Syntheses and structures of metal-metal triply bonded M2R6 compounds: Consideration of starting materials, stability, and structural parameters

Article abstract of DOI:10.1016/S0277-5387(98)00434-3

The syntheses and structures of four homoleptic metal-metal triply-bonded M₂R₆ compounds [Mo₂(CH₂CMe₂Ph)₆, 1; Mo₂(CH₂SiMe₂Ph)₆, 2; W₂(CH₂SiMe₂Ph)₆, 3; and W₂(CH₂Ph)₆, 4] are reported. The synthetic effort suggests that ditungsten compounds are inherently more difficult to prepare and more thermally sensitive than dimolybdenum compounds, probably as a result of the larger dimetal core the ligands must protect. The structural data confirm that dimetal hexaalkyls exhibit shorter M≡M distances than do dimetal hexaalkoxides, even in a matched pair case where steric differences are minimal.

Full text of DOI:10.1016/S0277-5387(98)00434-3

Polyhedron 18 (1999) 1303–1310
Syntheses and structures of metal–metal triply bonded M2R6 compounds:
consideration of starting materials, stability, and structural parameters
c
b,
Thomas M. Gilberta, , Cary B. Bauer , Robin D. Rogers
*
*
aDepartment of Chemistry and Biochemistry, Northern Illinois University, DeKalb, IL 60115, USA
bDepartment of Chemistry, The University of Alabama, Tuscaloosa, AL 35487, USA
cInstitute for Enzyme Research, University of Wisconsin, Madison, WI 53705, USA
Received 13 May 1998; accepted 6 November 1998
Abstract
The syntheses and structures of four homoleptic metal–metal triply-bonded M2R6 compounds [Mo2(CH2CMe2Ph)6, 1;
Mo2(CH2SiMe2Ph)6, 2; W2(CH2SiMe2Ph)6, 3; and W2(CH2Ph)6, 4] are reported. The synthetic effort suggests that ditungsten
compounds are inherently more difficult to prepare and more thermally sensitive than dimolybdenum compounds, probably as a result of
the larger dimetal core the ligands must protect. The structural data confirm that dimetal hexaalkyls exhibit shorter M;M distances than
do dimetal hexaalkoxides, even in a matched pair case where steric differences are minimal.
1999 Elsevier Science Ltd. All rights
reserved.
Keywords: Metal–metal dimer; Triple bond; Alkyl; Homoleptic; X-ray structure; Distal ligand
1. Introduction
respectively) [10], but the generality of these reactions has
not been examined.
Mo2(CH2SiMe3)6 was the first M2X6 compound (M5a
transition metal) to be synthesized, and the first shown
crystallographically to contain an unsupported metal–metal
triple bond [1]. Since then, numerous triply bonded M2X6
dimers containing pseudohalide ligands have appeared [2–
4]. However, the number of homoleptic dimetal hexaalkyls
has remained small; only M2(CH2CMe3)6 (M5Mo [5], W
[6]), W2(CH2SiMe3)6 [7], and Mo2(CH2Ph)6 [8] have
been added to the list, and only the last two were
characterized crystallographically.
One reason relatively few M2R6 compounds are known
is that suitable preparation methods have proved difficult
to find. Mo2(CH2CMe3)6, M2(CH2SiMe3)6 (M5Mo, W),
and Mo2(CH2Ph)6 were initially prepared from metal
halides and alkyllithium or Grignard reagents in 2–20%
yield. These reactions did not scale up well. More recently,
Rothwell prepared Mo2(CH2SiMe3)6 (74% yield) [9] and
Mo2(CH2Ph)6 (29%) [8] from Mo2(O-i-Pr)6, Chisholm
prepared W2(CH2CMe3)6 from NaW2Cl7(THF)5 (55%)
Our interest in this area arose from our Raman studies of
the metal–metal stretching frequency in M2X6 dimers
[11]. We needed a variety of M2R6 compounds, and felt
this warranted investigating the utility of ‘‘MoCl3(dme)’’
and NaW2Cl7(THF)5 in preparing them. We have found
that ‘‘MoCl3(dme)’’ is a good starting material for other
dimolybdenum hexaalkyls, but that preparing ditungsten
hexaalkyls from NaW2Cl7(THF)5 is problematic. Hexa-
(alkoxy)ditungsten compounds are much better precursors
for these. We report here four syntheses of dimetal
hexaalkyls containing phenyl groups, and the crystal
structures of the four compounds Mo2(CH2CMe2Ph)6, 1;
Mo2(CH2SiMe2Ph)6, 2; W2(CH2SiMe2Ph)6, 3; and
W2(CH2Ph)6, 4. The structural data confirm that dimetal
hexaalkyls exhibit the shortest metal–metal bond distances
of any member of the M2X6 class, even when comparing
compounds which are nearly isomorphous. Steric consid-
erations appear to determine most of the core and peripher-
al structural parameters.
[6],
and
we
prepared
Mo2(CH2CMe3)6
and
Mo2(CH2SiMe3)6 from ‘‘MoCl3(dme)’’ (73 and 84%,
2. Experimental
*
Corresponding authors.
