2616
J . Phys. Chem. 1991, 95, 2616-2623
AI( 2P)(S1H,} Complex and Photoreversible OxMatlve Addltion/Reductlve Eliminath Reactlon AI('P){SIH,} * SIH3AIH. 1. Al(2P){SlH,) Complex Michael A. Lefcourt and Geoffrey A. Ozin* Lash Miller Chemical Laboratories, University of Toronto, 80 St. George St., Toronto, Ontario, Canada M5S IAl (Received: April 26, 1989; In Final Form: August 9, 1990) Matrix-isolation investigations were carried out on samples consisting of individual aluminum atoms dispersed throughout silane (SiH,)-containing matrices and other related substrate materials at 12 K. UV-vis and EPR spectroscopies helped in the determination of a ground-stote complex between zP AI atoms and SiH4 molecules appearing upon matrix formation. The substitution of SiD, for SiH4 caused narrowing of the absorptions associated with aluminum in both the experimental techniques employed, giving further support for the existence of the ground-state complex. Further characterization of the complex was aided by results obtained from EPR spectral simulations and a series of ab initio self-consistent-field geometry optimization calculations. The EPR spectral simulations allowed g values, hyperfine splittings, and atomic orbital spin densities to be reported for the complex along with the ab initio results which supported the choice of a three-center, two-electron AI(v2-SiH4)side-on "agostic" interaction scheme as the most probable bonding picture for the aluminum/silane ground-state complex.
Introduction Since the important pioneering matrix isolation work by Turner and co-workers' in which interactions of the coordinatively unsaturated chromium species Cr(CO)5 with various ligands including methane were observed, several other matrix studies yielding similar phenomena have been r e p ~ r t e d . ~ With - ~ respect to ligand-free matrix-isolated metal atom systems involving saturated hydrocarbons, however, observations of ground-state adducts are extremely rare. In fact, only three papersm report the existence of any kind of ground-state reactivity between matrixisolated metal atoms and methane, all of which have drawn considerable skepticism including later refuting evidencee" concerning two of theme6*' This controversy centered around the reported observation6,' of ground-state aluminum atom insertion (as opposed to complexation as in adduct formation) into a C-H bond of methane, an event that was not reproducible in later work."' The open-shell ground-state configuration of aluminum, 3s23p', although orbitally equivalent to the excited states of other atoms that are known to be quenched in the presence of methane, was shown not to be responsible for the incorporation of aluminum into a C-H bond of methane. The photolytically generated zS state of aluminum (3s24s1),however, was shown to efficiently insert into the C-H bond of methane, presumably via an oxidative addition mechanism. The third reference cited8 is the only reported case of adduct formation between an isolated ground-state metal atom and methane. In this investigation iron atoms and dimers were reported as forming hydrogen-bonded complexes with methane in argon matrices. As well, the photogenerated insertion product methyliron hydride (CH3FeH) was proposed to form via the ground-state Fe(CH4) complex. Unfortunately, however, a number of discrepancies exist between this work and other, more thorough examinations of thisI2 and a related system (Mg'O). In light of the unusual reactivity of aluminum atoms toward the C-H bonds of methane, along with the deeper concern regarding the determination of the agostic-type bonding presumed to exist in metal/alkane reactive intermediates, we decided to (1) Turner, J. J.; Burdett, J. K.; Perutz, R. N.; PoIiakoff, M . Pure Appl. Chem. 1977, 49, 27 1. (2) Kasai, P. H. J . Am. Chem. Soc. 1982, 104, 1165. (3) Howard, J. A.; Mile, B.; Tse, J. S.; Moms, H. J. J . Chem. Soc.,Dolron Trans. 1987,83, 3701. (4) Kasai, P. H.; McLeod, D., Jr. J . Am. Chem. SOC.1979, 101, 5860. ( 5 ) Chenier, J. H. B.; Howard, J. A.; Mile, B. J . Am. Chem. SOC.1987, I09, 4109. (6) Klabunde, K. J.; Tanaka, Y . J . Am. Chem. SOC.1983, 105, 3544. (7) Jeong, G.;Klabunde. K. J. J . Am. Chem. Soc. 1986, 108, 7103. (8) Kafafi, 2.H.; Hauge, R. H.; Margrave, J. L. J . Am. Chem. Soc. 1985,
introduce a "twist" to the established alkane activation scenario found in the literature. On the basis of what was known about the metal/methane matrix system studied to date,we anticipated that the longer, weaker silicon-hydrogen bonds of silane (SiH4) along with the availability of the empty 3d silicon atomic orbitals of silane would open up favorable electronic and structural channels for enhanced interactions between ground-state metal atoms and the substrate as compared to the situation for methane. Such circumstances seemed appropriate, especially in view of the potential ground-state reactivity of aluminum, to witness A1 atom/silane molecule binding.
Experimental Details The basic apparatus used in the matrix isolation experiments has been described previ0us1y.l~ In brief, aluminum atoms and molecular matrix gases were deposited on cryogenically cooled spectroscopic surfaces (NaCl for UV-vis, CsI for infrared, and a sapphire rod for EPR measurements) positioned on the tips of helium-cooled cryostats. These cold tips were individually housed inside separate vacuum chambers operating in the range (1-5) X lo-' Torr when optimum cold tip temperatures were attained (12 K). The metal was monatomically vaporized from a length of aluminum wire (A.D. McKay, 0.020-in. diameter, 99.9% pure) wrapped around a tantalum filament (A.D. McKay, 0.010-in. thickness, 99.95% pure). The filament was positioned across two electrodes inside a small furnace allowing resistive heating of the aluminum wire to take place, thus causing the aluminum to vaporize. Removal of aluminum oxide impurities on the surface of the metal (prior to matrix deposition) was accomplished by low-temperature heating of the wirswrapped filament. The matrix gases used for the experiments in the current study along with those in part 2 of this investigation include SiH4 (Matheson, semiconductor purity), SiD4 (Merck Sharpe and Dohme, 99.3% isotopic purity), CH3SiH3 (Petrarch Systems), and argon (Matheson, 99.9995% purity). The following spectrometers were employed for the UV-vis, IR, and EPR, respectively: a PerkinElmer 330 model (2500-190-nm range with resolution of 0.3 nm), a Perkin-Elmer 180 grating instrument (4000-200-cm-' range with resolution of 1.0 cm-I), and an X-band Varian E4 instrument (0-6000-G range with resolution of 0.01 G). Two theoretical software packages residing on a Gould 9705 minicomputer were used for data analysis and interpretation. The first of these was an EPR spectral simulation package (SIM14Ak4) that enabled the online graphical superposition of digitized experimental data with computationally generated spectra in order
107, 6134.
