Formation of digallium sites in the reaction of trimethylgallium with silica

Article abstract of DOI:10.1021/om051034o

The room-temperature, gas-solid reaction of volatile GaMe₃ with a nonporous silica was studied by elemental and gas-phase analysis, in situ IR and ¹H, ¹³C, and ²⁹Si solid-state NMR, and extended X-ray absorption fine structure (EXAFS) spectroscopy. Most of the grafting (～85%) occurred on Q³ sites, O₃SiOH, but a small amount (～15%) of siloxane (O₃SiO-SiO₃) bond cleavage was also observed. The major, if not the only, gallium product has the empirical formula ≡SiOGaMe₂, but it is not an isolated site. The Ga K-edge EXAFS of GaMe₃-modified silica, recorded at 10 K, reveals that each Ga has a Ga neighbor at 2.97-2.99 A. The sites are best described as [GaMe ₂(μ-OSi≡)]₂. To strengthen this assignment, a molecular analogue, [GaMe₂(μ-OSiPh₃)]₂, was characterized by both single-crystal X-ray diffraction and EXAFS. The Ga ₂O₂ rings in the molecular complex and the silica-supported gallium dimer have very similar dimensions. The gallium dimer is formed on the silica surface regardless of the extent of partial dehydroxylation (varied by pretreatment in vacuo at 100 and 500°C). This result is interpreted in terms of a vicinal disposition for the majority of Q³ grafting sites.

Full text of DOI:10.1021/om051034o

Organometallics 2006, 25, 1891-1899
1891
Formation of Digallium Sites in the Reaction of Trimethylgallium
with Silica
Ziyad A. Taha,† Eric W. Deguns,‡ Swarup Chattopadhyay,† and Susannah L. Scott*,†,‡
Department of Chemical Engineering and Department of Chemistry and Biochemistry, UniVersity of
California, Santa Barbara, California 93106-5080
ReceiVed December 3, 2005
The room-temperature, gas-solid reaction of volatile GaMe3 with a nonporous silica was studied by
1
elemental and gas-phase analysis, in situ IR and H, 13C, and 29Si solid-state NMR, and extended X-ray
absorption fine structure (EXAFS) spectroscopy. Most of the grafting (∼85%) occurred on Q3 sites,
O3SiOH, but a small amount (∼15%) of siloxane (O3SiO-SiO3) bond cleavage was also observed. The
major, if not the only, gallium product has the empirical formula tSiOGaMe2, but it is not an “isolated”
site. The Ga K-edge EXAFS of GaMe3-modified silica, recorded at 10 K, reveals that each Ga has a Ga
neighbor at 2.97-2.99 Å. The sites are best described as [GaMe2(µ-OSit)]2. To strengthen this assignment,
a molecular analogue, [GaMe2(µ-OSiPh3)]2, was characterized by both single-crystal X-ray diffraction
and EXAFS. The Ga2O2 rings in the molecular complex and the silica-supported gallium dimer have
very similar dimensions. The gallium dimer is formed on the silica surface regardless of the extent of
partial dehydroxylation (varied by pretreatment in vacuo at 100 and 500 °C). This result is interpreted in
terms of a vicinal disposition for the majority of Q3 grafting sites.
liberating methane.12-14 The proposed reaction is shown in
eq 1,
Introduction
Gallosilicates catalyze the dehydrogenation1,2 and aromati-
zation3,4 of alkanes and the selective reduction of nitrogen oxides
by hydrocarbons.5 A gallosilicate catalyst is used in the BP-
UOP Cyclar process, by which liquefied petroleum gas (LPG,
primarily propane and butane) is converted to BTX (benzene/
toluene/xylene).6 The catalytic activity and selectivity depend
on the extent of dispersion of the gallium in the silicate
framework. High dispersion may be achieved by grafting a
volatile molecular precursor onto the surface. GaMe3 has been
used in this way to prepare gallosilicate catalysts,7-11 in which
the resulting Ga sites were assumed to be isolated.
tSiOH + GaMe3 f tSiOGaMe2 + MeH
(1)
where tSiOH denotes a sterically accessible Q3 site on the silica
surface. In previous studies, the gallium product was formulated
as a three-coordinate dimethylgallium(III) fragment, with a
trigonal coordination environment similar to that of GaMe3 both
in the gas phase15 and in solution.16 The hydrolytic stability of
the methyl ligands in GaMe3/SiO2 led one group to propose
extensive grafting of GaMe3 onto siloxane bridges (eq 2),
generating unreactive Si-C bonds,13
GaMe3 adsorbs reversibly on silica at cryogenic tempera-
tures.12 Above 273 K, it reacts with the surface hydroxyl groups,
tSiOSit + GaMe3 f tSiOGaMe2 + tSiMe (2)
* To whom correspondence should be addressed. E-mail:
sscott@engineering.ucsb.edu. Fax: 1-805-893-4731.
† Department of Chemical Engineering.
where tSiOSit represents a pair of vicinal Q4 sites and
tSiMe is a T1 site (monomethylated silicon atom). This hypo-
thesis was refuted in a subsequent investigation, although a
minor contribution from reaction 2 could not be ruled out.14
Grafting reactions analogous to eqs 1 and 2 have long been
proposed for AlMe3, whose deposition on silica can be used to
create Lewis acid sites that activate metallocene catalysts for
olefin polymerization.17 These grafted sites have been described
in many studies as three-coordinate tSiOAlMe2 and (tSiO)2Al-
Me.18-22 However, we recently concluded that AlMe3 reacts
‡ Department of Chemistry and Biochemistry.
(1) Giannetto, G.; Montes, A.; Gnep, N. S.; Florentino, A.; Cartraud,
P.; Guisnet, M. J. Catal. 1993, 145, 86-95.
(2) Me´riaudeau, P.; Sapaly, G.; Naccache, C. J. Mol. Catal. 1993, 81,
293-300.
(3) Thomas, J. M.; Liu, X.-S. J. Phys. Chem. 1986, 90, 4843-4847.
(4) Choudhary, V. R.; Devadas, P.; Kinage, A. K.; Sivadinarayana, C.;
Guisnet, M. J. Catal. 1996, 158, 537-550.
(5) Pushkar, Y. N.; Sinitsky, A.; Parenago, O. O.; Kharlanov, A. N.;
Lunnia, E. V. Appl. Surf. Sci. 2000, 167, 69-78.
(6) Mowry, J. R.; Anderson, R. F.; Johnson, J. A. Oil Gas J. 1985, 83,
1288-1293.
(13) Tubis, R.; Hamlett, B.; Lester, R.; Newman, C. G.; Ring, M. A.
Inorg. Chem. 1979, 18, 3275-3276.
(7) Bayense, C. R.; van Hooff, J. H. C.; de Haan, J. W.; van de Ven, L.
J. M.; Kentgens, A. P. M. Catal. Lett. 1993, 17, 349-361.
(8) Seidel, U.; Koch, M.; Brunner, E.; Staudte, B.; Pfeifer, H. Micropor.
Mesopor. Mater. 2000, 35-36, 341-347.
(14) Morrow, B. A.; McFarlane, R. A. J. Phys. Chem. 1986, 90, 3192-
3197.
(15) Beagley, B.; Schmidling, D. G.; Steer, I. A. J. Mol. Struct. 1974,
21, 437-444.