Unless otherwise noted, all reactions and manipulations
0277-5387/99/$ – see front matter
PII: S0277-5387(98)00434-3
1999 Elsevier Science Ltd. All rights reserved.

1304
T.M. Gilbert et al. / Polyhedron 18 (1999) 1303–1310
were performed under an inert atmosphere employing
standard Schlenk or glove box techniques. Solvents were
evaporated in vacuo. Ether and hydrocarbon solvents were
distilled from purple potassium diphenyl ketyl. Me3SiCl
was distilled from CaH2. ‘‘MoCl3(dme)’’ [10] W2(OCy)6
[12], W2(OCMe3)6 [6], and LiCH2Ph [13] were prepared
by published methods. LiCH2CMe2Ph and LiCH2SiMe2Ph
were prepared analogously to LiCH2CMe3 [14].
boiling to ambient temperature. 1H NMR (C6D6): d 7.50–
7.18 (m, 5H, Ph H); 1.76 (s, 2H, CH2); 0.34 (s, 6H, CH3).
2.3. Synthesis of W (CH2SiMe2Ph)6 , 3
2
A stirring slurry of W2(OCy)6 (0.962 g, 1.00 mmol) in
pentane (100 ml) was cooled to 2208C and treated
dropwise over 45 m with a pentane solution (100 ml) of
LiCH2SiMe2Ph (0.956 g, 6.12 mmol). The yellow
cyclohexoxide slowly dissolved, giving a red, translucent
solution. After the addition of alkyllithium was completed,
the slurry was stirred for 1 h at 2208C, then allowed to
warm slowly to ambient temperature over 1 h. After 36 h
of stirring, the solution was filtered, and the solvent was
evaporated, giving a mixture of orange and white solids.
This was dissolved in ether, and the stirring solution was
treated with 3 ml Me3SiCl. Over the course of 1 h, white
solid precipitated. The solution was filtered to remove the
solid, and the volatiles were evaporated. The resulting
red-orange semisolid was dissolved in pentane and cooled
to 2308C. The orange powder which precipitated was
filtered out, washed with cold pentane and dried (0.356 g,
0.282 mmol, 28%). The sample for X-ray study was
crystallized from a saturated heptane solution at 2308C. 1H
NMR (C6D6): d 7.49–7.17 (m, 5H, Ph H); 1.76 (s, 2H,
CH2); 0.35 (s, 6H, CH3).
NMR spectra were obtained on a Bruker WA-200
spectrometer at ambient temperature; chemical shifts are
reported as ppm downfield of tetramethylsilane.
2.1. Synthesis of Mo2(CH2CMe2Ph)6 , 1
Solid LiCH2CMe2Ph (2.24 g, 16.0 mmol) was added in
small portions from a powder addition funnel to a stirring
slurry of ‘‘MoCl3(dme)’’ (1.46 g, 5.00 mmol) in 1:1
ether/pentane (120 ml) at 2408C over 30 m. The mixture
was then allowed to warm slowly to ambient temperature
over 2 h, giving a dark red solution over a goopy grey
precipitate. The solvent was evaporated, giving orange-
brown solid. This was extracted with toluene until the
extracts were colorless. The extracts were filtered through
a 5 cm pad of silica gel in a frit, giving a dark red-brown
solution and removing a sizable quantity of solid. The
solvent was evaporated, and the residue dissolved in 4320
ml portions of boiling heptane. The heptane extracts were
filtered, combined, and cooled to 2308C. The precipitated
product was filtered out, washed with cold pentane, and
dried, giving yellow microflakes (0.270 g, 0.272 mmol,
11%). The sample for X-ray study was crystallized by
slow cooling of a saturated heptane solution from near
boiling to ambient temperature. 1H NMR (C6D6): d 7.24–
7.10 (m, 5H, Ph H); 2.02 (s, 2H, CH2); 1.36 (s, 6H, CH3).
2.4. Synthesis of W (CH2Ph)6 , 4
2
A stirring solution of W2(OCMe3)6 (1.61 g, 2.00 mmol)
in ether (75 ml) was cooled to 2108C and treated dropwise
over 30 m with an ethereal solution of LiCH2Ph (70 ml of
0.180 M solution, 12.6 mmol). The solution reddened, then
became a red slurry as red microcrystals precipitated. After
the addition of alkyllithium was complete, the slurry was
stirred for 30 m at 2108C, then allowed to warm slowly to
ambient temperature over 1 h. The solvent was evaporated,
giving orange-brown solid. This was triturated with hep-
tane, filtered out, washed with pentane and dried, giving
the product as a deep red powder (1.63 g, 1.78 mmol,
89%). It is crucial to avoid heating this material at any
point, as it decomposes nearly completely after 1 h at
508C. The crystal for X-ray study was grown from a
2.2. Synthesis of Mo2(CH2SiMe2Ph)6 , 2
A pentane solution (100 ml) of LiCH2SiMe2Ph (2.50 g,
16.0 mmol) was added dropwise over 30 m to a stirring
slurry of ‘‘MoCl3(dme)’’ (1.46 g, 5.00 mmol) in 1:1
ether/pentane (60 ml) at 2408C. The mixture was then
allowed to warm slowly to ambient temperature over 2 h,
giving a dark red solution over a goopy grey precipitate.