(9) Parnis, J. M.;Ozin, G. A. J . Am. Chem. Soc. 1986, 108, 1699. (IO) Parnis, J. M. Ph.D. Thesis, University of Toronto, 1987. ( I 1) Parnis, J. M.;Ozin,G. A. J. Phys. Chem. 1989, 93, 1204. (12) Ozin, G. A.; McCaffrey, J. G. J . Am. Chem. Soc. 1982, 104, 7351.
0022-3654/91/2095-2616$02.50/0
(13) Parnis, J. M.; Mitchell, S. A,; Garcia-Prieto, J.; Ozin, G. A. J. Am. Chem. SOC.1985, 107, 8169. (14) SIM14A: h z o s , G.; Hoffman, B.; Franz, C. Northwestern University Chemistry Department, Evanston, IL.
0 1991 American Chemical Society
The Journal of Physical Chemistry, Vol. 95, No. 7, 1991 2617
A1(2P)(SiH4]Complex
TABLE I: UV-Visible Electronic Assignments for the AI/Ar, AI/CH,, and AI/SiH, Matrix Systems origin/transition AI (2S+2P)
A11(2D+2P)
~
29 1 281 285
i3 ~ g~- - 3 ~( , ) 399 266
AI/SIHq
260
360
460
s ~ O 660
7a0
(&nm
Figure 1. UV-vis spectra of aluminum atoms deposited in each of pure argon (A), methane (B; from ref IO, p 248), and silane (C) matrices at 12 K. The respective metal/substrate dispersion ratios for these samples were 1:2250, 1:4500, and 1:5000.
that accurate EPR spectral parameters ( g values, A values) could be extracted from the best-fit computations. The second interpretive computer package used was a set of ab initio self-consistent-field molecular orbital routines that enabled geometry optimizations to be carried out with a variety of basis sets. Known as the MONSTERGAUSS set of program^,'^ the package consisted of STO-3G, 3-21G*, 6-31G*, and 6-31G** basis sets" among others and included the choice of either restricted or unrestricted Hartree-Fock (RHF, UHF) formalisms for the geometry calculations. Results A. UV-Visible Absorption Studies. Figure IC shows the UV-vis spectrum obtained when aluminum atoms are individually isolated in an otherwise pure silane matrix. The dilution ratio of metal atoms to substrate molecules indicated in the figure caption (1 5000) reasonably ensured that the atoms were monomeric upon matrix formation at the typical operating temperatures of between 10 and 12 K. Supporting spectra of aluminum atoms isolated in pure argon matrices with metal/substrate ratios higher than this valueI8 revealed only very small amounts of dimeric metal compared to the quantities of individual atoms present on deposition. The total mass of metal deposited in the sample of Figure 1 C was approximately 6 kg while the amount of silane gas used was roughly 1.1 mmol. The length of time for the deposition was 66 min. As it can be seen from this spectrum several broad features are evident. The lowest energy feature, which extends roughly 100 nm between 450 and 350 nm, appears to contain two components: an intense band centered at 383 nm and a second, slightly less intense absorption at 430 nm. The higher energy absorption region (15) Monstergauu programs resident on the U. of Toronto Gould 9705 minicomputer: Peterson, M. R.; Pokier, R. A. University of Toronto Chemistry Department, 1981. Program components: GAUSSIAN 80 integration,I6 SCF and analytical energy gradient routines, and a number of geometry optimization routines. (16) GAUSSIAN 8 0 Binkley, J. S.;Whiteside, R. A.; Krishnan, R.; Seeger, R.;DeFrees, D. J.; Schlegel. H. B.; Topiol, S.;Kahn, L. R.;Pople, J. A. Carnegie-Mellon University Chemistry Department, Pittsburgh, PA. (17) STO-3G: Hehre, W. J.; Stewart, R. F.; Pople, J. A. J . Chem. Phys. 1969, 51, 2657. 3-21G: Binkley, J. S.;Pople, J. A.; Hehre, W. J. J . Am. Chem. Soc. 1980, 102,939. 6-31G.: Hariharan. P. C.; Pople, J. A. Chem. Phys. feu. 1972, 16, 217. 6-31G**: Hariharan, P. C.; Pople. J. A. Theor. Chim. Acta 1973, 28, 213. (18) This Work.
matrix band positions, nm AI/CH,b AI/SiH,f 388O 430 36ad 383 34V 332' 31od 284 30V 25 1 302d 295e 700 406 402 283
AI/Ara 372 (sh) 338
'As reported in ref 20. b A s reported in ref 10. CUnstable minor sites. dMajor site. CMinorsite. fBand positions only correlated with Ar and CHI matrices; electronic assignments are not implied.
consists of what looks like a relatively sharp, narrow absorption at 251 nm superimposed with that of a second, broad feature with a maximum or well-defined shoulder at 284 nm. If the spectral features of this spectrum are roughly divided into three g r o u p s o low-energy group containing the two components in the 450350-nm range and two higher energy groups each containing one of the two maxima below 300 nm-rough correlations can be attempted with known aluminum atom gas-phase transition^.'^ The lowest energy gas-phase value of 394.5 nm19 corresponding , 3/2, transition to the AI atom ( 3 ~ ~ 4 ~ ( ~3 ~)~~3 ps' ) ~(jP= falls within the region defining the low-energy group and is very close to that of one of the component maxima (383 nm). In the cryogenic rare gas matrices, Ar, Kr, and Xe, the 2S 2P aluminum atom transition has been shown to occur at 338,362, and 370 nm, respectively20 (values for stable AI atom trapping sites only). The extremely broad dual-component feature of Figure 1 that essentially encompasses these rare gas reported wavelengths along with the gas-phase value seems to logically correspond to the atomic excitation in question, especially when one considers that the transition results in the population of the highly diffuse Al atom 4s orbital which is much more susceptible to perturbations from the surrounding matrix cage2' than aluminum's lower lying orbitals. The next absorption in order of increasing energetics is much more difficult to correlate with gas-phase or rare gas matrix data. The band in question is the one evident at 284 nm, mostly obscured by the more intense superimposed absorption centered at 251 nm. Three gas-phase A1 atom transitions (not considering differences in j quantum numbers) lie in the spectral region between 309 and 265 nm inclusively, but none of the three is sufficiently close to the 284-nm position. The lowest of these gas-phase transitions, that corresponding with the ( 3 ~ ~ 3 d ' ) ~ D ( 3 ~ ~ 3 p ' transition, )~P which lies at 308.3 nmI9 (for all j values at this precision), has been found to correlate with the following 3-fold split absorptions in Ar, Kr, and Xe matrices: 291, 287, and 285 nm in Ar; 310, 303, and 298 nm in Kr; and 326, 319, and 316 nm in Xe.Zo Although the absorptions in the argon matrix are reasonably close to the 284-nm maximum in our AI/SiH4 system, the fact that these bands are red-shifted for the larger rare gas hosts (in accord with increasingly expanded matrix cages and decreasing repulsive interactions for increasingly larger substrates) runs counter to the silane situation in which a molecule considerably larger than argon has exhibited a blue shift with respect to this band's position in the rare gas matrices. Likewise, the highest lying absorption recorded in the Al/SiH4 matrix, that at 251 nm, is equally perplexing in that it seems very doubtful that it can correspond to any of the gas-phase transitions mentioned in reference to the 284-nm matrix absorption or to any of the even higher excited
-
-
-
(19) Moore, C. E. Atomic Energy Levels. Nurl. Bur. Stud. Circ. (U.S.) 1949, 1, 467. (20) Abe, H.; Kolb, D. M. Ber. Bunsen-Ges. Phys. Chem. 1983,87, 523. (21) Ammeter, J. H.; Schlosnagle, D. C. J . Chem. Phys. 1973, 59, 4784.
