ESR Spectra of Matrix-Isolated LiOp - American Chemical Society

Research Office (DAAG-2984K0086), and the donors of the. Petroleum Research Fund, administered by the American. Chemical Society. We also acknowledge ...
0 downloads 0 Views 506KB Size
6158

J. Phys. Chem. 1987, 91, 6158-6161

tunately, to date, no experimental data is available for comparison with these predictions. Our study of the H-(H2) complex is also important from the perspective of negative ion-molecule interactions. Contrary to common assumption, it was shown that electron correlation effects are of essential importance; in particular, they shorten the H2-Hdistance by ca. 0.4 A. They also yield ca. 40% of the interaction energy in the region of the potential. These results are analogous to our recent results on the H-(H20) complex.28 Furthermore, both the H-(H,) and H-(H20) studies point out the importance of using extended basis sets and the C P method to remove BSSE if quantitative results are desired.

Acknowledgment. We acknowledge the financial support of the National Science Foundation (CHE-8511307), the U S . Army Research Office (DAAG-2984K0086), and the donors of the Petroleum Research Fund, administered by the American Chemical Society. We also acknowledge the Harris Corp. for their generous computer system grant and the National Science Foundation for its San Diego Supercomputer time award. We also thank the San Diego Supercomputer Center for their computer time grant. G.C. also thanks the Polish Academy of Sciences within Program CPBOl. 12 for their partial support. We are grateful to N. Adams for providing us with his basis set for the hydrogen atom, to Dr. Dick Hilderbrandt for his help with the

Cray and SCS-40 versions of GAUSSIAN 82, and to Dr. Harvey Michels for his helpful suggestion concerning our manuscript. We are also indebted to Prof. F. W. Cagle for helpful discussion of the thermodynamic aspects of this problem. Appendix A. Cubic-Spline Fits to the MP4 Surface Both the “stretching“ and “bending” potentials were interpolated by using a clamped cubic-spline function. The interpolant in the interval xJ Ix xJ+,,SJ(x)is given by the equation

SJ(x)= aJ + bJ(x - x,)

+ cJ(x - x,)~+ dJ(x- xJ)3

Table VI1 contains the “stretching” potential interpolant where each xJ value is the R value (as defined in the main text) for each geometry of the MP4 surface. Table VI11 contains the “bending” potential interpolant. In Table VI11 each xJ value is the angle between the HH bond and the HH- bond. Since S has units of millihartrees, aJ does also, and b, has units of millihartrees per angstrom or millihartrees per degree, etc. These calculations correspond to the complex with a fixed r distance of 0.739 8, and a fixed distance between the middle H and the H- of 2.8355 8, (this is not R as defined in the main text) at various angles. Therefore, an xJ value of 180’ corresponds to the linear geometry of the H-(H,) complex. Registry No. H-, 12184-88-2; H2, 1333-74-0.

ESR Spectra of Matrix-Isolated LiOp D. M. Lindsay* and D. A. Garland Department of Chemistry, City University of New York, The City College, New York, New York 10031 (Received: May 7, 1987)

ESR spectra assigned to 6Li02and 7Li02molecules have been produced by codepositing atomic Li plus O2in Ar, Kr, and N2 matrices. For 7Li02in N2 (which gave the best spectral resolution) g,, = 2.0077 (2), gyy= 2.0014 (l), and g,, = 2.0677 (1) with A,, = Zt0.66 (6) G, A, = i2.67 (3) G, and A,, = A2.24 ( 5 ) G, where x is the C2 axis and y is perpendicular to the LiO, plane. Point spin calculations imply a ’A2 ground state for LiOz with isotropic and dipolar hyperfine constants: a = -1.42 G and T,, = +2.08 G, Tyy= -1.25 G, T,, = -0.82 G. The ESR data show a complete charge transfer from Li to O2and predict that the first excited 2Bl state lies at 5100 Zt 400 cm-l, in good agreement with the results of molecular structure calculations