The solvent was evaporated, giving yellow-black solid.
This was extracted with toluene until the extracts were
colorless. The extracts were filtered through a 5 cm pad of
silica gel in a frit, giving a clear, orange-yellow solution
and removing a sizable quantity of solid. The solvent was
evaporated, and the residue triturated with pentane. The
bright yellow powder was filtered out, washed with cold
pentane, and dried. A second crop was isolated by
evaporating the mother liquor to 1/3 the original volume,
and cooling it to 2308C. (Total yield: 0.970 g, 0.892
mmol, 36%). The sample for X-ray study was crystallized
by slow cooling of a saturated heptane solution from near
1
saturated toluene solution at 2308C. H NMR (C6D6): d
7.08–6.54 (m, 5H, Ph H); 3.64 (s, 2H, CH2).
2.5. Crystal structure determinations
Crystals were removed from the glovebox under a layer
of heavy mineral oil for microscopic examination. Selected
crystals were rapidly attached with silicone grease to a
glass pin held in the goniometer head, and were both
frozen in place and protected from the atmosphere by
immediately cooling them in a stream of cold nitrogen gas.
Space groups were determined by inspection of reflec-
tion data and confirmed by the successful solution of the

T.M. Gilbert et al. / Polyhedron 18 (1999) 1303–1310
1305
Table 1
Crystal data and structure refinement
1
2
3
4
Color and shape
Crystal size, mm
Formula
Temperature, K
Crystal system/space group
Reflections used for unit cell
(over full u range)
Yellow plate
0.3030.2030.10
C60H78Mo2
293 (2)
Triclinic/P1
430
Yellow fragment
0.1230.1030.10
C54H78Mo2Si6
173 (2)
Monoclinic/P21 /n
8201
Orange fragment
0.0930.2030.23
C54H78W2Si6
293 (2)
Monoclinic/P21 /n
5432
Red fragment
0.1030.2030.30
C42H42W2
173 (2)
Rhombohedral/R3
1768
¯
¯
Unit cell dimensions:
(A, deg)
a
b
c
a
b
g
12.4296 (14)
12.562 (2)
18.856 (2)
91.648 (5)
92.152 (5)
115.600 (4)
2
12.890 (10)
18.003 (7)
13.394 (8)
90
111.51 (3)
90
11.4155 (6)
13.7875 (7)
18.6577 (9)
90
92.16 (1)
90
14.7534 (6)
14.7534 (6)
13.3371 (10)
90
90
120
˚
Z
2
2
3
u range, deg
Index ranges
1.82–23.25
213#h#7
212#k#13
218#l#20
7933
6266 (Rint 50.0328)
6259
0.509
0.94–0.65
6259/0/572
0.0030 (4)
0.0649, 4.2361
1.88–23.29
212#h#14
29#k#18
214#l#7
5309
3791 (Rint 50.0537)
3443
0.590
0.9962–0.7683
3788/0/281
0.014 (2)
0.1301, 3.2094
1.84–23.47
210#h#12
215#k#14
220#l#19
10032
4152 (Rint 50.2052)
4145
4.071
0.9745–0.5430
4145/0/287
0.0006 (2)
0.0593, 2.2655
2.21–23.29
24#h#16
213#k#13
213#l#14
2058
805 (Rint 50.0305)
764
6.886
Reflections collected
Independent reflections
Observed reflectionsa
Absorption coefficient, mm21
Range of relat. transm. factors
Data/restraints/parameters
Extinction coefficient
SHELX-93 weight parametersb
R indices [I.2s(I)]c
R indices (all data)
GOFd
0.9183–0.5963
803/0/67
0.0085, 34.6981
R150.0388, wR250.1020 R150.0660, wR250.1699 R150.0508, wR250.1057 R150.0260, wR250.0574
R150.0484, wR250.1224 R150.0702, wR250.1781 R150.0790, wR250.1291 R150.0282, wR250.0594
0.951
1.038
1.043
1.128
a
Corrected for Lorentz/polarization effects and empirically corrected for absorption (c-scans); Io .2s(Io).
b
c
d
SHELXTL weight w5[s2(Fo)21(a*P)21b*P]21, where P5(Max (Fo2,0)12F2c )/3, and a and b are the values given.
R5SuuFou2uFcuu/SuFou; wR25hSw(F2o 2F2c )2 /Sw(Fo2)2j1 / 2
.
GOF5[Sw(F2o 2F2c )2 /(No 2Nv)]1 / 2; No 5number of observations, Nv 5number of variables.
structures. Data for 1–4 were collected on a Siemens
SMART diffractometer employing a CCD area detector,
using graphite monochromated Mo Ka radiation (l5
appear in Table 1. Solutions and refinements were carried
out using SHELXTL, refining on F2 values [15]. All
nonhydrogen atoms were refined with anisotropic tempera-
ture factors. Phenyl hydrogen atoms were placed in
calculated positions and allowed to ride on the bonded
carbon atom, with U(H)51.2*Ueqv (C). Methyl hydrogen
atoms were included as a rigid group with rotational
freedom at the bonded carbon atom, with U(H)51.2*Ueqv
(C). ORTEP-III [16] diagrams of 1–4 appear in Figs. 1–4,
respectively; selected distance and angle values appear in
Table 2.
