NEUTRONS & CATALYSIS by Juergen Eckert and Phillip J. Vergamini

atalysis—the ability of some ture of chemicals, could not function catalysts, is still. as it always has been, substances to alter the rate of without catalysts. It has been estimated, largely empirical. chemical reactions without for example, that catalysts are involved In the last few years, however, so- being consumed-was rec- at some point in the production of 60 to phisticated new analytical and compu- c tational techniques have helped stimu- ognized more than 150 years ago and 70 percent of all industrial chemicals. has been applied on an industrial scale Yet the store of knowledge about how late a renaissance in catalysis research. since the beginning of this century. catalysts work is surprisingly small. The Powerful economic forces have mo- Modem industrial chemistry, especially search for a catalyst for a particular re- tivated the study of catalysis as well: petroleum processing and the manufac- action, or for ways to improve existing the need for new sources of energy and Neutrons and Catalysis

chemicals, changes in the availability Unfortunately, these details of struc- of’ raw materials, potential restrictions ture and dynamics cannot easily be de- on the availability of noble-metal cat- termined in a “real-world” situation— alysts, and the desire for new products that is, during an actual catalytic reac- have pointed up the need for a clearer tion. Catalytic processes usually pro- understanding of catalytic processes. ceed under conditions that preclude the Much of the research into cataly- direct application of many powerful an- sis is directed toward metals. because alytical techniques-or at least make they catalyze many important reac- such application very difficult. Consid- tions. Metals may be catalytically ac- erable effort has therefore been devoted tive in the form of finely divided parti- to the study of so-called model systems, cles, organometallic compounds in so- which are designed to reproduce the lution, or ions bound to large biologi- critical relationships as accurately as cally active molecules, such as enzymes. possible. One useful model system is The catalysis may be heterogeneous in a single crystal of a metal for which the sense of involving more than one the surface arrangement of atoms is phase (solid metal and gaseous reac- known. Others that have been widely tants, for example) or it may be homo- used are synthetic molecules consisting geneous in the sense of involving only of a metal atom (or a cluster of metal one phase (such as a solution). What- atoms) surrounded by stabilizing lig- ever the form, when the metal binds to ands, usually carbonyl groups (CO) or a reactant molecule, it almost always other more complex organic groups. alters the chemical bonding in the re- When a reactant molecule such as ethy- actant, If that alteration is favorable to lene or benzene binds to such a syn- some particular reaction, then the metal thetic molecule, we can assume that, to is a catalyst for that reaction. some degree, the configuration of the To understand catalytic activity, or resulting complex resembles that of the to tailor a catalyst to do a specific job, same reactant adsorbed on a metal sur- we need to know the individual steps in face. The complex can be studied with the catalytic process in great detail, For several spectroscopic techniques, and its example, consider the hydrogenation of crystalline form can be characterized by ethylene to ethane, x-ray’ and neutron diffraction, which re- veal details of its architecture with great accuracy. The more closely the properties of the which can serve as a prototype of re- model system approximate the proper- actions used in producing synthetic fu- ties of the real-world system, of course, els. The production of synthetic fuel the better. As a result, model systems from coal, for example, involves various are often structurally modified to refine series of reactions, including the step- their properties and bring them closer in wise hydrogenation of carbon to form line with the more complex system of acetylene (HCCH), ethylene, and ethane, interest. However. such modifications as well as the stepwise hydrogenation can complicate the structural character- of carbon chains with more than two ization of the model system. For exam- carbon atoms. The hydrogenation of ple, as the model system becomes larger ethylene shown above is a particularly and more complex. the chances increase useful react ion to study because it can that some portions of the molecule will be carried out at moderate temperatures be disordered or less easily defined. The in the presence of a metal catalyst. The necessity of modeling the disorder can various steps to the reaction are repre- decrease the precision of the results for

116 Los Alamos Science Summer 1990 Neutrons and Catalysis

the metal-hydrogen interaction, which for a particular aspect of the problem, ing is in locating the all-important hy- is the feature of most interest. In effect, and neutron scattering is one of these. drogen atoms and highlighting the vibra- the results become slightly fuzzy and However, even the most intense neu- tional and rotational motions associated less precise. tron sources produce fluxes far below with them. This strength is a result of Besides being useful in the study of those commonly available from sources the fact that neutrons scatter as strongly catalysis, metal complexes are highly of photons (x rays, ultraviolet, visible from hydrogen as from most other el- suitable for theoretical studies of chemi- light, and infrared), and so neutron ements (see “Neutron Scattering—A cal bonding between the bound molecule scattering is not one of the principal Primer” by Roger Pynn). Although it (ligand) and the metal atoms. They tools of surface science. Nevertheless, is nearly impossible to “see” hydrogen are therefore of fundamental interest when the systems include hydrogen atoms in the presence of heavy met- to researchers studying chemical bonds als using x rays, x-ray diffraction can from first principles. Finally, metal- sometimes implicitly locate hydrogen cluster complexes can stabilize cer- atoms bound to or interacting with metal tain molecules that are unstable in pure atoms. If a site in a metal complex is form. For example, cyclobutadiene can usually filled, an apparent vacancy at be stabilized by binding to iron car- that site, together with other physical bonyl, Fe(CO)3; and ethylidyne, CH3-C and chemical evidence, can lead to the (a highly reactive intermediate formed inference that hydrogen occupies the po- in the hydrogenation of ethylene), can sition. Neutron scattering, however, is be isolated by reacting with cobalt car- needed to confirm the actual presence of bonyl to form the metal-cluster complex hydrogen. Thus, the structures of com-

CH3C–C03(CO)9. pounds of interest are typically deter- The kind of information available mined by first applying x-ray diffrac- through the study of model compounds tion to locate the heavier atoms and is illustrated by the case of the clus- then neutron diffraction to obtain pre- ter compound HFeCo3(CO)12. Diffrac- cise metal-hydrogen distances and bond tion studies show that the single hydro- angles. gen atom is located at a site of three- Historically, single-crystal neutron fold symmetry, that is, just outside A CLUSTER COMPOUND diffraction has been more difficult than the triangle formed by the three cobalt x-ray diffraction. Neutrons can travel atoms (Fig. 2). The vibrational spec- Fig. 2. The HFeCo3(CO)12 complex, which large distances through material without trum of hydrogen in this compound is contains a single hydrogen atom (blue) being scattered, so neutron diffraction very similar to that of hydrogen atoms located against an equilateral triangle of requires a much larger crystal. This chemisorbed on a nickel or a platinum cobalt atoms (red), can serve as a model problem has been partly alleviated by surface. Since the vibrational spectrum system for hydrogen atoms chemisorbed on the availability of more intense sources of a molecule or atom strongly reflects a metal surface. In particular, comparison of of neutrons. Furthermore, the time-of- the way in which it is bound to other vibrational spectra can help establish whether flight wavelength measurements possi- atoms, the similarity here allows the or not the hydrogen on the metal surface is ble at pulsed-accelerator-based neutron inference that hydrogen chemisorbed also located at sites with threefold symmetry. sources makes all neutrons in each pulse on a catalyst surface is located at a site (Adapted from a figure in an article by R. G. usable. Area detectors make it possi- of threefold symmetry. We can further Teller, R. D. Wilson, R. K. McMullan, T. F. ble to collect large volumes of data at infer that the catalytically active sur- Koetzle, and R. Bau. Journal of the American one time and make feasible full struc- face is the so-called (111) plane of the Chemical Society 100: 3071, 1978.) tural determination from polycrystalline metal, because that is the only crystal material. plane having threefold symmetry. This or molecules containing hydrogen—as For the observation of molecular vi- information could not have been easily do the more important types of com- brations, optical techniques (infrared obtained in any direct way. pounds involved in industrial catalytic absorption and Raman scattering) are How does one then study the model processes—neutron scattering is ex- far more common and much easier to systems? There are many experimen- tremely useful. use than neutron scattering. Once again, tal techniques, each especially suited The singular utility of neutron scatter- however. the difference in the nature of