introduction Except for LiO,, all the alkali metal superoxides occur as crystalline solids,’ readily prepared by reacting the metal with ~ x y g e n . ~Accordingly, -~ the gas-phase reaction of alkali metal atoms with O2has received considerable attention, dating back to the early flame-diffusion studies of Bawn and Evanss More recent ~ i n g l eand ~ , ~multiple* collision studies show that the reaction most likely proceeds via a “harpooning mechanism”, involving a complete transfer of the metal valence electron to 02.The superoxide anion has also been studied in alkali metal halide crystalsg and is often used as a probe of catalytic materials.’O The toxic (1) The existence of the superoxide anion (OF) in these compounds was first suggested by Pauling. See: Pauling, L. J. Am. Chem. SOC.1931, 53, 3225. Pauling, L. The Nature of the Chemical Bond; Cornell University Press: Ithaca, NY, 1960. (2) Vannerberg, N.-G. Prog. Inorg. Chem. 1962.4, 125. (3) Vol’nov, I. I. Peroxides, Superoxides and Ozonides of Alkali and Alkaline Earth Metals; Plenum: New York, 1966. (4) Hart, W. A.; Beumel, 0.F.;Whaley, T. P. The Chemistry of Lithium, Sodium, Potassium, Rubidium, Cesium and Francium; Pergamon: New York, 1973. (5) Bawn, C. E. H.; Evans, A. G.Trans. Faraday Soc. 1937, 33, 1580. (6) Lacmann, K.; Herschbach, D. R. Chem. Phys. Lett. 1970, 6, 106. (7) Mochizuki, T.; Lacmann, K. J . Chem. Phys. 1976,65, 3257. (8) Kramer, S. D.; Lehmann, B. E.; Hurst, G. S.; Payne, M. G.;Young, J. P. J. Chem. Phys. 1982, 76, 3614. (9) For example: Zeller, H. R.; Kanzig, W. Helu. Phys. Acta 1967, 40, 845. Shuey, R. T.; Kanzig, W. Helu. Phys. Acta 1967, 40, 873.

0022-3654/87 12091-6158$0 1S O ,I O I

,

effect of 0,-in biological systems is well-established,l’ and metal-oxygen bonding is of central importance to oxygen transport in the blood.I2 While the chemistry and biochemistry of metal superoxides have been actively pursued, the physical properties of these compounds have received relatively less attention. This situation may well change dramatically with the recent discovery of metal oxides exhibiting high-temperature superconductivity. In this context it is interesting to note that electron transfer between 02-moieties is thought to play a role in stabilizing the crystalline alkali metal superoxides and that solid KO2 is a semicond~ctor.’~ In this paper we present electron spin resonance (ESR) spectra for individual LiOz molecules isolated in Ar, Kr, and N2matrices. The superoxide species were produced by codepositing atomic lithium with excess O2and give spectra characteristic of a true ion pair, Li+02-. Computer synthesis of 7Li02spectra in nitrogen matrices allowed an accurate determination of all three 7Li hyperfine (hf) constants. A comparison of the measured hf with that predicted from a point spin model implies that the radical (10) For example: Wang, J.-X.; Lunsford, J. H . J . Phys. Chem. 1986, 90, 3890. (11) Sawyer, D. T.; Valentine, J. S. Acc. Chem. Res. 1981, 14, 393. Fridovich, I. Acc. Chem. Res. 1972, 10, 321. (12) Mingos, D. M. P. Nature (London), Phys. Sci. 1971, 230, 154. Collman. J. P. Acc. Chem. Res. 1977, 10, 265. (13) Khan, A. U.; Mahanti, S. D. J . Chem. Phys. 1975, 63, 2271.