(16) Muller, N.; Otermat, A. L. Inorg. Chem. 1965, 4, 296-299.
(17) Chien, J. C. W. Top. Catal. 1999, 7, 23-36.
(18) Peglar, R. J.; Hambleton, F. H.; Hockey, J. A. J. Catal. 1971, 20,
309-320.
(19) Morrow, B. A.; Hardin, A. H. J. Phys. Chem. 1979, 83, 3135-
3141.
(20) Bartram, M. E.; Michalske, T. A.; Rogers, J. W. J. Phys. Chem.
1991, 95, 4453-4463.
(9) Garc´ıa-Sa´nchez, M.; Magusin, P. C. M. M.; Hensen, E. J. M.; Thu¨ne,
P. C.; Rozanska, X.; van Santen, R. A. J. Catal. 2005, 219, 352-361.
(10) Hensen, E. J. M.; Garc´ıa-Sa´nchez, M.; Rane, N.; Magusin, P. C.
M. M.; Liu, P.-H.; Chao, K.-J.; van Santen, R. A. Catal. Lett. 2005, 101,
79-85.
(11) Kazansky, V. B.; Subbotina, I. R.; van Santen, R. A.; Hensen, E. J.
M. J. Catal. 2005, 233, 351-358.
(12) Mawhinney, D. B.; Glass, J. A.; Yates Jr., J. T. J. Vac. Sci. Technol.
A 1999, 17, 679-685.
10.1021/om051034o CCC: $33.50 © 2006 American Chemical Society
Publication on Web 03/18/2006

1892 Organometallics, Vol. 25, No. 8, 2006
Taha et al.
series. Caution! Neat GaMe3 is pyrophoric. Handle only small (<1
mL) quantities outside the gloVebox, and dispose of unreacted
material by controlled alcoholysis in a fume hood. Water was
double-distilled and then transferred into a high-vacuum glass
reactor equipped with a glass stopcock greased with Apiezon-H
(Varian). Before use, each liquid reagent was subjected to several
freeze-pump-thaw cycles in order to remove dissolved gases and
then introduced into the experimental reactor via gas-phase transfer
through a high-vacuum manifold (<10-4 Torr). In general, an excess
of GaMe3 was sublimed at room temperature and condensed onto
the silica with the aid of a liquid-N2 bath. After the mixture was
warmed to room temperature, the reaction of GaMe3 with silica
was allowed to proceed under vacuum, in the absence of solvent
or inert gases, for 20 min. Volatiles, including physisorbed GaMe3,
were desorbed at room temperature to a liquid nitrogen trap.
with the surface of a fumed silica to give four-coordinate
dialuminum surface sites, on the basis of the reaction stoich-
iometry.23
The organometallic chemistry of trivalent aluminum is similar
to that of gallium in many ways, including a proclivity for four-
coordination with heteroatom-donor ligands. The persistence of
undercoordinated metal (Al, Ga) sites on silica, despite the
abundance and proximity of potential oxygen donors, is therefore
unexpected. Coordination numbers have been evaluated for
hydrated alumino- and gallosilicates by their NMR chemical
shifts;24,25 however, both 27Al and 71Ga exhibit very broad,
frequently undetectable NMR resonances in the solid state under
the strictly anhydrous conditions required to prevent decomposi-
tion of their organometallic compounds.8,9,23 Furthermore, the
presence of water is likely to alter the metal coordination
number, since it binds strongly to Lewis acid sites.
IR Spectroscopy. Experiments were performed in an in situ IR
cell, equipped with a high-vacuum stopcock and either KCl or ZnSe
windows affixed to the Pyrex body with TorrSeal (Varian). Silica
was pressed at 40 kg/cm2 into a self-supporting disk 1.6 cm in
diameter, containing ca. 15 mg of silica. A thinner film was prepared
by spreading silica onto a ZnSe window, to increase the transpar-
ency below 1300 cm-1. Transmission infrared spectra were recorded
on a Shimadzu PrestigeIR spectrometer equipped with a DTGS
detector and purged with CO2-free dry air. Background and sample
spectra were recorded by co-adding 32 scans at a resolution of 2
We undertook an investigation of the reaction of GaMe3 with
silica in an effort to elucidate the structures of grafted group
1
III sites. Elemental analysis and IR, H, 29Si, and 13C solid-
state NMR, and extended X-ray absorption fine structure
(EXAFS) spectroscopy were employed. In particular, EXAFS
was used to probe the local environment of gallium. The
technique is indifferent to the degree of long-range order and
is thus well-suited to the study of structure in amorphous
materials, such as oxide-supported metal complexes. While the
EXAFS of Al is complicated by the need for UHV conditions
at the low X-ray energy of the Al K-edge (1.5596 keV), as well
as strong interference from the overlapping K-edge of Si (1.8389
keV), the Ga K-edge (10.3671 keV) is more accessible, and
high-quality spectra of gallosilicate materials are readily ob-
tained.26,27
cm-1
.
Solid-State NMR Spectroscopy. For NMR experiments, 50-
100 mg of silica was compacted into thick pellets and then coarsely
ground in a mortar. 1H, 13C, and 29Si solid-state NMR spectra were
recorded on a Bruker ASX-300 spectrometer. Samples were loaded
into 4 mm zirconia rotors in an Ar-filled glovebox equipped with
O2 and moisture sensors. 1H MAS NMR spectra were recorded at
300.05 MHz. The spectra were collected using a 2 µs 45° pulse, a
relaxation delay of 0.2 s, and an acquisition time of 49 ms. 13C
and 29Si CP-MAS NMR spectra were recorded at frequencies of
75.46 and 56.62 MHz, respectively. 13C spectra were collected using
a contact time of 2 ms and a relaxation delay of 2 s. 29Si spectra
were collected using a contact time of 10 ms and a relaxation delay
of 1 s. All samples were spun at 10 kHz. Water, adamantane, and
tetrakis(trimethylsilyl)silane were used as external chemical shift
Experimental Section
Materials and Reagents. The silica used in this work is Aerosil
380 (denoted A380), a nonporous, pyrogenic silica from Degussa,
with a BET surface area of (340 ( 2) m2/g and an average primary
particle size of 7 nm. The thermal pretreatment temperature of the
silica is indicated by its appended number. For example, A380-
500 indicates a sample of A380 treated at 500 °C. A standard
pretreatment procedure was followed in order to ensure reproduc-
ibility. A380-500 was prepared by calcining the silica under 300
Torr of O2 at 500 °C for at least 3 h. The silica was then partially
dehydroxylated at 500 °C for at least 4 h under dynamic vacuum
(<10-4 Torr). To prepare silica at a lower dehydroxylation
temperature, the calcination step was omitted. The silica was simply
heated to the appropriate temperature for 4 h under dynamic
vacuum. These thermal treatments do not alter the surface area of
the silica, but they do standardize the number of surface hydroxyl
groups.
1
references for H, 13C, and 29Si, respectively. 13C and 29Si spectra
were baseline-corrected and treated with 50 Hz line broadening.
71Ga MAS NMR spectroscopy was attempted on a Bruker
DMX500 spectrometer, using a 1.6 µs 90° pulse, a relaxation delay
of 2 s, and an acquisition time of 4.2 ms. The sample was spun at
14 kHz.