Los Alamos Science Summer 1990 117 Neutrons and Catalysis

the interaction between the scatterer and the ring hydrogens is replaced by a The Hydride Ligand the probe makes neutron-scattering vi- methyl group), the remaining ring hy- brational spectroscopy advantageous in drogens can be replaced by deuterium The first example we want to dis- certain cases. First, absorption of pho- atoms. Then, as far as neutron scatter- cuss in detail is the interstitial hydride tons in optical spectroscopy depends on ing is concerned, the deuterium atoms ligand—a single hydrogen atom bound the symmetry properties of the vibra- are much less “visible” than the three to a metal atom or atoms. Hydride lig- tional mode being excited in the sample, hydrogen atoms on the methyl group. ands are usually formed on metal sur- whereas no such symmetry-based se- (In optical spectroscopy, isotopic substi- faces when molecular hydrogen disso- lection rules apply to inelastic neutron tution alters the frequency of vibration ciates and are referred to as terminal, scattering. (We use inelastic to refer to but leaves the intensity of absorbed or doubly bridging, triply bridging, and the fact that the neutron loses or gains scattered photons virtually unchanged.) so forth. depending on whether they energy during the scattering process. Another application of isotopic substi- are bonded to one, two, three, or more The change in energy corresponds to a tution in neutron scattering involves the metal atoms. change in the vibrational energy of the differential spectra of isotopic species, In large cluster complexes with many scattering molecule. ) Hence, in principle examples of which will be described in metal atoms, hydride ligands may also (though not necessarily in fact), all vi- the following sections. occupy interior, or interstitial, sites. brations of a molecule can be observed We have chosen to describe three ex- Among the large metal-cluster com- by inelastic neutron scattering. amples of neutron-scattering studies on pounds of this type that have been syn- The factors determining the intensity metal complexes, each of which may thesized, two—the octahedral clusters of a given excitation are a second, and serve as a model system for a particu- of cobalt and ruthenium—stand out for perhaps more important, difference be- lar step in the hydrogenation reaction their remarkable simplicity. Both these tween neutron-scattering and optical vi- shown in Fig. 1. The first example is compounds have a cluster of six metal brational spectroscopy. Large-amplitude a hydride ligand in an octahedral clus- atoms that form the octahedral site, a vibrations by nuclei with high neutron- ter of metal atoms, a model system that single hydride ligand, and several car- scattering cross sections (such as hydro- may help us understand the motion of bonyl groups outside the metal cluster gen) typically give rise to intense inelas- hydrogen atoms between the surface and that serve to stabilize the molecule. tic neutron-scattering bands; whole-body the region just below the surface (once This kind of hydride coordination looks very much like that observed in Vibrations of molecules are typical ex- the H2 molecule has dissociated on the amples. Such motions, however, usually surface). The other two examples—an bulk metals, where interstitial sites of involve only small changes (if any) in ethylene-diosmium complex and a set octahedral or tetrahedral symmetry may the polarizability or the dipole moment of molecular-hydrogen complexes— be occupied by hydrogen. A hydrogen of the molecule, which are the factors may be regarded as models for the bond atom in a metal is of course surrounded that govern intensities in Raman scat- activation that precedes the actual re- by many more metal atoms than the tering and infrared absorption. Thus, action on the surface. The complexes six of a cluster compound. There are optical and neutron-scattering methods that bind molecular hydrogen are par- six atoms as nearest neighbors in an are remarkably complementary. ticularly important in this context be- octahedral site, but further shells of The utility of inelastic neutron scatter- cause they represent a “capture” of the metal atoms occur at ever increasing ing can be greatly enhanced by replac- long-sought intermediate in perhaps the distances. However, if the hydrogen ing certain atoms, whose vibrations are most fundamental reaction, the disso- atom is located just below the surface to be highlighted, with isotopes of dif- ciation of hydrogen molecules. As we of the metal, that is, between the first ferent neutron-scattering cross sections. shall show below, elastic and inelas- two layers of metal atoms, the number Such isotopic substitution is particularly tic neutron-scattering studies of these of more distant neighbors is minimized valuable for studying hydrogen, because compounds have provided remarkably in one direction. The interstitial hydro- the neutron-scattering cross sections ap- detailed information on the nature of gen in the cluster may therefore be a propriate to inelastic neutron scattering the chemical bond between the dihydro- better model for “subsurface” hydro- for hydrogen and deuterium differ by gen, or molecular hydrogen, ligand and gen than for hydrogen within the bulk more than an order of magnitude. For the metal center, including evidence for of the metal. Such a system may help example, to distinguish the motions of back-donation of electron density from answer a question raised earlier in the the methyl group in toluene (C6H5CH3, the metal to the antibonding orbital of discussion of hydrogenation-that is, a benzene derivative in which one of the hydrogen molecule. where the hydrogen is likely to move,