˚
0.71073 A). Data collection and refinement parameters
3. Results and discussion
3.1. Mo2(CH2CMe2Ph)6 , 1
Treatment of ‘‘MoCl3(dme)’’ with three equivalents of
LiCH2CMe2Ph provides yellow 1 in poor yield. The low
yield is surprising given that the identical reaction between
the trichloride and either LiCH2CMe3 or LiCH2SiMe3
gives the hexa(neopentyl)- or hexa(trimethylsilylmethyl)-
dimolybdenum compounds in .70% yield [10]. Once
Fig. 1. ORTEP-III diagram of one of the independent molecules of
Mo2(CH2CMe2Ph)6, 1. Hydrogen atoms were removed for clarity.

1306
T.M. Gilbert et al. / Polyhedron 18 (1999) 1303–1310
isolated, 1 shows thermal and air stability similar to those
of the trimethylalkyl dimers.
In common with Mo2(CH2SiMe3)6 [1], 1 crystallizes
with more than one dimer in the asymmetric unit. The two
independent molecules exhibit identical structural parame-
ters within experimental error (Table 2). The distances and
angles associated with the M2C6 core are entirely con-
sistent with those reported for the hexa(tri-
methylsilylmethyl) dimer and Mo2(CH2Ph)6 [8], with
slight differences attributable to steric issues. For example,
the average M–C distances in Mo2(CH2SiMe3)6 [2.131
˚
˚
(?) A], 1 [2.135 (5, 9, 6) A] [17], and 2 [2.113 (5, 11, 3)
˚
A] are experimentally identical, and shorter than that in
Mo2(CH2Ph)6 [2.162 (2) A] [18]. We think the difference
˚
arises from the need for the six ligands to ‘‘protect’’ the
dimetal core; in Mo2(CH2Ph)6, containing the smallest
ligand, the metal–carbon distance lengthens to better allow
the benzyl group to do this. Evidence for this theory
appears in the angle data: while the Mo–Mo–C angles of
all four dimolybdenum hexaalkyls are similar [97.89 (14,
34, 6)8 for 1; 99.7 (2, 12, 3)8 for 2; 100.6 (?)8 for
Mo2(CH2SiMe3)6; 97.61 (6)8 for Mo2(CH2Ph)6], the Mo–
C–C(Si) angles track the relative sizes of the groups
attached to the methylene carbon [125.0 (3, 9, 6)8 for 1;
122.0 (3, 16, 3)8 for 2; 121.1 (?)8 for Mo2(CH2SiMe3)6;
98.9 (1)8 for Mo2(CH2Ph)6]. The much more acute angle
in the hexabenzyl compounds denotes phenyl rings which
are tipped back toward the dimetal core as a protective
mechanism. It should be noted that this tipping is entirely
steric in origin; no short contacts between the molybdenum
and the 2- and 6-carbons of the phenyl ring were observed
in Mo2(CH2Ph)6, and thus no h3 benzyl–metal interac-
tions exist.
Fig. 2. ORTEP-III diagram of Mo2(CH2SiMe2Ph)6, 2. Hydrogen atoms
were removed for clarity.
In our study of M2(OR)6 compounds containing tertiary
alkoxides [19] we noted consistent differences between the
bond and angle values for proximal alkoxides (those where
the alkoxide group lies over the metal–metal triple bond,
and for which the M–M–O–C torsion angle is near 08)
and those for distal alkoxides (those where the alkoxide
group points away from the metal–metal triple bond, and
for which the M–M–O–C torsion angle is near 1808). It
would have been valuable to determine whether this holds
for dimetal hexaalkyls, but intriguingly, every such dimer
characterized by diffraction methods, including 1–4, crys-
tallizes in a nearly ‘‘all distal’’ motif. The Mo–Mo–CH2 –
C torsion angles in 1 average 144.5 (8, 3.4, 6)8. This value
is lowered from the ideal 1808 (probably by the steric
crowding that would arise from three perfectly distal
ligands on one metal), but is far larger than the expected 08
for a proximal ligand. This provides a distinct contrast to
M2(OR)6 compounds, where only one all-distal conforma-
tion has been observed.
Fig. 3. ORTEP-III diagram of W2(CH2SiMe2Ph)6, 3. Hydrogen atoms
were removed for clarity.
Compound 1 and Mo2(OCMe2Ph)6 [19] are ‘‘peripher-
ally identical’’: they contain ligands identical save for the
bonded atom. They thus provide a matched pair for
comparison of the effect of the bonded atom on the dimetal
Fig. 4. ORTEP-III diagram of W2(CH2Ph)6, 4. Hydrogen atoms were
removed for clarity.