Mass Balance. Methane was quantified in situ by IR spectro-
scopy. A calibration curve in the range 5-30 Torr was constructed
using the intensity of the CH4 deformation mode at 1306 cm-1. At
the end of each experiment, the GaMe3-modified silica was weighed
in air and then stirred overnight in 1.67 M HNO3 containing 35%
aqueous H2O2 (0.1 mL/mL of sample solution). The solution was
filtered and its gallium content analyzed by ICP. A calibration curve
was constructed by diluting an atomic absorption standard solution
(998 ppm of Ga, Fluka).
GaMe3 (colorless liquid) was purchased from Aldrich and
transferred in the glovebox into an all-glass reactor sealed with
two high-vacuum Teflon stopcocks (Young valves) arranged in
(21) Anwander, R.; Palm, C.; Groeger, O.; Engelhardt, G. Organome-
tallics 1998, 17, 2027-2036.
(22) Puurunen, R. C.; Root, A.; Sarv, P.; Viitanen, M. M.; Brongersma,
H. H.; Lindblad, M.; Krause, A. O. I. Chem. Mater. 2002, 14, 720-729.
(23) Scott, S. L.; Church, T. L.; Nguyen, D. H.; Mader, E. A.; Moran,
J. Top. Catal. 2005, 34, 109-120.
(24) Bayense, C. R.; Kentgens, A. P. M.; de Haan, J. W.; van de Ven,
L. J. M.; van Hooff, J. H. C. J. Phys. Chem. 1992, 96, 775-782.
(25) McManus, J.; Ashbrook, S. E.; MacKenzie, K. J. D.; Wimperis, S.
J. Non-Cryst. Solids 2001, 282, 278-290.
(26) Meitzner, G. D.; Iglesia, E.; Baumgartner, J. E.; Huang, E. S. J.
Catal. 1993, 140, 209-225.
(27) Nishi, K.; Shimizu, K.; Takamatsu, M.; Yoshida, H.; Satsuma, A.;
Tanaka, T.; Yoshida, S.; Hattori, T. J. Phys. Chem. B 1998, 102, 10190-
10195.
Synthesis and Single-Crystal X-ray Diffraction of [GaMe2-
(µ-OSiPh3)]2. The model complex was synthesized using Schlenk
techniques under a dry nitrogen atmosphere. Solvents were rigor-
ously dried and deaerated prior to use by passage over molecular
sieves and activated alumina. To 4.6 g (0.017 mol) of Ph3SiOH in
40 mL of toluene at room temperature was added a solution of 1.9
g of GaMe3 (0.017 mol) in 20 mL of toluene with stirring.
Immediate evolution of methane gas was observed. After the
mixture was stirred for 24 h, toluene was removed in vacuo (ca.
100 mTorr) to give [GaMe2(µ-OSiPh3)]2 as a white powder.
Recrystallization from a mixture of hexane and benzene afforded
a crystalline solid. Yield: ca. 70%. Solution NMR spectra were

Reaction of Trimethylgallium with Silica
Organometallics, Vol. 25, No. 8, 2006 1893
Table 1. Stoichiometries (mmol/g of silica)a for GaMe3
Uptake on Silica and Subsequent Hydrolysis of Grafted
GaMe3/SiO2
recorded on a Bruker DPX200 spectrometer. 1H NMR (CDCl3, δ):
7.15-7.50 (m, 30H, Ph); -0.53 (s, 12H, GaMe). 13C{1H} NMR
(CDCl3, δ): -2.12 (GaMe); 127.75 (s, 12 C, m-C phenyl); 130.07
(s, 6 C, p-C phenyl); 135.08 (s, 12 C, o-C phenyl); 135.75 (C6H5)
(s, 6 C, Si-C phenyl). Elemental analysis was performed under
N2. Anal. Calcd for C40H42Ga2O2Si2: C, 64.03; H, 5.64. Found:
C, 63.55; H, 5.54.
GaMe3/A380-500
grafting hydrolysis
Ga uptake 1.04 ( 0.04 (6)
GaMe3/A380-100
grafting hydrolysis
1.55 ( 0.09 (7)
MeH 0.88 ( 0.04 (9) 1.96 ( 0.07 (4) 1.35 ( 0.05 (8) 2.84 ( 0.10 (5)
tSiMeb 0.16 ( 0.06
0.12 ( 0.08 0.20 ( 0.10 0.26 ( 0.07
A diffraction-quality single crystal of [GaMe2(µ-OSiPh3)]2 was
grown by slow diffusion of hexane into a benzene solution of the
compound under N2. A colorless crystal of approximate dimensions
0.20 × 0.15 × 0.08 mm was mounted on a glass fiber and
transferred to a Bruker CCD platform diffractometer. Data collection
(20 s/frame, 0.3°/frame for a sphere of diffraction data) and
determination of unit cell parameters were accomplished using the
program SMART.28 The data were collected at 120 K, using an
Oxford nitrogen gas cryostream system. Raw frame data were
processed using the program SAINT.29 Subsequent calculations
were carried out using the program SHELXTL.30 The structure was
solved by direct methods and refined on F2 by full-matrix least-
squares techniques. All non-hydrogen atoms were refined aniso-
tropically. Hydrogen atoms were located from difference Fourier
synthesis and refined isotropically.
a Reported as means and standard deviations; numbers in parentheses
represent the number of independent experiments. b Calculated as the
difference between the expected and observed methane yields (expected:
1.0 MeH/Ga for grafting and 2.0 MeH/Ga for hydrolysis).
number of scatterers in the ith shell at a distance R from the
absorber. The Debye-Waller factor, σi2, is the root-mean-squared
relative displacement of these scatterers, λ(k) is the mean-free path
of the photoelectron, and φ(k)i and F(k)i are the phase shift and
backscattering amplitude, respectively. The phase shift and back-
scattering amplitude functions were calculated using FEFF 8.20.34,35
Results and Discussion
X-ray Absorption Spectroscopy. Spectra at the Ga K-edge
(10.3671 keV) were recorded on beamline 2-3 (Bend) at the
Stanford Synchrotron Radiation Laboratory (SSRL), operated at
3.0 GeV with a ring current of 80-100 mA. X-rays were
monochromatized via reflection from a pair of Si(111) crystals,
through a 1 mm entrance slit. The incident beam was detuned 50%
to suppress harmonics. Samples were mounted at 45° to the beam,
to collect transmission and fluorescence spectra simultaneously. The
intensity of the incident beam was measured with a N2-filled ion
chamber detector installed in front of the sample. Transmitted
X-rays were detected in a second, Ar-filled ion chamber. Fluores-
cence from the sample was recorded at right angles to the beam,
using an Ar-purged Lytle detector without Soller slits.
Grafting Stoichiometry. When excess GaMe3 was sublimed
onto A380-500 and allowed to react at room temperature, a
portion of the organometallic complex chemisorbed on the silica
surface. The only volatile product detected by gas-phase IR
spectroscopy was methane: 0.88 ( 0.04 mmol/g of silica. The
amount of methane is equivalent to the number of accessible
hydroxyls on this silica.36 Analysis of the GaMe3-modified silica
after desorption of unreacted GaMe3 revealed the presence of
6.75 ( 0.28 wt % Ga, or 0.95 ( 0.04 mmol of Ga/g of sample.
After correction for the mass gained during grafting (9.7 wt %
for the GaMe2(-H) fragment), the Ga uptake was therefore 1.04
( 0.04 mmol of Ga/g of silica. Quantitative results are
summarized in Table 1.
