J . Phys. Chem. 1988, 92, 809-812 knowledge does not warrant further speculation. An interesting analogy exists between HxMOS2 and the quantity HR discovered by Hall and M a s ~ o t hfor ~ ~reduced molybdenaalumina catalysts. As shown by Cirillo et al.,39 these catalysts take up hydrogen in an analogous way to that in HxMoS2. Indeed, if the uptake is calculated as HYMoO2,it is easy to deduce from the data of LoJacono and Hall40 that Y = 0.8. Interestingly, Engelhardt and c o - ~ o r k e r shave ~ ~ shown that HR noticeably affects catalytic activity for hydrogenation and for metathesis, ~~
(38) Hall, W. K.; Massoth, F. E. J. Catal. 1974, 34, 41. (39) Cirillo, A. C., Jr.; Dollish, F. R.; Hall, W. K.J. Cutul. 1980,62, 379. (40) LoJacono, M.; Hall, W. K. J. Colloid Interface Sci. 1977, 58, 76. (41) Engelhardt, J. J. Mol. Catal. 1980, 8, 119 and earlier references cited therein.
809
enhancing the former and virtually eliminating the latter. The present work affords the possibility of at last understanding in greater detail the chemical effects attributed to pretreatment procedures in terms of chemical effect and delineating experiments that could be done to advance our understanding of the catalytic properties of reduced or sulfided molybdena-alumina catalysts as well as the interesting MoS, systems.
Acknowledgment. We acknowledge the donors of the Petroleum Research Fund, administered by the American Chemical Society, for support of this research through a supplementary Summer Research Fellowship for Z.Y.A. W.K.H. thanks the National Science Foundation for support. Registry No. H2, 1333-74-0; MoS2, 1317-33-5.
Activation of Methane on Iron, Nickel, and Platinum Surfaces. A Molecular Orbital Study Alfred B. Anderson* and John J. Maloney Chemistry Department, Case Western Reserve University, Cleveland, Ohio 441 06 (Received: May 26, 1987;
In Final Form: August 17, 1987)
Surface metal atom insertion into methane CH bonds (oxidative addition) has been studied for Fe( loo), rough Fe( loo), Fe( 1 lo), Ni( loo), Ni( 1 1 l), and Pt( 111) surfaces by using ASED molecular orbital theory and cluster models. All transition states have in common CH stretches of 0.4-0.5 A and three-centered metal-carbon, metal-hydrogen, and carbon-hydrogen bonding derived from two methane u orbitals, the empty u*, and metal s + d band orbitals. For iron the predicted order of activity is rough Fe(100) > Fe(100) > Fe(ll0) and for nickel the order is Ni(100) > Ni(ll1). The higher barriers for the more closely packed surfaces are found to be related to the stronger metal bonds in these surfaces. The calculated activation energies on Ni( 100) and Ni( 111) are close to recent experimental determinations, and a low value obtained for Pt(111) suggests more experiments are needed.
The activation of alkane CH bonds is the first step in hydrogenolysis and oxidation catalysis. It is clear from simple consideration of metal-hydrogen bond strengths (5250 kJ/mol) and C H bond strengths (434 kJ/mol for breaking the first CH bond in methane) that oxidative addition and not hydrogen atom abstraction will be the route followed on metal surfaces. In contrast, 0-defect centers at oxide surfaces do abstract hydrogen atoms from methane, forming gas-phase methyl radica1s.l The greater strength of an O H bond over a CH bond allows this to happen; a molecular orbital analysis has been published recently., Though kinetic studies of alkane reactions on metal surfaces are large in number, little is known about catalyst surface composition and structure or the structure and electronic factors responsible for C H activation by metals. The purpose of the present work is to explore mechanisms, activation energies, and orbital interactions associated with the oxidative addition of a methane CH bond to several idealized clean transition-metal surfaces. The metals chosen for our theoretical exploration, iron, nickel, and platinum, have been the topics of several systematic methane catalysis and surface science studies in recent years. A theoretical study of the oxidative addition of methane to an iron surface indicated the reaction was exothermic with an activation energy barrier of roughly 88 kJ/moL3 In that study, low barriers were also found for dehydrogenation of CH, ( x = 1-3) fragments, and this is reflected in the ability of iron particles to catalyze the high-temperature pyrolysis of natural gas to graphite fibers as (1) Driscoll, D. J.; Lunsford, J. H. J. Phys. Chem. 1985, 89, 4415. (2) Mehandru, S.P.; Anderson, A. B.; Brazdil, J. F.; Grasselli, R. K. J . Phys. Chem. 1987, 91, 2930. (3) Anderson, A. B. J. Am. Chem. Soc. 1977, 99, 696.
