Electronic Spectroscopy of Rhodium Mononitride

Journal of Molecular Spectroscopy 199, 18 –25 (2000) Article ID jmsp.1999.7972, available online at http://www.idealibrary.com on

Electronic Spectroscopy of Rhodium Mononitride Scott G. Fouge`re, Walter J. Balfour, Jianying Cao, and Charles X. W. Qian Department of Chemistry, University of Victoria, Victoria, British Columbia, Canada V8W 3V6 Received May 26, 1999; in revised form August 24, 1999

We report the first observation of gas-phase electronic spectra for rhodium mononitride. The RhN molecules have been produced in the reaction of laser-ablated rhodium metal with ammonia. Many vibronic bands have been studied in the 400 –700 nm region using laser-induced fluorescence. Rotational analyses of the stronger of these, together with excited state lifetime measurements and Rh 14N–Rh 15N isotopic shifts, identify three electronic systems in the region: [15.1]1–X 1S 1, [19.5]0 1–X 1S 1, and [22.4]0 1–X 1S 1 with (0, 0) bands near 15 071, 19 489, and 22 385 cm 21, respectively. The 1S 1 symmetry for the ground state agrees with theoretical predictions. Dispersed fluorescence spectra have been recorded which reveal the presence of electronic states at T 5 553, 1740, and 3920 cm 21. © 2000 Academic Press Key Words: rhodium nitride; laser-induced fluorescence; dispersed fluorescence; spectroscopic constants; electronic structure. INTRODUCTION

EXPERIMENTAL PROCEDURE

The past decade has seen a blossoming of interest in the spectroscopy of small molecules containing transition metals, from both the experimental and theoretical point of view. While the focus has been on diatomic carbides, nitrides, and oxides, there have also been studies on diatomic hydrides and halides as well as some polyatomic systems. The impetus for activity on the experimental side has come with the introduction of laser-ablation/jet-cooling techniques which produce more tractable spectra than earlier conventional high-temperature sources. There have been parallel theoretical advances in the treatment of these molecular systems (1). The present paper reports the discovery and characterization of the second-row (4d) transition metal mononitride, RhN. For this row of the periodic table, gas-phase spectra have now been found for YN (2), ZrN (3), NbN (4), MoN (5), and, most recently, for RuN (1). Correlation of molecular data within such a series is providing a better understanding of the nature of the bonding between a transition element and nitrogen. This information is important, given the central role transition elements play in catalytic processes such as ammonia synthesis, nitrogen fixation, and pollution control. One of the elements most active catalytically is rhodium. However, to date the only Rh-containing diatomic system to have been extensively characterized spectroscopically in the gas phase is RhC (6 – 8), although current work in this laboratory detected RhH and RhO in addition to RhN, and the electronic structures of RhC, RhN, and RhO were recently probed using anion photoelectron spectroscopy (9). Infrared absorption spectra of RhN trapped in an argon matrix were observed recently (10). An all-electron ab initio calculation for RhN was also published (11).

The molecules of interest were generated in the laser vaporization molecular beam source already successful in producing spectra of CrN, as previously described (12). For RhN, a rotating rhodium rod (5-mm diameter, 99.9% purity; Goodfellow) was used together with a stream of helium-doped 5% with 14 NH 3 and/or 15NH 3 (99%, Cambridge Isotopes). Laser-induced fluorescence (LIF) spectra were recorded in the region 370 –740 nm, where bands of interest were scanned with a step size of 0.001 nm and the signal averaged for 15 laser shots. LIF decay curves were measured at the bandheads and selected wavelengths by averaging the LIF signal for 3000 shots. Lowresolution dispersed fluorescence (DF) spectra were also obtained. Most of the strong features in the spectrum could be recorded without the aid of a dye amplifier; however, amplification was employed for the DF spectra of some of the weaker bands to enhance the signal quality. Known RhC spectra (6, 7) and the internal calibration of the scanning dye laser were used for wavelength calibration. The LIF bandhead positions should be good to 61 cm 21. Measurement errors within a band are estimated to be 60.1 cm 21; Rh 14N–Rh 15N isotopic shifts were measured from spectra containing both species, obtained by using a mixture of 14NH 3 and 15NH 3 gases. The DF spectral intervals have an estimated error of 615 cm 21. OBSERVATIONS

Twenty-three red-degraded RhN bands of appreciable intensity were observed between 700 and 400 nm, together with a similar number of weaker bands. The 400 –370 nm region is denser in features, but we did not study this region at higher resolution. In this region, there are many overlapping bands, where the correlation between Rh 14N and Rh 15N features is not 18

0022-2852/00 $35.00 Copyright © 2000 by Academic Press All rights of reproduction in any form reserved.

