Using JCP format - University of Rochester

JOURNAL OF CHEMICAL PHYSICS

VOLUME 114, NUMBER 14

8 APRIL 2001

Spectroscopy and reactivity of size-selected Mg¿ –methanol clusters James I. Lee,a) David C. Sperry,b) and James M. Farrarc) Department of Chemistry, University of Rochester, Rochester, New York 14627-0216

共Received 28 December 2000; accepted 22 January 2001兲 This work presents photodissociation spectra of Mg⫹共CH3OD兲n (n⫽1 – 5). Mass spectrometry of the parent cluster ions shows that C–H bond cleavage occurs in the ground electronic state. The branching ratios for products of photodissociation show strong selectivity; methyl loss is the exclusive quenching channel in n⫽1, while a unique pathway that eliminates CH3D occurs in n ⫽2. Methyl loss does not occur for clusters with more than three solvent molecules. The maximum of spectral intensity shifts to the red with increasing cluster size until halting at n⫽3. These data signal the formation of a solvent shell at a relatively small size. As the cluster size increases, ligand loss and D atom loss become overwhelmingly favored. We address the selectivity in the products in terms of the initial photoexcited state, nonadiabatic coupling to the ground state potential energy surface, and subsequent dissociation and product formation dictated by dynamics on the ground state surface. © 2001 American Institute of Physics. 关DOI: 10.1063/1.1355311兴

and Yang on M⫹共CH3OH兲 systems,9 product switching to form M⫹共OCH3兲共CH3OH兲n⫺1 occurs near n⫽4. Selegue and Lisy find a similar effect in M⫹共CH3OH兲n 共M⫹⫽Na, Cs兲 clusters, which undergo an intracluster reaction in the ground electronic state to eliminate dimethyl ether and form M⫹共CH3OH兲n⫺2 (H2O). The ‘‘mixed solvent’’ product cluster intensities increase as a function of cluster size, but do not surpass that of the parent M⫹共CH3OH兲n , suggesting that a full ‘‘product switch’’ does not occur.10,11 Spectroscopic investigations on solvated alkaline–earth metals reveal how the solvent ligand field perturbs the metal ion core valence electron as ion solvation evolves with increasing cluster size,4,7,8,12–15 as well as generating information on stepwise dissociation energies. In addition to the data on electronic structure, data on vibrational7,15–19 and rotational16,20 structure are also in the literature for selected systems. Thermochemical determinations of bond strengths using high pressure21–23 and high temperature24,25 mass spectrometry provide useful data for estimating the reaction energetics of various pathways in these systems. Theoretical investigations by Bauschlicher and co-workers have also proven to be invaluable aids, not only in comparison of metal–ligand binding energies,23,26–28 but also in assigning vertical transition energies29,30 and vibrational frequencies.31,32 The Mg⫹共CH3OH兲n system has been the subject of previous investigations. France et al. have probed the spectroscopy of the monomer and examined photodissociation of the larger clusters at selected wavelengths.33 Collision induced dissociation 共CID兲 experiments by Woodward et al. on Mg⫹共CH3OH兲3 exhibit both simple ligand loss and intracluster reactions.34 Lu and Yang9 recently performed mass spectrometric experiments with supporting ab initio calculations, to elucidate ground state reaction pathways. These data have proven helpful in interpreting the present results. In this publication, we report the photodissociation spectra of Mg⫹共CH3OD兲n for n⫽1 – 5, which primarily access

I. INTRODUCTION

The study of molecular aggregates or ‘‘clusters’’ has grown explosively since the early 1960s, aided greatly by experimental advances in mass spectrometry, molecular beam techniques, and lasers.1 One of the driving forces in cluster science is the appeal of studying gas phase species that provide insight into condensed phase behavior. By examining a microcosm of the bulk system, elucidating details of the many-body interactions becomes more tractable. In addition to providing a very convenient stepping stone from the gas phase to the bulk, clusters can exhibit their own unique properties as the discovery of fullerenes2 and metallocarbohedrenes 共Met-Cars兲 共Ref. 3兲 has demonstrated. Alkali and alkaline–earth metal cations possess several characteristics making them good candidates for metalcentered clusters. These species form clusters with many different polar solvents and the relatively strong electrostatic interactions enable multiple ligands of the given solvent to coordinate to the metal. These clusters also appeal experimentally because they form intense molecular beams in supersonic jets and are amenable to focusing and transport with standard ion optical techniques. By employing time-of-flight mass spectrometry, clusters of successive sizes can be massselected for study, thereby generating a stepwise progression of data. In some systems, reactions occurring in clusters above a certain size dominate the parent mass spectra; this phenomenon, known as ‘‘product switching,’’ is clearly demonstrated by Fuke and co-workers in M⫹共H2O兲n 共M⫽Mg, Ca兲,4–6 and in our experiments on Sr⫹共H2O兲n , 7 in which the cluster ‘‘core’’ switches from M⫹ to MOH⫹ over a small range of cluster sizes. In work from our laboratory on Sr⫹共CH3OD兲n clusters8,15 as well as studies reported by Lu a兲

Current address: Stanford Research Systems, 1290 D Reamwood Avenue, Sunnyvale, California 94089. b兲 Current address: Pharmacia and Upjohn, 7000 Portage Road, Mail Stop 4831-259-175, Kalamazoo, Michigan 49001. c兲 Author to whom correspondence should be addressed. 0021-9606/2001/114(14)/6180/10/$18.00

6180

© 2001 American Institute of Physics

Downloaded 23 Mar 2002 to 128.151.176.91. Redistribution subject to AIP license or copyright, see http://ojps.aip.org/jcpo/jcpcr.jsp

Reactivity of Mg⫹

J. Chem. Phys., Vol. 114, No. 14, 8 April 2001

6181

electronic states based on the strongly allowed 2 P← 2 S transitions of the cation. A striking feature of the parent ion mass spectra is a hydrogen atom loss channel suggestive of C–H bond cleavage that increases in intensity with cluster size. Two important features of the photodissociation data include a high degree of selectivity among the energetically accessible channels and a spectral red shift that halts at n⫽3, indicative of a solvation shell closing. II. EXPERIMENT