2618 The Journal of Physical Chemistry, Vol. 95, No. 7, 1991
Lefcourt and Ozin
- AI / (SIH4/Ar - 1/10) _ _ - - -AI / (SID4/Ar AI / (SIH4/Ar
- 1/100)
AI / (SIHq/Ar
- 1/10)
AI / SIH4 (pure)
ow 360
400
SW
(00
IW
8Wnm
Figure 2. UV-vis spectra of aluminum atoms deposited in pure and mixed silane/argon matrices (12 K): in pure Ar (A), in a 1:lOO SiHJAr matrix (E),in a 1 : l O SiH4/Ar matrix (C), and in pure SiH4 (D). The respective metal/substrate dispersion ratios for these samples were 1:2250, 1:7500, 1:7500, and 15000.
states listed in the comprehensive tables.19 The fact that the UV-vis absorptions in the Al/SiH4 matrix system are radically different insofar as wavelength positions (with respect to those in the gas phase and rare gas matrices) and bandwidths (compared to those in the rare gases) dictates that atomic conditions no longer exist and that the aluminum atom must form a chemical entity with the silane molecule that is effectively trapped in the cryogenic environment. This postulation is brought to the fore when the Al/SiH4 spectrum of Figure 1C is compared with the UV-vis spectra of both aluminum in argont8 and methaneto*"as displayed in Figure lA,B. Here, the dramatic contrast between the sharp and intense atomic bands of aluminum in these latter two gases and the broad structured absorptions in silane is readily observed. The seemingly low dilution ratio for aluminum in argon (see figure caption) was shown not to have an effect on the monoatomic dispersion of the metal throughout the matrix as the band assignments for A1 atoms vs A1 dimers clearly show in Table I. This table also shows the transitions for AI in methane alongside the meager list of AI/SiH4 band assignments. In both Ar and CHI the energies associated with the matrix splittings of degenerate levels and the existence of various trapping sites are indicated as well. In order to elucidate this somewhat unusual deviation from recognizable atomic spectra in the Al/SiH4 system, we performed a concentration study in which the silane matrix content was allowed to vary. The results of these investigations are shown in Figure 2. It seems somewhat apparent from these spectra that, upon the deposition of aluminum atoms into progressively doped SiH4/Ar matrices passing from pure argon through to pure silane a fairly smooth transformation exists. In the sample of spectrum B,where the silane is present in high dilution (with respect to the argon), spectral features corresponding to both extremes are evident. The ) ~ atom P band at 340 nm is present sharp ( 3 ~ ~ 4 ~ ) ()3~~ s~ 3 p ' AI both in this spectrum and in the silane-free sample of spectrum A; however, in spectrum B this absorption appears along with a broad low-energy shoulder that corresponds very well with the low-energy absorptions observed in spectra C and D. Comparison of these latter two spectra with each other reveals extremely similar absorption patterns with the main exception to their resemblance being the red shift of all the spectral components in spectrum D. The origin of these spectral shifts most likely arises from matrix cage expansion due to the complete replacement of the nearestneighbor argon atoms of spectrum C with silane molecules. The close resemblance between the two spectra strongly suggests that
-
-
1/10)
250 3do 460 sdo crdo 760 8d0nm Figure 3. UV-vis spectrum of aluminum atoms deposited in a 1:lO SiH,/Ar matrix (12 K) superimposed with a spectrum obtained from a similarly prepared sample in which perdeuteriosilane (SiD4)was used instead of SiH4.