ELECTRONIC SPECTRUM OF RhN

19

TABLE 1 Band Heads (nm) of Observed RhN Features a between 700 and 400 nm

a Although no rotational analyses have been performed for the weaker bands, identification and positive attribution to RhN can be made using DF intervals and isotopic observations. b w 5 weak, m 5 medium, s 5 strong, v 5 very.

in all instances apparent. Table 1 gives a list of RhN bandheads. In addition, a spectrum of RhH, a species whose electronic spectrum has not hitherto been reported, is obtained under the Rh 1 NH 3 conditions used. The RhH spectrum is readily distinguished from that of RhN by its more open rotational structure, evident in two wavelength ranges: 468 – 463 and

441– 437 nm. A separate publication (13) will deal with the characterization and analysis of this RhH spectrum and that of the corresponding deuteride. The structure of the observed jet-cooled RhN LIF bands, simple molecular orbital (MO) theory, and more sophisticated calculations all indicate that the RhN ground state symmetry is 1 1 S . Three electronic systems were identified between 700 and

Copyright © 2000 by Academic Press

20

FOUGE`RE ET AL.

B v 5 0.4680 2 0.0069(v 1 12). The corresponding Rh 14N equilibrium bond length, r9e [15.1] 5 0.1709 5 nm. (ii) The 512.95-nm System: [19.5]0 1–X 1S 1

FIG. 1. An LIF spectrum of the [15.1]1–X 1 S 1 (0, 0) band of Rh 14N near 663.4 nm.

400 nm. Their (0, 0) band heads are located near 663.35, 512.95, and 446.60 nm. A weak band with head at 429.15 nm is tentatively attributed to a fourth system. (i) The 663.35-nm System: [15.1]1–X 1S 1 This system is quite weak. The head at 663.35 nm is the strongest feature to the red end of the region surveyed: other bands of similar profile [see Fig. 1], and of comparable intensity, which form a progression with successive intervals of 661, 697, and 561 cm 21, are clearly the associated (1, 0), (2, 0), and (3, 0) bands. Irregular vibrational spacings are not uncommon in molecules such as RhN, where many electronic states lie close enough in energy to interact significantly. As may be noted by reference to the spectrum shown in Fig. 1, bands of this system have an R-, Q-, and P-branch structure indicative of a DV 5 1 transition; however, the relative intensity of the Q branch is markedly less than would be expected were the transition to be of 1P– 1S type. (See discussion below.) The other two systems are of 1S– 1S type. The bands of the 663.35 system differ from those in the 512.95- and 446.60-nm systems in two further respects. First, the excited state lifetimes found for the bands of the 663.35-nm system are considerably longer, and secondly, the bands of the 663.35-nm system show 2553 cm 21 “hot” bands not evident elsewhere in the spectrum. The 553-cm 21 interval is treated in more detail within the discussion of the DF spectra. Rotational analysis of the bands was straightforward, with no rotational perturbations noted. First combination differences show the expected slight combination defect from the upper state L doubling. Line positions with assignments are listed for the (0, 0) band in Table 2. Rotational constants are given in Table 3. For the ground state, only B 0 was measured. Its value of 0.5075 cm 21 gives r 00 5 0.1641 nm, in excellent agreement with the r e value, 0.1640 nm, obtained theoretically by Shim et al. (11). The B v values for the [15.1]1 state follow the equation:

There are four prominent bands in this system. Their heads occur near 512.95, 491.70, 472.95, and 456.20 nm, which are assigned from measured Rh 14N/Rh 15N isotopic shifts to (0, 0), (1, 0), (2, 0), and (3, 0) bands, respectively. A reproduction of the (1, 0) band of Rh 14N, showing its simple R- and P-branch structure, is given in Fig. 2. Local rotational perturbations are evident in these bands through line displacements, extra lines, and intensity disturbances. Three such perturbations may be seen in Fig. 2. They occur for the [19.5]0 1, v 5 1 upper J levels 12, 20, and 30. Application of simple perturbation theory for mutually interacting pairs of levels gives H 12 interaction matrix elements of 0.125, 0.63, and 0.88 cm 21, for these three instances, respectively. It seems clear that more than one state is responsible for the perturbations but, unfortunately, the jet-cooled spectra do not cover a large enough range of J values for these perturbing states to be characterized. Rotational assignments for the [19.5]0 1 –X 1 S 1 (0, 0) band are included in Table 2 and rotational constants for the system in Table 3. Two bands with v 5 1, X 1 S as lower state have been seen, allowing a value of 899 cm 21 to be determined for DG( 21) in the ground state [cf. the DF results discussed below]. The corresponding value reported for RhN in an argon matrix, where the frequency is usually a few wavenumbers lower, is 894.9 cm 21 (10). The vibrational spacings in the [19.5]0 1 state are sufficiently regular for rough estimates of v e and v e x e to be made: we find v e ' 874 cm 21 and v e x e ' 16.5 cm 21. (iii) The 446.60-nm System: [22.4]0 1–X 1S 1 The bands attributed to the [15.1]1–X 1 S and [19.5]0 1 – X S systems, as discussed above, account for substantially all the observed intensity in the spectrum to wavelengths longer than 450 nm. Toward shorter wavelengths, three prominent features are found (among others) which are assigned to a third electronic transition. The first of these bands falls near 446.60 nm and shows the very small isotopic shift expected for a (0, 0) band: the others are spaced at intervals typical of RhN vibrational spacings, 820 cm 21, then 798 cm 21, and have isotopic shifts typical of Dv 5 1 and Dv 5 2 transitions, respectively. The (0, 0) band of this [22.4]0 1 –X 1 S system is shown in Fig. 3. Its line assignments are included in Table 2 and the associated molecular constants in Table 3. Local perturbations at J9 5 21 and 35 are evident. The other bands of the system suffer similar rotational disturbances. 1

(iv) The 429.85-nm Band As may be seen from the spectrum in Fig. 4, there are two significant bands in the region between 429.5 and 431.5 nm. The head at 430.8 nm is that of the (1, 0) band of the

Copyright © 2000 by Academic Press

ELECTRONIC SPECTRUM OF RhN

21

TABLE 2 Rotational Assignments and Vacuum Wavenumbers (in cm 21) for the [15.1]1– X 1S 1(0, 0), [19.5]0 1–X 1S 1(0, 0), and [22.4]0 1–X 1S 1(0, 0) Bands of the Rh 14N Molecule

[22.4]0 1 –X 1 S 1 system. When the photomultiplier is allowed to record all-wavelength emission, the second band, near 429.85, is relatively weak. We obtained DF spectra following excitation in each of these bands. The results are quite different (see Fig. 5). When the 430.8-nm band is excited, strong emission is seen at a displacement of 1740 cm 21 from the exciting line; for the 429.85-nm band, the strongest emission occurs at a displacement of 3920 cm 21. We took advantage of this difference to record wavelength-filtered LIF spectra, collecting the emission separately at the two different displaced wavelengths. It is the results from these experiments that are displayed in Fig. 4. As discussed above, the 430.8 band comes from a DV 5 0 transition. Spectra at high dispersion indicate

that the transition giving rise to the 429.85-nm band has DV 5 1 with the lower state the ground state of the molecule. The upper state shows a local perturbation at N9 5 24 and has a B value of 0.4313(8) cm 21 and substantial L-type doubling. (v) Dispersed Fluorescence Spectra DF spectra of Rh 14N were obtained following individual band excitation to various levels in the three excited electronic states identified in the LIF spectra. Generally, the probe laser was set to coincide with the strong R-head region of a given band. The DF spectra are displayed in Figs. 6, 7, and 8. Each group of spectra reveals the presence of two or more low-lying

Copyright © 2000 by Academic Press

FOUGE`RE ET AL.

22 TABLE 3 Molecular Constants (cm 21) for Rh 14N

FIG. 3. An LIF spectrum of the [22.4]0 1 –X 1 S 1 (0, 0) band of Rh 14N near 446.6 nm. The J 5 19 level of the upper state is perturbed.