The ratio of evolved methane per chemisorbed Ga suggests
that a grafting reaction such as eq 1 therefore represents the
fate of 85 ( 5% of GaMe3. The difference between the Ga
uptake, 1.04 mmol/g of silica, and the methane yield upon
grafting, 0.88 mmol/g of silica, is attributed to a minor reaction
(15 ( 6%) of GaMe3 with siloxane bonds (eq 2). It contributes
0.16 ( 0.06 mmol of tSiMe/g of silica. A mass balance was
sought by quantifying the methyl groups liberated as methane
upon exhaustive hydrolysis of the GaMe3-modified silica. Only
a trace of methane was detected upon addition of excess water
vapor at room temperature. This is consistent with the well-
known stability of dimethylgallium(III) fragments to protonoly-
sis.14,37,38 However, after the material was heated to 200 °C for
3 h, 1.96 ( 0.07 mmol of MeH/g of silica was evolved,
corresponding to 1.88 MeH/Ga. If all gallium sites can be
described by the same empirical formula, tSiOGaMe2, regard-
less of their grafting origin via reactions such as eqs 1 and 2,
this result suggests that a very small amount of additional
The highly air-sensitive sample powders were packed under a
dry N2 atmosphere into aluminum plates with slots 10 × 4 × 2
mm, between windows of 12 µm polypropylene film (Chemplex
#475) affixed with double-sided tape to each side of the plate. To
avoid spectral artifacts due to thickness effects, [GaMe2(µ-OSiPh3)]2
was diluted with BN to ca. 10 wt % Ga. GaMe3/SiO2 samples did
not require dilution. Spectra were recorded at 10 K in an Oxford
Instruments liquid He flow cryostat. Three data sweeps (total
collection time ∼1 h) were averaged to improve the signal-to-noise
ratio. Fluorescence data were generally of better spectral quality
than transmission data and were therefore used in data analysis.
EXAFS Data Analysis. Data processing and analysis were
performed using WinXAS (version 3.1).31,32 XAS spectra were
background-corrected by subtracting a linear fit to the preedge
region extrapolated to the length of the entire spectrum and then
normalized by a seventh-degree polynomial fitted to the postedge
region. EXAFS spectra were k3-weighted and fitted by a polynomial
spline with 6 knots between 1.5 and 14.5 Å-1. A Bessel window
(R ) 4) was applied to the data before Fourier transformation to R
space. The data were fitted to single-scattering paths using the
EXAFS equation (eq 3) with least-squares refinement.33 Ni is the
(31) Ressler, T. J. Phys. IV 1997, 7, 269-270.
(32) Ressler, T.; Wong, J.; Roos, J. J. Synchroton Radiat. 1999, 6, 656-
658.
NiFi(k)
-2Ri
λ(k)
2
ø(k) ) S0
exp(- 2k2σi2) exp
sin(2kRi + φi(k))
(33) Sayers, D. E.; Stern, E. A.; Lytle, F. W. Phys. ReV. Lett. 1971, 27,
∑
(
)
kRi2
1204-1207.
i
(34) Ankudinov, A. L.; Ravel, B.; Rehr, J. J.; Conradson, S. D. Phys.
ReV. B 1998, 58, 7565.
(3)
(35) Ankudinov, A. L.; Bouldin, C.; Rehr, J. J.; Sims, J.; Hung, H. Phys.
ReV. B 2002, 65, 104107-11.
(28) SMART, version 5.1; Bruker Advanced X-ray Solutions, Inc.,
Madison, WI, 1999.
(29) SAINT, version 5.1; Bruker Advanced X-ray Solutions, Inc.,
Madison, WI, 1999.
(30) Sheldrick, G. M. SHELXTL, version 6.12; Bruker Advanced X-ray
Solutions, Inc.: Madison, WI, 2001.
(36) Taha, Z. A. Ph.D. Thesis, University of Ottawa, 2004.
(37) Beachley, O. T.; Kirss, R. U.; Bianchini, R. J.; Royster, T. L.
Organometallics 1987, 6, 724-727.
(38) Chesnut, R. W.; Cesati, R. R.; Cutler, C. S.; Pluth, S. L.;
Katzenellenbogen, J. A. Organometallics 1998, 17, 4889-4896.

1894 Organometallics, Vol. 25, No. 8, 2006
Taha et al.
IR spectrum of GaMe3 (2961 and 2911 cm-1).41 The methyl
deformation mode δas(CH3) appeared at 1420 cm-1, while
δs(CH3) was obscured by the intensely absorbing lattice modes
of the silica below 1300 cm-1. However, by using spectral
subtraction to remove contributions from silica lattice modes,
its overtone 2δs(CH3) was detected at 2407 cm-1. IR bands
below 1300 cm-1 were observed by preparing a thin film of
the silica supported on a ZnSe window (see Figure S1 in the
Supporting Information). All assignments are summarized in
Table S1; they are consistent with those reported by Morrow
and McFarlane,14 who demonstrated that the attributions of an
earlier study13 to tSiMe vibrations were in error.
Virtually identical IR spectra were recorded using Aerosil
A300-100 (Figure S2, Supporting Information), except that the
stretching modes of unreacted surface hydroxyls after treatment
of the silica with excess GaMe3 were more intense. This is
consistent with a report that the fraction of inaccessible hydroxyl
groups is greater on A380-100 than on A380-500.39
Figure 1. Room-temperature IR spectra of A380 silica (a) after
thermal pretreatment at 500 °C, (b) after reaction with excess
GaMe3, and (c) after exposure of the GaMe3-modified silica to
excess water vapor at 200 °C, in the ν(O-H) and ν(C-H) regions.
All spectra were recorded under vacuum, after removal of volatiles,
and are vertically offset for clarity.
When GaMe3-modified silica was exposed to excess water
vapor at room temperature, the stretching vibration of the t
SiO-H groups at 3747 cm-1 did not reappear, and only a trace
of methane was observed in the gas phase. When the sample
was heated at 200 °C in the presence of adsorbed water for 2
h, a large amount of methane was observed in the gas phase,
and almost all of the intensity attributed to methyl vibrations
disappeared. A sharp band at 3743 cm-1 appeared, accompanied
by a broad band centered at 3670 cm-1 (Figure 1c). The former
is assigned to regenerated silica silanols. The latter may contain
contributions from perturbed silanols as well as ν(GaO-H)
stretching modes, since its frequency is similar to that reported
for the hydroxyl-terminated surface of Ga2O3.42
After hydrolysis, two very small peaks remained in the C-H
stretching region, at 2983 and 2925 cm-1. They are assigned to
the asymmetric and symmetric methyl stretching modes of t
SiMe sites.14,43 Assuming that the extinction coefficients for the
stretching vibrations of different types of methyl groups are
similar, integration of the absorbance in the ν(C-H) region
before and after hydrolysis allows the fraction of the original
methyl groups present as non-hydrolyzable tSiMe to be
estimated at ∼5%.
tSiMe (0.12 ( 0.08 mmol/g) was formed at the elevated
temperature required to effect hydrolysis.