T.M. Gilbert et al. / Polyhedron 18 (1999) 1303–1310
1307
Table 2
Selected bond distances (A) and angles (8) for 1–4
˚
Atoms
1 (M5Mo)
2 (M5Mo)
3 (M5W)
4 (M5W)
M;M
2.1765 (8) (Mo1–Mo1a)
2.1768 (7) (Mo2–Mo2a)
2.138 (5) (Mo1–C1)
2.125 (5) (Mo1–C11)
2.138 (5) (Mo1–C21)
2.170 (2) (Mo–Mo9)
2.2587 (5) (W–W9)
2.2492 (9) (W–W9)
M–C
2.126 (6) (Mo–C1)
2.108 (5) (Mo–C11)
2.106 (5) (Mo–C21)
2.101 (7) (W–C1)
2.116 (7) (W–C2)
2.105 (7) (W–C3)
2.156 (6) (W–C1)
2.134 (4) (Mo2–C31)
2.125 (4) (Mo2–C41)
2.147 (5) (Mo2–C51)
C–C/Si
M–M–C
C–M–C
M–C–C/Si
1.545 (7) (C1–C2)
1.550 (7) (C11–C12)
1.551 (6) (C21–C22)
1.554 (6) (C31–C32)
1.555 (6) (C41–C42)
1.560 (7) (C51–C52)
1.868 (5) (C1–Si1)
1.875 (5) (C10–Si2)
1.876 (6) (C19–Si3)
1.856 (7) (C1–Si1)
1.845 (6) (C2–Si2)
1.860 (7) (C3–Si3)
1.492 (7) (C1–C2)
98.36 (14) (W–W9–C1)
117.92 (7) (C1–W–C19)
100.1 (3) (W–C1–C2)
97.83 (14) (Mo1–Mo1a–C1)
97.64 (14) (Mo1–Mo1a–C11)
97.82 (14) (Mo1–Mo1a–C21)
98.21 (13) (Mo2–Mo2a–C31)
97.47 (13) (Mo2–Mo2a–C41)
98.38 (13) (Mo2–Mo2a–C51)
116.2 (2) (C1–Mo1–C11)
120.9 (2) (C1–Mo1–C21)
117.5 (2) (C11–Mo1–C21)
118.8 (2) (C31–Mo2–C41)
116.6 (2) (C31–Mo2–C51)
118.9 (2) (C41–Mo2–C51)
124.9 (4) (Mo1–C1–C2)
125.1 (3) (Mo1–C11–C12)
126.8 (3) (Mo1–C21–C22)
124.2 (3) (Mo2–C31–C32)
124.6 (4) (Mo2–C41–C42)
124.3 (3) (Mo2–C51–C52)
98.9 (2) (Mo–Mo9–C1)
101.1 (2) (Mo–Mo9–C10)
99.0 (2) (Mo–Mo9–C19)
101.9 (2) (W–W9–C1)
102.2 (2) (W–W9–C2)
102.2 (2) (W–W9–C3)
117.4 (2) (C1–Mo–C10)
118.4 (2) (C1–Mo–C19)
115.9 (2) (C10–Mo–C19)
115.9 (3) (C1–W–C2)
114.9 (3) (C1–W–C3)
116.4 (3) (C2–W–C3)
123.8 (3) (Mo–C1–Si1)
121.7 (3) (Mo–C10–Si2)
120.6 (3) (Mo–C19–Si3)
125.8 (4) (W–C1–Si1)
122.7 (3) (W–C2–Si2)
122.0 (4) (W–C3–Si3)
core, although one must remember that the two are not
isomorphous. In the discussion below, we include only
parameters for the distal alkoxide ligands of
Mo2(OCMe2Ph)6 so as to compare only ligands of like
conformation.
The notable feature of the pair is that the core angles
match well. The Mo–Mo–O angles in Mo2(OCMe2Ph)6
average 97.10 (7, 20, 2)8, while the Mo–Mo–C angles in 1
average 97.89 (14, 34, 6)8. The Mo–O–C angles in
Mo2(OCMe2Ph)6 average 128.2 (2, 4, 2)8, while the Mo–
CH2 –C angles in 1 average 125.0 (3, 10, 6)8. That the
methylene group in 1 is larger than the oxygen in the
hexaalkoxide has little impact on the internal conformation
of the molecules.
the conclusions that could be drawn. For example, one
could argue that the Mo;Mo bond in Mo2(NMe2)6
lengthens to lessen steric contacts between dimethylamino
methyl groups on opposite metals, while this is unneces-
sary in Mo2(CH2SiMe3)6, where rotation about the
methylene group can decrease repulsions. The matched
pair here, however, provides clear evidence that hexaalkyls
exhibit inherently shorter metal–metal bond distances than
do hexaalkoxides.
To what should the phenomenon be attributed? Since
steric effects/ligand conformations appear irrelevant, there
are only two obvious candidates. One is the presence of O
p→M p* donation in the hexaalkoxides lengthening the
M;M bond; the other is the greater electronegativity of
oxygen vs. carbon forcing a larger effective positive charge
on the metals in the hexaalkoxides, in turn causing them to
repel each other and decrease metal orbital overlap [3].
The two represent limiting cases of the relationship
between electron density and bond length: in the former,
the metal–metal bond length increases because ligand
atoms add electron density to the metals, while in the
latter, the metal–metal bond length increases because
ligand atoms withdraw electron density from the metals.
As we discuss in the accompanying paper [19], collected
The one feature of the M2X6 cores which clearly
distinguishes the hexaalkoxide dimer from the hexaalkyl
dimer
is
the
metal–metal
bond
length.