The experimental apparatus for these experiments has been described in detail in previous publications;14,35 only a summary is presented here. Helium at approximately 1.5 atm is bubbled through room temperature CH3OD 共Cambridge Isotopes, 99% D兲. The vapor pressure of methanol at room temperature is about 100 Torr, making the gas mixture ratio roughly 10% CH3OD in He. This seeded vapor is adiabatically expanded through a solenoid-driven pulsed valve 共General Valve, Series 9兲 into a rotating rod-type36 laser vaporization source. Magnesium metal 共Alfa-Aesar, 99.8%兲 is ablated and ionized by focused second harmonic radiation from a Nd3⫹:YAG laser 共Continuum, NY-61兲. The seeded vapor is forced through the Mg⫹ plasma, creating a distribution of cluster sizes of the form Mg⫹共CH3OD兲n . The pressure in the cluster source chamber is typically 10⫺5 Torr during an experiment. The clusters are collisionally cooled as they travel down a source block channel and expand supersonically into the source chamber. They then enter the extraction region of a Wiley–McLaren-type37 time-of-flight mass spectrometer. The clusters reach a mass-independent spatial focus 1.5 m downstream from the center of the extraction region. The clusters are photolysed by radiation that transversely intersects the cluster beam at the spatial focus. Radiation is provided by a narrow bandwidth 共⬃0.2 cm⫺1兲 singly resonant optical parametric oscillator 共OPO兲 共Spectra Physics, MOPO-730兲. The OPO is pumped by the third harmonic of an injection seeded Nd3⫹:YAG laser 共Spectra Physics, GCR-190兲 running at 10 Hz. The OPO signal wave tuning range is from 440 to 690 nm and its idler wave tuning range is from 730 to 1830 nm. The OPO is also equipped with a frequency doubling option 共FDO-900兲 that converts the oscillator output into ultraviolet radiation. A frequency-doubled dye laser 共Continuum, TDL-50兲 pumped by the second harmonic of a Nd3⫹:YAG laser 共Continuum, YG580兲 was used to cover the degeneracy gap of the OPO/ FDO system for the Mg⫹共CH3OD兲2 spectrum. Photodissociation creates a distribution of product and unphotolyzed reactant ions that are separated using a reflectron-type mass spectrometer. A set of off-axis microchannel plates detects and amplifies the signal. The data are received by a transient recorder 共DSP, 2100AS兲, typically averaged over 100 laser shots 共DSP, 4101兲 and sent from the CAMAC crate to a PC. The photochemical product or ‘‘daughter’’ ions are produced from a single ‘‘parent’’ cluster ion and are normalized to the integrated intensity of the unattenuated parent. Photodissociation pulse energy is monitored throughout the experiment and kept at a constant value of 1 mJ/pulse, with a spot size of diameter 5 mm. Power

FIG. 1. Parent mass spectrum of Mg⫹共CH3OD兲n , n⫽0 – 7.

dependence studies on the product channels have shown a linear response in this photodissociation power regime. III. RESULTS A. Parent ion mass spectra

A mass spectrum of Mg⫹共CH3OD兲n clusters, where n ranges from 0 to 7, is plotted in Fig. 1. Within each cluster size n, we observe a triad of peaks corresponding to the 24, 25, and 26 amu isotopes of Mg; the intensities of these isotopomers reflect their natural abundance. Careful examination of our mass spectra shows evidence for elimination of a hydrogen atom from parent clusters in the ground electronic state. Figure 2 shows a detailed view of the mass spectra for n⫽1 – 5. The data show peaks corresponding to loss of one, two, or three mass units from the parent ion. As the cluster size increases, so do the number and intensity of these peaks: at n⫽5, the peak corresponding to a loss of one mass unit is roughly 50% of the parent ion peak. As we will discuss later, the deuterium atom in Mg⫹共CH3OD兲 allows us to determine that loss of a single mass unit must come from C–H bond cleavage; for loss of two mass units, we cannot distinguish between the loss of two hydrogen atoms or a single deuterium atom. Reports in the literature on methanol van der Waals clusters and atomic metal ions solvated by methanol show that reactions can occur in the ground electronic state to cause cleavage of bonds within individual solvent molecules. With isotopically unsubstituted methanol, mass spectral measurements cannot distinguish between methoxide ( – OCH3) or hydroxymethyl ( – CH2OH), the former resulting from O–H cleavage, the latter from C–H cleavage. Armentrout and coworkers have shown C–H collisional activation is both energetically and statistically favored overactivation of the O–H bond in CH3OH. 38 This is consistent with the methanol bond strengths; breaking the O–H bond requires 10 kcal/mol

Downloaded 23 Mar 2002 to 128.151.176.91. Redistribution subject to AIP license or copyright, see http://ojps.aip.org/jcpo/jcpcr.jsp

6182

J. Chem. Phys., Vol. 114, No. 14, 8 April 2001

FIG. 2. Detailed view of Mg⫹共CH3OD兲n parent mass spectra for n⫽1 – 5. Clusters composed of the most abundant isotope 24Mg⫹ are indicated by *. Loss of 1, 2, and 3 mass units are indicated by C, B, and A, respectively. The cluster size, n, is indicated above each trace.

more energy than the C–H bond does.39 In the collisioninduced dissociation 共CID兲 experiments performed by Woodward et al.,34 a H atom loss series similar to what we observe is reported; combination loss of a ligand and 1, 2, or 3 H atoms is observed in Mg⫹共CH3OH兲3. Loss of 2 and 3 mass units was also noted by Lu and Yang9 in reaction of Mg⫹ 2 with methanol clusters. Experiments on both the Sr⫹共CH3OH兲n and Sr⫹共CH3OD兲n systems in our laboratory unequivocally show that the C–H bond is broken in the parent mass spectra.8 We observe the same pattern of data in the parent mass spectra of the Mg⫹共CH3OD兲n system. Given the one amu spacing between the peaks, the data strongly suggest that the peaks correspond to C–H bond fission products. Lu and Yang9 and Woodward et al.34 each report a H atom elimination channel to form parent ions with apparent stoichiometry MgOCH⫹ 3 共CH3OH兲n that becomes prominent at n⬎5 using their ‘‘pick-up’’ type cluster sources. Our data with isotopically-labeled methanol clearly show a peak one mass unit lighter than the parent ion, which can only correspond to C–H cleavage in the Mg⫹共CH3OD兲n system. Although the full set of data do not allow us to rule out O–D bond cleavage, observation of single mass unit loss is only consistent with C–H bond cleavage. This result is in contrast to the data of Lu and Yang,9 who claim that hydrogen atom loss is exclusively from oxygen. Study of the Mg⫹共CD3OH兲n isotopomers will allow us to positively determine if the

Lee, Sperry, and Farrar

FIG. 3. Photodissociation mass spectra of Mg⫹共CH3OD兲n , n⫽1 – 5. The parent ion is indicated by *. Cluster size and photolysis wavelength are indicated above each trace. This figure uses a shorthand notation, and refers to the photochemical products in terms of what ligand, ligand fragment, or combination was lost from the parent;–Me, loss of the methyl fragment;– (Me⫹D), loss of CH3 and D fragments;–mL, loss of (CH3OD) m ; – (L⫹D), loss of CH3OD and D fragments.

peaks multiple mass units lighter than the parent ion are from loss of multiple H atoms or from a combination of H and D loss. B. Photodissociation mass spectra

Mass spectra of the fragments produced upon photolysis of a given cluster size at selected wavelengths are presented in Fig. 3. We refer to parent ions with 1, 2, 3, etc., solvent molecules as monomer, dimer, trimer, etc., respectively, and use the notation introduced in Fig. 3 to label our products. Loss of a methyl radical, deuterium atom, or methanol ligand is labeled as –Me, –D, or –L, respectively. Combinations of these channels are denoted by linking the channels with a ⫹ sign, and multiple ligand loss is denoted by –nL, where n equals the number of ligands lost. In the monomer, we observe a single product, –Me, over the entire spectral range investigated. In contrast, photolysis of the dimer creates four distinct products; –Me and –(Me⫹D) are created through intracluster reactions, –L corresponds to simple ligand loss, and –(L⫹D) represents a combination of ligand and D elimination. The –(Me⫹D) product, stoichiometrically equivalent to elimination of singly deuterated methane, is of particular interest. This product is unique to photodissociation of the dimer, appearing in no other Mg⫹共CH3OD兲n pho-