0022-3654/88/2092-0809$01.50/0
studied recently by T i b b e t t ~ . ~Early work using iron surfaces was unsuccessful in yielding activation energie~,~ probably because contaminants lowered the activity. In the case of nickel, activation energies are higher on some single-crystal surfaces than on films and supported metal particles. Estimates are 88 kJ/mol on Ni(1 51738and 53 f 5 kJ/mols on Ni( 11 l), and 0,9 37,8910and 27 f 5 kJ/molI0 on Ni(100). On a Ni film 42 kJ/mol has been reported,5 and independent studies of methane activation by silica-supported nickel yield similar barriers, 29" and 2512 kJ/mol. Activation of n-alkanes on platinum has been studied.13J4 Activation energies are about 46 kJ/mol on Pt( 1 lO),I4 and the inactivity of Pt(l1 l)I3 has been interpreted to mean that n-alkane (4) Tibbetts, G. G. J. Cryst. Growth 1984,66, 632; J . Crysr. Growth 1985, 73, 431. (5) Wright, P. G.; Ashmore, P. G.;Kemball, C. Trans. Faraday SOC.1958, 54, 1692. (6) Schouten, F. C.; Kaleveld, E. W.; Bootsma, G. A. Surf. Sci. 1977, 63, 460
(7) Lee, M. B.; Yang, Q. Y.; Tang, S . L.; Ceyer, S . T. J. Chem. Phys. 1986,85, 1693. (8) Beebe, T.P., Jr.; Goodman, D. W.; Kay, B. D.; Yates, J. T., Jr. J . Chem. Phys., in press. (9) Schouten, F. C.; Gijzeman, 0. L. J.; Bootsma, G. A. Surf. Sci. 1979, "V
3
61, I .
(10) Hamza, A. V.;Madix, R. J. Surf. Sci. 1987, 179, 25. (1 1) Gaidai, N. A.; Babernich, L.; Guczi, L. Kinet. Katal. 1974, 15,974. (12) Kuijpers, E. G.M.; Jansen, J. W.; van Dillen, A. J.; Geus, J. W. J. Catal. 1981, 72, 75. (13) Firment, L. E.; Somorjai, G. A. J. Chem. Phys. 1977, 66, 2901. (14) Szuromi, P. D.; Engstron, J. R.; Weinberg, W. H. J . Phys. Chem. 1985, 89, 2491.
0 1988 American Chemical Society
810
The Journal of Physical Chemistry, Vol. 92, No. 3, 1988
Anderson and Maloney
TABLE I: Atomic Parameters Used in the Calculations: Principal Quantum Number, n, Ionization Potential, IP (ev), and Slater Orbital Exponents, € (au) P
S
atom
n
IP
Fe Ni Pt C
4 4 6
H
1
9.37 9.135 10.5 15.09 12.1
2
d
E
n
IP
E
n
IP
CI
E1
c2
1.7
4 4 6 2
6.97 5.94 6.46 9.76
1.4 1.5 2.254 1.618
3 3 5
10.5 11.5 11.1
0.5366 0.5683 0.5629
5.35 5.75 6.013
0.6678 0.6292 0.6461
1.8 2.554 1.658 1.2
C H activation energies must be greater than 67 kJ/mol.I4 Because of various experimental difficulties in the early work, only the recently determined activation energies, 51 and 53 kJ/mol for Ni( 111) and 37 and 27 kJ/mol for Ni( loo), are likely to be a c c ~ r a t e . ~The ~ ~other ~ J ~ values are in greater doubt. Theoretical Method The atom superposition and electron delocalization molecular orbital (ASED-MO) method15used in this paper is a semiempirical theoretical approach based on partitioning molecular charge density distribution functions (pmol) into atomic (pa) and nonperfectly-following (pnpf)components. For a diatomic (a-b) example Pmol
= Pa
+ Pb + Pnpf
2.396
0-0-0 0-0-0 Fe(100)
Fe/Fe(lOO)
(1)
Since pa and pb are rigid distributions centered on nuclei a and b, pnpfis flexible; as the atoms w m e together to form the molecule, it is a positive function in the region between the nuclei, corresponding to the bond charge, and it is negative near the nuclei, indicating depletion. From the electrostatic theorem there are two nonzero components to the force on nucleus a in the molecule, an attractive force due to pnpfand a repulsive force due to Pb and nucleus b. Integrating these forces yields the potential energy curve E ( R ) where R is the internuclear distance:
