J. Phys. Chem. 1991, 95, 7145-7153
7145
FEATURE ARTICLE Bonding In Complexes of Hg(6s6s1S,) and Hg(6s6p3P,) with Rare-Gas Atoms and Small Molecules: From Physical to chemical Interactions M.-C.Duval, B. Soep,* Laboratoire de Photophysique MolPculaire du CNRS and Institut de Physico- Chimie MolPculaire de I’UniversitP Paris-Sud, Bat. 21 3, UniversitP de Paris-Sud, 91405- Orsay Cedex, France
and W . H. Breckenridge* Department of Chemistry, University of Utah, Salt Lake City, Utah 84112 (Received: April 2, 1991; In Final Form: June 17, 1991) The van der Waals’ bonding of atoms and molecules has become an increasingly important topic of study in the past few years in many areas of physical chemistry. One of the challenges of such studies is to characterize where weak, “physical” (multipole) interactions are sufficient to describe the bonding, and where stronger “chemical” (electron-sharing) interactions begin to be important. We show in this Feature Article that the bonding in complexes of ground-state Hg(6s6s’So) and excited-state Hg(6s6p3PI) with rare-gas atoms and small molecules varies from purely physical to strongly chemical. In complexes of the filled-shell Hg(6s6s’So) state with rare gases and molecules such as H2, CH4, NH,, or H 2 0 the bonding and structure can be understood within the context of weak, long-range multipolar interactions and short-range Pauli repulsions: “physical” (van der Waals) bonding. On the other hand, such interactions in the analogous complexesof the excited Hg(6s6dPJ state can vary from essentially physical (with rare gases or CH4) to chemical (with H20, NH3, H2, or Hg).
I. Introduction The van der Waals’bonding of atoms and molecules has become an increasingly important topic of study in the last few years in many areas of physical chemistry. One of the challenges of such studies is to characterize where weak, “physical” (multipole) interactions are sufficient to describe the bonding, and where stronger “chemical” (electron-sharing) interactions begin to be important. A great deal of spectroscopic and dynamical information has now been accumulated concerning the van der Waals complexes of rare gas atoms and small molecules with group IIA and IIB metal atoms in their ground and excited states.’-” In (1) Duval, M . 4 . ; Soep, 9. Submitted for publication. (2) Breckenridge, W. H. Acc. Chem. Res. 1989, 22, 21. (3) Breckenridge. W. H.; Benoist d’Azy, 0.;Duval, M.-C.; Jouvet, C.;
Soep, 9. In Stochasticity and Intramolecular Redistribution of Energy; Lefebvre, R., Mukamel, S.,Eds.; North-Holland: Amsterdam, 1987; p 149. (4) Brcckmridge, W. H.; Duval, M. C.; Jouvet, C.; Soep, B. J. Chim. Phys. 1987, 84, 38 1. (5) Duval, M.-C.; Benoist DAzy, 0.;Breckenridge, W. H.; Jouvet, C.; Soep, B. J . Chem. Phys. 1986,85,6324. (6) Breckenridge, W. H.; Jouvet, C.; Soep, B. J . Chem. Phys. 1986,84, 1443. (7) Breckenridge, W. H.; Duval, M.-C.; Jouvet, C.; Soep, B. Chem. Phys. Lett. 1985, 122, 181. (8) Jouvet, C.; Soep, B. J . Chem. Phys. 1984,80, 2229. (9) Jouvet, C.; Boivineau, M.; Duval, M.-C.; Soep, B. J. Phys. Chem. 1987, 91, 5416. (IO) Yamanouchi, K.; Isogai, S.;Tsuchiya, S.;Duval, M.-C.; Jouvet, C.; Benoist d’Azy, 0.; Soep, 9. J . Chem. Phys. 1988,89, 2975. (1 I ) Duval, M.-C.;Jouvet, C.; Soep, B. Chem. Phys. Lett. 1985,119,317. (12) Fuke, K.; Saito, T.; Nonose, S.;Kaya, K. J . Chem. Phys. 1987,86, 4745. (13) Duval, M.-C.; Soep, B. Chem. Phys. Lett. 1987, 141, 225. (14) Yamanouchi, K.; Fukuyama, J.; Horiguchi, H.; Tsuchiya, S.;Fuke, K.; Saito, T.; Kaya, K. J . Chem. Phys. 1986, 85, 1806. (15) Fuke, K.; Saito. T.; Kaya, K. J . Chem. Phys. 1984, 81, 2591. (16) Yamanouchi, K.; lsogai, S.; Okunishi, M.; Tsuchiya, S. J . Chem. Phys. 1988,88, 205. (17) Duval, M.-C.; Soep, 9.;van Zee, R.; Bosma, W.; Zwier, T. J . Chem. Phys. 1988, 88, 2148. (18) Jouvet. C.; Soep, B. Chem. Phys. Lett. 1983, 96,426. (19) Jouvet, C. Thbe de Doctorat d’Etat, Universit6 Paris-Sud, 1985. (20) Duval, M.-C. These de Doctorat d’Etat, Universit€ Paris-Sud, 1988. (21) Tsuchizawa, T.; Yamanouchi, K.; Tsuchiya, S.J . Chem. Phys. 1988, 89, 4646.
0022-3654/91/2095-7145$02.50/0
this paper we examine the interactions of the ground state nsns’S,-, (closed shell) or nsnp3PI(open shell) configurations of such metal atoms. The open-shell cases offer a particularly interesting variety of interactions ranging from purely physical to strongly chemical bonding with atoms or molecules, and we focus attention on the most widely studied excited atomic state: Hg(6s6p3PI). The “bonds” in such complexes have been found to vary from extremely weak linkages due only to long-range dispersive interactions (a few cm-l) to cases in which there is either strong bonding (up to lo4 cm-I) or chemical reaction. We shall attempt to correlate and rationalize the observations on such systems, using simple physical and chemical (molecular orbital) ideas. We will also show how these considerations may be related to several other areas of inquiry, including reaction and energy-transfer dynamics and the bonding of small molecules at metal atom sites in coordination complexes and on surfaces. Since the emphasis of this article will be on providing an explanation of the range of bonding phenomena observed, experimental details will not be discussed but may be found in the original references. 11. Complexes with Ground-State Hg(6s6s1So)Atoms A. HgRG Complexes (RG = Rare Gas). The bonds of RG
atoms with the filled-shell ground-state Hg(6s6s1&) atom are quite weak. Shown in Table I are the bond energies (0,)and equi(22) Wallace, I.; Bennett, R. R.; Breckenridge, W. H. Chem. Phys. Lett. 1988, 153, 127. (23) Bennett, R. R.; McCaffrey, J. G.; Wallace, I.; Funk, D. J.; Kowalski, A.; Breckenridge, W. H. J . Chem. Phys. 1989, 90,2139. (24) Funk, D. J.; Kvaran, A.; Breckenridge, W. H. J . Chem. Phys. 1989, 90, 2915. (25) Funk, D. J.; Breckenridge, W. H. J . Chem. Phys. 1989, 90,2927. (26) Bennett, R. R.; Breckenridge, W. H. J . Chem. Phys. 1990,92, 1588. (27) Bennett, R. R.; McCaffrey, J. G.; Breckenridge, W. H. J . Chem. Phys. 1990, 92, 2740. (28) Kvaran. A.; Funk, D. J.; Kowalski, A.; Breckenridge, W. H. J. Chem. Phys. 1988,89, 6069. (29) Bennett, R. R.; Breckenridge, W. H. To bc submitted for publication. (30) Bennett, R. R. Ph.D. Thesis, University of Utah, 1989. (31) Wallace, I.; Funk, D. J.; Breckcnridge. W. H. To be submitted for
publication.
0 1991 American Chemical Society
7146 The Journal of Physical Chemistry, Vol. 95, No. 19, 1991
Duval et al.