0 1987 American Chemical Society

The Journal of Physical Chemistry, Vol. 91, No. 24, 1987 6159

ESR Spectra of Matrix-Isolated LiO,

TABLE I: Magnetic Parameters for 'LiO, and 6LiOz in N,, Kr, and Ar Matrices" matrix

N2 Kr Ar

isotope 7 6 7 6 7 6

g1

2.0677 2.0679 2.0577 2.0581

(1) (2) (1) (1)

2.0592 (1)

g2 2.0077 2.0084 2.0095 2.0092 2.0084 2.0083

g3

(2) (3) (2)

(I) (1) (1)

2.0014 2.0014 2.0024 2.0027 2.0017 2.0022

(I) (2) (2) (2) (1) (2)

AI 2.24 (5) 0.85 2.3 ( I )

A2 0.66 (6) 0.25

'43 2.67 (3) 1.01 3.0 ( I ) 3.1 (1)

"Experimental uncertainties (see text) in parentheses. Units of A are gauss. 6 L i 0 2hf constants are 'LiO, data divided by the ratio of nuclear g values, g7/g6= 2.641 from ref 21.

has a ,A2 ground state and a geometry close to that found by ab initio calculation^.^^ Experimental g values are compared with dataIsJ6 previously reported for N a 0 2 CsO, and are discussed in the context of a crystal field model for the alkali metal superoxides.

-

Experimental Section Lithium superoxide molecules were prepared by codepositing atomic lithium, either naturally abundant (Alfa, 99.9%) or isotopically enriched (Oak Ridge National Laboratory: 98% 6Li and 99.9% 7Li), with oxygen (Matheson, >99.99%) and excess argon, krypton, or nitrogen (Airco, 99.9995%). Superoxide samples were formed on a liquid helium cooled sapphire or copper substrate, as described elsewhere.17J8 The lithium was evaporated from a Knudsen effusion source. The metal flux was periodically monitored with a quartz crystal microbalance (Veeco, QM-301). Typically, the ratio of matrix gas:oxygen:lithium was 50051, with about 0.1 pmol/h of lithium incident on the substrate and total deposition times of 5-6 h. In some experiments, the 0,:Li ratio was increased to about 20:l in order to minimize the ESR intensities of atomic nitrogen and lithium, which overlap the superoxide transitions in certain regions. The superoxide intensity was approximately doubled by a 3-h photolysis with a 150-W Xe lamp. Annealing had little beneficial effect on the superoxide spectra. Spectra definitively attributable to L i 0 4 were observed only in N 2 matrices at T = 27 K. Matrix temperatures were measured and controlled with a Scientific Instruments Model 5500 temperature controller and a GaAs sensor. Most spectra were recorded on an IBM-Bruker E R 200D ESR spectrometer (microwave power =1 mW) using 100-kHz magnetic field modulation (peak-to-peak amplitude = l G) and phasesensitive detection. Typical scan rates were 0.1 G/s with a time constant of 0.1 s. ESR spectra could also be processed by averaging repetitive scans on a multichannel analyzer (Tracor Northern, TN 1710). Some analog scans were later digitized for ease of comparison with simulated spectra. The resonance field of the cavity plus substrate and matrix sample was periodically measured with a microwave frequency counter (HP 5245L plus HP 5255A plug-in). Individual spectra were calibrated with a proton magnetometer (Micronow, Model 515). Relative and absolute field positions are judged accurate to f0.2% and f0.5 G , respectively. Spectra and Analysis Figure l a shows the ESR spectrum observed for 7Li02in a nitrogen matrix. The spectrum, similar to those observed for N a 0 , Cs02,15316319 may be described by an orthorhombic g tensor whose principal values are conventionally labeled g , > g2 > g , in Figure 1. Each g feature is split into quartets by a small hf -+

(14) O'Neil, S.V.; Schaefer, H. F.; Bender, C. F. J . Chem. Phys. 1973,

!a1 OBSERVED SPECTRUM

/ 9 A

l32CC G

I3210

,

13300G

,

1

13320

Figure 1. (a) Observed and (b) simulated ESR spectra for 'Li02 in an N2 matrix. Stick spectra pertain to calculated transition fields. The resonance field for a free electron is He = 3312.2 G.