Downloaded 23 Mar 2002 to 128.151.176.91. Redistribution subject to AIP license or copyright, see http://ojps.aip.org/jcpo/jcpcr.jsp

J. Chem. Phys., Vol. 114, No. 14, 8 April 2001

tochemical reaction. As discussed below, the unique structure of Mg⫹共CH3OD兲2, determined with Hartree–Fock calculations by Lu and Yang,9 is the key to understanding how this reaction to occurs. The trimer photodissociation spectrum displays mass peaks corresponding to –Me. –L, –(L⫹D), and –(L⫹Me) products. The tetramer photoproducts are –(L⫹D), –(L⫹Me), and –2L. The pentamer photolysis signal was very weak, and mass resolution was poor. Only a single loss channel, corresponding to –(L⫹D) remained significantly above the noise level of the experiment. We emphasize that the photolysis products from the tetramer and pentamer all involve loss of a solvent molecule. France et al.33 investigated the photolysis of ⫹ Mg 共CH3OH兲n for n⫽1 – 6. For the singly-solvated cluster, a photodissociation spectrum was reported between 27 000 and 33 000 cm⫺1; for larger clusters, dissociation was probed at selected wavelengths. In the monomer, the data show that MgOH⫹, Mg⫹, and small amounts of MgO⫹ and CH⫹ 3 are produced by photolysis. In contrast, we observed exclusive loss of methyl radicals, corresponding to MgOH⫹ formation in the monomer, indicating that the photochemistry observed by France et al. is more extensive and presumably less selective. In each cluster size studied, their experiment accesses more photochemical channels than our experiment. The enthalpy change for CH⫹ 3 production from Mg⫹⫹CH3OH is reported as 76.3 kcal/mol.17 Adding the enthalpy of breaking the electrostatic bond in Mg⫹共CH3OH兲 共Ref. 27兲 places the endothermicity to production of CH⫹ 3 from Mg⫹共CH3OH兲 at approximately 113 kcal/mol, requiring at least two 350 nm photons. Furthermore, the charge transfer 共CT兲 state for the monomer was calculated to occur at 51 100 cm⫺1 above the ground state,27 also requiring two 350 nm photons to access. Therefore, the fragments observed in Ref. 33 undoubtedly come from absorption of multiple photons. In contrast, our power studies show a linear response, suggesting that multiphoton events are not significant dissociation pathways. The results presented here for the branching ratios in larger clusters show both differences and similarities with previous studies. The data of France et al.33 show ligand and methyl loss in larger clusters, as well as combinations of these two channels. However, no H atom elimination is observed for clusters larger than the monomer. In contrast, Woodward et al. observe several reactions involving H atom elimination in their CID experiments on Mg⫹共CH3OH兲3, including the loss of a solvent molecule and an H atom.34 Consistent with the latter experiments, the results we report here show that methyl loss decreases in importance with increasing cluster size, and several channels involving O–D bond cleavage become increasingly important in larger clusters. A propensity for larger clusters to eject hydrogen atoms bound to solvent oxygen atoms upon photon absorption has been observed in other systems. Fuke and co-workers report H elimination in Mg⫹共H2O兲1,2 clusters, as well as combination ligand plus H loss in Mg⫹共H2O兲2–5 . 12 The photoproducts from the Sr⫹共methanol兲n system8 exhibit behavior qualitatively similar to the Mg⫹共CH3OD兲n results reported here, in that in both systems, the methyl loss channel disappears,

Reactivity of Mg⫹

6183

FIG. 4. Total photodissociation spectra for Mg⫹共CH3OD兲n , n⫽1 – 5. Ab initio estimates of the vertical transition energies in the monomer by Bauschlicher and co-workers are indicated by vertical bars. In order of increasing transition energy, the ab initio calculations locate the (2) 2 A ⬘ 3p ␲ -, (1) 2 A ⬙ 3p ␲ -, and (3) 2 A ⬘ 3p ␴ - like states of Mg⫹共CH3OD兲.

while reactions involving –L and –D become dominant with increasing size. Differences among these studies likely result from differences in ion source conditions, interaction geometries between ions and radiation, and product mass spectral resolution. In the next section, we report our results on the photochemical products of the larger Mg⫹共CH3OD兲n clusters, comparing and contrasting them with other similar studies in the literature. C. Photodissociation spectroscopy

The total photodissociation spectra of Mg⫹共CH3OD兲n , are plotted in Fig. 4, and the corresponding channel-resolved spectra are plotted in Figs. 5 and 6. The symmetry species of the monomer is predicted to be C s by ab initio calculations.9,44 Sodupe and Bauschlicher32 have calculated vertical transition energies, transition moments, and orbital populations for Mg⫹共CH3OH兲. We use their calculations as a guide in assigning the electronic transitions we observe. Each of the transitions is to a 2 P-derived electronic state of the cation

Downloaded 23 Mar 2002 to 128.151.176.91. Redistribution subject to AIP license or copyright, see http://ojps.aip.org/jcpo/jcpcr.jsp

6184

J. Chem. Phys., Vol. 114, No. 14, 8 April 2001

Lee, Sperry, and Farrar

FIG. 5. Branching ratio data for Mg⫹共CH3OD兲2. The figure uses the shorthand notation adopted in Fig. 3. Note the shoulder in the high energy band.

as demonstrated by the orbital population calculations. The electronic bands are rather broad, each covering about 4000 cm⫺1. The highest energy band peaks near 38 000 cm⫺1, in good agreement with the theoretical prediction placing the (3) 2 A ⬘ (3p ␴ ) transition energy in the range from ⬍38 700 to 40 400 cm⫺1. The lower energy band peaks near 30 000 cm⫺1, which is within the predicted 29 429–30 700 cm⫺1 range for the (1) 2 A ⬙ (3 p ␲ ) transition. Given that the band positions we observe agree favorably with the calculated transition energies, it is surprising that we do not find a distinct (2) 2 A ⬘ (3p ␲ ) band in the range 27 550–28 600 cm⫺1 as Sodupe and Bauschlicher predict. We believe both p ␲ -like transitions are within the lower energy band, closely spaced and unresolvable as discussed below. The spectrum of Mg⫹共CH3OH兲2 displays bands that peak near 24 000 cm⫺1 and 30 000 cm⫺1. The higher energy band has a shoulder near 29 000 cm⫺1. The low energy band and the shoulder of the high energy band are assigned to p ␲ -type transitions, while the high energy band is assigned to a p ␴ -type transition. Our reasoning is based on the assignments given by Fuke and co-workers to the similar Mg⫹共H2O兲n system12 and is discussed below. In the trimer, we observe a single strong band peaking at 24 000 cm⫺1, and a weak band beginning at ⬃35 000 cm⫺1 with a high energy tail, shown in Fig. 4. Hartree–Fock calculations on Mg⫹共CH3OH兲3 by Lu and Yang9 show that the ligands all bind to the same side of the metal, most probably in a C 3 v configuration. This kind of ligand arrangement was predicted by theory in Mg⫹共H2O兲n to decrease the p←s type transition energy spacing29 and was confirmed by experiment.12 Misaizu et al. observed a coalescence of all three of the p←s-type transitions into a single band in Mg⫹共H2O兲3. 12 The same decrease in transition energy spacing occurs with CH3OD as the solvent, and coupled with the increase in the density of states gives rise to a single strong band in

FIG. 6. Branching ratio data for Mg⫹共CH3OD兲n , n⫽3,4. The figure uses the shorthand notation adopted in Fig. 3.