In
˚
Mo2(OCMe2Ph)6, the Mo;Mo distance is 2.2388 (6) A,
˚
˚
while in 1 the distances average 2.1767 (8, 2, 2) A, 0.06 A
shorter. It was recognized in the mid-1970s that the Mo;
Mo bond length of Mo2(CH2SiMe3)6 (2.167 A) appeared
shorter than that in Mo2(NMe2)6 [2.214 (3) A] [20] and in
Mo2(OCH2CMe3)6 [2.222 (2) A] [21] but the paucity of
˚
˚
˚
data and the structural diversity of the molecules limited

1308
T.M. Gilbert et al. / Polyhedron 18 (1999) 1303–1310
structural evidence suggests that M–O p interaction
contributes little to the bonding picture in M2(OR)6
species. Furthermore, postulating an effective charge in-
crease in M2(OR)6 compounds is crucial in explaining
photoelectron spectroscopic experiments to be reported
elsewhere [22]. So it appears the metal–metal bond is
shorter in 1 because the M(CH2CMe2Ph)3 moieties do not
repel each other as much as the M(OCMe2Ph)3 fragments
in Mo2(OCMe2Ph)6 do.
silicon atom in 2, the structures of 1 and 2 are strikingly
similar. In particular, the M–C–C angles in 1 are ex-
perimentally identical to the M–C–Si angles in 2, indicat-
ing that the larger silicon atom is too far removed from the
dimetal core to affect it. The size difference manifests
itself solely in the M–M–C–C(Si) torsion angles, which
are 144.5 (8, 3.4, 6)8 for 1 and 126.2 (7, 2.7, 3)8 for 2.
This is straightforward to rationalize based on steric
considerations: the ligand with the larger silicon atom (and
therefore longer CH2 –X and X–CH3 distances) can more
readily bend back toward the dimetal core and lower the
torsion angle.
3.2. Mo2(CH2SiMe2Ph)6 , 2, and W (CH2SiMe2Ph)6 , 3
2
Treating ‘‘MoCl3(dme)’’ with LiCH2SiMe2Ph gives
dimer 2 in acceptable yield [though still quite a bit smaller
than that of Mo2(CH2SiMe3)6] after minimal workup.
Preparing the ditungsten homologue 3 is considerably
more difficult. We attempted several times to prepare it
from NaW2Cl7(THF)5, the most generally useful tungsten
(III) halide starting material, but were unable to isolate
more than a few crystals of the hexaalkyl dimer. Using
W2(OCMe3)6 proved no more successful. Ultimately we
found that treating W2(OCy)6 with the alkyllithium fol-
lowed by Me3SiCl to remove the LiOCy byproduct led to
3 in acceptable yield, on a useful scale.
The original report of the structure of Mo2(CH2SiMe3)6
stated that the ditungsten analogue W2(CH2SiMe3)6 ex-
hibited nearly identical cell parameters and the identical
space group and thus was isostructural [1]. However, the
full details have never been published, and Chisholm and
coworkers later found that they were unable to refine data
from a crystal with these cell parameters. They solved the
structure using a crystal in which the tungsten dimer
crystallized in a different space group, with different cell
parameters [7]. Compounds 2 and 3 provide a complement
to this curiosity: they crystallize in cells with quite
different parameters, but in the same space group.
Comparisons between 2 and Mo2(CH2SiMe3)6 and 3
and W2(CH2SiMe3)6 demonstrate that the added steric
bulk of the phenyl groups in the dimethylphenylsilylmethyl
systems causes little if any molecular distortion in the
M2C6 core. The core distances and angles in each pair of
compounds are identical within experimental error.
3.3. W (CH2Ph)6 , 4
2
With the publication of the synthesis and structure of
Mo2(CH2Ph)6, Rothwell and coworkers demonstrated that
stable M2R6 dimers could be formed with relatively small
R groups [8]. Surprisingly, the dimolybdenum compound
was not stabilized by h3 interactions between the benzyl
group and the metal, implying that perhaps other small
groups, such as CH2CF3 or CH2CHMe2, might give stable
dimers. We have investigated this issue peripherally by
synthesizing the hexa(benzyl)ditungsten compound 4. In
contrast to 3, W2(OCMe3)6 is an excellent starting material
for this dimer. Complex 4 precipitates from the reaction
mixture, simplifying its isolation considerably, although
for best yield removal of solvent and trituration of the
residue is necessary. The advantage of using the hexa(t-
butoxide) dimer in place of the hexa(cyclohexoxide) dimer
is that the LiOCMe3 byproduct is easily removed by the
trituration step. We have been unable to prepare 4 from
Na2W2Cl7(THF)5.
The two homologues are nearly isostructural save that
the Mo;Mo bond length is shorter than the W;W bond
˚
length by the usual 0.07 A. It appears the M–M–C angle
in 2 is slightly more acute [99.7 (2, 12, 3)8] than that in 3
[102.1 (2, 2, 3)8], but the data are insufficient to demon-
strate this unambiguously. Furthermore, the corresponding
values for Mo2(CH2SiMe3)6 [100.6 (?)8] and
W2(CH2SiMe3)6 [101.7 (10, 21, 12)8] are experimentally
identical. Providing some corroboration for the difference,
however, are the M–M–C–Si torsion angles, which aver-
age to 126.2 (7, 2.7, 3)8 for 2 and 118.3 (9, 1.6, 3)8 for 3.