Similar results were obtained at a lower silica pretreatment
temperature, corresponding to a higher initial density of surface
hydroxyl groups (but no adsorbed water). The methane yield
upon reaction of A380-100 with excess GaMe3 was 1.35 ( 0.05
mmol/g of silica. This is consistent with a population of
accessible hydroxyls that is ca. 45% greater on Aerosil pretreated
at 100 °C compared to 500 °C.36 The material contained 9.66
( 0.57 wt % Ga, or 1.36 ( 0.08 mmol of Ga/g of sample.
Adjusting for the mass gained upon grafting (13.7 wt % for the
GaMe2(-H) fragment), the Ga uptake was 1.55 ( 0.09 mmol/g
of silica. The difference between the amount of chemisorbed
Ga, 1.55 mmol/g of silica, and the methane yield upon grafting,
1.35 mmol/g of silica, is again attributed to the reaction of
GaMe3 with siloxane bonds (13 ( 6%), producing 0.20 ( 0.10
mmol of tSiMe/g of silica. Thus, the ratio of grafting pathways
(silanol vs siloxane) does not depend on the silica pretreatment
temperature. Exhaustive hydrolysis with excess water vapor at
200 °C resulted in liberation of 2.84 ( 0.10 mmol of MeH/g
of silica, corresponding to 1.83 ( 0.07 MeH/Ga. The missing
methane is again ascribed to the formation of tSiMe during
hydrolysis (0.26 ( 0.07 mmol/g of silica).
Infrared Characterization. Infrared spectra of a self-
supporting disk of A380-500, before and after its reaction with
GaMe3, are shown in Figure 1. The sharp band at 3747 cm-1
due to the stretching mode ν(SiO-H) of the isolated surface
hydroxyl sites of silica disappears almost completely during the
reaction. Only a weak, broad absorption band remains, centered
at approximately 3665 cm-1 (Figure 1b). It is assigned to internal
silanols, perturbed by the silica matrix,39 that are unreactive
toward GaMe314 and many other inorganic hydrogen-sequester-
ing agents, including AlMe3, TiCl4, and VOCl3.23,40 They are
therefore described as inaccessible silanols. The IR evidence
suggests that the accessible hydroxyl groups of silica are
completely consumed by reaction with excess GaMe3 at room
temperature.
Solid-State NMR. The 1H MAS NMR spectrum of GaMe3-
modified A380-500 (Figure 2a) consists of a broad signal with
a maximum at -1.0 ppm. This peak is assigned to Ga-CH3
and is shielded relative to the corresponding signals in the
molecular analogues [Ga(CH3)2(µ-OSiPh3)]2, -0.55 ppm, and
{[(c-C5H9)7Si7O11(OSiMePh)2)]2(Ga(CH3)2)4}, -0.20 ppm.44
The signal for Si-CH3 sites is expected to appear in the same
region; however, it was not resolved until after hydrolysis (see
below). The 13C CP-MAS NMR spectrum of the same sample
is shown in Figure 3a. A peak at -4.4 ppm is assigned to Ga-
CH3, also shielded relative to the signals for [Ga(CH3)2(µ-
OSiPh3)]2, -2.12 ppm, and {[(c-C5H9)7Si7O11(OSiMePh)2)]2-
(Ga(CH3)2)4}, -1.91 ppm.44 A shoulder at ca. -7 ppm is
tentatively assigned to tSiCH3 sites. For comparison, a chemi-
cal shift of -5.2 ppm was reported for monomethylated T1 sites
(SiO)3SiCH3 on silica.45
(41) Kvisle, S.; Rytter, E. Spectrochim. Acta, Part A 1984, 40A, 939-
951.
(42) Vimont, A.; Lavalley, J. C.; Sahibed-Dine, A.; Otero Area´n, C.;
Rodr´ıguez Delgado, M.; Daturi, M. J. Phys. Chem. B 2005, 109, 9656-
9664.
(43) Morrow, B. A.; McFarlan, A. J. J. Non-Cryst. Sol. 1990, 120, 61-
71.
After evacuation of volatiles (i.e., methane and unreacted
GaMe3), two new bands were observed in the ν(C-H) region
at 2968 and 2914 cm-1 (Figure 1b). They are assigned to the
asymmetric and symmetric stretching modes, respectively, of
methyl groups bonded to Ga, by comparison to the gas-phase
(44) Gerritsen, G.; Duchateau, R.; van Santen, R. A.; Yap, G. P. A.
Organometallics 2003, 22, 100-110.
(45) Tao, T.; Maciel, G. E. J. Am. Chem. Soc. 2000, 122, 3118-3126.
(39) Morrow, B. A.; McFarlan, A. J. Langmuir 1991, 7, 1695-1701.
(40) Rice, G. L.; Scott, S. L. Langmuir 1991, 7, 1695-1701.

Reaction of Trimethylgallium with Silica
Organometallics, Vol. 25, No. 8, 2006 1895
Figure 2. 1H MAS NMR spectra of (a) GaMe3-modified A380-
500 and (b) the sample in (a) after exposure to water vapor at 200
°C, followed by evacuation at the same temperature.
Figure 4. 29Si CP-MAS NMR spectra of (a) GaMe3-modified
A380-500 and (b) the sample in (a) after exposure to water vapor
at 200 °C, followed by evacuation at 200 °C.
Chart 1. Structures Considered for Silica-Supported
GaMe3, Consistent with the Empirical Formula tSiOGaMe2
Figure 3. 13C CP-MAS NMR spectra of (a) GaMe3-modified
A380-500 and (b) the sample in (a) after exposure to water vapor
at 200 °C, followed by evacuation at the same temperature.
Addition of water vapor to GaMe3-modified A380-500 and
heating at 200 °C for 2 h followed by evacuation at the same
1
temperature resulted in a dramatic decrease in the H NMR
signal intensity at -1.0 ppm and the appearance of a new signal
at +0.7 ppm (Figure 2b). The latter is assigned to Ga(µ-OH)-
Ga, by comparison to the chemical shift of 0.77 ppm reported
for [tBu2Ga(µ-OH)]3.46,47 The tSiOH signal appeared as a broad
shoulder centered at 2 ppm.48 The persistence of a small signal
at -1.0 ppm is consistent with the presence of tSiCH3 sites
that are stable toward hydrolysis.49 In the 13C CP-MAS NMR
spectrum of the same sample, the signal at -4.4 ppm completely
disappeared, exposing a small peak at -8.1 ppm (Figure 3b).
It is assigned to nonhydrolyzable tSiCH3 sites on the surface.
The 29Si CP-MAS NMR spectrum of GaMe3-modified
A380-500 shows a broad signal at -106 ppm assigned to both
(SiO)4Si and (SiO)3(GaO)Si sites (Figure 4a). The resonance at
-60 ppm is assigned to tSiMe. For comparison, a chemical
shift of -62 ppm was reported for T1 sites on silicas treated
with a variety of methylating agents.45 After hydrolysis, the peak
centered at -105 ppm appears narrower and more intense
(Figure 4b), presumably due to the contribution of more
efficiently cross-polarized Q3 sites (SiO)3SiOH formed by the
hydrolytic cleavage of SiOGa linkages.50 The tSiMe signal
shifts downfield, to -49 ppm. Its new position suggests that it
corresponds to (tSiO)2Si(OH)(Me) sites, for which a chemical
shift of -52 ppm has been reported.45 Thus, the nonhydrolyzable
Si-C bonds are found in sites that also possess a gallium
neighbor, (tSiO)2(GaO)SiMe, to which the prehydrolysis peak
at -60 ppm is attributed.