interaction with the single ( I = ,/,, ml = f3/,) 'Li nucleus. ESR transitions were analyzed with the high-field expression ( i = 1, 2, or 3)2032' ge

Hi = -(He - AimI) gi

(1)

where g, = 2.0023 and He are respectively the g value and resonance field for a free electron and A , are the principal values of the orthorhombic hf tensor. As discussed below, the principal values of g and A are coincident and correspond to the molecular symmetry axes of LiO,. Owing to the smallz1nuclear quadrupole moments of both 6Li and 7Li, no measurable nuclear quadrupole interaction is to be expected for either superoxide. Figure 1b shows a simulated, powder ESR spectrum for 7Li02. The simulation algorithm, based upon an elliptic integral approximation, is discussed elsewhere.1*~z2The line shape function was a Lorentzian derivative with a peak-to-peak width, AHpp,of =1.2 G. Table I gives the g, and A , used in the simulation. Estimated errors (given in parentheses) represent the range over which the parameters could be varied without obtaining a noticeably poorer match to the observed spectrum. Even though the hf splitting about gzis not resolved, the A , constant could be extracted with a high degree of certainty. A slight discrepancy between observed and simulated features in the g3 region arises at least in part from a weak, underlying spectrum of atomic n i t r ~ g e n . ~ Characteristic , alkali metal tetroxide spectraz4were observed in one experimental run with 'Li in Nz. These give g, = 2.0410 (4) with A I = 1.86 ( 6 ) a n d g2 = 2.0181. F i g u r e 2

59, 3608.

(15) Adrian, F. J.; Cochran, E. L.; Bowers, V. A. J . Chem. Phys. 1973, 59, 56.

(16) Lindsay, D. M.; Herschbach, D. R.; Kwiram, A. L. Chem. Phys. Lett. 1974, 25, 175.

(17) Lindsay, D. M.; Herschbach, D. R.; Kwiram, A. L. Mol. Phys. 1976, 32, 1199.

(18) Kernisant, K.; Thompson, G . A,; Lindsay, D. M. J . Chem. Phys.

1985,82, 4739.

(19) Lindsay, D. M. Ph.D. Thesis, Harvard University, 1974. Garland, D. A. Ph.D. Thesis, City University of New York, 1986.

(20) Abragam, A,; Bleaney, B. Electron Paramagnetic Resonance of Transition Metal Ions;Oxford University Press: London, 1970. (21) Weltner, W. Magnetic Atoms and Molecules; Van Nostrand: New York, 1983. (22) Lindsay, D. M.; Thompson, G . A,; Wang, Y . J . Phys. Chem. 1987, 91, 2630. (23) Lindsay, D. M. J . Chem. Phys. 1984, 81, 3356. (24) Lindsay, D. M.; Herschbach, D. R.; Kwiram, A. L. J . Phys. Chem. 1983, 87, 21 13.

6160 The Journal of Physical Chemistry, Vol. 91, No. 24, 1987 f a J O B S E R V E D SPECTRUM

Lindsay and Garland TABLE 11: Comparison of Experimental 8 and A (Figufe 3) for LiO, in Several MatricesQ N2 Kr Ar

0.0655 0.0556 0.0569

0.0058 0.0071 0.0061

4580 5400 5270

43700 38100 43400

0.0009 0.0003 0.0004

0.0011 0.0008 0.0008

"Experimental g shifts from Table I; 6 and A from eq 2.

TABLE 111: Comparison of Experimental ('Li02 in N2)Dipolar Hyperfine Constants with Those Calculated from a Point Soin Model'

( b ) SIMULATED SPECTRUM

~~

A

exptl 2A2

2B,

2AI/

;3200

G

,

Txx

TYY

TZ,

Tx,

-1.42 (3)

+2.08 (7) +2.19 +3.62 +0.23

-1.25 (4) -1.22 -3.02 -0.89

-0.82 (6) -0.97 -0.63 +0.66

0.00 0.00 0.00 -1.53

'Units are gauss. 2A2 and 2B, geometries from Table I of ref 14. 2A" geometry from ref 2 8 . T,, and Tyy calculated for 2B,,should be compared with experimental T,, = -1.25 and Tyy = +2.08.