118 Los Alamos Science Summer 1990 Neutrons and Catalysis

(a) (b) The Central Octahedron

Site

co

AN OCTAHEDRAL CLUSTER

- Fig. 3. (a) The location of the hydride ligand in the anion [HC06(CO)15] has been determined with considerable certainty by neutron diffraction + from a single crystal containing the cation [(Ph3P)2N] . The hydride ligand (blue) is located at the center of an octahedron of cobalt atoms (red); the cobalt atoms, in turn, are surrounded by twelve carbonyl ligands (green and yellow). The shapes at each atomic position are thermal ellipsoids, which indicate the extent and direction of the thermal motion of the atoms about their equilibrium positions. (The surface of each ellipsoid defines the volume in which the atom is contained 50 percent of the time.) (b) There are a number of alternative sites close to the central octahedron of cobalt atoms that may, in some compounds, serve as the location of the hydride ligand (two possibilities are shown in light blue). The alternative sites are either external or internal to the octahedron of cobalt atoms, and the hydrogen atom can be doubly or triply bridged to those atoms. However, such sites have considerably less symmetry than the central octahedral site, and the vibrational spectra of hydrogen when located at such a site would be quite different from the single excitation that is actually observed in the spectrum of [HC06(CO)15] . (Adapted from a figure in an article by D. W. Hart, R. G. Teller, C.-Y. Wei, R. Bau, G. Longoni, S. Campanella, P, Chini, and T. F, Koetzle,Journal of the American Chemica/ Society 103: 1458, 1981.) after dissociation, relative to the metal must be used for the diffraction stud- A neutron-diffraction study of a sin- catalyst’s surface. ies. Also, cluster compounds are most gle crystal of [(Ph3P)2N ][HCo6(CO)15],, The first step, however, in under- commonly ionic species, and a suit- carried out on the high-flux-beam reac- standing the microscopic properties of able counter ion -a large, oppositely tor at Brookhaven National Laboratory, hydrogen in a metal cluster is to attempt charged ion—must be added to pro- showed quite convincingly that the hy- to locate it by diffraction studies. The duce sufficiently large single crystals. In drogen is located approximately at the only reliable way to do this is by use the present case, the complex of inter- center of a somewhat irregular octa- of neutron beams, for the reasons dis- est is [HCO6(CO) 15]-, and the counter hedron of cobalt atoms (Fig. 3) even cussed earlier. Because hydrogen is a ion used to produce the single crystal is though other locations. such as three- + minor component of the rather large [(Ph3P)2N] , in which Ph is the phenyl fold coordination inside or outside one metal-cluster molecules, single crystals group, C6H5. of the triangular faces of the octahedron,

119 Neutrons and Catalysis

are possible. The vibrational spectrum either case, a second vibrational line at of hydrogen in a regular octahedral site higher frequency would be expected. would show a single excitation, a triply The data are not conclusive in this re- -1 degenerate hydrogen-metal stretching spect, but if the band at 950 cm is the mode. If, however, the hydrogen were symmetric stretching vibration for the to also move significantly off center, doubly or triply bridged hydrogen, then additional peaks would appear in the the high-frequency shoulder just above vibrational spectrum. It is in just such 1100 cm–1 has about the expected fre- cases that vibrational spectra are of quency for the asymmetric stretch of great value in obtaining structural infor- triply bridged hydrogen. mation. An inelastic neutron-scattering The spectrum thus appears to reveal study of the cesium salt of the same an instance of the fluxionality of the cobalt cluster showed primarily a single hydride ligands in cluster compounds. excitation at a frequency of 1056 recip- Fluxionality-commonly detected in rocal centimeters (cm– ]), confirming the nuclear-magnetic-resonance studies— central location of the hydride ligand in refers to the movement of hydrogen the octahedral site. from one site to another. Because the Prompted by the results of some in- movement occurs on a time scale that is frared spectroscopic studies that showed many orders of magnitude greater than interesting changes in the spectra of the the time scale of a typical vibration, - [HC06(CO)15] cluster as the crystalline the hydrogen can be “caught” vibrating environment was altered, we recently rapidly at more than one site. However, investigated the vibrational spectrum if the binding energy is much larger at of the cluster combined with the very one site than at others, such fluxionality much smaller counter ion K+. The data, is unlikely. shown in Fig. 4, were obtained by the In any case, the remarkable result of differential technique on the Filter Dif- our studies is that the position of the ference Spectrometer at the Manuel hydride ligand in these metal clusters Lujan, Jr. Neutron Scattering Center apparently depends on the nature of the (LANSCE) at Los Alamos. Two sam- counter ion used to crystallize the com- ples were measured, one with hydrogen cm -1 is not present in the spectra of crystals pound. This fact suggests that the bind- and one with deuterium as the ligand. containing Cs+ as a counter ion and may be ing strengths for hydrogen at the vari- Because the scattering cross section for the stretching vibration of doubly or triply ous sites differ by only small amounts deuterium is much smaller than that for bridged hydrogen located at an alternative and may, in fact, be affected by the hydrogen, the vibrational spectrum of site (such as those shown in light blue in charge balance between the complex the deuterated compound serves essen- Fig. 3b). If the latter assignment is correct, ion and its counter ion. Such a conjec- tially as a “blank” to be subtracted from the high-frequency shoulder just above 1100 ture is needed to explain the observed that of the protonated compound. The cm -1 would correspond to the asymmetric change in fluxionality of the hydrogen resulting differential spectrum is free of stretching vibration of hydrogen at a triply atom in the cluster. Moreover, the con- all the many vibrational modes of this bridged site. These data were obtained by jecture is in agreement with nuclear- large molecule that do not involve mo- using the Filter Difference Spectrometer at magnetic-resonance observations of the - tion of the hydrogen and thus highlights LANSCE. [HC06(CO)15] ion in solution, which the vibrational modes that do involve show that the hydrogen can easily leave hydrogen. band in the region between 1050 and the octahedron and exchange with pro- The features shown in the differential 1100 cm-1 may certainly be assigned tons of the solvent molecules. spectrum can be immediately identified to the stretching vibration of hydrogen The factors that govern fluxionality of with hydrogen vibrations and suggest at the interstitial site, but the band at the hydride ligand in cluster compounds that all the hydrogen atoms are by no 950 cm–] must then be indicative of may, of course, differ considerably from means located at the center of the oc- hydrogen at a different site—one bridg- those that determine the diffusion of hy- tahedron of cobalt atoms. The broad ing either two or three cobalt atoms. In drogen between the metal surface and,

120 Los Alamos Science Summer 1990 Neutrons and Catalysis

(a) Ethylene for example, the subsurface layer. If one wishes to hydrogenate, say, ethy- lene on a metal surface. it is important

dissociation to react with the ethylene rather than diffuse rapidly into the bulk

portant for understanding the diffusion of hydrogen along a metal surface or between the surface and the bulk.

The Ethylene-Metal Complex We now focus on another question in the hydrogenation reaction, namely (b) Ethane the formation of a complex between ethylene and the metal and the result- ing bond activation that is necessary for ethylene to take up hydrogen. Hydro- genation changes the ethylene molecule from an arrangement in which all the atoms are coplanar and the carbon- carbon bond is a short double bond that includes n-bonding electrons (Fig. 5) to an ethane molecule in which the atoms are grouped tetrahedrally and the carbon-carbon bond is the longer sin-

Here again, we shall examine a model compound in which the precise. arrange- ment of the ethylene molecule and the metal atoms can be studied by neutron diffraction. Ethylene molecules can interact with a metal surface in several ways. Per- haps the most common configuration is GEOMETRY OF ETHANE AND ETHYLENE in Fig. 6a, in which the planar ethy- Fig. 5. (a) All the atoms of ethylene lie in a single plane, and almost all of its electron orbitals lene molecule is parallel to the surface are concentrated close to that plane. The exception is the double bond between the two carbon carbon-carbon double bond interact di- rectly with the electrons on a single other hand, are arranged in two overlapping tetrahedral groups that surround each carbon atom bonded complex (Fig. 6b), has also been

arrangement allows the two CH3 groups to rotate with respect to one another. observed: in this case the double bond

Los Alamos Science Summer 1990 121 Neutrons and Catalysis

between the carbon atoms is reduced to formed between each carbon atom and one of two adjacent metal atoms. Ob- viously, this last complex could be an intermediate species in the hydrogena- tion reaction,