the mechanism responsible for the unusual optical bands has to essentially dominate once one silane molecule occupies a cage position around an aluminum atom. The ratio of silane to argon in spectrum C dictates that, in the case of a statistical dispersion of silane molecules throughout the sample, one SiH4molecule per 12-membered tetradecahedral cagezt would be the prevailing situation. Further spectral evidence for the existence of an aluminum atomsilane ground-state complex is shown in Figure 3, which displays an enlargement of the Al/( 1:10 SiH4/Ar) spectrum from Figure 3 superimposed with a spectrum from a sample in which the silane was replaced with perdeuteriosilane, SiD4. It is evident that in the presence of SiD4 there is a definite narrowing in all of the observed optical bands. The narrowing was found to be significant: approximately 475 cm-' in total for the broad lowenergy feature (the apparent two overlapping bands inclusive) and roughly 150 cm-' in total for the single Bharp high-energy band (centered at 251 nm in the ordinary silane matrix and a t 244 nm in SiD4). Determination of these narrowings involved the manual measurement of full width at half-maximum (fwhm) values while taking into account the existence of sloping baselines. The importance of the band narrowing found here for the SiD,-containing sample is that it constitutes convincing evidence for an aluminumsilane interaction in which some degree of binding takes place since the narrowing is characteristic of a change in the vibronic coupling of the species giving rise to the absorptions. The differing frequencies of the normal modes of vibration for SiH4and SiD4 can only manifest themselves in the UV-vis band narrowing (due to the diminished vibrational spacings of the absorptions as well as a possible shift in the Franck-Condon maximum) if these molecular species form complexes of some sort with the monatomically isolated aluminum. The type of interaction most easily chosen to describe the system would be that of charge transfer since 3d orbitals are present on both the A1 atom and the Si atom of the silane molecule along with partially filled 3p orbitals. Charge transfer could therefore m u r from the metal to the substrate via a 3p to 3d orbital donation. However, without a t least a qualitative knowledge of atomic orbital spin densities (derived from EPR spectral data; see section B), it is somewhat premature to attempt to identify the nature of the aluminumsilane interaction. Before leaving the discussion of Figure 3 it should be added that, in the absence of deposited metal, the matrix gas in both cases (SiH4 and SiD4) diluted in argon (1:lO) did not yield any UV-vis absorptions. B. Electron Paramagnetic Resonance Studies. Chemical systems containing aluminum atoms give rise to EPR signals since the metal is paramagnetic with one naturally occurring nAl htope having a nuclear spin of 5/2. The gas-phase aluminum 2Pground state is triply degenerate however, a situation which would presumably result in very short spin-lattice relaxation times ( T I )for aluminum atoms in the solid state, particularly in rare gas matrices where perturbations of the orbitals of the metal are expected to be minimal. The most significant result arising from fast spinlattice relaxation from the perspective of the EPR experiment is
The Journal of Physical Chemistry, Vol. 95, No. 7 , 1991 2619
AI(*P)(SiH,} Complex
A
3180
3380
JS'MQAUSS
Figure 4. EPR spectra of aluminum atoms deposited in each of pure argon (A), methane (B; from ref 10, p 280), and silane (C) matrices at 12 K. Hyperfine field positions along with gHand g, values for the AI/Ar and AI/CH4 samples are indicated by stick spectra. Asterisk indicates methyl (CH,)radical impurity bands while denotes silyl (SiHJ radical bands. The total quantity of metal present in each sample was approximately IO times the amount used in the analogous UV-vis preparations or between roughly 60 and 100 p g . The respective metal/substrate dispersion ratios for these samples were approximately 1:3200, 1:6500, and 1:3200.
a high degree of broadening inherent in the resonance bands which, in some cases, causes the bands to be unobservable. Although it has been shown that aluminum atoms in rare gas matrices are in fact amenable to spectroscopic observation via EPR spectrmpy,l0J' early reports indicated that this was not the case.2223 Knight and Weltner2) attributed the presence of EPR transitions in aluminum/rare gas matrices (prepared from pure AI and A1203) to the A1-0 molecule at low metal concentrations and to an unknown weakly bound molecular species AI-X when higher concentrations of dopant were used. Due to the rather detailed experimental and theoretical analyses carried out by Ammeter and Schlosnagle,21however, the EPR transitions observed in the earlier study were designated as arising from single aluminum atoms occupying axially symmetric rare gas matrix cage sites. It was determined that the orbital angular momentum associated with the free 2P gas-phase atom is almost completely quenched in the solid due to the presence of a strong ground-state matrix cage interaction which relegates the unpaired electron to the relatively unperturbed AI 3p, orbital. The axiality of the substitutional site was proposed as originating from the dynamic Jahn-Teller effect, a conclusion that has since been substantiated by a recent magnetic circular dichroism study of rare gas-isolated aluminum atoms.24 The use of silane in our EPR studies did not cause any advance concern with respect to overly extensive line broadening from spin-lattice relaxation since well-resolved spectra of aluminum atom resonances in the rare gases mentioned above and in the methane matrices have been shown to be easily obtainable.lOJl Figure 4 shows the EPR spectra, obtained in derivative mode, of monatomic aluminum isolated in each of pure Ar, CH4,Ioand SiH4 matrices at 12 K. As in the corresponding UV-vis spectra (Figure l ) , the EPR results display noticeable similarities between the AI/Ar and AI/CH4 systems but a dramatic contrast when these results are compared with those obtained for the AI/SiH4 system. In pure Ar (spectrum A in Figure 4), the aluminum atoms (27Al, I = 5/2, natural abundance = 100%)exhibit an axial hyperfine (22) Jen, C. K.;Foner, S.N.; Cochran, E. L.; Bowers, V . A., Phys. Rea 1958, 112, 1169. (23) Kn.ight, L. B.,Jr.; Weltner, W., Jr. J . Chem. Phys. 1971.55, 5066. (24) Grinter, R.; Singer, R. J. Chem. Phys. 1987, 113, 87.
3180
3380
3580 OAUSS
Figure 5. EPR spectra of aluminum atoms deposited in pure and mixed silane/argon matrices (12 K): in pure Ar (A), in a 1:IOO SiH4/Ar matrix (B), in a 1:lOO SiH4/Ar matrix (C), and in pure SiH, (D). Hyperfine field positions along with g, and g, values for the AI/Ar sample are indicated by stick spectra. The respective metal/substrate dispersion ratios for these samples were 1:3200, 1:3600, 1:3900, and 1:3200.
sextet, the line positions of which are indicated by stick spectra. The four sharp components denoted by the asterisks are believed to be due to trace methyl radical impurity originating from the backstreaming of diffusion pump oil along the vacuum manifold. In spectrum B, a very similar axial derivative profile is apparent for the methane-containing sample, again showing well-resolved parallel and perpendicular line-shape components. The only real exception to the resemblance of the spectrum to that recorded in argon is that the methyl radical impurity bands are greatly enhanced. The increase in intensity of these bands mast likely arises from methane cracking caused from trace, reactive A102molecules formed from small quantities of molecular oxygen present on matrix formation. In pure silane however, Figure 4C, instead of k i n g presented with a readily identifiable axial hyperfine pattern, one observes instead a rather unique spectrum composed of six severely broadened derivative line shapes superimposed with a second pattern, that of a narrow quartet (situated roughly between the second and third bands of the sextet) indicated by the square markers. A brief literature search revealed fairly q ~ i c k l ythat ~~,~ this latter grouping is most likely due to silyl radicals (*SiH3), which seems quite plausible in light of the trace methane cracking believed to occur in the sample of spectrum B. The previously reported g values for this species26(photolytically generated from parent SiH4 which was isolated in argon matrices) are gll = 2.003 and g, = 2.004, quite close to the values we were able to obtain from spectrum C of gll = 2.004 and g, = 2.006. However, the proton hyperfine splittings differed somewhat between the two studies. Our values of All = 17 MHz and A, = 23 MHz are inverted in magnitude with respect to the values previously reported26 for this radical of All = 24 MHz and A, = 20 MHz. A detailed discussion of the computer simulation of EPR spectra required in order to obtain these g and A values follows shortly. The broad sextet observed in Figure 4 is undoubtedly due to aluminum ( I = 5/2); however, the spectrum is very unusual in that it resembles no previously recorded spectrum of aluminum atoms in any other environment. To probe these unique aspects of the EPR spectra profile, we carried out a silane concentration study as in the corresponding UV-vis investigations. The results of this work appear in Figure 5. (25) Morehouse, R. L.; Christiansen, J. J.; Gordy, W . J . Chem. Phys. 1966, 45, 1751. (26) Raghunathan, P.; Shimokoshi, K. Spectrochim. Acta 1980,36A, 285.