TABLE I: B o d D&clllees md Dirroehtioa EaerPies of Various Comlexes HnX'
HgX HgHeC
polarizability o f X (As)-' 0.21 0.40 1.64 2.48 4.04 0.81 2.60 1.49 2.26 1.76 1.95
hard-sphere diameter ofX,bA
ground state re
2.6 2.8 3.4 3.6 4.1 2.9 3.8 2.7d
(4.6) 3.9 3.99 4.07 4.25 X
4.0 3.6 3.4 4.17
2.6d 3.5 3.7 2.9
De (8) 46 142 180 254 46 180 300 j: 50 250 25 110 130 350
*
O+
re (3.67) 3.47 3.36 3.52 3.25
(n) state
1 (Il
De (28) 67 350 520 1460 710 2850 6100 660
De
re
(no bound
4.92 4.70 4.57 4.47
(strongly bound) 3.2 2.4 2.2 3.08
+ 2 ) state
x
+ 1.0
4.5
500
2.2 5.36
states) 13 53 96 172 24 110
6100
50
500
refs 11, l 6 , 9 9 15, 16 15, 16 15 1 1 , 14 4, 6, 19 13, 20 1 17
IO 12
70,71, 100, 101 5.1 3.63 10700 9000 'Values of re in A, Dl in cm-I. bFrom ref 34. 'The values quoted are r,, Do. dStockmayer potential. 'The *O+" state and 'I" state are the Os+ and I, states of Hg2 which correlate with Ho(~P,) + Hg(ISo).and are spin-orbit components of the u ~ u , * ~ , ( ~ I lM.O. , ) configurati~n.'~/From ref 98.
librium bond lengths (re)for such molecules. The bond strengths increase regularly with the polarizibilities of the RG atoms. As discussed by Buckingham et al.,32the only possible longrange attractive force between two filled-shell atoms is due to the dispersion interaction caused by correlations between the fluctuating multipolar charge distributions of the two atoms. At shorter range, this attraction will be suddenly overwhelmed by a strong repulsion resulting from the overlap of the filled-shell electron distributions (Pauli repulsion). As can be seen in Table I, the attractive dispersion forces (which increase with polarizability) appear to dominate the bond-strength trends from HgNe to HgXe. Since the re values increase only from 3.9 to 4.2 A in this series of molecules, the internuclear distances at which the "hard-sphere" repulsive force suddenly appears must increase more slowly than do the attractive dispersive forces." For Ar, Kr, and Xe, the d-shell and lanthanide contractions apparently modulate the increase in the hard-sphere diameters more than the increase in polarizabilities. B. Hg-Molecule Complexes. The interaction of a closed-shell atom like Hg(6s6s'q) with a molecule is obviously more complex, but at very long range the dominant attractive forces will be nearly isotropic and resemble those of an inert gas atom with the same polarizability. In fact, spectroscopiccharacterizations of metal atom/rare-gas potentials are valuable as models for metalatom/molecule potentials at large distances (for example, to characterize a high impact parameter collision process). Further, however, it appears3*that the same balance between the orientation-averaged dispersive attraction and Pauli repulsion may also play the dominant role in determining dissociation energies (De) and center-of-mass equilibrium distances (re) for van der Waals complexes between filled-shell atoms and filled-shell molecules. For example, the De and re values for the ground-state Hgmolecule and Hg-RG complexes listed in Table I are qualitatively consistent with the polarizabilities and hard-sphere radii of the molecules and the analogous RG atoms. Note particularly the similarities between Hg-CH4and Hg-Kr, and between Hg-H, and HgNe. The relatively large De and small re values for Hg-H20 and H g N H 3 can be rationalized by the relatively small hardsphere radii of these species, given their polari~abilities.~~ This could be due to the "tight" bonding of the electrons on the electronegative N and 0 atoms, as compared to, say, isoelectronic CHe Anisotropic interactions in ground-state Hg-molecule complexes cannot be ignored, of course, since they will determine any preferred molecular geometry of the van der Waals complex. This is an active area of study in its own right, but is quite important for the laser-excitation studies of jet-cooled Hg-molecule com~
~
~~~~~~~
(32) Buckingham, A. D.; Fowler, P. W.; Hutson, J. M. Chem. Rev. 19%8, 88, 963. (33) Liuti, G.; Dirani, F. Chem. Phys. ~ I I 1985, . 122, 245. (34) Hirschfelder, J. 0.;Curtiss, C. F.; Bird, R. B. Molecular Theory of Gases ond Liquids; Wiley: New York, 1954. H o n d h k of Chemistry ond Physics, 66th ed.; CRC Press: Boca Raton, FL, 1985.
plexes to be discussed below, since due to the Franck-Condon principle the excited Hg*-molecule complexes will be initially created with the geometry of the ground-statecomplex at the time of excitation. For filled-shell atom molecule interactions, anisotropy can obviously result from repulsive forces but also from long-range inductive interactions in which the electron density of the atom is distorted due to the multipolar charge distribution of the molecule. The latter effect appears to be quite important in determining the equilibrium geometries of weakly bound van der Waals comple~es.'~From experiments to be discussed below, it appears that the potentials for the Hg-H20 and Hg-NH3 ground-state complexes, for example, are quite different. The Hg-NH, equilibrium geometry appears to be of C,, symmetry with the N-atom lone pair pointing toward the Hg atom, indicating a fair degree of anisotropy. The Hg-H20 ground-state interaction potential, however, shows little anisotropy, with the H 2 0 molecule almost freely rotating. This conclusion has also been reached very recently in the analysis of microwave spectra of the Hg-H20 van der Waals complex.3s As pointed out by H ~ t s o n 'for ~ the similar case of the nearly isotropic intermolecular potential of Ar-H20, the polarizability tensor of H 2 0 is surprisingly isotropic, and the C, axis component of the quadrupole tensor of H 2 0 is nearly zero, so that there is little reinforcement of the dipolar field. The maximum Hg-H20 interaction anisotropy has in fact been estimated' to be only -20 cm-' at the equilibrium distance of 3.5 A. On the other hand, the polarizability tensor component for NH, along the C,, axis (which is larger than that for H 2 0 along its C, axis) is greater than that perpendicular to it, and the large C3,-axis component of the quadrupole tensor for NH, reinforces the dipolar field." For Hg-NH,, then, the induction forces appear to lead to a preferred C,, geometry, although the extent of the anisotropy is not known from experiments in the cold jet ( - 5 K), since no ground-state librational mode "hot bands" were observed." It should be noted, however, that the Ar-NH3 potential is thought to be fairly isotropic,'8 and the larger anisotropy for Hg-NH3 may be due to the fact that the polarizability of the Hg atom is more than 3 times greater than that of the Ar atom. Consistent with this, De for HgNH, is more than twice that of ArNH3 and the center-of-mass re is almost 0.5 A smaller for HgNH3.'7*38 Since CH4 has no dipole or quadrupole moment, only the repulsive interactions could lead to strong anisotropy, and there is no evidence that this is appreciable at the ground-state bonddistance of 4.0 A.13*zoIt is also likely that the potential of the very weakly bound ground-state Hg-H2 complex has very little
+
(35) Endo, Y. Private communication. (36) Hutson, J. M.J . Chem. Phys. 1990. 92, 157. (37) Miirk, T. D.; Castleman, Jr., A. W. Experimental Studies on Cluster Ions. Ado. At. Mol. Phys. 1985, 20, 65. (38) Nelson, D. D.; Fraser, G. T.; Peterson, K. I.; Zhao, K.; Klemperer, W.; Louas, F. J.; Suenram, R. D. J . Chem. Phys. 1986,85, 5S12. Werner, H.-S.; Meyer, W. Mol. Phys. 1976, 31, 855.
The Journal of Physical Chemistry, Vol. 95, No. 19, 1991 1141
Feature Article Hunds case (cl
A m
&w!
Molecule Q
Molecule (nl
.... .... .... .... .... .... .... ..--g
1
-......
.........y-
'P,
-o+
:. : -
2 I
-"
0-
o+ -:.'
o--
.......,.,..........