The OS2(CO)8(C2H4) complex (Fig. 7) does in fact have an ethylene ligand try. Further, the vibrational spectrum of this complex is very similar to that of ethylene chemisorbed on platinum at temperatures below 100 kelvins, and the complex can serve as a model for that system. X-ray diffraction studies of OS2(CO)8(C2H4) show both that the carbon-carbon distance is longer than a normal ethylene double bond and that the four-membered ring formed by the two osmiums and two carbons is twisted and nonplanar. This last observation implies that the hydrogens have proba- bly also twisted out of their plane with the carbons. However, as we already pointed out, it is very difficult to di- rectly determine the positions of the hydrogen atoms with x rays. Spectro- scopic evidence also suggests unusual structural features within the bridging ethylene ligand. This evidence may or may not be consistent with the x-ray observations but cannot be interpreted in an unambiguous way. Therefore, a TWO KINDS OF ETHYLENE LIGANDS knowledge of the detailed structure, par- ticularly the positions of the hydrogen Fig. 6. Ethylene can form two distinctly different types of ligand bonds with metal atoms. (a) atoms, is necessary to resolve questions regarding the bonding in this compound. Neutron-scattering measurements show that two hydrogens, an osmo- two sigma-like bonds between the carbon atoms and two adjacent metal atoms. As a result, the nium atom, and the other carbon atom carbon-carbon bond in the ethylene becomes a single u bond and the hydrogen atoms move out are arranged approximately tetrahedrally of the molecular plane and assume an approximate tetrahedral arrangement about the carbon around each ethylene carbon atom. This atoms. observation is consistent with the elon- gation of the carbon-carbon bond, as to a metal-cluster compound results in mium carbonyl cluster. Simply replac- well as with the spectroscopic evidence rearrangement of the bonding electrons ing the two osmium atoms with hydro- and theoretical calculations that allow in the smaller molecule. gen atoms would produce the geometry The planar configuration of the ethy- found in ethane, the product of hydro- bon atom and an osmium atom. In this lene molecule is obviously drastically genation of ethylene. The twist ob- example, then, the bonding of ethylene altered by its association with the os- served in the cluster-bonded ethylene

122 Los Alamos Science Summer 1990 Neutrons and Catalysis

is tending toward the bond angles nor- may play: the P–W–P axis is distorted mally found in ethane. and the organic portion of one phos- In this example, the osmium cluster phine appears to fill the hole left by the can be considered a model of either an absent molecule of hydrogen. isolated fragment of a metal surface (as Neutron-scattering techniques have in heterogeneous catalysis) or an indi- played a decisive role in characterizing vidual catalytic molecule (as in homoge- the dihydrogen ligand of the complexes neous catalysis). In either case, osmium in terms of both its structural and dy- is not necessarily unique in completing namical properties. This information has with ethylene. Other metal atoms have then been used to work out a detailed different electrons at different energy quantitative picture of the bonding of levels, so the degree of activation and AN ETHYLENE-BRIDGED COMPLEX the hydrogen molecule to the metal, as distortion may differ from one complex will be described in the following sec- to another. However, all of them should tions. have a tendency to activate ethylene of ethylene ligand (Fig. 6b) may be found to one degree or another by forming a in the osmium complex Os2(CO)8(C2H4). A Sigma-Bond Complex. In the early complex of this kind. Not only has the carbon-carbon bond in the stages of the investigation of the com- ethylene ligand lengthened, but the ligand has plexes, it was absolutely essential to lo- cate the dihydrogen ligand and ascertain Binding of Molecular twisted, allowing the hydrogen atoms (blue) and the two osmium atoms (red) to assume a whether, in fact, it retained its molecu- Hydrogen to a Metal more tetrahedral grouping about each central lar identity. Although some initial evi- In this final example we shall go back carbon atom (green). dence for the molecular-hydrogen bind- one step in the hydrogenation reaction ing came from x-ray diffraction, conclu- and focus on the reaction of molecular may represent the long-sought interme- sive evidence required the use of neu- hydrogen with a metal atom, the reac- diate in the oxidative addition of hy- tron diffraction because of its sensitivity tion that precedes its dissociation into drogen to a metal. Since then, many to scattering from hydrogen. The first hydrogen atoms. As mentioned ear- additional molecular-hydrogen com- structure determined for a molecular- lier, most small molecules can bind plexes with central metal atoms other hydrogen complex is shown in Fig. 8. chemically to complexes containing than tungsten and ligands other than The complex is the same as the one one or more metal atoms, often in ways tricyclohexylphosphine have been iden- we have been discussing except the that roughly resemble the chemisorbed tified. The hydrogen in these complexes tricyclohexylphosphine ligands (PCy3) state of the molecule. The coordinated is apparently reversibly bound to the have been replaced with less bulky tri- molecule and the metal atom or atoms metal, as can be demonstrated by pass- isopropylphosphine ligands (P(i-Pr)3, share electrons to some extent; as a re- ing hydrogen gas into a solution of, for where i -Pr represents CH(CH3)2). Prob- sult, some bond angles or bond lengths example, the precursor of the above ably the most important features of this in the bound molecule are changed. tungsten-tricyclohexylphosphine com- structure are the two equal W–H dis- Molecular hydrogen has always been plex at room temperature. The solution, a notable exception; until recently it was which is originally purple. turns yellow, H–H distance by some 10 percent over found to bind only dissociatively, that and light yellow crystals of the H2 com- is, as two individual hydrogen atoms. plex can be precipitated from it. If the These facts clearly suggest the forma- Observation of chemically bound molec- hydrogen stream is replaced by a chemi- tion of a three-center metal-dihydrogen ular hydrogen would offer enormous cally inert gas such as argon, the purple bond (that is, some of the electrons are potential for understanding on the basis color returns, implying the dissociation shared between the metal atom and the of first principles the process that even- of H2 from the complex. An interest- two hydrogen nuclei) and a substantially tually results in dissociative binding of ing feature of these compounds is that weakened H–H bond. hydrogen. formation of a stable hydrogen complex The dihydrogen ligand was also found A few years ago, G. J. Kubas and apparently requires organophosphine to have a well-defined equilibrium ori- collaborators isolated the tungsten com- ligands that are large and bulky. The entation (Fig. 8c), one in which the plex W(CO)3(PCy3)2H2 (where Cy is structure of the purple precursor con- H–H axis is parallel to the P–W–P axis cyclohexyl, a 6-carbon alkane ring) that tains a clue to the role these ligands of the complex. This fact might be ex-

123 Neutrons and Catalysis

A DIHYDROGEN COMPLEX Tri-isopropylphosphine Ligand Fig. 8. (a) W(CO)3(P(i-Pr) 3 )2H2, the first molecular-hydrogen complex to have its structure determined, has two bulky tri- The Central Region isopropylphosphine ligands (orange, blue, and H 2 green) located on opposite sides of a tungsten atom (red). The central region between the opposing phosphorus atoms (orange) contains three carbonyl ligands (green and yellow) and the molecular-hydrogen ligand (blue). The fact that the H-H bond length (0.82

A) is longer than in free H2 (0.74 A) and the fact that the two W–H bond lengths are equal (1.89 A) suggest a three-center metal-dihydrogen bond and a substantially weakened H–H bond.