2620 The Journal of Physical Chemistry, Vol. 95, No. 7, 1991
n
Lefcourt and Ozin
n
I C3v
3180
33kO
QAUSS
-
H AXIAL
In comparing the spectra of Figure 5, it is immediately apparent that upon the addition of even relatively small quantities of silane to the substrate gas mixture the resulting spectra differ greatly from that obtained in the complete absence of the molecule. Spectra B-D exhibit the broad hyperfine sextet introduced in the proceding spectrum along with the characteristic silyl radical bands. The fact that spectrum B displays the unusual deposition spectrum when the silane is quite dilute is possible evidence for the migration of aluminum atoms in the early stages of matrix formation to preferentially seek out the silane molecules. The unique characteristics, particularly the breadth, of the A1 atom resonances in silane-containing matrices prompted us to use perdeuterated silane again, SiD,, as in the UV-vis studies. The rationale for performing this exercise as an EPR experiment is that a ground-state complex of the type we have been proposing would, upon complete deuteration, be expected to show band narrowing due to the much smaller nuclear magnetic moment of deuterium (0.857 45 kN)compared to that of hydrogen (2.792 85 c(N),27 a ratio of roughly 1/3. For systems in which the superhyperfine splitting is not resolvable, the three bands per deuterium nucleus (I = 1) overlap with the main-line resonances (Le., A1 atoms) such that the broadening will be roughly 2 X (1/ 3 ) of that found in an equivalent I = system since the magnitude of two deuterium splittings has to be incorporated within the overall line width. Figure 6 shows the EPR derivative spectrum obtained upon SiD4 substitution. As hoped, the component resonance of the hyperfine sextet show considerable narrowing, especially in the two high-field bands which owe their intensities to the close overlapping of the parallel and perpendicular hyperfine components in this region. Manual measurement of the line indicates that the Al/SiD4 widths are reduced to the anticipated values of two-thirds that of the magnitudes of the corresponding Al/SiH4 widths. The other noticeable deviations from the spectra of Figure 5 are mainly due to the spin multiplicity of 1 for deuterium. The silyl group region between roughly 3330 and 3360 G in Figure 5 is considerably different in the Al/SiD4 spectrum with portions of the expected seven-band SiD3 spectrum (3 X (21) 1) revealed. C. Electron Paramagnetic Resonance Spectral Simulation. Through the use of appropriate computer software packages in conjunction with digitized EPR spectra, a means is achieved by which reasonable guesses can be made concerning the geometry of paramagnetic species. The computer simulations of EPR spectra reported in this work were carried out with the SIM14AI4 package mentioned in the Experimental Section. Briefly, the package allows the online graphical superposition of digitized experimental data with computationally generated spectra obtained from spin Hamiltonians that are constructed from user input. The spin parameters thus obtained can be reported more confidently (particularly for nonisotropic systems) than values obtained via manually measured distances directly from the original spectra. Execution of the package involves several preliminary “manual” approximations with the ‘SIM” program to obtain a fairly good
+
~
(27) Atkins, P.W. Physfcol Chemfsrry. 2nd ed.;W. H. Freeman and Co.: San Francisco, 1982; p 645.
c2v
-
ORTHORHOMBIC
. .
3580
Figure 6. EPR spectrum of aluminum atoms deposited in a 1:lO SiD4/Ar matrix (12 K). Relative deposition quantities of the matrix components are analogous to those in the 1 : l O SiH4/Ar sample.
n
‘dl H cs - OUTHORHOMBIC Figure 7. Four possible general interaction schemes that could describe Al/SiH4 ground-state complexation. These geometries were employed to computationallymodel experimentallyobtained EPR spectra and as basic starting structures to a series of ab initio SCF MO calculations.
m
c3,
-
unls
visual correspondence between the experimental and computed resonance positions, line widths, and intensities. Subsequently, these data are used as input to the “FIT” program, which utilizes a Simplex optimization routine that iterates until convergence criteria imposed by the user is attained. The result can be, and is often, negligibly disparate with respect to the original data. Concerning the Al/SiH4 matrix system, it was possible, by entering appropriate values for the various terms in the program-resident spin Hamiltonian, to successfully model the EPR transitions that would be expected to arise for each of the possible general interaction schemes depicted in Figure 7. The modeling was accomplished by designating the g and A tensors as either axial or orthorhombic and by incorporating varying degrees of hydrogen atom superhyperfine coupling into the fit depending on the number of H atoms coordinated to the aluminum atom (1, 2, or 3) as indicated in Figure 7 by the dotted lines. The particular goodness-of-fit criterion we used to rate the various optimized spectra was simply the root-mean-squared (rms) deviation from the experimental data points:
(rms units in gauss). The four modeled interaction geometries of Figure 7 yielded surprisingly low rms values and therefore very satisfactory fits with respect to some of the experimental data curves. Figure 8 shows the results of one of these optimizations. Here, the particular fit portrayed is that employing the C, orthorhombic interaction represented by geometry IV in Figure 7 in which the A1 atom is shown to bind sideon to a silicon-hydrogen bond. The modeling for this system therefore utilized orthorhombic g and A tensorial data along with superhyperfine coupling to one hydrogen nucleus. The experimental data to which the optimization was targeted was the A1 atom EPR spectrum obtained from a 1:lO SiH4/Ar matrix, a sample which provided a fairly well resolved band profile for both the A1 atom and silyl group reSOnanCeS compared to other samples observed in the EPR investigations. The clarity of the bands and their field positions enabled reasonably good ‘first-guess” parameters to be measured directly from the spectrum for entry into program “SIM”. The experimental data are represented by the solid line in Figure 8 while the final optimized spectral profile (using program “FITw) is indicated by the dashed line. Visual assessment of the superposition clearly shows that this optimization is excellent; in fact, its rms deviation from experiment was calculated by using the above formula to be a minuscule 0.030 G ( N = 500, X = 0.8 G). It should be stressed, however, that all of the geometries of Figure 7 gave very low rms values a t the termination of their respective optimizations, and the value of 0.030 G for structure IV was only slightly lower than the other rms deviations. The value for the second-best result was 0.038 G, obtained for the C, system represented by geometry I1 in Figure 7. This situation demonstrates that the C, side-on geometry of Figure 7 cannot be designated as the most probable structure for the AI/SiH4 groundstate complex based on the EPR simulation results alone.