'Po
Fipn 1. A schematic diagram showing the correlation of atomic with molecular spin-orbit levels in C, symmetry (e& for Hg-RG molecules) for Hund's case a and Hund's case c.
anisotropy, similar to the Ar-Hl p~tentiaI.'*~.'~*~~ A recent high-level ab initio calculation of the analogous Mg-H2 ground-state potential did show a very slight preference for C, geometry, however, even at the calculated M g H 2 re and De of 4.9 A and 28 cm-'.'O 111. Complexes with Electronically Excited Hg(6s6dP1)
Atoms A Hg-RC Complexes. Because the atomic spin-orbit coupling constant (-4265 cm-')for the Hg(6s6$PJ) multiplets41is larger than the bond strengths of the Hg-RG diatomic molecular states which correlate with these multiplets, a Hund's case c treatment of the electronic interactions at the equilibrium bond distances is appropriate.5 This means that the only good quantum number is 51, the component of the total electronic angular momentum along the bond axis. Shown in Figure 1 is a correlation diagram for the Hund's case a and c extremes. For example, the Hg(6s6p3Pl) atom can form two electronic states with the argon atomas The first (A) state, with 51 = 0 and (+) symmetry, corresponds to a pure "II" orientation of the Hg(6p) orbital, similar to a Hund's case a 311 state (where the spin-orbit interaction is small compared to the bond strength). The second (e) state, with a = 1, cannot be described so simply because of spin-orbit mixing with the high-lying 51 = 1 (E)state correlating with Hg(6s6$P2). In effect, in going from low spin-orbit (case a) to high spin-orbit (case c) interactions, the multiplets of the 311Qand 32Q states with 511 0(-) and 51 = 1 mix strongly, and the B(51=l) state can in fact be approximated5as 50% Z (po) and 50% II (phi) orientations of the Hg(6p) orbital with respect to the rare-gas atom. Shown in Figure 2 are the potential curves for the 51 = 0-,0+, 1, and 2 states of HgAr correlating with the Hg(6s6p3PJ) multiplets, either determined experimentally or calculated by using ~ can be seen in a simple case c type model for the b ~ n d i n g .As this diagram and in Table I, the A(O+) states have much larger D i s and smaller r i s than the X(O+) ground states, while exactly the reverse is true for the B( 1) states. This can be rationalized (39) LeRoy, R. J.; Hutson, J. M. J . Chem. Phys. 1986,86, 837. (40)Augsbcrger, J. D.; Dykstra, C. Chem. Phys. Lrrr. 1989, 158, 399. (41) Hay, P.J.; Dunning, Jr., T. J . Chem. Phys. 1976. 65. 2679.
s7m
97 700
37 6ooO
3
9
4
6
6
7
PO
8
R (LyMmr) Figure 2. Potential energy curves of the electronic states of HgAr correlating with the Hg(6~6p'P~,,,~) asymptotic atomic states.'
qualitatively if the Z orientation of the diffuse Hg(6p) orbital is attractive at very long range but becomes repulsive even for fairly large r as the RG atom filled shell begins to overlap repulsively with the diffuse 6p orbital. The II orientation of the Hg(6p) orbital, on the other hand, leads to a more attractive potential than the ground state because the dispersive interaction with the diffuse 6p electron (as well as with the somewhat contracted, and more positive, 6s electron "core") can extend to much smaller r before hard-sphere repulsion sets in. Thus the A(O+) states are reasonably attractive and the B( 1) states are only weakly bound since they have 50% Z character, which contributes repulsively even at large internucleardistances. This model is consistent with recent spectroscopicmeasurements on the M-RG states correlating with metal M(mnplP,) states, where the pure I F states have only very shallow minima at extremely large r.21-24*42 B. Hg-Molecule Complexes. The analogous interactions of excited Hg(6s6p3Pi)atoms with molecules will of course be much more complicated. At long range, however, where anisotropic interactions are still negligible, "pseudodiatomic" A(O+) and B( 1) ~~
(42) Wallace, 1.; Kaup, J.; Breckenridge, W. H. To be submitted for
publication.
Duval et al.
7148 The Journal of Physical Chemistry, Vol. 95, No. 19, 1991
type states can be identified which eventually connect to the polyatomic potential surfaces (which are extremely difficult to characterize experimentally and theoretically for the heavy atom mercury). One of the great values of the spectroscopicinformation collected on the HpRG A(O+) and B( 1) states is that it provides asymptotic model potentials at long range for molecules with similar polarizability. (i) Hg-H2 Complexes. The interaction of Hg(6s6p3PI) with H2 is quite interesting, in that the “A(O+)” II-type state and the “B( 1)” mixed 2-II state can in some ways be regarded as extremes of “chemical”versus “physical” interaction of Hg(6s6p3PI)with a molecule, although paradoxically both states have been observed via detection of a chemical product: HgH. As can be seen from Table I, the B(‘1“) state of HgH2 has a small De and large relative r, consistent with what would be expected for an H2 polarizability lying between those of the Ne and Ar atoms. However, the HgH2(“B”) state was characterized not by direct laser-induced fluorescence (LIF), as for the HgNe and MgAr cases, but by monitoring (by LIF) the formation of ground-state HgH with a second pulsed dye laser delayed a few nanoseconds. Details can be found in the but by analysis of partially resolved rotational structure for the H g H 2 and HgD2 B X bands observed with HgH(HgD) “action” spectroscopy, it was possible to show that the HgH2(B) state lives at least 5 ps before the HgH product is somehow formed. We believe this behavior indicates that the HgH2(B) state is quite similar to the dispersion-bound HgNe(B) or HgAr(B) states, except that inefficient pathways exist (nuclear tunneling or rotational coupling to the A(O+) state) which allow eventual chemical reaction (predissociation, in spectroscopicterms) to form HgH. Thus, although there is a weak long-range attraction, the exchange repulsion between the Hg(”,) “B( 1)” orbital configuration and H2 sets in at fairly large r (-4 A), and a reaction “barrier” is created, similar at these internuclear distances to the repulsive potentials encountered with the nonreactive neon or argon atoms. The observation of a chemical barrier is consistent with recent ab initio calculations by Bemier and Millie43vuof the HgH2“B(1)” excited triplet state surfaces. The relevant HgH2 52 = 1 state correlates to two spin-orbit states, one of BI(A”) and one of B2(A’) symmetry for C,(C,) geometries. The (Q = 1, A’, B2) state was found to be quite repulsive, and a small potential barrier was observed in the (Q = 1, A”, B,) adiabatic surface when careful calculations using a relativistic pseudopotential and a perturbative treatment of spin-orbit coupling were Finally, it is interesting to note that the form of the observed Q = 1 potential surface, i.e., that of an outer van der Waals well, an activation barrier, then eventual chemical interaction with the metal atom to break the H-H bond and form metal-hydrogen bonds, is very similar to those proposed by Lennard-Jones and others for hydrogen chemisorption on metals.45 It has in fact been shown that first-row transition-metal surfaces have not only substantial s-character but some pdensity even at the Fermi level energies. Saillard and HoffmanMhave shown, for example, that there is a strong interaction of the s,p bands on Ni( 11 1) with the u* antibonding orbital of “end-on” H-H at short distances. In sharp contrast to the structured HgH “action” spectrum associated with excitation to the HgH2 B( 1) state, only a very broad continuum was observed to the red of the Hg(ISo-)P1) atomic line, where HgH2 A(O+) excitation should be observed.6 From the width of the continuum, it was concluded that the HgH2 A(O+) state predissociated within -0.2 ps to form HgH + H. Such an efficient process is consistent with strong attractive chemical interaction of Hg(6s6p3Pl) in the A(O+) (II) orbital configuration with the H2 molecule, with no repulsive potential barrier. This observation is consistent with qualitative mechanistic arguments as well as ab initio studies of the interaction of group +
(43) Bernier, A.; Millie, P. J . Chem. Phys. 1988,88, 4843. (44) Bernier, A. Thhe, UniversitC Paris-Sud, 1987. (45) Mucller, W. M.; Blackledge, J. P.; Libowiz, G. G. Metal Hydrides; Academic Press: New York, 1968; p 52. (46) Saillard, J . Y.; Hoffmann. R. J . Am. Chem. SOC.1984, 106, 2006.