Mg⫹共CH3OH兲3. The calculated structures for the tetramer and pentamer are similar9 and the observed single strong absorption band persists in these clusters as well. The weak band found in the trimer is also present in the tetramer, but is slightly red shifted. The pentamer does not display this weak feature, but it is possible that the signal level for this cluster is too small to discern such features. The most striking feature of the tetramer and pentamer spectra shown in Fig. 4 is the fact that the band near 24 000 cm⫺1 shows no red shift relative to the trimer. This observation is a strong indication that the solvent shell closes at n ⫽3; the addition of the fourth and fifth ligands only weakly perturbs the cluster chromophore because these molecules are at a significantly larger distance from the cluster core. This result agrees with the structure calculations reported by Lu and Yang, which find a three ligand first solvation shell is the most favorable configuration for the tetramer and pentamer.9 IV. DISCUSSION A. Spectroscopy

As discussed in the previous section, the positions of the two bands in our monomer spectrum agree with the theoret-

Downloaded 23 Mar 2002 to 128.151.176.91. Redistribution subject to AIP license or copyright, see http://ojps.aip.org/jcpo/jcpcr.jsp

J. Chem. Phys., Vol. 114, No. 14, 8 April 2001

ical prediction.32 Interestingly, we do not resolve a distinct band arising from a transition to the (2) 2 A ⬘ state. Sodupe and Bauschlicher predict that such a transition will occur near 28 000 cm⫺1; the calculated transition moments for the (2) 2 A ⬘ (3p ␲ ) and (1) 2 A ⬙ (3 p ␲ ) states suggest that the corresponding bands will have nearly the same intensity. Figure 4 shows that within the error limits of the theoretical calculations, these two transitions may be overlapping. The monomer spectrum reported by France et al. over the spectral range from 27 000 to 33 000 cm⫺1 is taken with colder monomers than we produced, and with sufficient resolution to resolve vibrational structure attributed to the Mg⫹ –O stretch in the excited electronic states.33 This high resolution spectrum, the envelope of which is in excellent agreement with the lower energy band of the data in Fig. 4, is assigned to transitions to the (2) 2 A ⬘ (3 p ␲ ) and (1) 2 A ⬙ (3p ␲ ) states. The reported spectrum does not include the wavelength region near 38 000 cm⫺1 where a transition to the (3) 2 A ⬘ (3p ␴ ) state can be accessed. The spectral features that are assigned to the (2) 2 A ⬘ (3 p ␲ ) band have intensities that are less than expected on the basis of the transition moments reported by Bauschlicher et al. The spectrum we report is of broader spectral coverage, but of resolution insufficient to distinguish the (2) 2 A ⬘ (3 p ␲ ) and (1) 2 A ⬙ (3p ␲ ) states. In assigning the spectra of the larger clusters it is helpful to refer to the experimental work of Misaizu et al.12 on the related Mg⫹共H2O兲n systems, and supporting calculations. Qualitatively, the photodissociation spectra are strikingly similar to our Mg⫹共CH3OD兲n spectra. In both systems, two obvious bands are present in the monomer, and these bands coalesce into a single band with increasing cluster size. Misaizu et al. explain their Mg⫹共H2O兲n results in terms of their own ab initio calculations4 as well as those performed by Bauschlicher and co-workers.27,29,32,44 We use calculated structures of Mg⫹共CH3OH兲n with n⫽1 – 5 recently published by Lu and Yang9 in a similar manner. The structure of lowest potential energy for the dimer is found to have both methanol ligands on the same side of the metal. This geometry is very different from clusters that are centered on Sr⫹ and ligated by similar molecules. Unlike the heavier alkaline earth metals, Mg⫹ lacks low-lying d orbitals and does not undergo the sd hybridization that allows a linear O–M–O bond through reduction of metal–ligand repulsion.29 Instead, the electron is polarized away from the ligands in sp hybridization, leading to a bent configuration in both Mg⫹共H2O兲2 共Ref. 44兲 and Mg⫹共CH3OH兲2. 9 As calculations of the vertical transition energies in the Mg⫹共H2O兲1.2 systems have shown,29,32 this bent geometry lowers the monomer (2) 2 A 1 (3p ␴ ) and (1) 2 B 2 (3 p ␲ ) states in energy by a similar amount, but raises the (1) 2 B 1 (3 p ␲ ) state in energy. While calculations of the transition energies in Mg⫹共CH3OH兲2 are unavailable, the structure and photodissociation spectrum of this species are very similar to Mg⫹共H2O兲2. Thus we believe the electronic structure of Mg⫹ is perturbed by H2O and CH3OD in a similar fashion. Examination of our dimer spectra shows that bands assigned to p ␴ and p ␲ states in the monomer are shifted by 8000 and 6000 cm⫺1, respectively, in the dimer. Close inspection of

Reactivity of Mg⫹

6185

the high energy band in Mg⫹共CH3OH兲2 in Fig. 5 shows a small shoulder near 28 000 cm⫺1. Bauschlicher et al. found that the (1) 2 B 1 (3p ␲ ) state shifts 200 cm⫺1 to the blue in the dimer, while the (1) 2 B 2 (3p ␲ ) and (2) 2 A 1 (3p ␴ ) states red shift 2700 and 3900 cm⫺1, respectively. Therefore, the shoulder may correspond to a slightly blueshifted (2) 2 A ⬘ p ␲ -like state of the monomer, while the lowest energy band at 24 000 cm⫺1 corresponds to a redshifted (1) 2 A ⬙ p ␲ -like state. We assign the highest energy band at 30 000 cm⫺1 to a redshifted (3) 2 A ⬘ state of p ␴ character. The trimer structure calculations of Lu and Yang find that configurations of C 3 v and C s symmetry are of similar energy.9 In the C 3 v structure, each oxygen atom is bound directly to the metal, while in the C s configuration, the third ligand binds to the OH hydrogen atoms via a dipole–dipole interaction and is thus more removed from the metal center. Both structures place all three ligands on the same side of the metal. Even though the C 3 v isomer is more sterically hindered, it is found to be more stable by about 1.7 kcal/mol. An analogous C 3 structure was reported as the most favorable configuration for Mg⫹共H2O兲3. 44 The C s structure should produce a spectrum similar in structure and band position to that of the dimer, as the third ligand is further away from the metal and its field is screened somewhat by the first two solvent molecules. The spectra of the dimer and trimer are dramatically different however, and the red shifts from the positions of the highest energy bands are large. For these reasons, we interpret our data in terms of the more stable C 3 v symmetry structure. In C 3 v symmetry the p ␲ -like orbitals are degenerate and lie above the methanol molecules in a plane perpendicular to the symmetry axis. The p ␴ -like orbital will experience a stronger perturbation as electron density points toward the three solvent molecules. If the p ␴ -like orbital redshifts 6000 cm⫺1 from its position in the dimer, while the p ␲ -like orbitals experience little redshift because they are not as directly perturbed by the ligand field, a single band results as we observe. Misaizu and co-workers point out in their interpretation of the Mg⫹共H2O兲3 spectrum that closely spaced p←s transitions12 give rise to the single band they observe; we believe the same reasoning holds here. The tetramer and pentamer spectra each show no red shift relative to the trimer, providing significant evidence for a solvent shell closing at n⫽3. This observation agrees with Lu and Yang’s tetramer calculations, which find that the most stable configuration is a structure in which the fourth ligand is bound to a three-member solvent shell.9 This observation is also consistent with both calculation44 and experiment12 on the similar Mg⫹共H2O兲n systems, which show that the first solvent shell closes at n⫽3. The dimer and trimer exhibit ligand loss, intracluster reactions, and their combination, while in the tetramer and pentamer, all dissociative channels occur with accompanying ligand loss. As discussed in the next section, the change in the product distribution is likely a result of the solvation shell structure of the clusters, causing energetic photofragments originating from the innermost solvent molecules to induce evaporation of more weakly bound solvents from the cluster surface. The trimer and tetramer display weak bands in the ultraviolet region, above 35 000 cm⫺1. It is possible that a state