The Journal of Physical Chemistry, Vol. 95, No. 7, 1991 2621
A1(2P)(SiH4]Complex
7 9 ' ndk-n
TABLE III: Qualitative EPR Atomic Orbital Spin Densities (in Percentages) Calculated from tbe Spectral Parameters in Table I1 via the Morton and Preston Axial Approximation (Ref 30) orbital spin
density
AI/Ar
PAI(~S)
total
3180
*I / (SlHq/*r
-
1/10)
COMWTLR SlMUUttON RMS = 0.030
Figure 8. EPR spectrum of aluminum atoms deposited in a 1:lO SiH,/Ar matrix ( 1 2 K) superimposed with a computer-optimizedspectral simulation resonance pattern. The stick spectra pictorially indicate the orthorhombic g and hyperfine splitting parameters obtained from the simulated spectrum. TABLE II: EPR Spectral Parameters for Each of the AI/Ar,b AI/CHi and AIISiH, Matrix Systems4 parametelb AI/Af AI/CHdd AI/SiH4 gl = 1.999 2.000 2.001 g H 1.963 g2 g, = 1.982 g, 1.954 136 A,(AI) = 1I7 AII(A1) 143 96 A,(AI) = 88 A,(AI) 102 A3(AI) = 77 A W Al(H) = 13 AZ(H) = 18 A3(H) = 16
OFor the latter two systems, the values were extracted from computer-optimized simulationsinvolving the software packages mentioned in the text. Hyperfine values ( A ) in MHz. buncertainty in g (fO.OO1) and A (AI MHz). CAs reported in ref 21. dAs reported in ref IO.
Due to solution-phase precedents for three-centered coordination between single metal atoms (Mn,28*29 Cr28and silicon-hydrogen bonds along with the results of some a b initio self-consistent-field molecular orbital (SCF MO) calculations (subject of section D) that support the choice of a three-centered bonding scheme between AI and SiH4, we have chosen to report the EPR spectral parameters obtained from the computer modeling of structure IV in Figure 7 as the currently accepted values for the Al/SiH4 ground-state species. Table I1 lists these computer-obtained values alongside similarly calculated quantities for the AI/CH4 matrix systemlo and the previously reported values for AI in Ar (average values from two slightly energetically different sites2'). As it can be seen from the table, axial parameters are reported for AI/Ar and Al/CH4. The axial hyperfine quantities for these systems along with the orthorhombic Al/SiH4 values are shown in magnitude only-the A, values are actually negative quantities for the AI/Ar sample according to Ammeter and Schlosnagle.2' Although the silane system was modeled orthorhombically, the comparable values obtained for g2 and g3 as well as for A2(AI) and A,(AI) show that the system can be described as being approximately axial. This situation allowed us to calculate qualitative atomic orbital spin densities for the AI/SiH4 complex using the Morton and Preston axial approximation.M This treatment utilizes the isotropic and dipolar coupling contributions as obtained from (28) Schubert, U.;Muller, J.; Alt, H.G. Organometallics 1987, 6, 469. (29) Schubert, U.; Scholz, G.; Muller, J.; Ackermann, K.; Wkle, B.; Stransfield, R. F. D.J. Organomet. Chem. 1986, 306, 303. (30) Morton, J. R.; Preston, K. F. J. Magn. Reson. 1978, 30, 577.
AI/SiH,
93.2 0.5
80.1 0.4 1.1 -82
98.3 0.5
pAl(3p)
PH(1S)
AI/CHA
-
99
-
94
the appropriate All and A, quantities. From these equations the assignment of negative quantities to the A, parameters of the Al/Ar system can be understood. A1 atoms (as well as B and Ga) quenched in the axial crystal fields provided by rare gas matrices yield EPR spectra that are analogous to those arising from 22 molecules.31 The lone electron of aluminum occupies a "pun orbital in this situation such that Ai, = 0 and Adip >. 0.31 For these relationships to hold A, must be negative while A, positive. By equating axial hyperfine and superhyperfine quantities with the experimental orthorhombic values for the Al/SiH4 system via All = A I and A, = ' / 2 ( A 2+ A3), approximate Ai, and Adi contributions were obtained (due to the fact that the AI/SiH4 EPI! simulation was close to being axial). These values, when divided by the free atom, or unperturbed isotropic and dipolar coupling constants for aluminum (Aho = 391 1 MHz; = 83.1 MHz, including a multiplicative angular factor, a,to distinguish p orbitals from d and f functions, et~.~O) yield the Al atomic orbital fractional contributions to the SOMO (singly occupied molecular orbital) wave function. Table I11 lists these qualitative atomic orbital spin densities for the Al/SiH4 matrix system alongside the quantities calculated for aluminum in argon2' and methane.1° Unlike the case for aluminum, the small spin density calculated for the hydrogen 1s orbital (due to H atom superhyperfine splitting of the A1 atom hyperfine resonances) was obtained from entirely positive experimental A values (along with the free atom Aho value for hydrogen of 1420.4 M H Z ~ ~ ) . Upon looking at the atomic orbital spin densities listed in Table 111, it is readily apparent that there is a decreasing trend in pA1(3p) from the argon, through to the silane matrix. When the fractional contributions accounted for in the table are summed for the Al/SiH4 complex (pAI(3p),PA@), and pH(Is)), the total only amounts to roughly 82%, which, compared to the quantities much closer to 100%for A1 in Ar and CH4, strongly suggests that most of the approximately 18% unallocated spin density should reside in the silicon 3d atomic orbitals. The experiment that would clarify this matter would be the EPR investigation of A1 atoms in an enriched 29Sisilane matrix since this isotope has a spin of and would therefore manifest itself in the superhyperfine splitting of the aluminum bands in the case of an interaction complex. We calculated approximate 29SiAll and A, values for the axial case of 26 and 15 G, respectively. Due to the exorbitant cost of enriched 29Si silane, however (-$30000 g-' for only 50% isotopic enrichment), in conjunction with our current incapability of resolving 29Siresonances in ordinary silane with the signal-to-noise ratio of our instrumentation (natural abundance = 4.7%), studies involving this magnetic isotope have not been pursued. Before leaving this section, it should be added that the use of electric quadrupole parameters for aluminum (I = 5/2) in the spectral simulations carried out in this study were omitted due to the degree of success attained without them. 27Al in reality does have an electric quadrupole moment of +O. 15 b (+O. 15 X cm2).32 D. Ab Initio Self-Consistent-Field Molecular Orbital Calculations. As earlier mentioned, a second theoretical tool in addition to the EPR spectral simulation package that supports the choice of a three-center, two-electron bonding scheme for the Al/SiH4 ground-state complex was a series of a b initio SCF M O calculations carried out on the four basic interaction geometries por(31) Weltner, W., Jr. Magnetic Aroms and Molecules; Van Nostrand Reinhold Co.: New York, 1983; p 62. (32) Reference 31, p 341.