I

13350

Figure 2. (a) Observed and (b) simulated E S R spectra for 6Li02 in an N2 matrix. He= 3322.3 G. (a)

a

Geometry and Axis System

Yt

(b) Molecular Orbital Scheme

Discussion Figure 3a shows the axis system and e ~ p e c t e d ~ " ' ~geometry -~~~* (C, point group) for Li02 The schematic molecular orbital (MO) scheme, shown in Figure 1b, pertains to an ionic Li+Oz- superoxide molecule with ground-state symmetry 2Az. As discussed below, the small lithium hf structure indicates a near complete electron transfer from the metal to the oxygen molecule. Accordingly, L i 0 2 should be viewed as an 02-anion perturbed by the electrostatic interaction of the nearby Li' cation which lifts the degeneracy of the superoxide r* (and r) orbitals. Since the cation interacts most strongly with the in-plane r,* component, this orbital is stabilized relative to ry*and the unpaired electron will occupy the out-of-plane ry*orbital leading to a 2Azground state. For this situation, the superoxide g shifts, Agff= g,, - g,, are given bY3I 32

Ag,, = -X2/62

(2)

Agzz= 2X/6

C

---+--+-

IO(a,))

Figure 3. (a) Geometry and axis system and (b) schematic molecular orbital scheme for Li02.

compares observed and simulated spectra for 6Li02in a nitrogen matrix. The line width function is also Lorentzian, but with AHpp = 2.9 G. In simulating 6Li02,the gi were varied to give the best agreement with the observed spectrum, but the hf constants were set equal to the corresponding 7Li values divided by the ratio of nuclear g factors, g7/& = 2.641.2' The best fit gi (given in Table I) are in excellent agreement with those determined from the 7Li02 spectra. A l s o given in Table I are parameters for L i O , in Kr and A r matrices. Lithium superoxide spectra were noticeably poorer in Kr and Ar, when compared to the results obtained in nitrogen matrices. This is due in part to overlap with atomic ESR transitions, which obscure the g, and g3regions of 7Li02and 6Li0,, respectively. In addition, however, the spectral resolution is significantly better in nitrogen. Computer-simulated line shapes and intensities for the g, region of 7Li02in Ar and Kr were in relatively poor agreement with the observed profiles. Nevertheless, unambiguous assignments could be made by comparing the ESR spectra of the two isotopic superoxides. For NaO, CsO,,Ar and/or Kr matrices gave better resolved ESR spectra than did N2.I9By contrast, nitrogen was the superior matrix material for all the alkali metal tetroxide molecules except for Cs with which it reacts.24

-

where X = 150 cm-' is the spin-orbit coupling constant of oxygen33 and 6, A are defined in Figure 3b. Since A >> 6, eq 2 implies Agzz >> Ag,, > Agyy= 0, so that the g,, gz, and g3 data of Table I correspond to g,,, g,, and gyy,respectively. Note that were the radical ground state 2B, (as postulatedI6 for the heavier alkali metal superoxides), then the assignment of x and y axes should be interchanged. Table I1 gives experimental values for 6 and A. These were obtained from eq 2 and the measured g,, data of Table I. Several points are worth considering. First, the measured A are largely ( ~ 6 % independent ) of the matrix. More important, the average A for LiO, (41700 f 2600 cm-I) is close in value to the corresponding values for N a 0 2 Cs02. For Kr matrices, for example, 1800 cm-I. Thus, A is the data of ref 16 give A = 38200 relatively insensitive to the alkali metal, as would be anticipated for a parameter associated primarily with the 0,- moiety. Columns 6 and 7 of Table I1 compare Agyydetermined directly from the measured g2 data with those predicted from the g, parameter and eq 2. The two data sets agree to within experimental error. The crystal field parameter 6 represents the energy