Downloaded 23 Mar 2002 to 128.151.176.91. Redistribution subject to AIP license or copyright, see http://ojps.aip.org/jcpo/jcpcr.jsp

6186

J. Chem. Phys., Vol. 114, No. 14, 8 April 2001

Lee, Sperry, and Farrar

derived from 4s or 3d metal ion configurations is redshifted by solvation into the wavelength range of our observations, just as the 3p bands are. The dominant spectral feature of the trimer, based on transitions to states with substantial 3p character, is redshifted with respect to its counterpart in the monomer by 13 000 cm⫺1. An even larger redshift of electronic states based on 2 S(4s) or 2 D(3d) atomic ion configurations would be required to produce absorption in the trimer at 35 000 cm⫺1. In the tetramer, this assignment is inconsistent with evidence for a solvation shell closing at n⫽3. If electronic states of 2 P(3 p) character are minimally affected by the addition of a ligand to a second solvation shell, we would also expect that electronic states based on 2 S(4s) or 2 D(3d) atomic ion configurations would only experience small redshifts under similar circumstances, although the larger spatial extent of the 4s or 3d orbitals might modify this view. While such a possibility cannot be dismissed, significant theoretical support would be required to justify the assignment. Accessing a charge transfer 共CT兲 state is another possibility for the assignment of the weak bands in the trimer and tetramer. Bauschlicher and Partridge predict that ligand-tometal CT excitation is possible in Mg⫹共CH3OH兲, and will be accessed at 146 kcal/mol 共51 100 cm⫺1兲.27 The 3p ␴ band in the monomer shifts roughly 13 000 cm⫺1 to the red in the trimer. If the charge transfer state is lowered by this amount it should be detected near 38 000 cm⫺1, which is where the weak UV band falls in the trimer, lending support to a CT band assignment for this feature. Yeh et al. report detecting small amounts of CH3OH⫹ formed through a multiphoton process in Mg⫹共CH3OH兲, providing direct evidence that the CT state is accessible.17 The strongest argument against assigning the weak bands to a CT excitation is that the products observed from the photolysis in these bands are not CH3OH⫹ or CH⫹ 3 . Instead, the charge remains on the metal in the photoproducts. However, as Kleiber and Chen point out, it is also possible to observe metal cation/neutral ligand 共or ligand fragment兲 products after accessing the CT band.47 They list the expected dissociation pathways following CT excitation as 共 M ⫹ – L 兲 ⫹h ␯ → 共 M – L ⫹ 兲 * →M ⫹L ⫹

→M ⫹ ⫹L.

共1兲 共2兲

Process 共1兲 is a result of direct photodissociation of the CT state and process 共2兲 represents ‘‘frustrated’’ CT dissociation, in which a CT state is accessed initially, but that state then undergoes internal conversion to a state below the M⫹L⫹ asymptote, resulting in reverse CT. This process is known to occur in condensed phase organometallic photochemistry.47 Duncan and co-workers provide examples of both pathways in their photodissociation experiments on Mg⫹ 共acetone兲.17 They find excitation to a CT state produced not only a CH3CHO⫹ product, but the Mg⫹ is also formed, through process 共2兲. A theoretical investigation of the trimer and tetramer clusters in which ab initio calculations locate vertical transition energies to available excited states will be invaluable in assigning the weak bands we observe. Experimental exami-

TABLE I. Enthalpies of reaction.

Reaction ⫹

⫹

Mg CH3OH→Mg ⫹CH3OH Mg⫹共CH3OH兲2→Mg⫹共CH3OH兲⫹CH3OH Mg⫹共CH3OH兲3→Mg⫹共CH3OH兲2⫹CH3OH Mg⫹共CH3OH兲4→Mg⫹共CH3OH兲3⫹CH3OH Mg⫹CH3OH→Mg⫹OCH3 Mg⫹共CH3OH兲2→Mg⫹OCH3共CH3OH兲 Mg⫹共CH3OH兲3→Mg⫹OCH3共CH3OH兲2 Mg⫹共CH3OH兲4→Mg⫹OCH3共CH3OH兲3 Mg⫹OH→Mg⫹⫹OH Mg⫹O→Mg⫹⫹O CH4→CH3⫹H CH3OH→CH3⫹OH CH3OH→CH3O⫹H

Enthalpy 共kcal/mol兲

Reference

37.7 29.2 21.8 17.4 74.6 44 23 7.4 76 46 104.7 92 104

9 9 9 9 9 9 9 9 25 25 39 39 39

nation of the smaller clusters spectra deeper in the UV will demonstrate whether these bands are red shifted from high transition energies. If such a deep UV band is found in the monomer, knowing its transition energy will be paramount to an assignment, because the CT band position is known.32 B. Photochemical products

The products created by photolysis show a high degree of selectivity. In the monomer, the only product observed over the entire spectral range investigated is loss of a methyl fragment. This behavior is dramatically different from Sr⫹共CH3OD兲, which displayed methyl, ligand,15 and deuterium atom8 loss channels. Examining the bond enthalpies of these channels for Mg⫹共CH3OD兲 in Table I, the energetics suggest that simple solvent evaporation ( – L) should be favored over C–O bond cleavage 共–Me兲; the latter channel is still more favorable than O–D bond cleavage 共–D兲 by about 20 kcal/mol. This last channel requires at least a 26 000 cm⫺1 photon. The results demonstrate that a very selective photochemical reaction takes place to yield a nonstatistical distribution of products. The manner in which metal atoms and ions cleave bonds in simple alcohols and hydrocarbons has been the subject of several related experimental studies in other laboratories. Davis et al. investigated collisions of electronically excited Ba( 1 S 0 , 1 D 2 ) with CH3OH in crossed molecular beams, and found that BaOCH3 was produced exclusively from both electronic states of the atom, despite the fact that the BaOH product is favored energetically by 30 kcal mol⫺1.40 The potential energy barrier to insertion of the large Ba atom into the C–O bond is high enough to prevent formation of the statistically favored BaOH product through a three-center intermediate. In a related study, Kleiber and co-workers.41,42 investigated the systems Ca⫹共CH4兲 and Mg⫹共CH4兲 by photodissociation spectroscopy. The Ca⫹共CH4兲 spectrum is highly structured and solvent evaporation is the only dissociative channel. On the other hand, photodissociation of Mg⫹共CH4兲 ⫹ produces Mg⫹, MgCH⫹ 3 , and MgH in a 60:33:7 branching ratio and the spectrum is completely unstructured. The differences between these experimental results are rationalized by Wong et al. in terms of metal cation size and concomitant