Note. r 0 (X 1 S 1 ) 5 0.1642 nm; r e ([15.1]1) 5 0.1705 nm; r e ([19.5]0 1 ) 5 0.1712 nm; r e ([22.4]0 1 ) 5 0.1715 nm.

electronic states (or substates) and the associated vibrational progressions provide information on their vibrational spacings. Data pertaining to the DF spectra are collected in Table 4. What may be noted by comparing the spectra in Fig. 6 with those in Figs. 7 and 8 is that the DF spectra recorded for the [15.1] state as emitting state show one extra progression. This progression has its origin anchored at ;555 cm 21 above the ground state. Since the [19.5] and [22.1] states are 0 1 states, while the [15.1] state has V 5 1, the inference is that the state with T ' 555 cm 21 is accessible by an allowed transition from V 5 1 but not from V 5 0 1, i.e., the [0.55] state probably has V 5 0 2 or 2. The DF spectra also indicate that there are other electronic states which are low-lying, with T 5 1740, 3920,

1

1

FIG. 2. An LIF spectrum of the [19.5]0 –X S (1, 0) band of Rh N near 491.7 nm. Local rotational perturbations are evident through positional and intensity disturbances in P(13), P(21), and R(29). 1

14

4955, and 7245 cm 21. Such a proliferation of low-energy states for RhN is in sharp contrast to what is found in similar DF experiments on RhC (8) and RhO (14). Lai and Wang (9) studied the photoelectron (PE) spectra of the anions RhC 2, RhN 2, and RhO 2 and obtained information on low-lying states of the corresponding neutral species. Their spectra show the same higher density of such states for RhN versus RhC and RhO as do our spectra. They find evidence for RhN states with T 5 560, 970, 1900, 3000, 7800, and 12 300 (6100 cm 21), interpreting the state with T 5 560 cm 21, which is clearly the state seen in our spectra at 553 cm 21, as the lowest (V 5 0 1) component of a 3P state, and those at T 5 970 and 1900 cm 21 with its associated 3P 1 and 3P 2

FIG. 4. Wavelength-filtered LIF spectra of Rh 14N near 430 nm. The spectrum in (a) was recorded with a monochromator setting of 510.0 nm; that in (b) under identical conditions, but with a monochromator setting of 521.4 nm.

Copyright © 2000 by Academic Press

23

ELECTRONIC SPECTRUM OF RhN

14

FIG. 5. Rh N dispersed-fluorescence spectra. In scan (a), the excitation laser wavelength was 430.960 nm; in scan (b), the excitation laser wavelength was 429.856 nm.

substates, respectively. Our experiments cast some doubt on this interpretation, for we see no evidence for a state at 970 cm 21, and should have been able to, if its V value is 1. We do see a level at 900 cm 21, which we assign with some confidence to the v 5 1 level of the ground state, and prefer the same assignment for the 970-cm 21 feature in the PE spectrum. Much of the apparent intensity for the 970-cm 21 PE feature is the result of overlap by the considerably stronger 560-cm 21 feature. The values of 7800 and 12 300 cm 21 were assigned by Lai and Wang on the assumption that the transitions measured came from an excited state of the RhN 2 ion. If, instead, we adopt the alternative interpretation that the observed PE peaks originate with ground state RhN 2, then the corresponding RhN

FIG. 6. Rh 14N dispersed-fluorescence spectra following excitation to various vibrational levels of the [15.1]1 state.

FIG. 7. Rh 14N dispersed-fluorescence spectra following excitation to various vibrational levels of the [19.5]0 1 state.

levels lie at 3950 and 8000 cm 21, respectively. The experimental evidence from both the DF and anion PE spectroscopy then points to low-lying RhN electronic states at approximately 550, 1800, and 3950 cm 21. A level at 8000 cm 21 is not detected in the DF spectra but could just be weak there. DISCUSSION

We used the molecular orbital diagram in Fig. 9, together with electronic selection rules and the ab initio predictions of Shim et al. (11, 15), in an attempt to assign our observed LIF and DF states. Cross-correlating with recent observations in the spectrum of the isoelectronic PdC molecule (16) has also proved rewarding. The 12s MO is principally a 5s s (Rh)

FIG. 8. Rh 14N dispersed-fluorescence spectra following excitation to various vibrational levels of the [22.4]0 1 state.

Copyright © 2000 by Academic Press

24

FOUGE`RE ET AL.