(b) The preferred orientation of the H2 ligand is parallel to P–W–P axis, suggesting that there is a barrier to rotation of the H2 ligand about the W–H2 axis. (c) A potential-energy curve for rotation of the H2 ligand in a plane perpendicular to the W–H2 axis with one minima for the identical orientations of O and I (b) H2-Ligand Dynamics 180 degrees from the P–W–P axis. Because the ground-state wave functions (dashed lines) for each potential well overlap (shaded areas), there is tunneling between potential wells and, as a result, the energy levels split. plained on the basis of interactions be- tween the dihydrogen ligand and other ligands bound to the metal that would make alignment perpendicular to the P–W–P axis (and parallel to the OC– W–CO axis) energetically less favorable than alignment parallel to the P–W–P axis. We also observed that the hydro- gen atoms of the bulky organophos- (c) Potential-Energy Barrier to Rotation phines formed a pocket around the re- gion of the dihydrogen ligand, but the orientation of these organic groups is very accommodating and would not be expected to constrain the H2 molecule. Theoretical analysis is necessary to derive a more quantitative picture of the metal-ligand bonding than that indicated by the structural results. The most fun- damental types of calculations can, in fact, derive structural parameters such Neutrons and Catalysis

as the H–H or W–H distances; com- tion reactions. These theoretical predic- parison with experimental values then tions, of course, require experimental serves as a check on the validity of the confirmation. theory. The current problem, however, is sufficiently complex that structural Rotational Dynamics. Hydrogen in information is used as input to simpli- the side-on coordination mode can un- fied theoretical models. Whether or not dergo a remarkably wide variety of lig- the theoretical model is derived from and dynamics, including torsional os- first principles or from a combination of cillations, or librations, about its equi- structural data and a theoretical model, librium orientation and much slower it is highly desirable to have other ex- Orbitals 180-degree reorientations by tunnel- perimental information on the nature of ing through the rotational barrier. Es- the chemical bonding that can be used BONDING OF tablishing the presence of a significant to gauge the theoretical picture. MOLECULAR-HYDROGEN LIGAND electronic energy barrier to rotation The nature of the bonding between would provide confirmation of metal- the dihydrogen ligand and the transition Fig. 10. Although the main bonding between to-H2 backbonding. Such a barrier is metal is of major significance because the dihydrogen Iigand and the metal atom too small to be observed by standard the complex represents the first exam- is due to an interaction between an empty nuclear-magnetic-resonance techniques. ple of a sigma-bond complex, that is, Inelastic neutron scattering, however, a complex in which the ligand binds hydrogen molecule (here shown in a cross- is highly sensitive to hydrogen mo- through interaction of a metal center sectional view), there is evidence for some tions because of the very large neutron- with a o-bonding electron pair. The- backbonding, which is an interaction between scattering cross section of protons and oretical studies of this three-center, an antisymmetric metal-atom orbital and the the typically large amplitude of the mo- two-electron bond indicate that both tions. In fact, this technique is routinely the bonding and antibonding orbitals of these interactions weaken the H-H bond used to study rapid rotational motion hydrogen (Fig. 9) may be involved. The and strengthen the M-H bond. The former (for example, of methyl groups and of interaction donates electron density from the solid or liquid hydrogen or molecular

o-bonding orbital of H2 to the metal atom, hydrogen in zeolites). whereas the latter interaction puts electron The nature of the rotational motion of

density from the metal atom into the H2 the bound hydrogen molecule may be antibonding orbital. described with the aid of a diagram that shows the energy levels that the dihy-

primary interaction between H2 and the drogen ligand may occupy as a function metal atom is donation of electron den- of the height of the barrier hindering the MOLECULAR-HYDROGEN ORBITALS rotation. These levels are the solutions orbital in the metal atom (Fig. 10); how- to the Schrodinger equation chosen to Fig. 9. The usual theoretical picture of ever, the same studies indicate that, to represent the rotational motion of the H-H bonding has the two electrons in the a lesser degree, backbonding between bound hydrogen molecule. In particular. hydrogen molecule occupying a low-energy a metal orbital and the H2 antibonding the equation includes only one angular degree of freedom because we assume map (here pictured schematically in a cross- bonding stabilizes the side-on orienta- that the relatively strong three-center sectional view) generally occupies the space tion shown in Figs. 8 and 10 rather than metal-dihydrogen bond keeps the hydro- between the two hydrogen nuclei. However, an end-on orientation (in which the H2 gen ligand essentially in a plane during molecule would have its bonding axis its rotational motion. The complex may, orbital that is usually unoccupied and whose pointed straight at the tungsten atom in fact, be the first example of hydro- electron-density map has a node between the with one hydrogen atom much closer gen rotation with only one degree of two hydrogen nuclei. (The plus and minus to the metal atom than the other). The rotational freedom, a situation first de- patterns in the antibonding orbital are there side-on coordination ultimately facili- scribed by Pauling as an approximation to indicate the antisymmetric nature of the tates cleavage of the H–H bond to give for solid hydrogen. If any mixing with orbital.) dihydride complexes in oxidative addi- vibrational modes can also be neglected,

Los Alamos Science Summer 1990 125 Neutrons and Catalysis

the Schrodinger equation for the rota- tional motion is

where B is the rotational constant (B = h2/21, where I is the moment of inertia of the molecule for the rotation in ques-

the O–C–W–H2 axis, V2n represents the barrier-height potential energy for a po-

are, respectively, the wave function and energy of the allowed rotational states. In the present case, as we’ve already

pointed out. crystal-structure studies Barrier Height (V2/B) as well as theoretical calculations have shown the dihydrogen ligand to have a ROTATIONAL ENERGY-LEVEL DIAGRAM well-defined orientation parallel to the P–W–P axis. The hydrogen molecule Fig. 11. A dumbbell molecule (such as hydrogen) constrained to rotate in a plane has one in this complex should then have two rotational degree of freedom and rotational statesJ = O, 1, 2, 3, . . at energies of 0, B, 4B, 9B,...

equivalent orientations located at po- if there is no barrier to the rotation (that is, if V2 = O). On the other hand, if V2 is very high (that tential minima that are 180 degrees is, beyond the right side of the figure), the molecule will occupy a set of equally spaced torsional apart (Fig. 8c), and we may assume oscillator levels. For intermediate barrier heights we find a series of split Vibrational states. The that the term with n = 1 (a simple observed transitions (indicated by arrows) are of two types: transitions within the Vibrational double-minimum potential) will dom- inate. Equation 1 can then be reduced ground to the first excited Vibrational state called torsional transitions. Because photons do not to the Mathieu equation, for which couple with nuclear moments, optical spectroscopy cannot be used to observe the tunneling solutions are tabulated. The resulting transitions directly and can be used to observe only the torsional transitions between levels energy-level diagram as a function of = O). The observed values for the transition 1 barrier height V2 is shown in Fig. 11, energies are scaled by a value for B of 49.5 cm–’ rather than the 60 cm - value for free H2 to

in which both the energies and barrier reflect the increased H-H bond length (0.82 A) relative to free H2 (0.74 A). The transitions shown

heights are given in terms of B. are for a complex with a barrier height V2 equal to 15.7 B. The energy levels corresponding to