-

*

(25) Andrews, L. J. Chem. Phys. 1969, 50, 4288. (26) Andrews, L.; Smardzewski, R. R. J . Chem. Phys. 1973, 58, 2258. (27) Hiiber, H.; Ozin, G. A. J . Mol. Spectrosc. 1972, 41, 595. (28) Billingsley, F. P.; Trindle, T. J . Phys. Chem. 1972, 76, 2995. (29) Grow, D. T.; Pitzer, R. M. J . Chem. Phys. 1977, 67, 4019. (30) Alexander, M. H. J . Chem. Phys. 1978, 69, 3502. (31) Kanzig, W.; Cohen,M. H . Phys. Reu. Lett. 1959, 3, 509. (32) Kasai, P. H. J . Chem. Phys. 1965, 43, 3322. (33) Moore, C. E. Natl. Stand. Ref Data Ser. (U.S.,Natl. Bur. Stand.) 1971, No. 35.

The Journal of Physical Chemistry, Vol. 91, No. 24, 1987 6161

ESR Spectra of Matrix-Isolated L i 0 2 TABLE IV: Comparison of Crystal Field 13 Values Estimated from a Point Charge Model with the Corresponding ESR Parameters for LiOz CsOz"

-

Li Na

5400 2770 2580 2490 2870

1.64 1.96 2.18 2.48 2.59

K Rb

cs

7700 5100 3900 2800 2500

R(M-02) is distance from cation to midpoint of O2 bond; R(0-0) = 1.33 A, except for CsO, where R(0-0) = 1.28 A; data from ref 25 and 37-39. Measured 6 from eq 2 with Ag,, from Table I1 and ref 16.

separation between the ground 2A2and first excited 2Bl states of Li02. As shown in column 3 of Table 111, 6 is relatively invariant (=7%) to the matrix material. Moreover, the measured value, 6 = 5100 f 400 cm-l, is in excellent agreement with that predicted from ab initio calculations, 6 = 4900 cm-l.I4 As a corollary of the g tensor assignment discussed above, the measured hf constants A , , A2, and A , may also be labeled A,,, A,, and Ayy,respectively, where x , y , and z are indicated in Figure 3a. Although the ESR spectra do not directly determine the signs of these constants, the most likely a ~ s i g n m e n t ' ~is, ' A,, ~ = +0.66 G, A, = -2.67 G, and A,, = -2.24 G. Table I11 gives the corresponding isotropic (a) and dipolar ( T J hf parameters, where (1 is a unit tensor) A = a1 T. The measured isotropic constant (a = -1.42 G) is both small (-1% of atomic value) and negative as observed for similar ionic radicals's~16and also predicted by the exchange polarization model of Adrian and Jette.34 Table I11 also compares the measured TZiwith those calculated from a point spin model similar to that described e l ~ e w h e r e . ' ~The ,~~ unpaired electron is assumed to reside entirely as point spins ( p = located on the four lobes of the oxygen 2p orbitals at an average distance ( r ) = f0.54 A from each nucleus. Pertinent oxygen-oxygen and metal-oxygen bond lengths were taken from the molecular structure calculations of ref 14 and 28. It should be noted that (1) experimental x and y axes should be interchanged when comparing the calculated 2Bl data and (2) for a 2A" state (for which T,, # 0), the measured A , , A2,and A, are related to the T,,, Tyy,T,,, and T,, by expressions similar to eq 9 of ref 24. Thus, the predicted 2Arfhf constants are35A, = f 1.90 G, A2 = f2.18 G, and A3 = -2.67 G, quite different from the measured data of Table I. In contrast to the situation for 2B, and 2Ar', the measured hf constants are in very good agreement36with those calculated assuming the 2A2geometry of ref 14. Accordingly, the hf analysis implies an ESR assignment of 2A2for the ground state of Li02.