No 71Ga signal was detected for the unhydrolyzed sample
under MAS conditions, even after 48 h of acquisition. This
absence of signal intensity is attributed to locally distorted Ga
environments, resulting in large quadrupolar coupling constants.
Similar negative results have been reported previously for other
GaMe3-modified oxide materials.7,9
EXAFS Analysis of GaMe3-Modified Silica. Several struc-
tures consistent with the empirical formula of the gallium sites,
tSiOGaMe2, are shown in Chart 1. Model A was proposed in
previous studies of the reaction between GaMe3 and amorphous
silica and a zeolite.10,13,14 Molecular organogallium(III) alkoxides
and siloxides are well-known to aggregate, or to form adducts
with Lewis bases, to render gallium four-coordinate.44,51-54 The
(46) Naiini, A. A.; Yong, V.; Han, Y.; Akinc, M.; Verkade, J. G. Inorg.
Chem. 1993, 32, 3781-3782.
(47) Storre, J.; Klemp, A.; Roesky, H. W.; Schmidt, H.-G.; Noltemyer,
R.; Fleischer, R.; Stalke, D. J. Am. Chem. Soc. 1996, 118, 1380-1386.
(48) Bronnimann, C. E.; Zeigler, R. C.; Maciel, G. E. J. Am. Chem. Soc.
1988, 110, 2023-2026.
(49) Uhl, W.; Hannemann, F. J. Organomet. Chem. 1999, 579, 18-23.
(50) Haukka, S.; Root, A. J. Phys. Chem. 1994, 98, 1695-1703.

1896 Organometallics, Vol. 25, No. 8, 2006
Taha et al.
Figure 5. Fourier transform magnitude EXAFS in non-phase-
corrected R space for (a) GaMe3-modified A380-100 and (b)
GaMe3-modified A380-500, at the Ga K-edge. The spectra are
vertically offset for clarity.
Figure 6. Structure of [Me2Ga(µ-OSiPh3)]2. Thermal ellipsoids
are drawn at the 30% probability level. Hydrogen atoms are omitted
for clarity. Selected bond lengths (Å): C(1)-Ga, 1.953(4); C(2)-
Ga, 1.943(3); Ga-O#1, 1.976(2); Ga-O, 1.9800(19); Ga-Ga#1,
2.9813(7); O-Si, 1.656(2). Selected angles (deg): C(2)-Ga-C(1),
127.50(18); C(2)-Ga-O#1, 110.95(14); C(1)-Ga-O#1, 109.76(13);
C(2)-Ga-O, 105.85(12); C(1)-Ga-O#1, 111.23(13); O#1-Ga-
O, 82.21(8).
predominance of electron-deficient three-coordinate Ga centers
on silica (model A) therefore seems unlikely. Coordination of
a neighboring siloxane oxygen is represented in model B.
However, the considerably more electron-rich siloxane oxygens
of silsesquioxanes do not coordinate to dimethylgallium(III)
siloxides; instead, dimeric structures with bridging siloxide
ligands are formed.55 A silica analogue, with bridging siloxides
derived from the hydroxyl groups of the silica surface, is shown
in model C. If the spatial distribution of surface hydroxyls on
silica is not conducive to the formation of C, a digallium
structure may be formed with bridging methyl ligands, D.
Evidence for the presence of one or more of these structures
was sought using EXAFS.
The k3-weighted Fourier-transformed EXAFS spectra for
GaMe3-modified A380-100 and A380-500 silicas, recorded at
10 K, look very similar (Figure 5), suggesting that the same
sites dominate on both surfaces. The proposed models shown
in Chart 1 were evaluated by curve-fitting these spectra. For
each model, the coordination numbers were fixed at integer
values and interatomic distances were constrained to be >1.8
Å.56 The goodness of fit was evaluated by the magnitude of
the residual and, more importantly, the plausibility of the fitted
bond distances and Debye-Waller factors.
When model A was refined to the data, the Debye-Waller
factor for its Ga-O path (N ) 1) was σ2 ) -9.6 × 10-4 Å2
(see Table S2, Supporting Information). Since negative thermal
disorder is not possible, this result suggests that a first
coordination sphere consisting of only one O and two C
neighbors cannot generate enough intensity for the peak at 1.6
Å in R space. Including a second, independent Ga-O path (i.e.,
model B) generated physically meaningful values for all
Debye-Waller factors (see Table S3). This model accounts for
the first coordination sphere of Ga (i.e., C2O2) but provides no
intensity for the peak at 2.7 Å.
Table 2. Crystal Data for [(GaMe2)(µ-OSiPh3)]2
formula
fw
C20H21GaOSi
375.18
cryst syst
triclinic
space group
a, Å
P1h
8.7861(11)
9.5628(12)
12.8261(16)
69.910(2)
74.453(2)
64.222(2)
902.3(2)
2
b, Å
c, Å
R, deg
â, deg
γ, deg
V, Å3
Z
µ(Mo KR), mm-1
cryst size, mm
total no. of data points
no. of unique data points
GOF
1.593
0.2 × 0.15 × 0.08
6534
3886
1.058
R1, wR2 (I > 2σ(I))
0.0402, 0.1098
In a recent XAS study of GaMe3 adsorbed in ZSM-5,9 a peak
at R ) 2.9 Å was assumed to arise from a Si or Al backscatterer
of the zeolite support, although no curve fitting of this path
was attempted. Modifying model B to include a Si atom at this
distance caused the Debye-Waller factor expand to an unre-
alistically large value (∼0.04 Å2). However, the presence of a
Ga backscatterer (i.e., model C) resulted in a good fit to both
the imaginary component and the FT magnitude. Our hypothesis
that the EXAFS contains a Ga-Ga path led us to synthesize a
molecular dimer of dimethylgallium and to analyze its EXAFS
under similar conditions.
X-ray Diffraction Study and X-ray Absorption Spectro-
scopy of [GaMe2(µ-OSiPh3)]2. Although [GaMe2(µ-OSiPh3)]2
has been previously reported57 and its dimeric nature inferred
by measurement of its molecular weight, structural parameters
were not determined. It was synthesized by the reaction of
equivalent amounts of Ph3SiOH and GaMe3. The results of
single-crystal X-ray diffraction are shown in Figure 6. Crystal
data and details of the structure refinement are reported in Table
2. The molecule consists of two identical GaMe2 fragments
bridged by two triphenylsiloxide ligands, such that the geometry
(51) Schmidbaur, H. Angew. Chem., Int. Ed. Engl. 1965, 3, 201-211.
(52) Murugavel, R.; Voigt, A.; Walawalkar, M. G.; Roesky, H. W. Chem.
ReV. 1996, 96, 2205-2236.
(53) Keys, A.; Barbarich, T. J.; Bott, S. G.; Barron, A. R. Dalton Trans.
2000, 577-588.
(54) Oliver, J. G.; Worrall, I. J. J. Chem. Soc. A 1970, 845-848.
(55) Duchateau, R.; Dijkstra, T. W.; van Santen, R. A.; Yap, G. P. A.
Chem. Eur. J. 2004, 10, 3979-3990.
(56) Constraining the bond distances to g1.8 Å was necessary because
of a low-R shoulder on the first maximum in R space, possibly due to
incomplete background removal of the atomic XAFS. Without this
constraint, path lengths for some scatterers in the first coordination sphere
of Ga were refined to physically implausible values: i.e., R < 1.7 Å.