In (2) E,(R) is the repulsive energy due to atom superposition and Enpf(R)is the attractive energy due to electron delocalization. The E,(R) function is determined with Slater atomic orbitals from the literature.16 Enpfis approximated as the difference between the valence electron orbital energies for the atoms calculated from measured first ionization potentials” and the molecular orbital energies of the valence electrons in the molecule. The molecular orbital energies are functions of the atomic ionization potentials, Slater orbital exponents, and molecular structure as calculated by using a modified extended Hiickel Hamiltonian. Atomic ionization potentials are sometimes changed in heteronuclear molecules so that the calculations produce reasonable bond ionicities, and sometimes exponents are changed to yield resonable bond lengths. When the orbital energy difference approximation is used for Enpf,the charge density function of the more electronegative atoms is usually used in calculating each pairwise repulsion energy for nonhydrides. Thus in this study E , were calculated by using the carbon atom density for the Fe and Ni surfaces, but for the Pt surface the Pt density was used because of improved predictions for diatomic PtC. For metal-hydrogen bonds the metal density was used in all cases because of improved diatomic hydride predictions. The metal occupations were assumed to be dn+lsl,which corresponds most nearly to the bulk band occupations. Parameters used in this study are in Table I. The ASED-MO method is a simple way to predict certain molecular data from atomic data. Its structure and energy predictions are often quite accurate, but in general it is best used to (15) Anderson, A. B. J . Chem. Phys. 1975, 62, 1187. (16) Clementi, E.; Raimondi, D. L. J . Chem. Phys. 1963, 38, 2868. Richardson, J. W.; N i e u w p r t , W. C.; Powell, R.R.; Edgell, W. F. J. Chem. Phys. 1962, 36, 1057. Basch, H . ; Gray, H. B. Theoret. Chim. Acra 1966, 4 , 361. (17) Lotz, W. J . Opt. Soc. Am. 1970, 60, 206. Moore, C. E. Atomic Energy Leuels; National Bureau of Standards Circular ( U S . ) 467; US.
Government Printing Office: Washington, DC, 1958.
E2
1.8 2.0
Ni(100)
Ni(ll1)
Pt(ll1) Figure 1. Cluster models used for the calculations. Shaded atoms a r e those associated with methane in Figure 2.
establish and explain chemical trends. Surfaces were modeled by using the clusters in Figure 1. Atoms with which methane interacts directly in the transition state are shaded. Adsorption studies were performed assuming higb-spin molecular orbital occupations with lower levels in the d band doubly occupied and upper levels singly occupied. The Fe,3 and Fe14 clusters had 38 unpaired electrons, the Fell cluster 30, the Nil, 4, and the Nilo and Ptlo clusters 6. The cluster structures are bulk superimposable and based on well-known lattice constants.I8 By use of transition-state structures from the Fe13,Nil,, Nile, and Ptlo clusters, activation energies were recalculated by using Fe18,Nils, NiI5,and Pt15 clusters with, respectively, 52, 6, 6, and 8 unpaired electrons. Activation energies changed by less than 3 kJ/mol for all but Ni(100), where a decrease of 6 kJ/mol was found. These clusters increased the coordination of the shaded atoms in Figure 1. Our calculated H3C-H bond length and strength are 1.21 8, (1.10) and 531 kJ/mol (434), where the experimental values are given in parentheses. In finding transition states, certain parameters (Table 11) were varied and others are derived from them. The methyl group was kept rigid and tilted from normal in steps of 5’ or less. The C height and position on the x axis (origin on the nearest metal atom) were vaired in steps of 0.05 8, or less and the CH bond bending from the tetrahedral direction was varied in steps of 2O or less. (1 8) ASM Metals Reference Book, 2nd ed.; American Society for Metals: Metals Park, OH, 1984.