Downloaded 23 Mar 2002 to 128.151.176.91. Redistribution subject to AIP license or copyright, see http://ojps.aip.org/jcpo/jcpcr.jsp

J. Chem. Phys., Vol. 114, No. 14, 8 April 2001

reaction mechanisms.43 Ab initio calculations show that insertion of Mg⫹ into the C–H bond leads to a dramatic increase in electronic energy as the C–H bond is stretched. This bond stretching attraction mechanism45 and fast internal conversion to the ground electronic state results in rapid dissociation and an unstructured spectrum. The potential energy barrier for insertion of the larger Ca⫹ species into the C–H bond results in the more favorable, but longer-lived ligand loss process, and consequently a structured dissociation spectrum. These studies identify the size of an atomic ion and its 3s-orbital occupancy as key factors governing the ability of the metal to insert into C–H bonds. The issue of s-orbital occupancy is consistent with observations by Armentrout and co-workers showing that a vacant 4s orbital available to accept electrons in a covalent interaction greatly facilitates oxidative addition in transition metals.38,46 The atomic ion size issue is less critical for insertion into C–O single bonds, which are typically 30% longer than C–H bonds. The observation of selective methyl loss in Mg⫹共CH3OD兲 is consistent with the close approach of Mg* ⫹ to the C–O bond. The lone frontier electron of the metal is in a 3p orbital, leaving the 3s orbital vacant. The s orbital of the metal then readily accepts an electron from the ␴ orbital of the C–O bond, creating a covalent Mg* ⫹ – O interaction, and drastically weakening the C–O bond. Furthermore, the 3p ␲ orbitals are of the appropriate symmetry to overlap the antibonding orbital of the C–O bond, which increases the likelihood of breaking this bond. Photodissociation studies of Sr⫹共CH3OH兲n and ⫹ Sr 共CH3OD兲n from our laboratory8,15 provide results consistent with the bond insertion picture appropriate for Mg⫹, although the accessibility of low-lying 4d-orbitals in Sr⫹ results in a more complex set of electronic states and symmetries. In the dissociation of the monomer, the product channels –L, –Me, and –H 共D兲 are observed. The results are consistent with a state-specific dissociation mechanism that favors methyl loss when orbitals of p character are accessed. The Sr⫹* 5p-like cation appears to insert into the C–O bond, and the –Me product is dominant 共80:20 ratio兲 over –L; no –H occurs upon accessing these bands. These bands are also broad and structureless, indicative of a fast dissociation process. In contrast, in the d-like bands, ligand and H loss dominate, and the spectrum shows partially-resolved vibrational structure, consistent with a slower dissociation process. The dynamics that govern dissociation processes in metal ion–solvent systems are quite complex, involving the initial Franck–Condon excitation, evolution of the system on excited state potential energy surfaces, nonadiabatic interactions among the excited states and the ground electronic state, and final product formation.47 The fact that extensive photochemical reactions involving covalent bonds in the solvent molecules accompany the expected cleavage of the much weaker electrostatic bonds between metal and solvent provides clear evidence that insertion of metal ions into solvent molecule bonds is an important route for dissociation in such systems. The role that the coupling between vibrational and electronic motion plays in the dissociation of alkaline

Reactivity of Mg⫹

6187

earth ion–molecule systems has been discussed in the context of bond-stretching attractions and their role in weakening bonds to facilitate their activation.45 Ab initio calculations on the simple Mg⫹共H2兲 and Mg⫹共CH4兲 systems42,48 have revealed specific details of such interactions, elucidating the excited state motions and couplings with the ground electronic state that govern dissociation. Both theory and experiment provide evidence consistent with insertion of the 3 p-orbital of Mg⫹* into ␴ * (H–H) or ␴ * (C–H) antibonding orbitals of the solvent molecule. Bond stretching lowers the energy for this insertion process; in the case of CH4, a steric effect based on repulsion with electrons in the CH3 radical limits the amount of bond stretching, as discussed by Breckenridge.49 Although ab initio theoretical studies of the excited states involved in C–O bond cleavage in small solvated metal ion clusters have not been reported, a number of fruitful analogies with the simpler systems can guide our discussions. Molecular orbital correlations and comparisons of the features of the potential energy surfaces governing collisions of metals ions with methanol also provide insight into critical dynamics that govern dissociation in clusters. As we have already noted, the Mg⫹* system satisfies the important criterion of s-orbital vacancy to accommodate the electron density from the cleaving bond. The accessibility of a ␴ * (C–O) antibonding molecular orbital in CH3OH and the observation of products consistent with C–O bond cleavage make an analogous interpretation of our photodissociation data reasonable. Armentrout and co-workers38 draw a close analogy between the Co⫹共CH3OH兲 potential energy surface, where the formation of the HO–Co⫹ –CH3 insertion complex is a prelude to further reaction, and that for Co⫹共H2O兲, in which O–H bond insertion governs the subsequent chemistry. We have already noted the close similarities of the photodissociation spectra and electronic states in Mg⫹共CH3OH兲 with those in Mg⫹共H2O兲. In the latter case, one can use the correlation diagram analysis of Margrave and co-workers on the Mg共H2O兲 system50 to deduce important features of the potential surface. Applying that analysis to the Mg⫹共H2O兲 system suggests that metal insertion into the O–H bond proceeds smoothly, without a potential energy barrier. Thus, the H–Mg⫹ –OH insertion complex correlates adiabatically with the ground state Mg⫹共H2O兲 electrostatic complex. Analogously, the correlation diagram argument provides support for the claim that the HO–Mg⫹ –CH3 species evolves adiabatically on the ground state surface from Mg⫹共CH3OH兲. The analogies between the electronic states, bond stretching factors, and reaction products reported in the Mg⫹共H2兲 and Mg⫹共CH4兲 systems are useful in rationalizing the spectra and product distributions in the Mg⫹共CH3OH兲 system. The exclusive production of MgOH⫹ by methyl ejection over the full wavelength range is consistent with a bond-stretching interaction picture, followed by internal conversion and dissociation governed by ground state dynamics. The fact that methyl loss is the exclusive product formation channel, despite the fact that solvent evaporation is energetically favored, suggests that the dynamics of dissociation are nonstatistical. The dissociation process is independent of the