I1 M(sp) excited atomic states with the H-H It had been that if the metal atom approaches the H-H bond side-on the half-filled porbital (HOMO) can overlap well with the empty antibonding u* orbital (LUMO) on H2 to form half a r-bond. Simultaneously, the filled u-orbital on H2 can overlap with the half-filled s-orbital on the metal atom, forming a u2u* 3-electron “half-bond”. Ab initio calculations5’of the interaction of Be(2s2p1Pl) and Mg(3s3p1P1)with H2 showed, indeed, that the IB2 states corresponding to r-orientation of the porbitals had no activation barriers for insertion of the atom into the H-H bond, and that both u donation and ‘K back-donation interactions were important. The extensive set of calculations of the triplet HgH2 surface^^^,^ also support the idea that the HgH2 ‘A(O+)” state accessed in the laser excitation experiments is strongly chemically bound. For the (Q = O+, A’, A,) spin-orbit component of the HgH2(’B2) state which correlates with HgCP,), there is no barrier for insertion , into the H-H bond and the 1,l eV local minimum in the C potential surface corresponds to an HHgH bond angle of 70°. The C , *(’B2,0+) interaction of Hg(6s6p3P,) with H2 is an extreme example of strong chemical rather than physical bonding and involves both a-donation into the Hg(6s) orbital by a lone pair of electrons as well as r-back-donation of the Hg(6p) orbital into an empty low-lying orbital with the correct symmetry. Not only for complexes of molecules with Hg(6s6p3PI),but also for metal-ligand chemical bonding in general as well as for the interactions of small molecules with metal surfaces or catalysts, the u-donor and r-acceptor capabilities of the molecule must be considered carefully to understand the bonding.M,u*s5Study of Hg(3P)-molecule or other excited metal atom molecule “van der Waals” complexes may therefore have wider relevance than may be generally appreciated. Along these lines, the reaction of Hg(6s6p3P1)with H2may be regarded formally46 as the formation of a Hg-H2 A(O+) metal-ligand complex, leading to oxidative addition and H-H bond activation, with HgH + H as final products. (ii) Hg-cH, Complex. From Table I, it can be seen that both the A(O+) and B(l) states of the Hg-CH, molecule which correlate with Hg(6s6p3P1)have bond strengths and bond lengths similar to the same states of HgKr, and that Kr has nearly the same polarizability and “hard-sphere” diameter as CH4. Also, both of these Hg-CH4 excited states are long-lived ( > l P s), since they were characterized by laser-induced fluorescence, and no product HgH was detected when either state was excited. Thus, in the regions of the potential surfaces which are Franck-Condon accessible from the weakly bound Hg-CH, ground state, the methane molecule acts remarkably like an inert gas atom with the same polarizability. This must mean that for both states (as r decreases) the Pauli repulsive forces begin to dominate before any possible chemical attraction can occur. Recently obtained Cd(’P,) “action” spectra of the analogous Cd*CH4(A(O+) X(O+)) transition show the same ‘pseudodiatomic” similarities between the CdCH4 (A(O+)) and CdKr(A(O+)) excited states.% It is likely that in bent H-Hg-CH3 geometries similar to those at which the bound HHgH “A(O+)”spin-orbit state is most stable, the H-Hg-CH3 “A(O+)” state will also be lower in energy than
-
+
-
(47) Brcckenridge, W. H.; Umemoto. H. in The Dynamics of the Excited Stare; Lawley. K., Ed.;Advances in Chemical Physics, Vol. 50; Wiley: New York. -1982. (48) Brcckenridge, W. H. In Reacriom of Small Transient Species; Clyne. M., Fontijn, A., Eds.; Academic Press: New York, 1983. (49) Callear, A. B.; McGurk, J. C. J. Chem. Soc., Faraday Trans. 2 1973, 69, 97. (50) Brcckenridge, W. H.; Adams, N.; Simons, J. Chem. Phys. 1981,56, 3327. (51) Blickensderfer, R.; Jordan, K.; Adams, N.; Breckenridge, W. H. J . Phys. Chem. 1982,86, 1930. (52) Adams, N.; Breckenridge, W. H. To be published. (53) Chaquin. P.; Sevin, A,; Yu, H. J . Phys. Chem. 1985, 89, 2813. (54) Shustorovich, E.; Baetzold. R. C.; Muetterties, E. L. J . Phys. Chem. 1983,87, 1100. (55) Cotton, F. A.; Wilkinson, G. Advanced Inorganic Chemistry, 4th ed.; Wiley-Interscience: New York. 1980. (56) Wallace, I.; Breckenridge, W. H. To be submitted for publication.
----.
The Journal of Physical Chemistry, Vol. 95, No. 19, 1991 7149
Feature Article asymptotic Hg(3PI)+ CH4. Thus,in contrast to the H2case, there must be an activation barrier for insertion into the C-H bond for the II (A(Ot)) as well as the Z II (B(1)) potential surfaces with CH4. The HgCH,(A(O+)) van der Waals state observed is more strongly bound and does have a slightly smaller center-of-mass bond distance than the HgKr(A(O+)) state, but this could just indicate that a slightly more favorable geometric orientation of the Hg(p) orbital with respect to the directional C-H bond electron densities can be achieved at intermediate distances compared to structureless Kr. There is evidence for some anisotropy in the Hg-CH,(A(O+)) interaction potential, since weak, low-frequency progressions were observed and assigned to "librational" (hindered rotor) motion of CH4.13920In any case, there is no evidence for strong chemical interaction in the HgCH,(A(O+)) state. This conclusion is consistent with the known collisional quenching behavior toward Hg(6s6p3PI)of CH4versus H2. The quenching of Hg(3PI)by H2 occurs at essentially every collision (aQ 5* 30 A2), with chemical products (HgH + H, Hg + 2H) produced in nearly 100% yield.4749 In contrast, the quenching of Hg(6s6p3Pl) by CH, is very inefficient (a = 0.3 A2),even though the same chemical channels are availabpe, since the C-H bond strength in CH4 is virtually identical with the H-H bond strength in H2.47*48The height of the activation barrier for chemical interaction appears to decrease with decreasing C-H bond strength, since for a series of alkane hydrocarbons the apparent quenching cross sections can be qualitatively rationalized by assigning "per-bond" cross sections of -0.05, -3.0, and 16.0 A2 for primary, secondary, and tertiary C-H bonds, respectively, at room temperat~re.,~This would be roughly consistent with an activation barrier height on the order of 1000-1500 cm-I for the chemical reaction of Hg(6s6$P1) with the C-H bonds in CHI. Finally, it should also be noted that the analogous Zn(4s4p3Pl) and Cd( 5s5p3PI)states are also quenched very inefficiently by CH, even at higher temperatures, but react readily with H2 to produce ZnH and CdH, r e s p e c t i ~ e l y . ~ ~ * ~ ~ ~ ~ * ~ " One reason for the activation barriers for M(nsnp3Pl) states reacting with CHI may be that, unlike H2, a true "side-on" attack of the C-H bond, with *-orientation of the p-orbital, cannot be accomplished without overcoming repulsive forces due to the steric hindrance of the other C-H bond^.'^^^ Also, the sideon gorbital overlap with the lowest lying a* antibonding orbital of CH4 may be spatially much less favorable than for H2, as shown by Saillard and Hoffmann in their comparison of H2 and CH4 II interactions with transition metal d-orbitak6 The fundamental interactions of Hg(3PI)with C-H bonds may even have practical implications. Crabtree and co-workerss9have discovered (by accident, as many such findings occur) that Hg('P,) excitation in the gas phase in the presence of substrates containing C-H bonds can (under the right conditions) be utilized for highly selective syntheses of a variety of organic molecules. (iii) Hg-NH3 Complex. (e) General Bonding Considerations. As one moves from CH, to the isoelectronic molecule NH3, it is obvious from Table I that radically different behavior is observed for the complexes with Hg(6s6p3Pl).17 Both the "A(O+)" and "B(1)" states are relatively strongly bound for NH3 (-6ooo cm-') and have a very short center-of-mass equilibrium internuclear distance of 2.2 A. However, as shown in Figure 3, the pseudodiatomic potential curves are in fact quite different for the two states, the B( 1) state being much less bound at larger r than the A(O+) state, which of course is the "normal" situation for these two types of electronic states at intermediate distances. We believe that this behavior is consistent with metal-ligand chemical bonding between Hg('PI) and NH3, due mostly to the good overlap between the filled, high-lying, and properly directional lone-pair molecular orbital centered on the nitrogen atom with that of the half-filled Hg(6s) atomic orbital, and that the Hg(6p) T orbital is essentially nonbonding. The MO picture representing this bonding scheme is shown in Figure 4, along with the C,,
9
II
+
0 0
W
I 77
3.08
2.83
4.94
6.00
r, (3,
Figure 3. The pseudodiatomic potential curves of the Hg-NH3 A(A1) and B(E) electronic states correlating with Hg(6s6p3P,)+ NH3. The vertical dotted lines show the FranckCondon-accessibleregions of the two excited-state potential curves from the vibrationally cold HgNH3(XIA,) ground state. Spectroscopic parameters used: A(Al) wo = 276 cm-I, w,g0 = 3.32 cm-I, D,,' = 5600 cm-I, re = 2.2 A; B(E) wo = 338 cm-1, w,g0 = 5.1 cm-I, D,,' = 5600 cm-', re = 2.2 A.