The Journal of Physical Chemistry, Vol. 92, No. 3, 1988 811
Activation of CHI on Fe, Ni, and Pt Surfaces
TABLE II: Calculated Transition-State Properties and Metal Bond Orders“
surface Fe( 100) Fe/Fe(100) Fe(ll0) (A) Fe(ll0) (B) Ni(l11) Ni(100) Pt(ll1)
hc, 8,
2.25 2.18 2.35
2.28 2.15
CH3 tilt, deg 25
CH bend, deg
-0.57 -0.05 -0.30
37 12
22 34
16
35
34 27 30
xc,
A
0.00
-0.10
12
2.18
0.08
17
2.30
0.00
15
22
CM bend, deg 25 22 11 8 12
19 15
M-H, 8, 1.65 1.5 1.66 1.58 1.57 I .57 1.63
M-C, 8, 2.25 2.25 2.35 2.30 2.15 2.18 2.30
M-M, overlap 2.20 2.10 2.90 2.90 1.66 1.36 2.10
9cw
0.09
0.14 0.01 -0.04 0.02 0.17 0.18
ACH, 8, 0.36 0.44 0.46 0.54 0.51 0.40 0.36
E,,
kJ/mol 32 21 118 135
64 29 43
“Structure parameters are with reference to Figure 2. hc is the distance between C and the surface atom nuclear plane, and xc is the displacement left or right. The CH3 tilt is from the vertical, and bends are from the tetrahedral direction. The C-M bend and M-H and M-C distances are derived; other structure variables were optimized. The M-M, overlap includes next-neighbor contributions and is calculated in the Mulliken definition. qCH is the net charge of CH,, ACH is the CH bond stretch, and E, is the activation energy.
Fe(100)
Fe/Fe (100) -8
t
s
-9 -10-111
W
Ni(ll1)
Ni(100) -1 5
-1 6
Figure 2. Calculated transition-state structures.
The CH bond was stretched in steps of 0.05 %, or less. Given the large number of variables, there is a possibility that the lowest transition state was not always found, but we believe the calculated activation energies have uncertainties of less than 1 kJ/mol. Methane Activation on Fe( 100) and F e ( l l 0 ) Surfaces Key numerical results are in Table 11, and transition-state structures are in Figure 2. The calculated activation energy for inserting an Fe(100) surface atom into a methane C H bond is 32 kJ/mol and occurs when the bond is stretched 0.36 A. On a “roughened” surface site, consisting of an Fe adatom placed in a bulklike position, the activation energy decreases slightly to 27 kJ/mol. The bonding in the transition state is best characterized as CH donation to the surface, and the electronic structure for the adatom case is in Figure 3. This figure is representative of all other transition states discussed in this work. The stretching causes one of the 3-fold degenerate t symmetry methane orbitals to become destabilized. Its bending of 22’ away from the tetrahedral direction contributes further to the destabilization and also leads to a small stabilization in one of the other t orbitals. The low-lying a orbital is destabilized by the distortion. Interactions with the surface consist in a small stabilization of the a orbital and mixing of both of the rehybridized orbitals from the t set with the Fe surface orbitals to form clearly defind C-.H--Fe and C-.Fe m bonding orbitals. The main occupied antibonding counterpart orbital energy level lies in the half-filled d band region (and in the doubly occupied region for Ni and Pt); participation of the CH u* orbital in it removes almost all H contribution so that this orbital has a C.-Fe u bonding character. There are additional higher lying unoccupied C-M antibonding orbitals which are not shown in Figure 3. The net CH4 charge is 0.14 in the transition state, indicating a net donation to the surface. The transition-state structures given in Figure 2 show how the methyl groups tilt with respect to the two Fe( 100) surfaces. The activated CH bonds are bent away from the tetrahedral directions, as are the newly forming metal-carbon bonds, allowing threecentered bonding. On these surfaces the deviations from the
Figure 3. Orbital interactions between CH, and the adatom on Fe(100). Energy levels of distorted methane with the cluster removed are in the CH,* column. Orbitals in the shaded metal band region are half-filled. Empty s and p band orbitals lie above them.