(57) Schmidbaur, H.; Schindler, F. Chem. Ber. 1966, 99, 2178-2186.

Reaction of Trimethylgallium with Silica
Organometallics, Vol. 25, No. 8, 2006 1897
Table 3. Comparison of Bond Distances Obtained by
Single-Crystal X-ray Diffraction and EXAFS Curve Fittinga
for [(GaMe2)(µ-OSiPh3)]2
EXAFSb
σ2 (Å2)
XRD
R (Å)
path
N
R (Å)
∆E0 (eV)
Ga-C
Ga-O
Ga-Ga
Ga-Si
1.943, 1.953
1.976, 1.980
2.981
2
2
1
1.94
1.99
2.98
0.0027
0.0070
0.0079
2.01
1.15
1.12
3.294, 3.303
a Errors for single-scattering fits, in the absence of systematic fitting
uncertainties, are generally to be accepted as follows: bond lengths, (0.02
2
Å; σ2, (20%; ∆E0, (20%. b S0 ) 0.74; residual 6.6.
2
∆E0 and S0 ) were allowed to refine freely. The curve fit is
shown in both k3-weighted k space and R space in Figure 7.
The same fitted parameters were obtained when the fit was
performed with either k3- or k1-weighted data. Path lengths for
the single-scattering model containing one C shell (N ) 2), one
O shell (N ) 2), and one Ga shell (N ) 1) are compared with
the bond distances obtained by X-ray diffraction in Table 3.
The agreement is well within the experimental uncertainty of
the EXAFS technique, validating our curve-fitting approach.
Allowing either the Ga-C or Ga-O paths (each with N ) 2)
to be refined independently (i.e., as two shells with N ) 1)
returned the same distances for each type of scatterer (within
0.01 Å). When the coordination numbers were also refined, their
fitted values were within 10% of the required integers (N is
strongly correlated with σ2).
The model described above generates no intensity for the
shoulder apparent in R space at 3.1 Å. Due to the phase
correction, this feature may represent a backscatter at 3.3-3.4
Å. We attempted to refine the Ga-Si path (N ) 2) by fixing
the Ga-Si distance at its crystallographic value, 3.30 Å.
However, the associated value of σ2 increased to >0.04 Å2
(Table S4, Supporting Information). This result implies that a
Ga-Si path does not contribute significantly to the EXAFS.
We also investigated whether the feature at 3.1 Å could arise
from a multiple scattering path (Ga-O-Ga, Ga-O-O, or Ga-
C-O). However, inclusion of any combination of these paths
was not justified by a statistically significant improvement in
the fit.
Structure of GaMe3-Modified Silica. The EXAFS of GaMe3
supported on A380-100 and A380-500 was analyzed by curve
fitting to the same single-scattering model used to fit the EXAFS
of [GaMe2(µ-OSiPh3)]2. The results are shown for A380-500
in Figure 8; parameters obtained from the fits are reported in
Table 4. The fitted values of the Ga-C and Ga-O paths and
their associated Debye-Waller factors are very similar to those
obtained for the molecular complex, and the inner-potential
shifts ∆E0 are all small, <|1.2| eV. The peak at 2.7 Å is
generated by the Ga backscatterer. When the coordination
numbers were refined, values close to the expected integers were
again found (2.2, 1.9, and 1.0 for C, O, and Ga shells,
respectively; see Table S5, Supporting Information), suggesting
that the sites are highly uniform. We were again unable to refine
a Ga-Si path to the EXAFS data. Silicon is often reported to
be a weak EXAFS backscatterer, and our previous attempts to
locate Si-containing paths for silica-supported metal complexes
were unsuccessful.58 Inclusion of multiple-scattering paths to
the refinement yielded no statistically significant results.
Figure 7. Ga K-edge EXAFS for [GaMe2(µ-OSiPh3)]2 in k3-
weighted k space (top; red line) and non-phase-corrected R space
(bottom; imaginary data given as black points and FT magnitude
as red points). The curve fit to the single-scattering model based
on the crystal structure is shown (blue line).
at gallium is distorted tetrahedral. The Ga-C distances, at
1.943(3) and 1.953(4) Å, are comparable to the average distance
reported for the oligomeric silsesquioxane analogue {[(c-C5H9)7-
Si7O11(OSiMePh2)]2(GaMe2)4}, 1.947 Å.39 The Ga-Ga non-
bonded distance of 2.98 Å is close to that reported for
[(c-C5H9)7Si7O10(OSiPh2O)(GaMe2)]2, at 2.99 Å.55 The bridging
triphenylsiloxide ligands are almost symmetrically disposed
between the two gallium atoms, with Ga-O and Ga-O#1
distances of 1.9800(19) and 1.976(2) Å, respectively. These
distances, as well as the C-Ga-C and O-Ga-O bond angles,
are also comparable to those previously observed for dimeth-
ylgallium(III) with silsequioxane ligands.44,55 Only the Si-O
distance, at 1.656(2) Å, is slightly longer than that reported for
the silsesquioxane complexes, 1.617(3) Å. This is likely a
consequence of the different silicon coordination environments.
The EXAFS of [GaMe2(µ-OSiPh3)]2 was recorded at the Ga
K-edge at 10 K, to directly compare EXAFS-derived structural
parameters with those obtained by single-crystal X-ray diffrac-
tion. The EXAFS spectrum is shown in Figure 7. Its resemblance
to the spectrum of GaMe3-modified silica, including the
prominent peak at 2.7 Å, is striking. The model used in curve
fitting was constructed from the crystallographic parameters for
[GaMe2(µ-OSiPh3)]2. Coordination numbers were fixed at
integer values, while bond distances were constrained to vary
by no more than (0.15 Å. All other parameters in the fit (σ2,
Finally, the fitted Ga-C path lengths are not compatible with
a digallium surface complex bridged by methyl ligands, model
(58) Deguns, E. W.; Taha, Z.; Meitzner, G. W.; Scott, S. L. J. Phys.
Chem. B 2005, 109, 5005-5011.

1898 Organometallics, Vol. 25, No. 8, 2006
Taha et al.
SiO2, the fitted Ga-C path length of 1.95 Å is consistent with
terminal methyl ligands. In sum, the EXAFS supports a siloxide-
bridged dimethylgallium dimer (model C in Scheme 1) for the
structure of GaMe3-modified A380 silica at 10 K.
Curve fitting of the EXAFS of GaMe3-modified A380-100
(Figure S7, Supporting Information) gave fit parameters identical
with those obtained for A380-500, within experimental uncer-
tainty (Table 4). Thus, the coordination environment of Ga is
independent of the initial hydroxyl density of the silica. The
Debye-Waller factors for the Ga-Ga paths in both silica-
supported samples are the same, 0.009 Å2, and the Debye-
Waller factor for the same path in [GaMe2(µ-OSiPh3)]2 is very
similar (0.008 Å2). This result requires that the Ga2O2 rings in
all three materials be of similar size and flexibility, even though
the orientation of the bridging “siloxide” ligand substituents is
trans in the molecular silanolate but is presumably cis for the
silica-supported complexes.
The simplest explanation for the uniformity of the grafted
sites is that the reactive hydroxyls are disposed vicinally on
the silica surface, regardless of the average hydroxyl density.