2622 The Journal of Physical Chemistry, Vol. 95, No. 7, 1991
Lefcourt and Ozin
H'
H'
-530.201914 H
-530.210166 H
.H
Ii L
-530.299028 H -530.388957 H Figure 9. Pictorial results of two ab initio SCF MO geometry-optimization calculations employing the 3-21GS basis set and the restricted Hartree-Fock (RHF) formalism. The first (upper) optimization shown is for a C, system in the 2A1state while the second (lower) optimization is for a C, system in the 2A' state. In both cases the starting structure is depicted on the left-hand side, and the final, optimized geometry is shown on the right. Single-point SCF starting energies and the final,
converged values are given in hartrees. trayed in Figure 7. Figure 9 shows the pictorial results for two such calculations in which geometry optimizations were carried out with the 3-21G* basis set." The two optimizations presented , structure and the second for a C, system, here, the first for a C were the most important calculations concerning the set of general structures in Figure 7 since these computations yielded the lowest energies for systems in which the final, optimized structures represented bound species. Neither of the two C,, geometries depicted in Figure 7 resulted in suitable A l S i or AI-H interatomic distances for bonding in their respective optimizations. The first system represented in Figure 7 is for the 2AI state and shows the starting C , geometry on the left along with the single-point S C F energy of the structure in hartrees while on the right the optimized system is shown with its associated energy. The optimization was confirmed to keeping C2,symmetry throughout the calculation in an attempt to locate an energy minimum for this particular geometry. In none of our interaction complex calculations, however, was a true energy minimum found (critical point of 0). It is evident upon looking at the particular bond lengths and angles shown for this system that the geometry-optimized structure does not differ very much at all from the initial geometry, a situation that is reflected in the converged energy which is only approximately 1/lo0 of a hartree lower (-26 kJ/mol) than the single-point value associated with the starting structure. Calculations carried out on systems having the equivalent starting geometry but inhabiting the 2Bland 2B, electronic states yielded final energies that were considerably higher (0.17 and 0.31 hartree, respectively) than the energy obtained in the 2AI optimization. The second optimization shown in Figure 9, that for a C, three-centered AI-Si-H system in the 2A' state, is much more interesting than the C, calculation since its final, converged energy is considerably lower than the starting value by almost 1/ 10 of a hartree (-230 kJ/mol). As well, the single-point S C F energy of the initial structure is much lower than that of either the starting or final C, geometries shown in the figure. With respect to the starting parameters shown for the C,system, the Si-H bond closest to the aluminum was purposely lengthened to 2.5 A from the normally accepted 1.48 A covalent bond length?3 This elongation was employed due to the fact that the choice of initial Si-H bond lengths of 1.48 and 2.00 A resulted in unbound aluminum and silane components during the respective optimizations. Also of note was the fact that the single-point starting energies for these much shorter bonded systems were somewhat higher than the analogous value associated with the optimization shown in the (33) Peterson, M. R.; 1905, 123, 399.
Csizmadia, I. G. J . Mol. Srrucr. (THEOCHEM)
ii
l
ai00 f& do om1 Figure 10. IR spectra (taken in transmission) of a 1:lO SiH4/Ar sample in the absence of aluminum (A), a 1:lOOO SiH,/Ar sample in the absence of aluminum (B), and a 1:lOOO SiH,/Ar sample in the presence of aluminum (C),all at 12 K. The total quantity of metal present in the last sample was approximately 50 p g , and the metal/substrate dispersion ratio was roughly 1:3100 or therefore 1:3 for AI/SiH4 alone. The asterisks in spectrum C indicate very weak bands attributable to the formation of trace amounts of the insertion product, silylaluminum hydride (SiH,AIH), probably from gound-state AI atoms. (See part 2 of this study.)
figure. A quick look at the final, optimized structure reveals that the interatomic distances did not change radically over the course of the optimization with one exception: the silicon-hydrogen bond closest to the metal. One realizes that the resulting distance of 3.57 A is somewhat large for bonding to actually be present, and therefore the structure should be cautiously interpreted as having a siliconaluminum bond and an aluminum-hydrogen bond instead of the original Si-H bond. Such a species would be expected to form upon the oxidative addition of a silicon-hydrogen bond to an aluminum atom, and the optimization suggests that the progression from the starting geometry chosen is energetically downhill. Clearly, although these theoretical treatments for the C , and C,geometry models are interesting they are not sufficiently conclusive to allow one to decide as to which one actually exists in practice. Much more work will be required to further elucidate this important point. E. Infrared Spectroscopic Studies. One problematic aspect concerning the determination of the Al/SiH4ground-state complex has been the lack of infrared data supporting its existence. Figure 10 shows the IR transmission spectra for several systems, the interpretation of which reveals this situation. Spectrum A is the scan of a 1:lO SiH4/Ar sample in the absence of aluminum. The band centered at 2180 cm-' is the T2asymmetric silane stretching frequency ( Y ~ ) , ~while ~ , ~ the ~ second major intense feature at ~ roughly 900 cm-l is the T2silane deformation mode ( v , ) . ~The broad band immediately to the high-energy side of each of these fundamentals could be the result of a site splitting or the combination of a fundamental with a phonon mode. The band at approximately 1860 cm-' is believed to be an overtone of the 900-cm-I fundamental. Spectra B and C of Figure 10 were recorded from samples containing highly dilute silane-l:1000 with respect to argon-in the absence and presence of aluminum atoms, respectively. The dispersion ratio of metal to substrate in spectrum C was ap(34) Owyoung, A.;Eshcrick, P.; Robiette, A. G.; McDowell, R. S.J . Mol. Specrrosc. 1981, 86, 209. (35) Nakamoto, K.Infrared and Raman Spectra of Inorganic ond Coordination Compounds, 3rd ed.; Wiley: Toronto, 1978; p 104.