tetrahedral directions are nearly symmetric; numerical values for structure parameters are in Table 11. The close-packed Fe( 110) surface is much less reactive. The activation energy for site A is 118 kJ/mol, and for site B it is 135 k J / m o l s e e Figure 2 and Table 11. The methyl tilts from vertical are less than for the (100) surface, and this is symptomatic of increased steric repulsions with the close-packed surface. The bending of the activated C H bond away from the tetrahedral direction is greater than on Fe(100), and the F e C bonds are closer to the tetrahedral direction despite the greater Fe-C distances. From Table I1 it can be deduced that the activity of a surface Fe atom toward the C H bond is inversely related to its bond order with its neighboring Fe atoms. The bond order, in turn, varies directly with the Fe surface packing density. Methane Activation on N i ( l l l ) , Ni(100), and P t ( l l 1 ) Surfaces Ni( 1 11) is close-packed, and the transition-state structures are similar to those for Fe(1 IO). The calculated activation energy, 64 kJ/mol, is close to 51 and 53 f 5 kJ/mol reported in ref 7 and 8. On the more open (100) surface of Ni we calculate an activation energy of 29 kJ/mol. This is as expected, being substantially less than for the close-packed surface. It is close to the experimental estimates of 37 and 27 kJ/mol from ref 8 and 10. The 43 kJ/mol activation energy calculated for Pt( 1 11) is close to the experimentally determined value of about 46 kJ/mol for the (1 10) surface14 and is smaller than implied by early studies of n-alkanes on Pt(l1 l ) . I 3 We do not know if the disagreement is caused by the approximations of the theory or uncertainty in the experimental measurements, surface composition, or structure.
812
J . Phys. Chem. 1988, 92, 812-818 These activities correlate with the inverse of the bond order between the activating surface atom and its neighbors. The same is true for the two Ni surfaces, indicating a contribution to the barrier comes from a weakening of metal bonding at the transition state. Hydrogen atoms have been noted to weaken iron bonding in clusters while carbon atoms strengthened iron overlap^.^ It is obvious that adsorption should affect metal bonding. The relative activation of close-packed and open surfaces toward methane could also have been anticipated. Our calculated activation energies for (1 11) and (100) Ni agree within about 20% with the recent experimental determinations. For pt( 111) we think the activation energy could be less than the experimental estimate. No experimental values exist for (100) and (1 10) Fe, but our results strongly suggest that the close-packed (1 10) surface will be much less reactive.
Conclusions The oxidative addition of methane to the iron, nickel, and platinum surfaces considered here is characterized by the insertion of a surface metal atom into a CH bond. Transition-state C H bond stretches amount to around 0.4-0.5 A. In the transition state two methane C H u orbitals hybridize with the metal s and d band orbitals to form metal-H and m e t a l 4 bonds and the antibonding counterpart to these u donation interactions is stabilized by mixing with the empty C H u* orbital to give additional C-metal bond order. Our finding of charge donation to the metal surfaces in most of the transition states conflicts with the conclusions of Saillard and Hoffmann,lg who used stylized structure models and extended Huckel calculations but never actually studied properties along reaction paths for activating methane. We have found that the close-packed iron(ll0) surface is a much weaker CH activator than the more open (100) surface and that an adatom on the (100) surface is the most active site of all.
Acknowledgment. We are grateful to the Gas Research Institute for granting support for this work. Registry No. Fe, 7439-89-6; Ni, 7440-02-0; Pt, 7440-06-4; CHI,
(19) Saillard, J.-Y.; Hoffmann, R. J . A m . Chem. SOC.1984, 106, 2006.
74-82-8.
Interaction of Pt with Ceria and Silica Y. Zhou, M. Nakashima,+and J. M. White* Department of Chemistry, University of Texas, Austin, Texas 78712 (Received: May 28, 1987; In Final Form: August 10, 1987)
The interaction of thin-film Pt with oxidized Ce foils, high surface area ceria powder, and thin films of silica has been studied by XPS, AES, and ISS. Vapor deposition of Pt on oxidized cerium led to some chemical interaction with Ce3+ at room temperature. On the basis of electron and ion spectroscopy results, we propose that there is migration of Pt into the ceria and partial reduction of the latter. When the Pt-covered sample was annealed above 750 K, these diffusion and reduction processes became more significant. On fully oxidized ceria powder, Pt maintained its metallic properties and its presence did not promote the partial reduction of ceria even when the sample was annealed at 800 K under vacuum. On a silica thin film, there is no evidence for any strong chemical interactions with Pt. All the data on silica can be understood in terms of the conversion of Pt thin films to particles upon heating to 850 K.