As one would expect, the smaller torsion angle correlates
with the larger M–M–C angle which determines it.
Nonetheless, it does not seem that the larger W;W core in
3 has a dramatic impact on the conformations of the
ligands. Combined with our comments above regarding
Mo2(CH2Ph)6, it appears ligand size dictates ligand con-
formation more than does the size of the core the ligand
must protect.
We thought that the increased W;W bond length in 4 as
compared to the Mo;Mo bond length in Mo2(CH2Ph)6,
and the consequent increased chance for decomposition by
a- or g-hydrogen abstraction by the metal, might mean that
4 would prove unisolable, but in fact the tungsten dimer is
stable at room temperature when pure. It is, however,
thermally sensitive; an NMR sample decomposed com-
pletely to uncharacterized materials within 1 hour at 508C.
Since the dimolybdenum compound is stable for several
hours under similar conditions, it appears that the larger
size of the ditungsten core in 4 does allow decomposition
routes to become viable.
Rothwell et al., noted that the methylene resonance in
the 1H NMR spectrum of Mo2(CH2Ph)6, appeared at d
3.70, a shift attributed to the deshielding caused by the
diamagnetic anisotropy of the Mo–Mo triple bond [8]. The
analogous resonance in 4 appears at d 3.64. The shift in
Despite the presence of the larger, more polarizable

T.M. Gilbert et al. / Polyhedron 18 (1999) 1303–1310
1309
each case is ca. d 1.6 from that of the parent hydrocarbon
(toluene), similar to the shift observed for the methylene
vs. Me3SiPh in 2 and 3 (ca. d 1.5; see Experimental). The
observation holds for M2(CH2SiMe3)6 (M5Mo: d 2.02;
M5W: d 1.90; (CH3)4Si: d 0.00) and M2(CH2CMe3)6
(M5Mo: d 2.75; M5W: d 2.52; (CH3)4C: d 0.90). Thus
the additional data support Rothwell’s hypothesis and
suggest that the methylene resonance in an M2(CH2R)6
compound will generally lie 1.6–1.9 ppm downfield of the
resonance in the organic parent.
W2(OCy)6 good for preparing some hexaalkyls, but form-
ing this material requires several steps. Even when it is
used, yields are not great. W2(OCMe3)6, while useful for
specific syntheses such as that of 4, does not have broad
applicability. Probably the bulky cyclohexoxy and t-butoxy
groups slow the attack by the alkyllithium and allow side
reactions to occur. Possibly W2(O–iPr)6, the homologue of
Rothwell’s dimolybdenum reagent, is the ideal starting
point for the synthesis of ditungsten hexaalkyls, but it
brings its own set of problems: it must be prepared and
stored at low temperature to avoid dimerization [23] As a
result, we have not tested its efficacy.
Dimer 4 is isomorphous with the dimolydenum com-
˚
pound. The W;W distance is 0.074 A longer than the
Mo;Mo distance, but this has only a minor impact on any
of the other parameters. We noted above that the Mo–C
bond distance in Mo2(CH2Ph)6 appeared slightly longer
than that in the other dimolybdenum compounds character-
ized. The same may hold for W2(CH2Ph)6: the W–C
The structural data expand the database of homoleptic
dimetal hexa(ligand) species and provide clear proof that
the metal–metal triple-bond distance in dimetal hexaalkyls
˚
is inherently 0.06–0.08 A shorter than that in dimetal
hexaalkoxides. This has been known for a long time, but
the matched pair of compounds reported here removes the
possibility that the difference arises from steric concerns
rather than electronic ones. The latter are clearly key, and
we believe that the larger effective positive charge on the
metals in the hexaalkoxides resulting from the greater
electronegativity of oxygen vs. carbon causes the phenom-
enon.
It is interesting that the structural differences involving
the alkyl ligands can be attributed to steric needs. One
might have suspected that silyl-substituted alkyl ligands
would display unique parameters arising from their rela-
tively electron-rich character, but this does not appear so.
Even more surprising is the fact that the hexa(benzyl)
compounds are not remarkably different from the hexa-
(primary alkyls), despite the electronic variations between
benzyl and other primary alkyls. This suggests that compu-
tational chemists should be able to model most dimetal
(hexaalkyls) as M2(CH3)6 or M2(CH2CH3)6 species,
simplifying the problem considerably.
˚
distance of 2.156 (6) A appears longer than that in 3.
However, the average W–C distance in W2(CH2SiMe3)6 is
˚
2.14 (4, 6, 12) A, experimentally indistinguishable from
the target value. This distance may be artifactually long,
though, as the diffraction data for the hexa(tri-
methylsilylmethyl) compound were noted to be limited and
of mediocre quality, so that the carbon atoms were refined
isotropically [7]. More data are needed to address this
point unambiguously.
The W–W–C angle [98.36 (14)8 vs. 97.61 (6)8 for
Mo2(CH2Ph)6] and W–C–C angle [100.1 (3)8 vs. 98.9
(1)8 for Mo2(CH2Ph)6] open slightly, reflecting, as in the
comparisons above, the need for the ligands to protect the
larger dimetal core in 4 by occupying more space. The
M–M–C–C (phenyl) torsion angles are essentially identi-
cal for the molybdenum and tungsten homologues [143.48
and 143.7 (5)8, respectively], and are similar to that
observed in 1.