2623
J. Phys. Chem. 1991,95, 2623-2628
proximately 1:3100 (therefore only 1:3 for AI/SiH, in the sample). Reducing the quantity of silane in the matrix narrowed the absorption bands considerably as is seen in Figure 10 in an attempt to reveal weak satellite bands near the two infrared-active fundamentals in the presence of aluminum (spectrum C). Satellite bands closely proximate to the fundamental vibrational frequencies could be attributable to perturbations of the tetrahedral geometry of silane due to the presence of complexed aluminum. As far as it can be detected, however, spectra B and C are identical with two exceptions: there appear to be two very small bands exclusive to spectrum C at approximately 2100 and 850 cm-I, both of which are very close on the low-energy side to each of the two fundamental vibrational frequencies. These weak bands are indicated in the figure by an asterisk. Upon obtaining spectrum C, much work was carried out enlarging these spectral regions (along with preparing alternate samples having varying dispersion ratios of metal/substrate) to convince ourselves that the weak bands were in fact inherent to the system and not anomalous and that they were reproducible. However, upon irradiation of the sample that gave rise to spectrum C (and the similar samples mentioned above) to photolytically generate the insertion product silylaluminum hydride, SiH,AIH (the subject of part 2 of this study), these bands
-
grew in intensity at equal rates, indicating that they were characteristic of the insertion product and not the ground-state complex. Hence, with the dilution conditions present in the sample of spectrum C, approximately 1:IOOO SiH,/Ar and 1:3 Ai/SiH4, the aluminum-silane complex has not manifested itself with "grating" IR detection. Summary
The experimental and theoretical results of this study demonstrate that A1 atoms in pure silane or silane/argon mixtures at 12 K yield a ground-state complex possibly containing a threecenter, two-electron AI($-SiH,) bonding interaction. Acknowledgment. The generous financial support of the Natural Sciences and Engineering Resehrch Council of Canada is greatly appreciated. The award of an Ontario Graduate Scholarship (M.A.L.) is deeply appreciated. Valuable computational assistance from Dr. Douglas McIntosh and Dr. Mike Peterson with various aspects of the EPR simulations and ab initio calculations are gratefully acknowledged. Registry No. AI, 7429-90-5; SiH4, 7803-62-5; Ar, 7440-37-1; SiH3AIH, 116233-51-3.
-
Ai(2P)(SiH4]Complex and Photoreversible Oxidative Addition/Reductive Elimination Reaction Ai(2P)(SiH4] SIHSAiH. 2. Ai(*P)(SiH,) SIHSAiH Reaction Michael A. Lefcourt and Geoffrey A. Ozin* Lash Miller Chemical Laboratories, University of Toronto, 80 St. George St., Toronto, Ontario, Canada M5S 1Al (Received: April 26, 1989; In Final Form: August 9, 1990)
Brief 400-nm photolysis of 1: 10 silane/argon matrices containing monatomically isolated aluminum results in the formation of the insertion product silylaluminum hydride (SiH,AIH), most likely via an oxidative addition mechanism. The techniques of UV-vis, EPR, and infrared spectroscopy enabled the identification and characterization of this molecule. The photolytic generation of this species paralleled the formation of methylaluminum hydride (CH3AIH) studied previously in our laboratory. EPR spectral simulations and ab initio self-consistent-fieldmolecular orbital (SCF MO) calculations were employed to help in the characterization of silylaluminum hydride. The EPR spectral parameters extracted from the raw data via the computer simulations, along with subsequently calculated atomic orbital spin densities, determined that the molecule was a bent, orthorhombic species. The SCF MO calculated bond angle at the AI atom was shown to be 1 18.80°. Secondary photolysis of the aluminum/silane matrix sample resulted in the conversion of the insertion product back to that of the ground-state complex most likely by a reductive elimination mechanism.
Introduction
To confidently report the existence of a matrix-isolated aluminum-silane AI(SiH4) ground-state complex, justification must be made to the extent that the spectral features presented in part 1 of this study do not correspond to the more easily envisaged entity, that of the insertion product silylaluminum hydride (SiH3AlH). This latter species is more abundantly precedented in the literature by several analogous matrix systems involving methane and a variety of metals such as Mn,4SS Fe,es to list a few. Each of these metals in atomic Cu,2*4*798 and Mg2*9*'0 (1) Parnis, J. M.; Ozin, G. A. J . Am. Chem. SOC.1986, 108, 1699. (2) Parnis, J. M. Ph.D. Thesis, University of Toronto, 1987. (3) Parnis, J. M.; Ozin. G. A. J. Phys. Chem. 1989, 93, 1204. (4) Billups, W. E.; Konarski, M. M.; Hauge, R. H.;Margrave, J. L. J. Am. Chem. Soc. 1980, 102,7393. ( 5 ) Ozin, G. A.; McCaffrey, J. G.; McIntosh, D. F. Pure Appl. Chem. 1984, 56, 1 1 1 . ( 6 ) Ozin, G. A.; McCaffrcy, J. G. J . Am. Chem. SOC.1982, 104, 7351. (7) &in, G. A.; McIntosh, D. F.; Mitchell, S. A. J. Am. Chem. Soc. 1981, 103. 1514.
form has been shown to oxidatively add to single C-H bonds of methane, resulting in products of the general form CH3MH (where M represents the metal atom). In a few instances, notably with copper, the product methylmetal hydride is subject to secondary photolysis and shows a wavelength-dependent fragmentation to form such radicals as CH3, MH, CH3M, and H.2,7*8With respect to the use of silane as a substrate, two prior gas-phase studies have demonstrated the ability of silane to be insertively attacked by the monatomic species 'PI Mg" and the zPSi ion.I2 Similar to the matrix reaction of Cu with methane, both of these gas-phase systems exhibited immediate fragmentation upon Si-H bond insertion; however, in these cases the exothermicity was at the origin of the fragmentation step. ~
~
(8) Parnis, J. M.; Mitchell, S.A.; Garcia-Prieto, J.; Ozin, G. A. J. Am. Chem. SOC.1985, 107, 8169. (9) McCaffrey, J. G.; Parnis, J. M.; Ozin, G. A.; Breckenridge, W. H. J. Phys. Chem. 1985, 89, 4945. (IO) McCaffrey, J. G. Ph.D. Thesis, University of Toronto, 1987. ( 1 1) Breckenridge, W. H.; Umemoto, H. J. Chem. Phys. 1984,81,3852. (12) Boo, B. H.; Armentrout, P. B. J . Am. Chem. Soc. 1987, 109. 3549.
0022-3654/91/2095-2623$02.50/00 1991 American Chemical Society