Downloaded 23 Mar 2002 to 128.151.176.91. Redistribution subject to AIP license or copyright, see http://ojps.aip.org/jcpo/jcpcr.jsp

6188

J. Chem. Phys., Vol. 114, No. 14, 8 April 2001

initial electronic excitation, governed instead by the details of the nonadiabatic interactions between ground and excited states and by motion on the ground state surface. We now turn to the dimer and larger clusters. The dimer photolysis produces –Me, – (Me⫹D), –L, and – (L⫹D) channels. There is an intriguing dependence to the branching ratios as a function of photolysis wavelength. In the higher energy band of the dimer, the channels that eliminate a deuterium atom in conjunction with another fragment dominate. The channels that do not involve combination loss dominate in the lower energy band. The fact that there is this pairing of similar behavior between the ‘‘combination’’ channels leads us to suspect that the deuterium atom loss occurs by a similar mechanism in both these channels. In the – (L⫹D) channel, the deuterium atom must obviously come from a ligand other than the one lost by evaporation. As their wavelength dependence is similar, it is possible that the deuterium atom from the – (Me⫹D) channel is also taken from the other ligand. The product from such a reaction mechanism is a methoxy/ hydroxy solvated cation of the form H3COMg⫹OD, while a ligand solvated monoxide H3CODMg⫹O is produced by elimination of CH3D from the same ligand. Using the values in Table I, we can estimate the minimum endothermicities for production of H3COMg⫹OD and H3CODMg⫹O at 9.6 and 40.3 kcal/mol, respectively. The energetics suggest the production of H3COMg⫹OD via the double ligand cleavage mechanism is much more favorable. The evidence that the methyl fragment and deuterium atom are not lost from the same ligand aids in understanding why the – (Me⫹D) channel occurs only in the dimer. The dimer must have a unique structural element that allows the reaction to take place. The lowest potential energy dimer structure calculated by Lu and Yang9 shows that the methyl group of one ligand is placed in close proximity to the deuterium atom of the other ligand. Once a methyl radical is created by C–O bond cleavage, it is in an ideal position to abstract a deuterium atom to create CH3D. This geometry, which allows facile production of CH3D, is not found in the larger cluster structures.9 This type of concerted mechanism for the elimination of methane was proposed by Davis et al. from their work on collisions of Ba⫹CH3OH. 40 They point out that the barrier for such a reaction is high, as the C–O bond must be nearly broken for the reaction to proceed. Given the evidence for our proposed mechanism of C–O cleavage in the monomer, this requirement may well be met. Last, we turn to the products from photolysis of the trimer and the larger clusters. As mentioned, the trimer does not display the – (Me⫹D) channel, but a combination ligand and methyl loss channel is accessed in addition to the other products displayed by the dimer. Lu and Yang find that the barrier to ligand loss drops by nearly 16 kcal mol⫺1 in the trimer relative to the monomer.9 This more facile ligand loss may explain why the – (L⫹Me) channel becomes accessible in the trimer. Product trends similar to the dimer are observed in the trimer. The channels involving combination ligand and fragment loss display a similar wavelength dependence and dominate in the higher energy portion of the spectrum.

Lee, Sperry, and Farrar

In the tetramer and pentamer, the methyl loss channel is absent. The tetramer and pentamer clusters have absorption spectra very similar to that for the trimer, causing us to speculate that the larger clusters have photochemistry based on a trimer ‘‘core,’’ where C–O bond cleavage is known to occur. We would therefore expect the deposition of energy in the tetramer and pentamer to produce nascent photoproducts similar to those of the trimer. However, methyl radical emission is undetectable in the tetramer and pentamer. Although the endothermicity of the process to form MgOD⫹共CH3OD兲3 from the tetramer is not precisely known, we can make analogies with similar systems to estimate this quantity. Fuke and co-workers have shown that solvation of the metal hydroxide core in MgOH⫹共H2O兲n is significantly more energetically favorable than solvation of the metal in the corresponding Mg⫹共H2O兲n cluster, owing to the oxidation of the metal to the 2⫹ state in the hydroxide.5 In particular, the endothermicities for formation of MgOH⫹共H2O兲3⫹H and Mg⫹共H2O兲3⫹H2O from Mg⫹共H2O兲4 appear to be nearly identical. By analogy, we expect that the endothermicities for formation of MgOD⫹共CH3OD兲3⫹CH3 and Mg⫹共CH3OD兲3⫹CH3OD from Mg⫹共CH3OD兲4 should be comparable. However, one expects a significant activation barrier in excess of the endothermicity for the methyl radical channel, as we have discussed earlier. This activation barrier for C–O bond cleavage depends most sensitively on solvent molecules that interact directly with the metal center, i.e., those in the first solvation shell. Thus, we expect the activation barrier for methyl emission to lower by only a small amount as the second solvent shell assembles. With the first solvent shell closed, the fourth and fifth ligands attach to the cluster via a dipole–dipole interaction. The ligand loss barrier is sufficiently low, ⬃17 kcal mol⫺1,9 that decay through this channel appears to be the exclusive route to products. Interestingly, D-atom ejection from O–D bond cleavage, the barrier for which appears to lower with increasing cluster size, is still observed in clusters with a developing second solvation shell, with evaporation continuing to be the dominant dissociation channel. V. CONCLUSIONS

Our studies of Mg⫹共CH3OD兲n clusters have revealed interesting facets of their chemistry in the ground and excited states. In the ground state, the parent ions eliminate H 共and possibly D兲 atoms, this process becoming more facile with increasing cluster size. Spectroscopy of these clusters demonstrates that photon absorption to states based on excitation of p-orbitals on the metals is followed by nonadiabatic crossing to ground state potential energy surface, where specific dynamics govern the dissociation and product distributions. Such dynamics result in size-dependent, nonstatistical product formation, exemplified by selective C–O bond cleavage and concomitant methyl group ejection in the monomeric species. Methyl ejection decreases in importance with increasing cluster size, effectively halting when the first solvation shell forms at n⫽3, evidence for which comes both from our spectroscopic studies as well as theoretical structure calculations. Deuterium atom ejection resulting from O–D bond cleavage becomes increasingly important as the clus-

Downloaded 23 Mar 2002 to 128.151.176.91. Redistribution subject to AIP license or copyright, see http://ojps.aip.org/jcpo/jcpcr.jsp

Reactivity of Mg⫹

J. Chem. Phys., Vol. 114, No. 14, 8 April 2001

ters become larger, persisting to clusters with n⫽4 and 5. Several points of evidence lead us to conclude that a unique, concerted chemical reaction occurs in the dimer, in which CH3 and D are eliminated by bond scissions in two distinct ligands. The results suggest a number of topics for further theoretical study. The details of geometries and transition states for metal ion insertion into the C–O bond, possible surface crossings to the ground state, and the nature of the nonadiabatic couplings are all topics requiring further understanding. We hope the present data will encourage those studies. ACKNOWLEDGMENTS

We gratefully acknowledge support of this work under National Science Foundation Grant No. CHE-095-23401. Acknowledgment is also made to the Donors of the Petroleum Research Fund, administered by the American Chemical Society, for the partial support of this research. 1