4. Conclusion
Supplementary Material Available
The goal of this work was to find general and optimal
routes to dimolybdenum- and ditungsten hexaalkyls. While
‘‘MoCl3(dme)’’ appears to be a general reagent for such
syntheses, product yields are such that it cannot be termed
an optimal one. Advantages of the trichloride include its
easy preparation and the simple separation of the hexaalkyl
product from byproducts, but its giving unpredictable
yields cannot be overlooked when the required
alkyllithiums are expensive to prepare. If one is interested
in preparing a dimolybdenum hexaalkyl in the highest
possible yield, Rothwell’s route starting from Mo2(O–iPr)6
appears to be the best choice.
Structural data have been deposited in the Cambridge
Crystallographic Data Centre Structural Database [24].
Some data are also available from the primary authors
upon request.
Acknowledgements
We appreciate the interest in and comments regarding
this work of Professor Malcolm Chisholm, Indiana Uni-
versity, and Professor Richard Dallinger, Wabash College.
NaW2Cl7(THF)5 seems useful only for preparing the
hexa(neopentyl)- or hexa-(trimethylsilylmethyl)ditungsten
compounds. We had little success in preparing hexaalkyls
using more reducing phenyl-substituted alkyllithiums, and
we suspect this to be a general problem. We found
References
[1] F. Huq, W. Mowat, A. Shortland, A.C. Skapski, G. Wilkinson, J.
Chem. Soc., Chem. Commun. (1971) 1079.

1310
T.M. Gilbert et al. / Polyhedron 18 (1999) 1303–1310
[2] M.H. Chisholm, Polyhedron 2 (1983) 681.
[16] M.N. Burnett, C.K. Johnson, ORTEP-III: Oak Ridge thermal
ellipsoid plot program for crystal structure illustrations. Oak Ridge
National Laboratory Report ORNL-6895, 1996.
[17] Esds of averaged values are given as (a, b, c), where a is the average
esd, b is the standard deviation of the individual values from the
average, and c is the number of values averaged.
[18] We note that diffraction data for Mo2(CH2Ph)6 were collected at
120 K, lower than the temperature used for any other dimer. Since
bond distances normally decrease with lowered data collection
temperature, this supports the argument that the M–C bond distance
in the hexabenzyl. compound is unusually long.
[3] M.H. Chisholm, ChemTracts-Inorg. Chem. 4 (1992) 273.
[4] F.A. Cotton, R.A. Walton, Multiple bonds between metal atoms. 2nd
ed. Oxford: Clarendon, 1993.
[5] W. Mowat, G. Wilkinson, J. Chem. Soc., Dalton Trans. (1973) 1120.
[6] M.H. Chisholm, B.W. Eichhorn, K. Folting, et al., Inorg. Chem. 26
(1987) 3182.
[7] M.H. Chisholm, F.A. Cotton, M. Extine, B.R. Stults, Inorg. Chem.
15 (1976) 2252.
[8] S.M. Beshouri, I.P. Rothwell, K. Folting, J.C. Huffman, W.E. Streib,
Polyhedron 5 (1986) 1191.
[9] L. Chamberlain, J. Keddington, I.P. Rothwell, Organometallics 1
(1982) 1098.
[19] T.M. Gilbert, C.B. Bauer, A.H. Bond, R.D. Rogers, Polyhedron,
preceeding paper in this issue.
[10] T.M. Gilbert, A.M. Landes, R.D. Rogers, Inorg. Chem. 31 (1992)
3438.
[11] J.C. Littrell, C.E. Talley, R.F. Dallinger, T.M. Gilbert, Inorg. Chem.
36 (1997) 760.
[20] M.H. Chisholm, F.A. Cotton, B.A. Frenz, W.W. Reichert, L.W.
Shive, B.R. Stults, J. Am. Chem. Soc. 98 (1976) 4469.
[21] M.H. Chisholm, F.A. Cotton, C.A. Murillo, W.W. Reichert, Inorg.
Chem. 16 (1977) 1801.
[12] M.H. Chisholm, K. Folting, M. Hampden-Smith, C.A. Smith,
Polyhedron 6 (1987) 1747.
[22] J.R. Pollard, T.M. Gilbert, D.L. Lichtenberger, J. Am. Chem. Soc.,
submitted.
[13] M. Schlosser, J. Hartmann, Angew. Chem., Int. Ed. Eng. 12 (1973)
508.
[23] M.H. Chisholm, D.L. Clark, K. Folting, J.C. Huffman, M. Ham-
pden-Smith, J. Am. Chem. Soc. 109 (1987) 7750.
[14] R.R. Schrock, J.D. Fellman, J. Am. Chem. Soc. 100 (1978) 3359.
[15] G. Sheldrick, SHELXTL. A suite of computer programs for X-ray
structure determination, sold by Siemens Analytical Instruments,
Inc., Madison, WI, USA.
[24] F.H. Allen, O. Kennard, R. Taylor, Acc. Chem. Res. 16 (1983) 146.