-
(57) Breckenridge, W. H.;Wang, J..H. J . Chcm. Phys. 1987, 87, 2630. H.;Wang, J. Chcm. Phys. Ln.1986, 123, 23. (59) Brown, S.H.; Crabtree, R. H. J . Chcm. Educ. 1988,65, 290, and references therein. (58) Breckenridge, W.
Hg
NHS
HS*(;7N/" V H f
i
H
Figure 4. Molecular orbital energy level diagram for the A(A,) excited state of Hg-NH3 showing the chemical binding expected in the excited state due to Hg(6s)-NH3 (lone-pair) interactions. The left-hand side of the figure ignores spin-orbit interaction while the central portion includes it. The right-hand side shows the corresponding spin-orbit levels in C, symmetry for comparison with Hg-X diatomic systems. (SeeFigure 1.)
spin-orbit states correlating to the C,,(Q) pseudodiatomic designations we have used previously in this paper. The degenerate E spin-orbit state correlates to the "B(1)" state and the AI spin-orbit state to the "A(O+)" state designated in general terms in Table I. The a2a*-type bonding should be particularly strong in the Hg(3PJ)-NH3 case, since the 6s orbital of Hg lies at virtually the same energy and is of the same symmetry as the 2al lone-pair orbital of NH3, allowing for strong bonding/antibonding molecular orbital interaction. In fact, the ionization potentials of Hg(6s2 'So)and NH3 (the 2al orbital is the HOMO) are 10.4 and 10.2 eV, respectively. To emphasize this point further, NH3 is known to be a good a-donor ligand in coordination chemistrys4vs5because of its high-energy, directed lone-pair orbital, and the Hg atom
7150 The Journal of Physical Chemistry, Vol. 95, No. 19, 19'91 has the highest ionization potential of any metal atom, thus providing optimum conditions for strong covalent chemical interaction. We emphasize the u-donor aspect of the interaction, because the lowest lying unoccupied orbital on NH, of suitable symmetry for T back-donation from the Hg(6p) orbital is the antibonding 2e* orbital, but this is essentially a high-lying diffuse Rydberg-type orbital,@'and is not a good acceptor orbital. Also, there is no real evidence for any appreciable *-acceptor capability of the NH, ligand in coordination chemistry.55 Finally, we also note that the concept of a (a),(.*) "half-bond" is an old one61,62 but that there has been a resurgence of interest recently in this kind of bonding interaction.63*" We first present qualitative arguments whict successfully rationalize the different excited-state Hg-NH, potential curves, and then examine the detailed spectroscopic and dynamical experimental results of the collaborative study of the Hg-NH, system by Soep, Zwier, and c o - ~ o r k e r s within '~ the framework of our model. At moderate Hg('P,)-NH3 distances, as observed, the A(A,) surface should be more attractive than the B(E) surface, similar to the analogous surfaces of, say, the Hg-Kr, or Hg-CH4 molecules. In this region, the spin-orbit coupling of the asymptotic Hg(6s6p) atomic states is larger than the bonding interactions between Hg(,PI) and the NH3 lone-pair orbital, Hund's case c treatment of the interaction is still appropriate (see Figure 2), and the A(A,) and B(E) surfaces of HgNH, should be quite different in energy. As the donation of NH3 lone-pair electronic density to the Hg(6s) orbital becomes more important, however (at center-of-mass distances less than about 2.5 A), the Hg-NH3+ character of the interaction should become more important. The Hg(6p) orbital will polarize away from the increased electron density at the Hg atom and become more diffuse (i.e., will acquire more Hg(7p) "IT" character). The spin-orbit interaction thus will decrease, resulting in less splitting between the two spin-orbit (A, and E) component^.^^ (The splitting between the ,Po and 'PI states of the Hg(6s7p'P) configuration is only 145 cm-l compared to 1767 cm-' for the analogous Hg(6s6p) atomic configuration.66) It should be pointed out that a simple pseudodiatomic change from Hund's case c to case a coupling in the Hg-NH3 excited-state bonding (in which the Hg(6p) character of the "nonbonding" Hg(6pr) orbital is essentially unchanged from the asymptotic Hg(6s6p) atomic state) cannot rationalize the observations. The binding energies in the Hg-NH, excited states (6100 cm-I) are certainly greater than the atomic 3Po-3P1splitting (1767 cm-I), indicating that a case a description would be more appropriate near the well depths of such states. In such a case a limit, the "O+"(A,) and the O-(A2) state (correlating with Hg(,Po)) would certainly become similar in energy, but the "O+"(A,) and "l"(E) states would still be separated by approximately the 3P03Pl atomic state energy d i f f e r e n ~ e . ~ ~ * ~ * We now digress slightly to consider two contrasting examples of deeply bound states correlating with Hg(6s5p3Pl): Hg(3PI)-C0 and Hg(3Pl).Hg(1So).In both these cases, the observed splittings between the two pseudodiatomic "O+" and "I" states (see Figure 1) offer important information about the type of bonding involved, just as we claim to be the case for the analogous Hg(3Pl).NH3 states. Although the Hg-CO states have not yet been definitively characterized experimentally,12*20 a high-level theoretical calculation@has shown that the analogous strongly bound linear HgCO (60) Herzbcrg, G.Electronic Spectra and Structure of Polyotomic Molecules; van Nostrand Reinhold: New York, 1967. (61) Herzberg, G.Spectro o/Diotomic Molecules; van Nostrand Reinhold: New York, 1950. (62) Kauzmann, W. Quantum Chemistry; Academic Press: New York, 1957; p 399. (63) Clark, T. J . Am. Chem. Soc. 1988, 110, 1672. (64) Gill, P. M. W.; Radom, L. J . Am. Chem. Soc. 1988, 110, 4931. (65) We acknowledge P. Millie for first suggesting this possibility. (66) Moore, C. E. Aromic Energy Levels 1971, III. NSRDS-NBS 35. (67) Bennett, R.; Breckenridge, W. H. J . Chem. Phys. 1990, 92, 1588. (68) Bennett, R.; McCaffrey, J.; Breckenridge, W. H. J . Chem. Phys. 1990. 92. 2740. (69) Kato,W.; Jaffe, R. L.; Komornicki, A.; Morokuma, K. J . Chem. Phys. 1983, 78,4567.
