlonahzmg unambiguously such behaviour We shall proceed to show that this is so. Method Both the mteractlon energies of the Cu + H2 reactions and the separation of the parrwise additive and non-additive contributions thereof were studied using the PSHONDO-CIPSI series of programs developed at Toulouse University [7,8]. This implies use of the ab znztzomodel potential to represent the core electrons of the Cu atom as optlmlzed by Pehssier [9] The basis sets are reported m [5] and consist of a double-zeta quahty set of gaussmns for Cu and H. It was optlmlzed to reproduce the energy drfferences between the (3d ’ 4s 2, 2D) and the (3di04p1, ‘p) states of copper as well as some of the properties of the copper hydnde molecule. All of the results reported earlier for the potential energy surfaces of the Cu + H2 reactions [5] mclude extensive configuration mteractlon, and therefore our present two- and three-body results have to include a similar

39

level of electron correlation corrections to the SCF energies, l.e , using identical basis sets and extensions of the CI space. This was achieved by mtroducmg second-order perturbation energy techniques which start from a reference wave function that already describes reasonably well the potential energy surfaces at regions that correspond to bondmg situations at eqmhbrium The separation of additive and non-additive contributions to the mteraction energies for a system has been throughly discussed m Part I of this series [6]. We only remmd that if the mteraction energy itself is defmed for the CuH, system as: El,t(Cu-Hz)

= E,,,(CuH,) - E(Cu) - E(H,)

(1)

where E,,,(CuH,) 1s the total energy for the composrte CuH, system, and E(Cu) and E(H,) are the total energies calculated for the isolated copper atom and isolated hydrogen molecule, E,,,(Cu-H,) is then the interaction energy which tells us whether or not the hydrogen-copper atom mteraction for each CuH, configuration is favourable From eqn (1) we propose the separation of this total mteraction energy mto a pairwise additive energy plus a non-additive three-body mteraction energy * &1t(C~2)

= Eadd

+ &on-add

(2)

where E&, unphes the sum of the mteractions of the copper atom with each individual hydrogen atom. Calhng these atoms H, and H,,, this would be: &dd(CUH2)

=

E(CuH,) + E(CuH,)

(3)

Here Enon-add contams the effects on the CuH2 mteraction that are mdependent of the mdividual and localized mteractions withm CuH, and CuH, . In fact9 &on-add can be viewed as contaming the effect of the presence of the copper atom on the H2 molecule (as well as alternatively; the effect of H, on the CuH, subsystem and so on). These effects by defmition cannot be localized m a specrfic region of the CuH2 system In the following section we shall proceed to prove that this last component of the total mteraction energy is responsible for the catalytic cleavage of the H, bond by the Cu center.

Results As m Part I, we wish to demonstrate here how the non-additive analysis is used for understandmg the dissociative mteraction of the hydrogen molecule with a metal atom; m the present case, the Cu atom m either its ground (2S) state or its (2P) and ( 2D) excited states is considered. In Fig. 1, the mteraction energies (as have been reported elsewhere [ 51) of the Cu( 2S)-H2, CU*(~D)--H~ and CU*(~P)--H~ systems are depicted. As is shown, the (‘P)

360

320

28G

I 60

240

320

400 r Kh-H~)

4 60 dlstonca

5 60

6 40

7 20

8 00

(0 u I

!hg I. Total mteractlon energies of the C&-H2 reactlon with the copper atom m its ground state GS Cu(*S)+ Vi>, its first excited state FES CU*(~D) + Hz and second excited state SES CU*(~P) + Hz

state af Cu 1s able to capture the unpert~bed Hz molecule without any actlvatlon barrier. The (‘D) state also has a mmunum m the CdVapproach of the unrelaxed HZ, but only after surmounting a relatively small barrier. The ground state of copper (2S), in contrast, does not attract Hz as long as the H-H distance is kept fixed at Its equdrbrmm (0 749 ii) value, Nevertheless, If the H--H dz&ance IS allowed to relax wlthm this Czy symmetry, the Cu(*S)--Hz system does eventually reach a mmunal energy well after surmountmg an important energy barrier, as 1~,shown m Fig. 2, where the behavrour of CU(~P)--H, mteractlon curve owmg to the same relaxation of the H-H distance IS also depxted. In Rg. 2 the Cu-H distances are kept fured at the equlllbrmm va.Iue of the Cu(%) -t Hz curve m I?% I As seen m Figs I and 2, the behavlour of the three different states of copper for the capture and actlvatlon of Hz (through the H--H bond sc~~on) 1s quite dlssn-nllar One must ask then, whether this stems from the