V2 = 0 (left axis in Fig. 11) are those of The molecular hydrogen complexes (Fig. 8c). To satisfy the Pauli principle, a free rotor with one degree of freedom being discussed here have intermediate the degenerate states corresponding to 2 (Ej = BJ , where J is the rotational barriers, and for these, we find a series these two orientations must split into quantum number, yielding levels at en- of vibrational states, each of which is two states, each with a slightly different ergies of 0, B, 4B, 9B. . .). introduc- split relative to the torsional oscillator energy (Fig. 11 inset). The splitting is

tion of a barrier to this rotation (V2 > O) levels. This splitting arises from the fact called tunnel splitting because it is due changes the level spacing drastically and that the barrier is not overly high, al- to the overlap of wave functions through removes some degeneracies. In the limit lowing the amplitude of the libations a potential barrier. The size of the split- of very high barriers (suggested by the of the hydrogen molecule to be rela- ting decreases rapidly with increasing arrows on the right side of Fig. 11), the tively large—large enough, in fact, for barrier height and is thus an extremely states approach a set of equally spaced the wave functions that correspond to sensitive measure of barrier height. energy levels characteristic of essen- the molecule’s being located in either The two resulting states are charac- tially harmonic torsional oscillations. of the two potential minima to overlap terized by their symmetry. For exam-

126 Los Alamos Science Summer 1990 Neutrons and Catalysis

(b) W-P(i-Pr) Complex (a) W-PCY3 Complex 3

Experimental Confirmation. Now that we have selected an appropriate model for the rotational dynamics of the dihydrogen ligand in our system, it is 300 400 500 600 400 600 800 1000 a simple matter to relate the observed rotational transitions to the height of Frequency (cm-’) Frequency (cm-’) the rotational barrier. To observe both the high-frequency transitions to the ex- VIBRATIONAL-ROTATIONAL SPECTRA cited Vibrational state (the longer arrows in Fig. 11) and the very-low-frequency Fig. 12. The high-frequency transitions associated with torsional, or rotational, motion of transitions associated with rotational the dihydrogen Iigand have been identified for the two complexes (a) W(CO) (PCy ) H and 3 3 2 2 tunneling (the two short arrows in the (b) W(CO) (P(i-Pr) ) H by using the Filter Difference Spectrometer at LANSCE. Unrelated 3 3 2 2 exploded portion of Fig. 11), we had frequencies in the spectra were eliminated by taking the difference between the scattering to perform experiments on two spec- spectrum for the complex with a dihydrogen ligand and that for the complex with a dideuterium trometers, each located at a different ligand. The deformation modes include rocking and wagging of the dihydrogen ligand with neutron source. The high-frequency respect to the complex. One piece of evidence that the assignment are correct is the fact that, torsional transitions were measured for inelastic neutron scattering, the modes with the largest-amplitude motions of the hydrogen on the Filter Difference Spectrometer atoms have the highest intensity and the rotational modes involve more motion of the hydrogen at LANSCE by using two samples for atoms than the rocking modes. each complex, one of which had dideu- terium instead of dihydrogen ligands. pie, 180-degree rotation corresponds to however, do couple and are quite use- Vibrational modes involving mainly the an odd permutation of identical spin-½ ful for studying rotational transitions dideuterium ligand cannot be “seen” in particles (the protons), with respect to of this type. The neutron has a nuclear the presence of the many more modes which the total ground-state wave func- spin of ½, and a flip of the neutron spin that include hydrogen motion (that is, tion must be antisymmetric. The low- during the scattering process will cause those of the organophosphine ligands).

temperature wave function can be con- the total nuclear-spin state of the H2 The deuterium-substituted sample thus structed from linear combinations of served as a “blank” for subtracting all nuclear-spin and rotational wave func- A spin-flip neutron-scattering process the various vibrational modes except tions. Thus, a symmetric nuclear-spin then allows direct observation of the those of interest—the motions of the wave function (I = 1, where I is the ortho-para transition in hydrogen—for dihydrogen ligand. Figure 12 shows nuclear-spin-state quantum number) example, para-hydrogen (with I = O the results for two tungsten complexes: combines with an antisymmetric rota- and J even) changing to ortho-hydrogen one with tricyclohexylphosphine ligands tional wave function (J odd) and vice (with I = 1 and J odd). For a free hy- (PCy3) and one with the less bulky tri-

versa. These two cases correspond for drogen molecule with two rotational isopropylphosphine ligands (P(i-Pr) 3). zero barrier height to the two kinds of degrees of freedom, the transition that The low-frequency rotational tunnel-

H2 molecules referred to as ortho- and changes the rotational state from J = O ing spectra for three complexes (Fig. 13) para-hydrogen, respectively. For finite to J = 1 has an energy of 2B, where were obtained on a cold-neutron time- barrier heights, J is no longer a “good” B is the rotational constant. However, of-flight spectrometer at the High Flux quantum number to describe the en- if the molecule is constrained to rotate Reactor of the Institut Laue-Langevin ergy levels. The total nuclear spin of in a plane with only one degree of ro- in Grenoble, France. No “blank” sam- the molecule, however, must still change tational freedom, as is the case for our ple was necessary in this case, since in a transition between the two low- compounds, the transition has an en- the other ligands were not expected to est energy levels, that is, in a tunneling ergy of B for zero barrier height, that is, have observable excitations in the fre- transition. for free rotation. Moreover, as we dis- quency range of interest for this experi- We note that transitions in which cussed above, the energy for a tunneling ment (which is less than 10 cm-1). the total nuclear spin of the molecule transition rapidly becomes smaller with For the two tungsten complexes with

changes cannot be observed in optical increasing barrier height until, at infinite PCy3 and P(i-Pr)3 ligands, this type spectroscopy because photons do not barrier height, the splitting disappears of analysis yielded a significant bar- couple to the nuclear spin. Neutrons, and the two states become degenerate. rier height—one that was roughly 15