A. W. Castleman, Jr. and K. H. Bowen, Jr., J. Phys. Chem. 100, 12911 共1996兲. 2 H. W. Kroto, J. R. Heath, S. C. O’Brien, R. F. Curl, and R. E. Smalley, Nature 共London兲 18, 162 共1985兲. 3 S. Wei, G. B. C. , J. Purnell, S. Buzza, and A. W. Castleman, Jr., Science 256, 818 共1992兲. 4 F. Misaizu, M. Sanekata, K. Tsukamoto, and K. Fuke, J. Phys. Chem. 96, 8259 共1992兲. 5 M. Sanekata, F. Misaizu, and K. Fuke, J. Chem. Phys. 104, 9768 共1996兲. 6 M. Sanekata, F. Misaizu, K. Fuke, S. Iwata, and K. Hashimoto, J. Am. Chem. Soc. 117, 747 共1995兲. 7 D. C. Sperry, A. J. Midey, J. I. Lee, J. Qian, and J. M. Farrar, J. Chem. Phys. 111, 8469 共1999兲. 8 J. I. Lee, D. C. Sperry, A. J. Midey, J. Qian, and J. M. Farrar 共in progress兲. 9 W. Lu and S. Yang, J. Phys. Chem. A 102, 825 共1998兲. 10 T. J. Selegue and J. M. Lisy, J. Am. Chem. Soc. 116, 4874 共1994兲. 11 J. A. Draves and J. M. Lisy, J. Am. Chem. Soc. 112, 9006 共1990兲. 12 F. Misaizu, M. Sanekata, and K. Fuke, J. Chem. Phys. 100, 1161 共1994兲. 13 M. H. Shen and J. M. Farrar, J. Chem. Phys. 94, 3322 共1991兲. 14 S. G. Donnelly and J. M. Farrar, J. Chem. Phys. 98, 5450 共1993兲. 15 J. Qian, A. J. Midey, S. G. Donnelly, J. I. Lee, and J. M. Farrar, Chem. Phys. Lett. 244, 414 共1995兲. 16 L. N. Ding, M. A. Yong, P. D. Kleiber, W. C. Stwalley, and A. M. Lyrra, J. Phys. Chem. 97, 2181 共1993兲. 17 C. S. Yeh, K. F. Willey, D. L. Robbins, and M. A. Duncan, Int. J. Mass Spectrom. Ion Processes 131, 307 共1994兲. 18 C. S. Yeh, K. F. Willey, D. L. Robbins, J. S. Pilgrim, and M. A. Duncan, Chem. Phys. Lett. 196, 233 共1992兲. 19 C. T. Scurlock, S. H. Pullins, J. E. Reddic, and M. A. Duncan, J. Chem. Phys. 104, 4591 共1996兲.

6189

20

K. F. Willey, C. S. Yeh, D. L. Robbins, J. S. Pilgrim, and M. A. Duncan, J. Chem. Phys. 97, 8886 共1992兲. 21 I. N. Tang, M. S. Lian, and A. W. Castleman, Jr., J. Chem. Phys. 65, 4022 共1976兲. 22 A. C. Harms, S. N. Khanna, B. Chen, and A. W. Castleman, Jr., J. Chem. Phys. 100, 3540 共1994兲. 23 E. Kochanski and E. Constantin, J. Chem. Phys. 87, 1661 共1987兲. 24 E. Murad, J. Chem. Phys. 78, 6611 共1983兲. 25 E. Murad, J. Chem. Phys. 75, 4080 共1981兲. 26 C. W. Bauschlicher, Jr., S. R. Langhoff, H. Partridge, J. E. Rice, and A. Komornicki, J. Chem. Phys. 95, 5142 共1991兲. 27 C. W. Bauschlicher, Jr. and H. Partridge, J. Phys. Chem. 95, 3946 共1991兲. 28 M. Sodupe, C. W. Bauschlicher, Jr., and H. Partridge, J. Chem. Phys. 95, 9422 共1991兲. 29 C. W. Bauschlicher, Jr., M. Sodupe, and H. Partridge, J. Chem. Phys. 96, 4453 共1992兲. 30 M. Sodupe, C. W. Bauschlicher, Jr., and H. Partridge, Chem. Phys. Lett. 192, 185 共1992兲. 31 M. Sodupe and C. W. Bauschlicher, Jr., Chem. Phys. Lett. 212, 624 共1993兲. 32 M. Sodupe and C. W. Bauschlicher, Jr., Chem. Phys. Lett. 195, 494 共1992兲. 33 M. R. France, S. H. Pullins, and M. A. Duncan, Chem. Phys. 239, 447 共1998兲. 34 C. A. Woodward, M. P. Dobson, and A. J. Stace, J. Phys. Chem. A 101, 2279 共1997兲. 35 S. G. Donnelly, C. A. Schmuttenmaer, J. Qian, and J. M. Farrar, J. Chem. Soc., Faraday Trans. 89, 1457 共1993兲. 36 D. E. Powers, S. G. Hansen, M. E. Geusic, D. L. Michalopoulos, and R. E. Smalley, J. Chem. Phys. 76, 2866 共1983兲. 37 W. C. Wiley and I. H. McLaren, Rev. Sci. Instrum. 26, 1150 共1955兲. 38 Y.-M. Chen, D. E. Clemmer, and P. B. Armentrout, J. Am. Chem. Soc. 116, 7815 共1994兲. 39 CRC Handbook of Chemistry and Physics, 77th ed., edited by D. R. Lide 共CRC Press, Boca Raton, FL, 1997兲. 40 H. F. Davis, A. G. Suits, Y. T. Lee, C. Alcaraz, and J.-M. Mestdagh, J. Chem. Phys. 98, 9595 共1993兲. 41 J. Chen, Y. C. Cheng, and P. D. Kleiber, J. Chem. Phys. 106, 3884 共1997兲. 42 Y. C. Cheng, J. Chen, L. N. Ding, T. H. Wong, P. D. Kleiber, and D.-K. Liu, J. Chem. Phys. 104, 6452 共1998兲. 43 T. H. Wong, C. Freel, and P. D. Kleiber, J. Chem. Phys. 108, 5723 共1998兲. 44 C. W. Bauschlicher, Jr. and H. Partridge, J. Phys. Chem. 95, 9694 共1991兲. 45 I. V. Hertel, in Dynamics of the Excited State, Advances in Chemical Physics, edited by K. P. Lawley 共Wiley–Interscience, New York, 1982兲, Vol. L, p. 475. 46 P. B. Armentrout, Annu. Rev. Phys. Chem. 41, 313 共1990兲. 47 P. D. Kleiber and J. Chen, Int. Rev. Phys. Chem. 17, 1 共1998兲. 48 C. W. Bauschlicher, Jr., Chem. Phys. Lett. 201, 11 共1993兲. 49 W. H. Breckenridge, J. Phys. Chem. 100, 14840 共1996兲. 50 M. A. Douglas, R. H. Hauge, and J. L. Margrave, High. Temp. Sci. 17, 201 共1984兲.

Downloaded 23 Mar 2002 to 128.151.176.91. Redistribution subject to AIP license or copyright, see http://ojps.aip.org/jcpo/jcpcr.jsp