+

(34) Adrian, F. J.; Jette, A. N. J . Chem. Phys. 1984, 81, 2415. (35) Since A , = a Tyy,the Table I (AY = *2.67 G ) and Table I11 (T, = -0.89 G ) data imply u = -1.78 G or +$.56 G. We choose the former assignment in order to obtain u C 0. (36) The agreement between calculated and measured hf constants can be improved by increasing the oxygen-oxygen bond length from 1.38 8, to as much as 1.46 A. A lithium-oxygen bond of appoximately 1.68 8, was found to be optimum, however.

+

-

In concluding, we would like to comment on the trend in the crystal field parameter 6 for the series L i 0 2 Cs02. Table IV summarizes the measured 6 values for the alkali metal superoxides in Kr matrices and also makes a comparison with 6 estimated from a point charge model. The calculated 6 values assume a simple Coulombic interaction between a unit positive charge on the alkali metal cation and point, negative charges ( q = -0.25 e) placed at ( r ) = f0.54 A from each oxygen nucleus. The differing electronic distributions for the a,,* (ground 2A2state) and r X *(first excited 2Bl state) orbitals lead to a slightly greater stabilization of the in-plane ay*component, as noted earlier. For consistentcy, we adopted the geometrical parameters found from the vibrational analysis of Andrews and c o - ~ o r k e r s . ~ ~It. should ~ ~ - ~ ~be noted that these geometries pertain to the superoxide ground state. The 2B1geometries will be slightly different, but these data are generally not available.40 Also neglected are electrostatic terms involving the anion and cation polarizabilities. While important in magnitude, there is some uncertainty as to the details of including this additional i n t e r a ~ t i o n . ' ~The ~ ' ~most striking feature of the experimental 6 is its relative invariance to the nature of the cation for Na, K, Rb, and Cs. As expected, 6 decreases noticeably between Li02 and N a 0 2 but then levels off and actually increases in magnitude for Cs02. This reversal in the crystal field parameter was discussed in some detail in ref 16. It was also pointed out that since the ESR spectra do not determine the sign of 6, an alternative explanation might involve an inversion in the energy order of ay*and a,* for R b 0 2 and Cs02. By analogy with the observed behavior of the alkali metal monoxide^,^' it was suggested that an inner shell bonding mechanism might be important in determining the ground state of C S O ~ .However, ~~ C s 0 2 do not adequately support the ESR spectra of N a 0 2 either the inversion or the reversal hypotheses. Likewise, the results presented here, while significant in that they identify Li02 as an ionic radical with ground-state 2A2,do little to resolve the bonding behavior of the heavier alkali metal superoxides. A more complete understanding of these species awaits further experimental analysis combined with accurate molecular structure calculations on both R b 0 2 and Cs02.

-

Acknowledgment. We thank Lin Chu for her invaluable assistance in the computer simulations. This work was supported in part by the National Science Foundation under Grants C H E 83-07164 and RII 83-05241 and by The City University of New York PSC-BHE Faculty Research Award Program. Registry No. 6Li02, 12521-53-8; 'Li02, 110510-80-0; Ar, 7440-37-1; Kr, 7439-90-9; N,, 1727-37-9. (37) Andrews, L. J . Phys. Chem. 1969, 73, 3922. (38) Andrews, L. J . Chem. Phys. 1971, 54, 4935. (39) Andrews, L.; Hwang, J.-T.; Trindle, C. J . Phys. Chem. 1973, 77, 1065. (40) The ab initio data for Li02 (ref 14) indicate that 6 is underestimated by approximately 2000 cm-' as a consequence of assuming a ground-state geometry for the excited ,B, state. (41) Lindsay, D. M.; Herschbach, D. R.; Kwiram, A. L. J . Chem. Phys. 1974, 60, 315. (42) Recent LCAO-SCF calculations point out the stabilizing influence of the filled (n - l ) p orbitals in K 2 0 2and Rb202 See: Bravo, G.; Blaisten-Barojas, E. Chem. Phys. Lett. 1984, 108, 237. Allavena, M.; BlaistenBarojas, E.; Silvi, B. J . Chem. Phys. 1981, 75, 787.