Angewandte
Chemie
between 2g and 2b is only 0.6 kcalmolꢀ1, corroborating the
calculated position of the more intense HOMO-1!LUMO
ꢀ
validity of the experimental results. The agostic C H bonds
transition is in excellent agreement with the maxima obtained
in the experimental spectra. Overall, the experimental and
theoretical results suggest that the b-agostic bonding mode of
the phosphine induces a stronger ligand field than the g-
agostic bonding mode does.
are longer than the non-agostic ones: the elongation is
approximately 0.03 ꢀ for 2g and approximately 0.07 ꢀ for
ꢀ
~
2b. Additionally, the calculated nCH for the agostic C H bond
is 2560 cmꢀ1 for 2g and 2150 cmꢀ1 for 2b compared to 2960–
3050 cmꢀ1 for the non-agostic ones.[21] Weakening of the
To estimate the total stabilization energy associated with
the agostic interactions (DEagostic, Figure 3), hydride 1 was also
optimized. Subsequent computational removal of the molyb-
denum-bound hydride then gives 2vacant (Figure 3) as a plau-
sible model for a non-agostic structure, having a vacant
ꢀ
agostic C H bond clearly occurs in both complexes, its extent
being the highest in 2b.
The molecular orbital (MO) energy diagram and popula-
tion analysis[12] show that the compositions and energies of the
occupied MOs of both agostomers are quite similar
(Figure 4). The HOMO and HOMOꢀ1 are both largely
metal-based, harboring the four d-electrons. MOs involved
with the agostic interactions (HOMOꢀ12 for 2g; HOMOꢀ13
for 2b) are low in energy and of s symmetry with respect to
coordination site at Mo. A single-point calculation on 2vacant
,
which relaxes to 2g when fully optimized, revealed it to be
approximately 12 kcalmolꢀ1 higher in energy than 2g and 2b.
This provides an upper bound for DEagostic; its magnitude is in
agreement with estimates of strengths of no more than
15 kcalmolꢀ1 for agostic interactions.[2,6]
The weak agostic interactions are readily displaced
(Scheme 3). Not surprisingly, 2 binds acetonitrile essentially
irreversibly to form orange 2-MeCN, but it also binds the
ꢀ
the Mo···H axis; the contribution from the C H moiety is
more significant (21% for 2g; 25% for 2b) than that from Mo
(less than 5%).
Scheme 3. Displacement of the agostic interactions by MeCN and N2.
poorer ligand dinitrogen to give 2-N2. The extent of N2
coordination, under an atmosphere of N2 in fluorobenzene
solution,[22] is only 5–10% at room temperature.[12] Upon
cooling, the equilibrium shifts toward the N2 adduct, and at
ꢀ408C an approximately 2:1 ratio of 2-N2:2 is detected by
NMR spectroscopy. Warming to room temperature again
establishes 2 as the major species, showing that formation of
2-N2 is reversible. 15N and 31P NMR spectra[23] collected for
the isotopomer 2-15N2 provide evidence for the end-on
coordination of N2. Additional reactivity studies of 2 toward
small molecules are currently underway.
In summary, we have described the first example of an
organometallic complex for which two agostomers have been
individually isolated and characterized. These results illus-
trate the versatility of agostic bonding, and have led to the
unanticipated insight that agostomers may be differently
colored. Clearly, one crystal structure will not always provide
the complete “agostic picture”.
Figure 4. Energy level diagram for 2g and 2b, with selected MOs
depicted (isosurface value is 0.03). The double occupancies are only
illustrated for the HOMOs. Only the atoms of the agostic Mo-P-C(-C)-
H moieties are drawn as balls.
An important difference between the agostomers is that
the LUMO of 2b is 0.64 eV higher in energy than that of 2g.
This is probably caused by a higher contribution from the P
atom to the LUMO in 2b (10%) than in 2g (4%), owing to
Experimental Section
ꢀ
the significantly shorter Mo P bond length in 2b (D[d(Mo-
Synthetic details and characterization data, as well as more complete
computational details and atomic coordinates, are given in the
Supporting Information. First principles calculations were performed
using the DFT-D3(BJ) dispersion-corrected PBE exchange-correla-
tion functional[24] implemented in the Amsterdam Density Functional
(ADF 2013.01) program.[25] Scalar relativistic effects were taken into
account by the ZORA formalism to the Dirac equation.[26] TZ2P basis
sets with small cores were used for geometry optimization; all-
electron basis sets were used for property calculations.[27] The time-
P)] = 0.087 ꢀ, Table 1). The resulting different HOMO–
LUMO gaps in 2g and 2b provide the explanation for the
different colors observed, as further quantitatively investi-
gated by time-dependent DFT. Two transitions in the visible
region were found: HOMO!LUMO (2g: 698 nm; 2b:
517 nm) and HOMO-1!LUMO (2g: 574 nm; 2b: 472 nm),
drawn as sticks in Figure 2. For both agostomers, the
Angew. Chem. Int. Ed. 2013, 52, 1 – 5
ꢀ 2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
3
These are not the final page numbers!