41

H

-Cu -

H

angle

Fig 2 Angular dependence of the total mteractlon energy for the two main mteractlons of the Cu + Hz reaction, namely the ground state Cu(*S) + Hz (GS) and the first excited state Cu*(*D) + Hz (FES)

different natures of the Cu(?S)-H, CU(~P)-H and CU(~D)-H mdividual mteractions, z e , the pairwise additive contributions to the Cu-H, mteraction energy (as defined previously [6]), or does it have another source In Fig. 3 the mteraction energies of copper m the ( 2S), (2P) and (2D) states and a single hydrogen atom as a function of the Cu-H2 mtermolecular separation are shown. It should be noted m this case that the behaviour of all three potential energy curves is quite slmllar, ah of them leadmg to potential wells whose mmuna, m fact, do not he at very different mternuclear separations. This IS in contrast with the situation of the Cu + H2 mteraction curves, where radically different behaviour was evidenced for each configuration stemming from the (2S), (2D) and (2P) states respectively (Fig. 1). Therefore, the source of such differences must be sought elsewhere than in the pairwise energies.

42 240

> cl 5 402 w oA a Ip-40-

-60-

-120 J -160

I 80

245

3 20

400

460

r(Cu-H,)dstonce

640

560 (0

u

720

600

1

Fig 3 Attractive behavlour of Cu ground (?S) and two lowest excited (2D and 2P) states towards a single hydrogen atom The ordinate IS selected to be precisely that of Fig 1

The only other possible origm of these drfferences m behavlour of the (2S), (‘D) and (2P) states of the Cu atom as concerns H2 capture and scisslon 1s of course the three-body non-additive energy. This is obtamed, as defined m [6], by subtractmg twice the potential energies given m Frg. 3 (once for each CuH pan contamed) from the total mteractlon energies grven in Figs. 1 and 2. The three energy terms for the CU(~S) + H, mteractlon, z e , the total mteractlon energy, the pauwlse-addltme contrrbutlon thereof and the purely non-additive three-body energy correctron to the purely additive two-body energy are depicted m Fig. 4. This figure describes the H2 approach to the CU(~S) state as a function of the distance from the center of the molecule to the metal atom nucleus while keepmg the H-H separation frxed at the eqmhbnum value for the isolated hydrogen molecule. We fast of all notice that the purely additive two-body energies, (correspondmg to two tunes the CU(~S) + H mteractlon energy) present a smooth, deep descent mto a well-

43

240-

160-

j

60-

. 0 :

40-

> 0 0 LL w z w -40s

-6O-

-120-

-160’

I 60

2 40

320

400 r(Cu-Hz)

4 I 60 distance

’

6 40

560 (a

u

720

600

1

Fig 4 Energy terms for the ground state (GS) mteractlon CL@S) + H2 TE = total energy, AE = palrwlse addltlve component and NAE = non-additive three-body contrlbutlon

defined mmunum at around 2.6 au., which is m contrast with the total mteraction energy of CU(~S) with the unrelaxed H2 molecule which gives a consistently repulsive curve with no mmnnum. This completely different behaviour of the CU(~S) states towards the capture of two H atoms (very favourable) m comparison with the capture of an H2 molecule (quite unfavourable) can be understood by analyzmg the three-body non-additive contribution to the energy. In effect, the nonadditive energy is quite repulsive from the onset of the CU(~S)-H, approach, with a steep ascent that more than compensates the paxwise additive attraction well mto the distance where the latter has its mmmum, and only after the panwise energy begms to mcrease does the three-body energy begm to decrease, too late for the resultant total mteraction energy curve to have any potential well. Thus, the non-additive energy is solely responsible for the fact that while Cu(‘S) may easily capture H atoms, rt does not capture H2 molecules whose mtemal distance is not relaxed. This viewpomt was already