Such a spatial arrangement is consistent with reports of the
reactions of metal alkyls MR4 on silicas partially dehydroxylated
at e200 °C.61-65 It is plausible that such sites persist on silica
partially dehydroxylated at 500 °C, since vicinal Q3 sites in
amorphous silicas are unable to form intramolecular hydrogen
bonds,66 nor do they condense to form four-membered siloxane
rings at temperatures below 650 °C.67 Vicinal sites are indis-
tinguishable from truly isolated Q3 sites by either IR or 29Si
NMR spectroscopy.
A proposed mechanism for the major grafting reaction,
consistent with that originally envisaged in eq 1 but revised to
incorporate vicinal Q3 sites, is shown in Scheme 1. Reaction
with GaMe3 yields the digallium site I. Grafting of GaMe3 on
a small fraction of isolated Q3 sites is proposed to give the less
stable (tSiO)3SiOGaMe2, site II, which may interact with an
adjacent siloxane oxygen. Since the fitted coordination number
N for the Ga-Ga path was found to be 0.96 (Table S5), we
propose that site II spontaneously promotes the cleavage of an
adjacent siloxane bond by GaMe3 to give the digallium site III
and a methylated silicon. The amount of silicon methylation
observed during grafting (Table 1) allows us to estimate the
ratio of vicinal to isolated Q3 sites to be 6.8 ( 2.4 on A380-
100 and 4.5 ( 1.7 on A380-500.
The driving force for siloxane cleavage is presumably the
strong preference of gallium to achieve four-coordination
through bridging siloxide ligands, rather than through either
bridging methyl or siloxane ligands. Ga-induced siloxane
cleavage has precedent in the reaction of the dimer of tert-
butoxygallane with tetraphenyl-1,3-disiloxanediol: one of the
gallosiloxane products is a result of ring expansion to give a
[GaO4Si3]2 framework.68 Finally, hydrolysis of the Ga-O bonds
of site III accounts for the formation of (tSiO)2MeSiOH, which
Figure 8. Ga K-edge EXAFS for GaMe3-modified A380-500 in
k3-weighted k space (top; red line) and non-phase-corrected R-space
(bottom; imaginary data given as black points and FT magnitude
as red points). The curve fit to the single-scattering model for
[GaMe2(µ-OSit)]2 is shown (blue line).
Table 4. EXAFS Parameters for Single-Scattering Paths in
Model C, [(GaMe2)(µ-OSit)]2, Fitted to Data Obtained for
GaMe3-Modified A380-100 and A380-500a
GaMe3/A380-100b
GaMe3/A380-500c
path
Nd R (Å) σ2 (Å2) ∆E0 (eV) Nd R (Å) σ2 (Å2) ∆E0 (eV)
Ga-C
Ga-O
Ga-Ga
2
2
1
1.95 0.0045
1.98 0.0071
2.99 0.0090
0.65
-0.35
1.06
2
2
1
1.95 0.0032
1.98 0.0089
2.97 0.0088
1.19
-0.88
0.70
a Errors for single-scattering fits, in the absence of systematic fitting
uncertainties, are generally to be accepted as follows: bond lengths, (0.02
Å; σ2, (20%; ∆E0, (20%. b S02 ) 0.78; residual 6.1. c S02 ) 0.78; residual
6.6. d Coordination numbers were fixed at integer values during fit
refinement.
D. Examples of molecular complexes of Ga containing bridging
methyls are rare, and GaMe3 itself shows little tendency to
dimerize.16 However, in heterobimetallic complexes containing
Me2Ga(µ-Me)nM (M ) Zr, Nd, La; n ) 1, 2), the Ga-Cbridging
distances (2.033-2.142 Å) are always significantly longer than
the Ga-Cterminal distances (1.968-1.984 Å).59,60 For GaMe3/
(61) Ballard, D. G. H. AdV. Catal. 1973, 23, 263-325.
(62) Candlin, J. P.; Thomas, H. AdV. Chem. Ser. 1974, No. 132, 212-
239.
(63) Yermakov, Y.; Zakharov, V. AdV. Catal. 1975, 24, 173-219.
(64) Schwartz, J.; Ward, M. D. J. Mol. Catal. 1980, 8, 465-469.
(65) Amor Nait Ajjou, J.; Scott, S. L. Organometallics 1997, 16, 86-
92.
(66) Bunker, B. C.; Haaland, D. M.; Michalske, T. A.; Smith, W. L.
Surf. Sci. 1989, 222, 95-118.
(67) Morrow, B. A.; Cody, I. A. J. Phys. Chem. 1976, 80, 1998-2004.
(68) Veith, M.; Vogelgesang, H.; Huch, V. Organometallics 2002, 21,
380-388.
(59) Erker, G.; Albrecht, M.; Kru¨ger, C.; Werner, S. J. Am. Chem. Soc.
1992, 114, 8531-8536.
(60) Evans, W. J.; Anwander, R.; Doedens, R. J.; Ziller, J. W. Angew.
Chem., Int. Ed. Engl. 1994, 33, 1641-1644.

Reaction of Trimethylgallium with Silica
Organometallics, Vol. 25, No. 8, 2006 1899
Scheme 1. Proposed Grafting Mechanism for GaMe3 on
Vicinal and Isolated Q3 Hydroxyl Sites on the Silica Surface
Conclusions
The reactions of GaMe3 with the surface of a fumed silica,
partially dehydroxylated at either 100 or 500 °C, results in
grafted sites with the empirical formula tSiOGaMe2. Two
reaction pathways were inferred. The major pathway involves
the reaction of GaMe3 with vicinal surface hydroxyl groups. A
minor pathway involves grafting of GaMe3 onto isolated
hydroxyl groups, followed by reaction with a second equivalent
of GaMe3 via siloxane cleavage. The EXAFS of the GaMe3-
modified silicas contains clear evidence for a well-defined Ga-
Ga scattering path, identified as a rigid Ga2O2 ring by
comparison to the EXAFS of [GaMe2(µ-OSiPh3)]2. All of the
digallium fragments are formulated as [GaMe2(µ-OSit)]2,
tethered by two bridging siloxide ligands to the silica surface.
The mechanisms of transformations of these digallium sites (e.g.,
to supported gallium chlorides or hydrides by treatment with
HCl or H2, respectively)7,10,11,13 and their subsequent catalytic
activity, previously assumed to be noninteracting monogallium
sites, will need to be reinterpreted in this light. Finally,
considering the similarities in GaMe3 and AlMe3 chemistry, it
seems likely that the latter will also react to form four-coordinate
dialuminum sites on silica.23
Acknowledgment. This work was funded in part by the
U.S. Department of Energy, Basic Energy Sciences, Catalysis
Science (Grant No. DE-FG02-03ER15467). Portions of this
research were carried out at the Stanford Synchrotron Radia-
tion Laboratory, a national user facility operated by Stan-
ford University on behalf of the U.S. Department of Energy,
Office of Basic Energy Sciences. This work also made use of
MRL Central Facilities supported by the MRSEC Program of
the National Science Foundation under Award No. DMR00-
80034.
Supporting Information Available: Tables and figures giv-
ing additional IR spectra and EXAFS curve fits and a CIF
files giving crystallographic data for [Me2Ga(µ-OSiPh3)]2.
This material is available free of charge via the Internet at
http://pubs.acs.org.
causes the shift in the 29Si CP-MAS signal from -60 to -49
ppm.
OM051034O