Los Alamos Science Summer 1990 127 Neutrons and Catalysis

ROTATIONAL-TUNNELING SPECTRA

Fig. 13. The spectra for the low-frequency transitions associated with rotational tunnel- ing are shown here for the three complexes (a)

peak in each spectrum is an elastic-scattering line, whereas the peaks to both sides of that line are the inelastic-scattering transitions associated with rotational tunneling. The fact that the inelastic peaks have a doublet nature is most likely due to structural disorder in the

crystals. I I I I 4 -2 0 2 times our derived rotational constant -4

for bound H2. Using the barrier heights and the energy-level diagram (Fig. 11), we were able to calculate the frequen- cies expected for the high- frequency transitions associated with the torsional motion of both complexes. The cal- culated values are in good agreement with the experimental values measured with the Filter Difference Spectrometer (Fig. 12), which suggests that the sim- ple model of planar reorientation in a double-minimum potential is a reason- able description for the hydrogen motion in these systems. The crucial question at this point be- comes what interactions give rise to the W-P(i -Pr) Complex barrier to rotation. Two possible sources (c) 3 for the hindrance potential are electronic and steric effects. By an electronic ef- fect we mean that the dihydrogen ligand may be constrained in its orientation be- cause of the way the chemical bond is 0.73 cm-’ formed with the metal. In other words, the electron orbitals on the metal shar- ing electrons with those of the dihydro- gen ligand have a symmetry that deter- mines the orientation of the ligand. Steric effects refer to the interactions of the dihydrogen ligand with the sur- rounding atoms of the other ligands. I I These are nonbonding interactions that I I I o 1 2 may be described by van der Waals -2 -1 -1 forces between pairs of atoms. They Energy Transfer (Cm ) may be summed up for all pairs formed Los Alamos Science Summer 1990 128 Neutrons and Catalysis

by using either one of the H atoms on Table 1 the dihydrogen ligand and any one of the surrounding atoms. As the dihydro- Barrier Heights to Rotation for the Dihydrogen Ligand gen ligand is rotated, the sum of these interactions shows an angular variation, Observed Theoretical (keal/mole) which gives rise to an effective “steric” Complex (keal/mole) barrier. In an attempt to sort out the relative Molecular Metal Atom Ligand ab initio Sum effects of these two types of interac- Mechanics (pH ) tions, we performed separate measure- 3 ments on the two tungsten complexes W 0.6 1.8 2.4 2.2 with PCy and P(i-Pr) ligands; then PCy3 3 3 W P(i-Pr) 1.4 1.8 3.2 2.4 we replaced the tungsten atom in the j Mo 0.6 0.6 1.2 1.5 PCy 3 complex with a molybdenum atom PCy3 and took measurements on this third complex. Thus, we hoped to gauge the effects of changing the central metal tractions. Thus, the experimental evi- viewed as representing mainly steric atom and of replacing the large, bulky dence, at least in these cases, strongly effects. In this case, the pairwise, non-

PCy3 ligand with the less bulky P(i-Pr)3 suggests that the barrier to H2 rota- binding interactions between the hydro- ligand. tion is determined more by electronic gen atoms of the dihydrogen ligand and The peaks in the spectra of Fig. 13 than steric effects. To test this conclu- each of the other atoms of the molecule to the left and right of the strong elastic sion, Jeff Hay, John Hall, and Caroline are summed. The summation is repeated line represent the rotational-energy split- Boyle of the Theoretical Division at Los for each orientation of the ligand, gen- ting associated with the H2 molecule Alamos earned out two sets of calcula- erating a curve of potential energy as tunneling through the barrier from one tions: an ab initio calculation—that is, a function of orientation. This calcula- 180-degree orientation to the other. The from first principles—and a molecular- tion is not sensitive to the type of metal position of these lines is extremely sen- mechanics calculation. atom at the center of the molecule. The sitive to the height of the barrier. A The ab initio calculation treats pri- results show a barrier height of 0.6 kilo- comparison of the three spectra shows marily the electronic effects because calorie per mole for the complexes with that replacement of tungsten (Fig. 13b) a full set of one-electron wave func- PCy3 ligands and 1.4 kilocalories per with molybdenum (Fig. 13a) changes tions for the whole molecule is used to mole for those with P(i -Pr)3 ligands. the tunneling frequency by a factor of compute the relative energy of a given If one makes the assumption that just over 3, from 0.89 to 2.82 cm-]. On configuration. The barrier to rotation molecular mechanics treats only the the other hand, replacing the PCy3 lig- was obtained from the difference in steric effects and that the ab initio the- and in the tungsten complex with the total energies for the structure with ory accounts primarily for the direct less bulky P(i–Pr)3 ligand (Fig. 13c) the H2 aligned along the P–M–P axis electronic interaction between H2 and changes the frequency by less than 20 and the structure with the H2 aligned the metal, barrier heights from the two percent, from 0.89 to 0.73 cm-1. along the OC–M–CO axis. The calcu- calculations may be added to arrive at The change that occurs when the cen- lation is rather extensive, and the bulky an estimate of the effective total bar- tral metal atom is replaced may be taken organophosphine ligands must be sim- rier. These assumptions are not unrea- as reflecting the metal-dihydrogen bond- plified to make it possible at all. When sonable, because replacement of tung- ing directly; that is, it is essentially an unsubstituted phosphine (PH3) is used sten by molybdenum has no effect on electronic effect. Replacing the PCy3 as a ligand instead of tricyclohexylphos- the results of the molecular-mechanics ligands with P(i–Pr)3 ligands, on the phine or tri-isopropylphosphine, the cal- case, whereas the ab initio theory uses other hand, probably has little effect culation yields a barrier height of 1.8 the very small PH3 ligands, rather than on the electronic state of the metal and kilocalories per mole for the tungsten P(i-Pr) 3 or PCy3 ligands, and thus es- therefore on the metal-dihydrogen bond- complex and 0.6 kilocalorie per mole sentially ignores steric effects. The sum ing. These ligands most likely produce for the molybdenum complex. of the two calculations for each of the a steric component of the barrier to H2 The second type of calculation—the complexes is shown in Table 1 along rotation through direct, nonbonded in- molecular-mechanics type—may be with the corresponding barrier height

Los Alamos Science Summer 1990 129 Neutrons and Catalysis

Juergen Eckert earned his B.S. at Yale Uni- calculated from the observed inelastic H 2 by inelastic neutron scattering can be neutron-scattering data. versity and his Ph.D. at Princeton University used as a probe of the details of metal- in 1975. Most of the research for his doc- The calculated and observed barriers to-H2 binding. toral thesis, which involved neutron-scattering studies of the lattice dynamics of solid neon. to H2 rotation appear. at first glance, Given the fact that the directional to agree only qualitatively. If, how- was carried out as a member of the neutron- properties of the electron wave functions scattering group in the Physics Department of ever. one takes into account the various that help optimize the electron flow be- Brookhavcn National Laboratory. After earn- limitations of the theoretical calcula- tween the dihydrogen ligand and the ing his doctorate, he remained at Brookhaven tions, the agreement with experiment until 1979, when he accepted a staff member- metal atom also seem to be largely re- ship atl Los Alamos to initiate neutron-scattermg is remarkably good. For example, the sponsible for the barrier to H2 rotation, research on the newly commissioned pulsed neu- necessary structural information is not we feel that establishment of a signif- tron source (WNR) at LAMPF. The focus of known in detail for all three complexes his work has steadily shifted towards applica- icant electronic component to the bar- tions of neutron scattering to chemistry, albeit in this experiment, and both types of from a physicist’s perspective. He recently calculations are normally used for bar- backbonding between the metal atom spent one year as a visiting scientist at the In- rier heights that are a factor of ten or so stitut Laue Langevin in in Grenoble, France. where and molecular hydrogen. some of the work described here was carried out. higher than the one in this study. Fur- The latter conclusion, in particular, is thermore. comparison with experimental a truly remarkable result of our neutron- data does suggest that the molecular- scattering studies and illustrates the very mechanics calculation overestimates the fundamental details to which these cat- steric part of the barrier, since replace- alytic model systems can be studied