44

proposed m [6] as an alternative to the chemical pomt of view, and serves to supplement it. Followmg this analyas, m Fig. 5 the same descrlptlon of the total mteraction energy for H, capture by Cu(*D), the first excited state, 1s@ven, with its pau-wlse additive and three-body non-additive contrlbutlons. We remmd that the total mteractlon energy between Cu(*D) and H2 at the longer distances is mltlally repulsive, and only starts to decrease after surmounting a small barrier that arises from the avoided crossmg with an upper curve of the same symmetry, as depicted m Fig. 1. The same IS true for the piurwlse energy, because m Fig. 3 we see that for the *D state the Cu-H interaction IS mltlally repulsive until an avolded crossing, at around 5.8 a.u., converts it to an attraction that reaches a mmunum at around 3 a.u. Comparing both curves m Fig. 5, we see that the mitral repulsion is steeper for the pmwue curve, but this repulsion changes to an attraction much farther out; as for the total mteractlon energy, it contmues to be 260-

240-

2OQ-

160-

> c3

40-

(L w z w

Cu(*D)+H,

O-

-4o-

-6O-

-160 I 60

2 40

1 3 20

1 4 00

4 60

r(Cu-HL)dmtance

5 80 (a

5 40 u

7 20

6 00

1

Fig 5 Same energy terms TE, AE and NAE as m Fig 4 for the first excited state FES mteractlon Cu*( 2D) + H2

45

repulsive until the Cu(*D) atom-Hz midpomt of 4 a.u. is reached and only then begins to decrease. Now we look at the three-body term of Fig. 5, which clearly shows a compensatory character to the panwise energy. In fact, it is attractive for the longer distances where the two-body energies are repulsive, and becomes repulsive when the latter become attractive. In contrast with the situation found m the case of the % state, now the threebody repulsions for the mtermediate distances do not contmue to mcrease to the region where the pauwise curve has its mmlmal energy well, but rather begms to dunmish at somewhat longer Cu(*D)-H, separations Therefore the total mteraction energy, which m the present context can be considered a balance between the pauwise and non-additive curves (I e , 1s obtained from them by subtraction) is able to reach a mmimal energy well. The physical situation, then, is completely different than for the *S state of copper, because now the *D state is capable of capturmg the unrelaxed H2 molecule, after surmountmg a small barrier. The contrasting behaviour of the non-additive three-body energy for the different states of copper is nowhere more marked than m the case of Cu(*P). If we look at the Cu(*P)-H mteraction curve m Fig 3, we see that it is, if anythmg, less attractive than that of Cu(*D) or Cu(*S). Yet Cu( *P) spontaneously captures H2 while the other two states present barriers to such a capture [ 51. In Fig. 6 this is evident the total mteraction energy curve is smoothly attractive, so is the case of the purely additive curve, for which, however, the mnumal energy well lies at a substantially shorter distance (a full atomic unit smaller than the position of the total mteraction energy well) The three-body term, as always, compensates the two-body attractions, albeit m this case only shghtly, allowmg the total mteractions to follow the latter qualitatively. This stems from the much flatter ascent of the three-body repulsions as compared with the cases of Cu(*S) and (*D). The steepness of the non-additive repulsion m fact reaches its maxmum only at distances shorter than 4 a.u., duly correspondmg to the existence of a total mteraction energy well at such a Cu-H, separation. In Part I we analyzed the Li + H2 reaction, which m fact did not lead to spontaneous cleavage of the H, bond due to the high barrier unpedmg Hz capture by Li and the highly unstable nature of the L&I, system [6] For the case of Cu + Hz, on the other hand, all the three states studied do lead to Hz capture. In Fig. 1 we see the mmuna that are reached from the CzV approach of the unmodified H2 molecule to the (*P) and (*D) states of copper, although no mmmum is apparent for Cu(*S). When we allow for a relaxation of the H-H distance, however, the Cu(*S)H, system does present a mmunum, as shown m Fig. 2. The way such a muumum is reached implies leaving the common Cu(*S)-H distances fixed, while the HCuH angle is broadened until a lmear mmunal-energy HCuH structure is reached [5]. For the openmg of the angular structure, analysis of non-additive three-body terms leads to an explanation of how this mmnnum is achieved. In Fig. 7 the total mteraction energy (the same as that given m Fig. 2), as well as its twobody pairwise and non-additive three-body contributions, are given for the