--.--.-
Duval et al. A(O+) and B( 1) states are similar in energy at their equilibrium center-of-mass distances. In this situation, however, because of the strong x "back-bonding" donation of electron density from the Hg(6p) orbital to the low-lying empty T* orbital centered on the carbon atom of CO, there is net Hg+-CO- character in the bonding, opposite to the Hg-NH3+character invoked above. This results in an obvious reduction to spin-orbit coupling as the p~ electron density is transferred from a heavy to a light atom.69 For the strongly bound states of homonuclear Hg, correlating to Hg('P,) + Hg(%,), however, (see Table I) the $a* "half-bond" interaction is augmented by a very favorable p r - p ~one-electron *-bonding interaction between orbitals identical in energy, resulting in bond strengths7&', of 10700 cm-' and -9000 cm-' (for the A(O,+) and B(1,) states, which are spin-orbit states of the most strongly bound ug2a,*rU(,IIg)MO configuration)?, In this case, there is no good reason for the Hg(6p) orbital either to become more diffuse (as in Hg-NH,) or to transfer density to a good between the *-acceptor orbital (as in HgCO), so the two spin-orbit states correlating with Hg(,P,) + Hg('So), 1700 f 200 ~ m - ' , ~is' comparable to the Hg(3PI)-Hg(3Po) atomic splitting, 1767 cm-l, as expected. A case a treatment, with a spin-orbit interaction similar to that of the Hg(,P,) asymptotic state, is appropriate in this case, because the favorable p r interactions require the 6 p orbital to remain similar to the (relatively) "tight" Hg(6s6p) atomic p-orbital character. It thus appears that spin-orbit interactions, although notoriously difficult to deal with t h e ~ r e t i c a l l y ,may ~ ~ ~in~fact ~ ~ offer ~ * ~ important ~ experimental evidence relating to the bonding in these systems. Let us now examine more carefully the detailed experimental results of the Hg-NH, study within the framework of our model. As seen in Figure 3, the Franck-Condon-accessible portions of the "BI"(E) state are above the Hg('P0) asymptote, and in fact the B(E) spectra were all obtained by monitoring Hg(,Po) production via a second laser pulse ("action" spectro~copy).'~From the lack of observation of direct fluorescence, but the discrete rotational structure observed for the X(lAl) B(,E) "action" spectra, it was deduced that the lifetime of the B(E) state of HgNH, before predissociating to Hg('Po) is between 0.01 and 10 ns. Simulations of the rotational contours showed the expected perpendicular band structure for a E A, transition,I7 and t,, the electronic angular momentum in the upper state,@'was determined to be 0.92, showing that the Si = l pseudodiatomic character of the upper state is retained, at least at the large mean internuclear distances corresponding to energies above the Hg(,Po) asymptote. In contrast, the rotational contours of the Hg-NH3 "pure-stretch" transitions to the A(AI,O+)state showed the expected parallel band structure. But for the vibronic transitions in which one quantum of bending motion was also excited, the simulations were only consistent with perpendicular transitions, with upper-state vibrational angular momentum tV= 1, as expected for a degenerate bending mode in a C,, molecule. Thus the detailed rotational structures of the bands of the A(A,) and B(E) vibrational states above the HgCPo) asymptote are consistent with our pseudodiatomic view of the bonding, with essential separation of bending and stretching motions. The Franck-Condon-accessible region of the excited A(A,) state extends to energies below the Hg(,P0) predissociation asymptote (see Figure 3). Time-resolved measurements of fluorescence from the A(A,) vibrational states in this region showed lifetimes of 200-250 ns, about twice as long as the fluorescent lifetime of the asymptotic Hg(6s6p3PI)atomic state. This is in contrast to the analogous A(O+) states of the Hg-Ne, Hg-Ar, and Hg-Kr complexes, which were found to have lifetimes essentially identical with that of Hg(3Pl).1SThe Hg(,PP, ISo)atomic transition gains oscillator strength via spin-orbit mixing of Hg(6s6p1PI)character into the 'PI wavefunction. The slightly longer lifetimes observed
-
+
-
(70) Niefer, R.; Atkinson, J. B.; Krause, L. J . Phys. B 1983, 16, 3631; J . Phys. B 1983, 16, 3767. (71) Callear, A. B.; Lai, K.-L. Chem. Phys. 1982, 69, 1. (72) Hay, P. J.; Wadt, W. R.; Dunning, Jr., T. H. Annu. Rev. Phys. Chem. 1979, 30, 31 I . Mies, F. H.; Stevens, W. J.; Krauss, M. J . Mol. Spectrosc. 1978, 72, 303. Celestino, K. C.; Ermler, W. C. J. Chem. Phys. 1984,81, 1872.
The Journal of Physical Chemistry, Vol. 95, No. 19, 1991 7151
Feature Article
and electron donation by NH3, the ammonia ligand bond angles for the high-lying HgNH, A(AJ vibrational levels (u'= 17-19) were found to be essentially unchanged from free NH3, 106 f are therefore consistent with our model of the bonding, since the 3°.79 Thus it appears that even with loss of up to "half" an spin-orbit interaction should steadily decrease as r decreases and electron, the NH3 geometry does not change markedly, which the p-orbital becomes more diffuse. The lifetimes of the lower could indeed be consistent with the Hg-NH, observations. vibrational levels, which are unfortunately inaccessible (for Presumably the geometry of the ammonia molecule changes rather Franck-Condon reasons), should therefore have even longer dramatically between NH3+0.5and NH3+'.0. lifetimes. It has been known for some time73q74that when Hg(,P,) is Exploring further the analogy of NH3 as a ligand in transiexcited in the presence of NH3 and inert gas, a strongly red-shifted tion-metal complexes with that of NH3 bonded to Hg(6s6p3Pl), continuous emission at -3400 A is observed which has a very we note that the H -NH3 pseudodiatomic center-of-mass bond distances of -2.2 in the excited states correspond to a Hg-N long lifetime of -2 ps. Soep and c o - ~ o r k e r shave ' ~ shown that distance of -2.05 A if the NH3 bond angle remains 107". The this emission is probably from the A(AJ and B(E) states near metal-nitrogen bond distances in a variety of strongly bound their potential minima. The long lifetime, of course, is consistent with the drastic lowering of spin-orbit coupling in our model of coordination complexes of NH3 with transition-metal ionss1range the bonding. The emission has been traditionally associated with from 1.94 to -2.3 A, consistent with our contention that the Hg(,PI)-NH3 bonding is chemical in nature. the Hg(,Po)-NH3 ~omplex,'~ since it is also produced when (b) Relation to Bonding in Other Metal-NH, Interactions. We Hg('Po) interacts with NH3 at higher total pressures, but our hope that our postulated model for the bonding in Hg(,P,)-NH, bonding model would also predict that the HgNH3(A2) state correlating with Hg(,Po) should be uery close in energy to the complexes will inspire high-level theoretical calculations of the HgNH3(AI) state at their minima (O+,O- levels in the pseudorelevant potential surfaces. A few theoretical calculations of other diatomic analogy), so that collisional interconversion of the two neutral metal atoms interacting with ammonia have already been states should occur readily. Of course, if the Hg('P0)-NH3 (A,) reported, and we now discuss some examples of these efforts in state is long-lived radiatively (as is Hg('P0), which is metastable), relation to our proposed model of Hg('P,)-NH3 interactions. then any collisional equilibration of the close-lying A2 and the Bauschlicher and co-workers have calculated the binding between AI states would also tend to increase the "effective" lifetime of ground-state Cu atoms and NH3.82 The Cu( 3dI04s) configuration the AI ~ t a t e . ' ~ . ~ ~ should allow u2u* three-electron bonding similar to that which Careful analysis of the K-level rotational structure of both the we have proposed, but the ionization potential of Cu is 7.7 eV, X('A,)-A(E) and X('AI)-B(AI) transitionsI7 showed that there substantially less than the 10.4 eV for Hg, so that the u-acceptor was no major change in the Hg-NH3 geometrical orientation, and covalent interaction of the Cu(4s) orbital with the NH, lone pair it is likely that the ground state has the same C,, geometry which should be less favorable. Their calculations2 of a Cu-NH, bond maximizes the a bonding in the upper states. It should be noted strength of 2400 cm-' is consistent with our expectations. In fact, that there is much evidence that the NH, molecule bonds to metal however, these workers claim (using a Morokuma-type analysis surfaces, such as Ni(ll0) and Ni(l1 l), with the C,, axis perof the wave functions which others contend tends to overemphasize pendicular to, and the N atom pointed toward, the ~ u r f a c e . ~ ~ - electrostatic ~~ interactionss3)that there is little chemical bonding The vibrational structure of both the X('A1)-A(AI) and Xin this case but that polarization of the Cu(4s) orbital away from ('A,)-B(E) spectra of Hg-NH, showed no evidence of features the NH3 dipole can explain the "bonding." A u2u*-interaction, indicating excitation of localized NH, ligand vibration^.