TABLE 4 Summary of the Observed Low-Lying Levels of Rh 14N from Dispersed Fluorescence Spectra

nonbonding orbital. The 5p and 6p orbitals come from bonding and anti-bonding combinations of 4d p (Rh) and 2p p (N), respectively, while 11s and 13s orbitals are the corresponding bonding and antibonding pair from 4d s (Rh) and 2p s (N) with a small contribution for 5s s (Rh). The 7p orbital is derived from 5p p (Rh) and the 2d is a nonbonding 4d d (Rh)-based orbital. The 1S 1 ground state arises from the configuration . . .11s 25p 42d 412s 2. The four lowest energy one-electron pro-

motions and the states they give rise to are predicted to be 12s 3 6p, 1,3P; 2d 3 6p, 1,3P, 1,3F; 12s 3 13s, 1,3S; and 12s 3 14s, 1,3S. The ab initio calculations of Shim et al. suggest that the 12s and 6p orbitals lie close enough in energy that the 3 P r and 1P states arising from . . .11s 25p 42d 412s 16p 1 will be low-lying. We note that the 12s 3 6p electron promotion in RhC leads to its A 2 P state found near 10 000 cm 21 (7). The higher electronegativity of N over C is expected to lower the 6p energy giving a smaller 12s 3 6p energy gap in RhN. The ab initio prediction for the energy of the (12s 16p 1) 1 P is 4430 cm 21 and that of the corresponding 3P is about 2500 cm 21 lower. The configuration . . .11s 25p 42d 46p 2, which differs from that of the ground state by two electrons, gives rise to 3S 2, 1D, and 1S 1 states at somewhat higher energies. Calculations place this 3S 2 state at 8200 cm 21, the 1D at 14 000 cm 21, and the 1S 1 state still higher. It seems, therefore, safe to conclude that the RhN states having electronic energies between 0 and 5000 cm 21 come from the . . .12s 16p 1 configuration. The precise ordering of the various V substates will depend on the strength of the spin– orbit coupling in the 3P state and on interactions between substates having the same value of V. The 3P spin– orbit A constant is related to the one-electron parameter a p through A[ 3 P, s 1 p 1 ] 5 21 a p . A rough estimate of a p (RhN) may be obtained from experimental data available for the corresponding monocarbide (7). In the first excited 2 P(p 1) state of RhC, A 5 a p 5 1775.82 cm 21, which is 62% of the Rh atomic z (5d) parameter. Nitrogen being somewhat more electronegative than carbon, the Rh(4d p )–ligand(2p p ) mixing should be less in RhN than in RhC and result in an a p

FIG. 9. Molecular orbital correlation diagram for RhN.

Copyright © 2000 by Academic Press

25

ELECTRONIC SPECTRUM OF RhN

value for RhN closer to the atomic value than for RhC. We estimate A(RhN) to fall within the range 800 –1000 cm 21. The equal spacing of spin– orbit components of an isolated 3 P state will be disturbed by interactions with nearby states. Specifically, mixing with the ground state will displace the V 5 0 1 level upward, and mixing with the 1P state will push the V 5 1 level down. The V 5 0 2 and 2 components should remain relatively unaffected. The net result, for mixing that is not too strong, will be an energy ordering for the 12s 16p 1 configuration of 3P 02 , 3P 01 , 3P 1 , 3P 2 , 1P 1, which agrees with the ab initio calculation (15). The forgoing discussion clearly points in favor of the 3P 02 assignment for the state located 553 cm 21 above the v 5 0 level of the ground state, over the 3P 2 alternative. The level at 3920 cm 21 is suitably placed to be identified with the 1P 1 state calculated to lie at 4431 cm 21. However, the assignment of the DF level at 1740 cm 21 is less obvious. As Fig. 5 shows, it is seen strongly in emission when the 430.8-nm band is excited. The upper level, [22.4], v 5 1, of this transition has V 5 0 1, which should not emit strongly to the V 5 2 substate of 3P. On the other hand, the assignment of the 1740-cm 21 level as 3P 1 and the 553 cm 21 level as 3P 02 implies a spin– orbit splitting which seems too large. An alternative, preferred, explanation for the 1740-cm 21 level is the 3P 01 substate, considerably pushed up in energy through strong interaction with the ground state. We next turn to a discussion of the three excited states seen in the LIF spectra. The lowest in energy of these is an V 5 1 state with T 0 5 15 071 cm 21. As noted above, the [15.1]1–X system is weak and its bands have rather weak Q structure. This behavior is remarkably similar to what is found for the [17.9]1–X system in the isoelectronic PdC (16). There, a definitive electronic assignment of the upper state is possible because of measurable hyperfine splitting. Its [17.9]1 state is identified as 3S 11 from 2d 412s 113s 1. We believe the same assignment holds for [15.1]1 in RhN. The corresponding 1S 1 state must lie somewhat above the 3S 1 state and may explain either the [19.5]0 1 or [22.4]0 1 state. However, without the guidance of detailed calculations, more comment would be simply speculation. It is interesting to compare vibrational and rotational data in the isovalent systems RhN, IrN (17), PdC, and PtC (18). The ground state in all cases is formally triply bonded s 2p 4d 4: 1S 1. (The d electrons are nonbonding.) For the two diatomic nitrides, the 5d metal-containing species shows a larger vibrational spacing and a shorter bond length than the 4d one. A similar behavior is found for the two carbides. A reason for this may be, as suggested by Steimle (17, 18), that the 5d and 6s orbitals are more similar in radial extent than is so for 4d and