46 320 1

260-

240-

i

zoo-

2

2 > ” u w z w

160-

IZO-

60-

40-

O-

-4o-

-60

I 60

2 40

3 20

4 00 rCu-HZ

Fig 6 Same energy terms reaction Cu*(*P) + Hz

4 60

I 5 60

6’40

7 20

I 9 00

9 60

dlstonce

as m preceding

figures,

here for the second

excited

state SES

Cu( *S) + H2 reaction. We see that the non-additive term orlgmally describes a repulsion that leads to an activation barrier, but later descends into the linear conflguratlon mmunum. The total mteractlon energy naturally follows the behavlour of this three-body term m an exactly parallel fashion The other curve described m Fig 2 1s that correspondmg to the mteraction of the Cu* atom m the first excited state and the H, system, subJect to a snnllar HCuH angle openmg In this case, again, we also observe that the three-body energy term obviously controls the behavlour of the total mteraction energy, as shown m Fig 8 In contrast with Fig 7, now the threebody mteractlon does not have to overcome such a large barner and leads to a mmlmum at a shorter angle of 117“) so that it actually does increase somewhat when reaching the linear structure. The consequences of all these characterlstlcs of the Cu(*S), (*P) and (*D) interaction curves with H2 as concerns chemical and photochemlcal processes are discussed m [ 51

47

Ii-Cu-H

angle

Fig 7 Energy terms for the angular dependence of the GS reaction Cu(?S) + H2 The H-Cu-H angle IS at first simply closed, mamtammg the Cu-H distances flxed at their eqmhbrlum value untd (at an angle smaller than 40”) the H-H separation reaches its eqmhbnum value for Isolated H2 molecules Thereafter the angle IS formally reduced by removmg the H2 molecule from the CU(~S) site This explains the constant value of AE for angles larger than 40” and Its rapid chmb to zero energy for smaller angles when Hz approaches mfmlty

To conclude, we snnply want to pomt out that, from the present pomt of view, one could describe the H-H bond breakmg by a catalyst as a thud body (here represented by a smgle Cu atom m its different states) naturally described by an energy term that mdeed affects the mteractlon energy between It and the H2 molecule, but which ISnot contamed m the mdlvldual Cu-H mteractlons. The energy represented by this term 1sm fact non-local, and can be interpreted as bemg the effect of a witness thn-d-body as it modlfles the mtrmslc H-H mteractlon mdependently of its own mteractlons with each H moiety. What we have shown here 1s that to use such nonlocalized three-body energies as a guide of the whole chemical process (z e , H, capture and actlvatlon by the different Cu atom centers) leads to a completely coherent and reliable picture. Thus the goal orlgmally proposed m [6] of decomposmg the total energy mto a paxwise and a non-addltlve

48

160

60

0 E 1 B r

40

>c3 Lz w

0

* -40 w

-160 0

40

60

120 H-Cu-H

160 angle

200

Fig 8 Energy terms for the angular dependence of the FES reaction Cu*(‘D) + Hz As m Fig 7, for angles smaller than 40” the Hz molecule IS allowed to unthdraw from the Cu*(*D) site, thus explammg the departure of AE from its constant value

three-body contnbutlon, and then showmg that one may use the nonadditive term as a guide to adsorption and catalytic effects, 1s seen to be obtainable for the case of complex copper-hydrogen reactions References J Garcia-Prleto, M E Rulz and 0 Novaro, J Am Chem Sot , 107 (1985) 5635 G A Ozln,Acc Chem Res , 10 (1977) 21 See e g S 011~6 and G Henrlcl-01&i, Homogeneous Catalyszs, Elsevler, Amsterdam, 1978 See e g 0 Novaro, S Chow and Ph Magnonant, J Catal, 41 (1977) 91, correlation of this catalytic Tl complex action with the properties of the Tl atom ground and excited states IS presently being established by S Castlllo, M E Rulz and 0. Novaro, work m progress Papers on this sub]ect by the present authors as well as other members of our laboratories m Mexico and Canada have appeared m J Chem Phys, 80 (1984) 1529. J Chem Phys, 81 (1984) 5920,5 Phys Chem, 90 (1986) 279 M E Rulz, J Garcia-Prleto and 0. Novaro, J Mol Cat&, 33 (1985) 311 J C Barthelat, Ph Durand and A. Serafml, Mel Phys, 33 (1977) 159 B. Huron, P Rancurel and J. P Malrleu, J Chem Phys , 58 (1975) 5745 M Pehssler, J Chem Phys, 75 (1981) 775, rbzd , 79 (1983) 2099