^^^^ For however, would appear to be difficult to distinguish from such example, there were no progressions at higher frequencies which an 'electrostatic" interaction, since an electron in a u* orbital will would be expected if the H-N-H bond angle changed appreciably resemble a "polarized" s-electron on the side of the atom opposite in the excited state from that of the weakly bound NH3 molecule to the Cu-NH3 a-bond. in the ground-state. At first glance, this is some cause for concen, Bonding of NH3 at an atomic site in four- or five-atom metal since the Hg--NH3+ character of the bonding invoked in our model clusters has also been studied There is definite could cause changes in the H-N-H bond angle. The 107" bond evidence of electron density polarization away from the bonding angle of free ammonia changes to 120" when NH, is i o n i ~ e d , ~ . ~ metal ~ atom into the cluster as NH3 approaches, but even for and even the first Rydberg states of the NH, molecule, in which exactly the same system (All) one group rationalizes this as an a NH3+ core is surrounded by a diffuse electron cloud, are known electrostatic "dipole-polarization" effects4 while anothers3claims to be planar.@ One might therefore expect a significant bond angle it is consistent with a-donor chemical interactions. change in the NH3 molecule from ground state to excited HgNH,. Experimentally, chemisorption of NH, at metal surfaces appears However, it appears that even in coordination complexes where to lower the energy of the u lone pair (at least relative to the it has been demonstrated that the NH, ligand has a substantial strongly bonding N-H molecular orbitals), according to photonet positive charge (but still much less than +l), there is little electron spectroscopy measurementsas6 Such observations are change in bond angle. For example, a very careful X-ray decertainly qualitatively consistent with a-donor-type chemical termination of the electron density and molecular structure of the bonding, but it has been claimedss that this could also be due to CO(NH,)~~+-C~(CN complex79 ) ~ ~ - showed that the Co(NH& "electrostatic" stabilization of the NH3 lone pair due to penetration unit indeed had +3 net charge, but that this charge was equally of the NH, dipole into the electron density of the transition-metal shared by the NH3 units (mostly on the H atoms), and that the atom. Perhaps semantics (i.e., the definition of "electrostatic"), Co atom was nearly neutral. This is very consistent with the MO rather than demonstrable differences, separate the two models?83 theory of such complexes,ss.aOin which a set of bonding u-type (iv) Hg-HZO Complex. From Table I, it can be seen that the orbitals are formed between s, p, and d orbitals on cobalt and filled Hg-H20 'A(O+)" state is fairly strongly bound as well (-2900 NH3 lone-pair orbitals, resulting in substantial electron donation cm-I), but with a bond strength less than half that of Hg-NH, to the (formally) + 3 metal ion. Even with such strong bonding "A(O+)". Also, from a detailed analysis of the rotational structure of the Hg-H20 and Hg-D20 ('AO+" "XO+") vibronic tran-
x
-
+
(73) Callear, A. B. Chem. Rev. 1987, 87, 335. (74) Horiguchi, H.; Tsuchiya, S.Chem. Phys. 1986, 108, 153. (75) Madey, T. E.; Houston, J. E.; Seabury, C. W.; Rhodin, T. N . J. Yac. Sci. Technol. 1981, 18, 476. (76) Netser, F. P.; Madey, T. E. Sur- Sci. 1982, 119, 422. (77) Klauber, C.; Alvey, M. D.; Yates, Jr., J. T. Chem. Phys. k r r . 1984, 106,477. (78) DeKock, R. L.; Gray, H. B. Chemical Srrucrure and Bonding, Benjamin/Cummings: Menlo Park, 1980; p 288. (79) Iwata, M. Acra. Crysrallogr. 1977, 833, 59. (80) Figgis, B. N . introduction to tfgand Fields; Interscience: New York, 1966; p 172-202.
(81) Wilkinson, G., Ed. Comprehensive Coordination Chemistry; 1987; Vol. 2, pp 26-28. (82) Bauschlicher, Jr., C. W. J . Chem. Phys. 1986,84, 260. (83) Reed, A. E.; Curtis, L. A.; Weinhold, F. Chem. Reo. 1988,88,899. (84) Bagus, P. S.;Hermann, K.; Bauschlicher, Jr., C. W. J . Chrm. Phys. 1984,80, 4378. (85) Bagus, P. S.;Hermann, K.; Bauschlicher, Jr., C. W. J . Chem. Phys. 1984,81, 1966. (86) Granze, M.; Buzco, F.; Ertl, G.; Weiss, M. Appl. Sur/. Sci. 1978, I , 24. Seabury, C. W.; Rhodin, T. N.; Partell, R. J.; Mevill, R. P. S u r - Scf. 1980, 93, 117.
7152 The Journal of Physical Chemistry, Vol. 95, No. 19, 1991
c 2v
C-v 0BI,Bz
6'Pz
I
2
Bl,B2
I O+ A2
Q
c
1 63P,
0-
H20
0
Hg*
Figure 5. Molecular orbital energy diagram for the A(Al) excited state of Hg-H20 showing the chemical binding expected in the excited state due to Hg(6s)-H20 (lone-pair) interactions. The left-hand side of the figure ignores spin-orbit interaction while the central portion includes it. The right-hand side shows the corresponding spin-orbit levels in C ,, symmetry for comparison with Hg-X diatomic systems. (See Figure 1.) sitions,' it has been concluded that the "AO'" state has a C, (or near-C,) molecular geometry. The Hg-OH2 center-of-mass distance resulting from this analysis is small, 2.4 f 0.2 A. Because of severe congestion in the Hg-H20 and Hg-D20 B( 1) X(O+) spectra (perhaps partially due to the fact that in the Hg-H20 case the "B( 1) state" will really be two energetically separate states of Bl and B2 symmetry, in contrast to the degenerate Hg-NH, (E)state), no detailed information is available on the B( 1) state. We show, in Figure 5, a u2u* three-electron MO bonding scheme similar to that proposed for the Hg-NH3(A(O+)) state (see Figure 4). The half-filled Hg(6s) orbital overlaps with the 3al lone-pair MO on H 2 0 (which is located along the C2axis), and the Hg(6p) ?r orbital is again essentially nonbonding. However, in contrast to the NH, case, photoelectron spectra show that this 3a, lone pair orbital on H 2 0 lies -4 eV lower in energy than the NH, lonepair orbital,87and is therefore a much poorer a-donor than the NH, lone pair. The u2u* three-electron bond is thus substantially weaker for Hg-OH2 than for Hg-NH,. As shown in Figure 5, the 3al lone-pair orbital on H 2 0 is not the highest lying molecular orbital (HOMO) of H20, however. The lbl lone pair, which is perpendicular to the H 2 0 molecular plane, is about 2 eV higher in energy than the 3a, lone-pair?' and one might expect this lone pair to be a better u-donor, leading to a Hg-H20(A(O+)) geometry in which the Hg atom is above the H 2 0 molecular plane and the Hg-O axis is perpendicular to that plane (a "pinwheel" geometry). There are at least two factors which may contribute to the preference for C, versus pinwheel geometry. First of all, it is possible that in the pinwheel arrangement there is significant repulsive interaction between the diffuse Hg(6p) orbital and the electron density along the two 0-H bonds. The C, geometry, then, would allow a closer Hg-H20 center-of-mass approach even though the u2u* attraction might be somewhat stronger in the +
(87) Kimura, K.; Katsumata, S.;Achiba, Y.; Yamazaki, T.; Iwata, S. Handbook o/ He1 Photoelectron Specrra; Halsted Press: New York, 1981.