ment of the PCy3 with P(i-Pr)3 is found with such techniques. Apart from our to change the barrier height by 0.8 kilo- model systems, many more realistic cat- calorie per mole. which is four times alytic materials are being investigated the experimentally observed change. In by the same techniques—studies that are view of these considerations, we can often greatly aided by previous work on clearly conclude that the direct elec- model compounds. These more realistic tronic binding of the dihydrogen ligand systems include molecules adsorbed on to the metal contributes significantly dispersed metal particles, inside cavities to the barrier, at least one-half to two- of zeolites, or attached to many other thirds of’ the experimentally determined active substrates. Although the level of value. detail that can safely be inferred from The rotational tunnel splitting is an the “real” catalytic systems is somewhat extremely sensitive measure of the bar- lower than for the simpler model sys- rier height—in fact, it depends exponen- tems, significant progress can nonethe- tially on the value of the barrier. Such less be expected in understanding the sensitivity has clear advantages. For catalytic function of these materials on example, the observation of a higher an atomic scale. Neutron scattering will barrier in the tungsten complex than in certainly play an important role in these the molybdenum complex is an indica- studies. ■ tion of stronger binding of the hydro- gen molecule to the metal atom. This conclusion can be reached both by ob- Acknowledgments

serving the higher M–H2 infrared stretch frequency in the tungsten complex or It is a great pleasure to thank Greg Kubas, the person who first demonstrated the existence of by observing the difference in rotational molecular-hydrogen complexes, for our ongoing, tunnel splitting. However. the change fruitful collaboration. Of the many others who in infrared stretch frequency is only 10 have made significant contributions to the work and the ideas discussed in this article, we would percent, whereas the change in rota- in particular like to mention Larry Dahl, Werner tional tunnel splitting is more than 50 Press, Alberto Albinati, Guiliano Longoni, Tom Koctzle, Oren Anderson, and Jeff Hay). percent. It is therefore clear that rota- tional tunneling spectroscopy of side-on

130 Los Alamos Science Summer 1990 Neutrons and Catalysis

Further Reading Pier Luigi Stanghellini and Guiliano Longoni. 1987. Vibrational studies of interstitial hydro- R. K. Thomas. 1982. Neutron scattering from gen in metal carbonyl clusters. Journal of the adsorbed systems. Progress in Solid State Chem- Chemical Society Dalton Transactions 685–690. istry 14: 1 –93. Jacques Roziere and Antoine Potier. 1982. Li- C. J. Wright. 1985. Surface characterization aison metal-hydrogene-metal et spectroscopic de by the inelastic scattering of neutrons from ab- vibration. Bulletin de la Societe Chimique de sorbates. In The Structure of Surfaces, edited by France. Partie I. 1-339–346. M. A. Van Hove and S. Y. Tong, pp. 210-218. Springer Series in Surface Science, volume 2. Werner Press. 1981. Single-Particle Rotations in Berlin: Springer-Verlag. Molecular Crystals. Springer Tracts in Modem Physics, volume 92. Berlin: Springer-Vcrlag. T. J. Udovic and R. D. Kelley. 1988. Neutron scattering studies of hydrogen In catalysts. In P. Jeffrey Hay. 1987. Ab initio theoretical Hydrogen Effects in Catalysis, edited by Z. Paal studies of dihydrogcn coordination vs. oxida- and P G. Menon, pp. 107-182 New York: Mar- tive addition of H2 to five-coordinate tungsten cel Dekker. complexes. Journal of the American Chemical Society 109:705-710. G. A. Somorjai. 1986. Surface science and catalysis. Philosophical Transactions of the Gregory J. Kubas, Robert R. Ryan, Basil I. Royal Society of London A318:81–100. Swanson, Phillip J. Vergamini, and Harvey J. Wasserman. 1984. Characterization of the first Norman Sheppard. 1988. Vibrational spectro- examples of isolable molecular-hydrogen com- scopic studies of the structure of species derived

from the chemisorption of hydrocarbons on metal Cy, i-Pr). Evidence for a side-on bonded H2 li- single-crystal surfaces. Annual Review of Physi- gand. Journal of the American Chemical Society cal Chemistry 39:589–644. 106:451-452.

E. L. Muetterties, T. N. Rhodin, Elliot Baud, C. George Edmund Bacon. 1977. Neutron Scatter- F. Brucker, and W. R. Pretzer. 1979. Clusters ing in Chemistry. London: Butterworths. and surfaces. Chemical Review 79: 91–137. R. E. Lechner and C. Riekel. 1983. Applications R. R. Cavanagh, J. J. Rush, R. D. Kelley, and T. of neutron scattering in chemistry. In Neutron J. Udovic. 1984. Adsorption and decomposition Scattering and Muon Spin Rotation. Springer Phillip J. Vergamini received his M.S. in in- of hydrocarbons on platinum black: Vibrational Tracts in Modern Physics. volume 1 () 1. Berlin: organic chemistry from the University of Min- modes from N15. Journal of Chemical Physics Springer-Verlag. nesota, Minneapolis, in 1968 after earning his 80: 3478–3484. B.S. at the University of Wisconsin, Superior. Robert Bau, editor. 1978. Transition Metal Hy- He completed his Ph.D. in inorganic chemistry R. Whyman. 1980. Metal clusters in catalysis. drides. Washington, D.C.: American Chemical at the University of Wisconsin, Madison, in the In Transition Metal Clusters, edited by Brian Society. fall of 1971 and then joined the Isotope and Nu- F. G. Johnson, pp. 545–606. New York: John clear Chemistry Division at Los Alamos. His Wiley and Sons. H. Ibach and D. L. Mills. 1982. Electron En- research there involved application of spectro- ergy loss Spectroscopy and Surface Vibrations. scopic and x-ray crystallographic techniques to Juergen Eckert, Gregory J. Kubas, John H. Hall, New York: Academic Press. studying the synthesis and structure of inor- P. Jeffrey Hay, and Caroline M. Boyle. 1990. ganic and organometallic compounds. In the Molecular hydrogen complexes. 6. The barrier spring of 1980, he joined the neutron-scattering J. Eckert. 1986. Neutron vibrational spec- group at Los Alamos and took responsibility for troscopy: The use of hydrogen as a structural the construction and use of the Single-Crystal (M = W, Mo: R = Cy, i-Pr): Inelastic neutron and dynamical probe. Physica B+C 136: 150- Diffractometer, one of the first two LANSCE scattering, theoretical, and molecular mechan- 155. instruments to be placed into the international ics studies. Journal of the Americal Chemical users program. With that diffractometer he has Society 112: 2324. studied the structures of materials ranging from molecular-hydrogen complexes to the crystalline Gregory J. Kubas. 1988. Molecular hydrogen mineral in dinosaur bones. sition metals. Accounts of Chemical Research 21: 120–128.

J. Eckert, A. Albinati, and G. Longoni. 1989. Inel-

astic neutron-scattering study of K[HCO6(CO)15]: Implications for the location of the hydride. In- organic Chemistry 28: 4055.

Los Alamos Science Summer 1990 131