Introduction The strong interaction of noble metals with reducible oxides, particularly Ti02, has been widely studied. The most widely accepted mechanisms of interaction include (1) the encapsulation of noble metals by reduced oxide species from the support and (2) the electronic coupling (including chemical bonding) between the support and Ceria is technically important reducible oxide which has received relatively little attention in the literature. It is known as an important additive in automobile emission control catalysts containing noble metals as the active components. Ceria stabilizes the active precious metals in a finely dispersed state8 and acts as an oxygen storage agent.9 On ceria- and aluminasupported Pt catalysts, Summers and Ausen8 found that thermal aging of a catalyst in air decreased the apparent dispersion of Pt at low Ce loading. Just the reverse occurred at high Ce loading (>4.4 wt %). Unlike air, aging in H, increased the apparent Pt dispersion for low Ce loadings. Furthermore, aging in H2 at 1173 K leads to a reaction between Pt and ceria. Using XRD they identified Pt5Ce as a product. As an oxygen storage agent, Yao et aL9 found that the presence of noble metals lowered the reduction temperature of one kind of surface oxveen anion a t 773 K . They conclude that the precious metal and ceria cooperate to enhance the oxygen storage capacity. Even though there are several studies of noble metals on ceno surface characterization work has been done. Since ‘Permanent address: Japan Atomic Energy Research Institute, Tokaimura, Ibaraki-ken 319-1 1 , Japan.
0022-3654/88/2092-0812$01.50/0
Ce is a rare-earth metal, it and its oxides have properties reflecting 4f valence electrons. In our laboratory, we have investigated, using surface science tools, the oxidation of cerium metal using water and ~ x y g e n . ’This ~ ~ work ~ ~ forms a basis for interpreting the XPS measurements reported here. We have also investigated platinized (1) (a) Belton, D. N.; Sun, Y. M.; White, J. M. J . Phys. Chem. 1984,88, 1690. (b) Belton, D. N.; Sun, Y. M.; White, J. M. J . Am. Chem. SOC.1984, 106, 3059. (c) Belton, D. N.; Sun, Y. M.; White, J. M. J . Phys. Chem. 1984, 88, 5172. (2) Sadeghi, H. R.; Henrich, V. E. J . Cafal. 1984, 87, 279. (3) Bahl, M. K.; Tsai, S . C.; Chung, Y. W. Phys. Reu. B 1980, B21, 1344. (4) Kao, C. C.; Tsai, S.C.; Chung, Y. W. J . Coral. 1982, 73, 136. (5) Kao, C. C.; Tsai, S . C.; Bahl, M. K.; Chung, Y. W. Surf.Sci. 1980, 95, 1.
(6) Ocal, C.; Ferrer, S . J . Chem. Phys. 1986, 84, 1. (7) Beard, B. C.; Ross, P. N. J . Phys. Chem. 1986, 90, 6811. (8) Summers, J . C.; Ausen, S . A. J . Card. 1979, 58, 131. (9) Yao, H. C.; Yao, Y. F. Y . J . Catal. 1984, 86, 254. (IO) Foger, K.; Anderson, J. R. Appl. Coral., in press. (1 1) Sudhakar, C.; Vannice, M. A. J . Catal. 1985,95, 227. Sudhakar, C.; Vannice, M. A. Appl. Cafal. 1985, 14, 47. (12) Su,E. C.; Montreuil, C. N.; Rothschild, W. G. Appl. Curd. 1985, 17, 47. (13) Mitchell, M. D.; Vannice, M. A. Ind. Eng. Chem. Fundam. 1984,23, 88.
(14) Jin, T.; Okahara, T.; Mains, G. J.; White, J. M. J . Phys. Chem. 1987, 91, 3310. ( 1 5 ) Jin, T.; Zhou, Y.; Mains, G. J.; White, J. M. J . Phys. Chem. 1987, 91, 5931. (16) Praline, G.;Koel, B. E.; Lee, H. I.; White, J . M. J . Electron Spectrosc. Relut. Phenom. 1980, 21, 17. ( 1 7 ) Koel, B. E.; Praline, G.; Lee, I.; White, J. M.; Hance, R. L. J . Electron Spectrosc. Relat. Phenom. 1980, 21, 3 1
0 1988 American Chemical Society