Duval et al. pinwheel geometry. Second, electrostatic interaction between the H 2 0 dipole and the Hg(6s6p) state should be maximized in the C , geometry. Even if such electrostatic interactions are smaller than the u2u* interaction postulated here, they could tip the balance toward the Czvversus pinwheel geometries. There is ample expenmental evidence that when strongly bound in coordination complexes or when adsorbed at a metal surface , (or near-C,) geometry site, the H 2 0 molecule adopts a similar C with respect to the binding site. For instance, a neutron diffraction study8*showed unequivocally that the six H 2 0 molecules coordinated to Ni2+ ion in dilute aqueous solution were oriented in Ni-OH2 C, geometries, with a Ni+ bond distance of 2.1 A. The metal-ion/oxygen-atom distances of a variety of aquo complexes of metal ions8' range from 2.0 to 2.5 A, in fact, similar to the Hg-O distance of 2.2 f 0.2 A in the HgOH2(A(O+))state (2.4 f 0.2 A Hg-H20 center-of-mass distance). Similarly, ESDIAD experimental datag9vwon H 2 0 adsorbed on metal surfaces are consistent with binding via the oxygen atoms, with the H 2 0 symmetry axis oriented perpendicular to the surface (or nearly so). Theoretical calculations9' of H 2 0bound to Ni clusters support this view of the bonding but show that small tilt angles of up to 15O require very little energy. Theoretical calculations of stronger metal-atom H 2 0 interugh actions also indicate a preferred C, g e ~ m e t r y , B Z ~ ~ a l t h o in many cases a C, geometry (with no change in the H 2 0 subunit geometry) is assumed in such calculations. With regard to the nature of the bonding, some authors indicate a Lewis acid-base interaction with H 2 0 an electron d ~ n o r , ~ )very , ~ ' similar to our ( u ) ~ ( u * )molecular orbital picture of the bonding in the HgH20(A(O+))state. Others have characterized the bonding in electrostatic terms,829%93 emphasizing the penetration of the H 2 0 dipole and the polarization of electron density away from the metal atom. Again, it is difficult to see the real difference between this electron density pattern, and we have chosen to and a (.)*(.*)I adopt the molecular orbital point of view because it allows us to rationalize bonding patterns without extensive ab initio calculations. For example, with the electrostatic "penetrating dipole" model of metal-NH, or metal-H20 bonding it is difficult to rationalize the huge decreases in bond strengths from metals with (ns)' versus (ns)2 ground states. The bond strength of Na-H20 has been determined experimentally to be 1900-2200 cm-' (in good agreement with several theoretical calculation^)^^^% while that of Mg-H20 has been estimated in a very careful, high-level calculation to be only 300 f 100 ~ m - I . 9 ~This is consistent with the Hg-H20 ground-state experimental value of 300 f 50 cm-I, and it is likely that the dissociation energies of Zn-H20 and Cd-H20 ground-states will also be on the order of 300 cm-I. But as discussed above, the Cu-H20 bond energy has been calculated to be 1600 cm-1.82It is also likely that the Mg-NH,, Zn-NH3, and Cd-NH3 ground-state bond energies will be similar to that of Hg-NH3, 250 f 25 cm?, but the bond energies of NaNH3 and CuNH, have been calculated to be 2100 and 2400 cm-I, re~pectively.~~.~~ If the bonding is strictly electrostatic, it is difficult to see why there would be 5-10-fold decreases in bond energies in M-NH3 or M-H20 complexes in going from M(ns) to M(nsns) groundstate configurations. Although the polarizablity of Na(3s) is more (88) Enderby, J. E.;Neilson, G. W. In Water, A Comprehenriue Treatise; Plenum Press: New York, 1979; Vol. 6; pp 25-31. (89) Madey, T. E.; Netzer, F. P.SurJ Sci. 1982, 117, 549. (90) Madey, T. E.; Yates, Jr., T. J. Chem. Phys. Lett. 1977, 51, 77. (91) Bauschlicher, Jr., C . W. J . Chem. Phys. 1985,83, 3129. (92) Bentley, J. J . Am. Chem. Soc. 1982, 104, 2754. Bentley, J.; Carmichel, l. J . Phys. Chem. 1981, 85. 3821. (93) Bauschlicher, Jr., C. W. Chem. Phys. Left. 1987, 142, 71. (94) Trenary, M.; Schaeffer, 111, H. F.; Kollman, P. A. J . Chem. Phys. 1978. -.. - , 68. - -,4047. . ... (95) Schulz. C. P.;Haugstirtter, R.; Tittes. H.-U.; Hertel, I. V. Z . Phys. D 1988, 10. 279. (96) Daren, R.; Lackschewitz, U.; Milosevic, S.;Waldapfel, H.-J. Chem. Phys. 1990, 140, 199. (97) Sauer, J . ; Kathan, B.; Ahlrichs, R. Chem. Phys. 1987, 113, 201.
The Journal of Physical Chemistry, Vol. 95, No. 19, 1991 7153
Feature Article than twice that of Mg(3s3s), the Mg atom is still very polarizable ( a = 10.6 and one would not expect such a great change in interaction energies if the bonding were truly due to polarization of the s-electron(s) away from the NH3 or H 2 0dipoles, since the interaction energy is roughly proportional to au2, where a is the polarizability of the metal atom, and u is the dipole moment of the molecule.97 Even more difficult to explain with the electrostatic model (if indeed the Zn-H20 and Zn-NH, bond strengths are similar to those of Hg-H20 and Hg-NH,) are the much larger bond strengths of Cu-NH, and Cu-H20, since the polarizabilities of Cu and Zn atoms are virtually identical?' We also note that the calculated Cu-NH3 bond strength is greater than that of Na-NH,, even though the polarizability of Na (23.6 A,) is 3-4 times as large as that of Cu (6.8 f 0.5 A,), Finally, it is difficult to rationalize the fact that M-NH, bond strengths are always stronger than those of M-OHz complexes in the same theoretical stud (e.g. DJCuNH,) = 2400 cm-I, De(CuH20) = 1600 cm-')? despite the fact that the dipole moment of H 2 0 (1.85 D) is larger than that of NH, (1.47 D). Experimentally, the bond strength of HgNH, (A(O+)) is more than twice as large as that of HgOH2(A(O+)) (see Table I). Of course, our a-donor bonding model is quite consistent with the large bond energy differences, even for ground-state Cu-NH3 versus "Zn-NH,", since the ( ~ ) ~ ( aconfigurations *) for M(ns) states will be moderately bonding, while the ( ~ ) ~ ( a configu*)~ rations for M(nsns) states will be chemically nonbonding. The higher bond strengths for "open-shell" M-NH, versus M-H20
Y
(98) Handbook of Chemistry and Physics, 66th ed., CRC Press: Bcca Raton, FL, 1985. (99) Van Zee,R. D.; Blankespoor, S.C.; Zwicr. T.S.Chem. Phys. Lerr. 1989, 158,306.
(100) Zchnacker, A.; Duval, M.-C.; Jouvct, C.; Lardeux-Dedonder, C.; Solgadi, D.; Socp, B.; Bcnoist d'Azy, 0. J . Chem. Phys. 1987, 86, 6565. (101) Van Zte,R. D.; Blankcspoor, S.C.; Zwicr, T.J. Chem. Phys. 1987, 88, 4650.
states are also naturally rationalized by the poorer a-donor ability of H 2 0 versus NH,.
IV. Summary We have shown in this article that the bonding in complexes of ground-state Hg(6s6s'So) and excited-state Hg(6s6p3P,) with rare-gas atoms and small molecules can vary from purely physical to strongly chemical. In complexes of the tilled-shell ground-state Hg(6s6s'So) atomic state with rare gases and molecules such as Hz, CHI, NH3, or HzOthe bonding and structure can be understood within the context of weak, long-range multipolar interactions and short-range Pauli repulsions: "physical" (van der Waals) bonding. On the other hand, interactions in the analogous complexes of the excited Hg(6s6p3PI) state can vary from essentially physical (with rare gases or CH4) to chemical (with H20, NH,, Hz, or Hg). It has been postulated that threeelectron ($a*) bonding can be important in these chemical interactions. Future work in our laboratories will involve further characterization of the interactions of rare-gas atoms and small molecules with ground-state and low-lying excited-state metal atoms such as Cd, Zn, and Mg to explore the generalities of the bonding concepts discussed here. Extensions of such studies to the interactions of group I11 and transition-metal atoms, as well as to Rydberg states of metal atoms, are also planned.
Acknowledgment. This collaborative work was made possible by a US-France Cooperative Research Grant jointly funded by NSF and CNRS, for which we express our great appreciation. We also acknowledge general research support from NSF, CNRS, and the European Economic Community. We thank Mr. Steve Massick for preparing the figures. Finally, we are grateful for many stimulating conversations with C. Jouvet, Ph. Millie, and Jack Simons. Regishy NO. Hg, 7439-97-6; NH3, 7664-41-7; HZO, 7732-18-5; HZ, 1333-74-0; CHI, 74-82-8; Hgz, 12596-25-7.