5s. This greater similarity in Ir and Pt will facilitate sd hybridization. Significant sd hybridization should result in an orbital polarized toward the C and N ligands and give a strong bond. Alternatively, a stronger p bond may simply arise from greater metal 5d–ligand 2p overlap than 4d–2p overlap. A strengthening of bonding as one goes from the 3d to 4d to 5d series in the transition metals was also observed in larger transition metal complexes (19). ACKNOWLEDGMENTS The authors are most grateful to Dr. M. D. Morse (Utah) for helpful discussions and for the communication of unpublished results on the PdC spectrum. They also acknowledge correspondence on RhN prior to publication from Drs. L. Andrews (Charlottesville, VA) and I. Shim (Lyngby, Denmark). Funding for this work came in the form of operating and equipment grants from the Natural Sciences and Engineering Research Council of Canada.

REFERENCES 1. R. S. Ram, J. Lie´vin, and P. F. Bernath, J. Chem. Phys. 109, 6329 – 6337 (1998). [and references cited therein] 2. R. S. Ram and P. F. Bernath, J. Mol. Spectrosc. 165, 97–106 (1994). 3. J. He, C. Ma, and A. S.-C. Cheung, Chem. Phys. Lett. 295, 535–539 (1998). 4. Y. Azuma, G. Huang, M. P. J. Lyne, A. J. Merer, and V. I. Srdanov, J. Chem. Phys. 100, 4138 – 4155 (1994). 5. N. S.-K. Sze and A. S.-C. Cheung, J. Mol. Spectrosc. 173, 194 –204 (1995). 6. A. Lagerqvist and R. Scullman, Ark. Fys. 32, 479 –508 (1966). 7. B. Kaving and R. Scullman, J. Mol. Spectrosc. 32, 475–500 (1969). 8. W. J. Balfour, S. G. Fouge`re, R. Heuff, C. X. W. Qian, and C. Zhou, J. Mol. Spectrosc., in press. 9. X. Li and L.-S. Wang, J. Chem. Phys. 109, 5264 –5268 (1998). 10. L. Andrews, private communication, 1999. 11. I. Shim, K. Mandix, and K. A. Gingerich, J. Mol. Struct. (Theochem) 393, 127–139 (1997). 12. W. J. Balfour, C. X. W. Qian, and C. Zhou, J. Chem. Phys. 106, 4383– 4388 (1997). 13. W. J. Balfour, J. Cao, C. X. W. Qian, and C. Zhou, unpublished manuscript. 14. S. G. Fouge`re, W. J. Balfour, and C. X. W. Qian, unpublished results. 15. I. Shim, private communication, 1999. [I. Shim has reported some improved calculations on the energies of the low-lying RhN states. Her new calculations place the V components of the 12s 16p 1 3P state at 1826(0 2), 1956(0 1), 1958(1), and 2113(2), with the 1P from the same configuration at 4431 cm 21.] 16. J. D. Langenberg, L. Shao, and M. D. Morse, J. Chem. Phys. 111, 4077– 4086 (1999). 17. A. J. Marr, M. E. Flores, and T. C. Steimle, J. Chem. Phys. 104, 8183– 8195 (1996). 18. T. C. Steimle, K. Y. Jung, and B.-Z. Li, J. Chem. Phys. 103, 1767–1772 (1995). 19. W. A. Nugent and J. M. Mayer, “Metal-Ligand Multiple Bonds,” Wiley, New York, 1988.

Copyright © 2000 by Academic Press