Angewandte
Chemie
monotonously and reaches about 3 mB at 2 K, whereas multi-
field data sampled at 1, 4, and 7 T (Supporting Information,
Figure S11B) show only moderate nesting of magnetization
M(mBB/kT). Both features reveal large zero-field splitting of
chromium(IV), meaning that only a few magnetic substates
are low-lying in energy (Supporting Information, Figure S12).
A corresponding spin Hamiltonian simulation yields an
excellent global fit to both sets of magnetic data with weak
positive spin coupling, J = 2.0(Æ1) cmÀ1, and large zero-field
splitting with D = À53(Æ8) cmÀ1 (E/D = 0.04 Æ 0.04). For an
initial explanation, we suggest that the ligand field of the
chromyl units is dominated by the short chromium(IV) oxo
bonds so that the two sites may be regarded approximately as
pseudo-linear or distorted square-pyramidal complexes (note
that chromium(V) nitrido complexes show pseudo-linear
ligand-field splitting, irrespective of the nature of the
supporting coordination framework[23]). If then strong p
interaction with oxygen leads to splitting of the low-lying t2g
orbitals in dxy and quasi-degenerate first excited {dxz/px}, {dyz/
py} orbitals, spin–orbit coupling within the nearly-degenerate
orbital ground state of the resulting (dxy)1({dxz/px}/{dyz/py})1
configuration would explain the strong zero-field splitting of
Figure 3. ESI-MS(À) of a THF solution of 1 with a signal for [LCrIVO]À
(2’) and its putative structure (inset). a) The calculated signal for 2’
and b) the relevant section of the spectrum of 2’.
An unexpected observation was made in an ESI-MS
analysis, after 18O2 had been used for the oxidation of 1 in
THF, which in this context proved unique as a solvent (see
below): Instead of a simple shift of all peaks belonging to
[LCrIVO]À by m/z = 2 a complex signal was obtained that
indicated incorporation of more than one 18O atom (Figure 4).
CrIV O. But also the weak spin coupling in the dinuclear core
=
is plausible, because neither the magnetic dxy and d’xy orbitals
of Cr1 and Cr1’ nor the magnetic MOs {dxz/px} and {dyz/py} of
=
=
Cr1 O1 and Cr1’ O1’ moieties should have significant over-
=
lap owing to the misdirected orientation of the Cr O units in
the diamond core (Supporting Information, Scheme S2).
Electrostatic interaction between the half-filled t2g orbitals
of the two sites in this case should yield in fact (weak)
ferromagnetic spin coupling;[24] according to the classical
Goodenough–Kanamori rules, this holds also for residual
overlap between half-filled and empty orbitals.[25,26] Weak
ferromagnetic interaction is thus not unexpected for two
chromium(IV) oxo sites with 3d2 configuration arranged in
such a diamond core.
Figure 4. Section of the ESI-MS(À) spectrum of a THF solution of 2
prepared by 18O2 treatment of 1 showing the m/z range for 2’.
=
This suggested that not only the Cr O functionality was
isotopically labeled: 18O/16O exchange also occurred for O
atoms belonging to L3À. To confirm this, the reaction mixture
of [18O]-2 prepared this way was subsequently hydrolyzed
Complex 2 is stable for days in solution as well as in the
solid state at room temperature, but the ESI-MS(À) of a THF
solution not only showed the signal for its monomeric version
[LCrIVO]À at m/z = 815.09 but also signals for species with
higher O content (see the Supporting Information), which
might be due to disproportion of 2 under ESI conditions. As
1 (or its monomeric version) is highly sensitive towards O2, it
could not even be detected in ESI-MS studies: Owing to the
presence of O2 in the mass spectrometer, solutions of 1 in
THF only led to peaks indicating oxygenation, with the most
prominent peak at m/z = 815.09 corresponding to monomeric
2, which is denoted 2’ (Figure 3). All attempts to exclude O2 in
the mass spectrometer and thus to detect 1 by mass
spectrometry failed. Treatment of a THF solution of 1 with
two equivalents of PhIO instead of O2 also led to the
formation of 2, as shown by UV/Vis spectroscopy and ESI-MS
(see the Supporting Information), while N2O did not react. By
contrast, Copꢁret et al. recently reported a dinuclear chro-
mium(II) siloxide, [{Cr(OSi(OtBu)3)2}2], which did react with
N2O; however, to a corresponding chromium(III) complex
lacking terminal oxido ligands.[15] This illustrates well the
subtle influence the siloxide environment has on the CrII
reactivity.
with a diluted HCl solution to cleave the metal siloxide bonds
À
À
in complex 2 by protonation of the Si O functions, and
without further workup the solution was analyzed by mass
spectrometry. The ESI-MS(+) spectrum showed a prominent
peak at m/z = 773.16 assigned to [LH3Na]+ and again addi-
tional peaks that are shifted by m/z = 2, 4, and 6 were
observed (Figure 5). From these experiments it is evident that
in course of the oxidation of 1 by 18O2 18O atoms are also
incorporated into L3À, and isotopologues that bear up to four
18O atoms for [LCrIVO]À and up to three 18O atoms in case of
[LH3Na]+, respectively, result; a very similar incorporation of
18O was observed when using an excess of PhI18O as the
oxidant for a THF solution of 1. Incorporation of the 18O into
L3À might be explained by a shift of the silyl groups to
intermediate terminal oxido ligands; to our knowledge this
type of reactivity has been observed neither for alkoxide nor
siloxide complexes so far. There is only precedence for
Claisen-like rearrangements in case of allylic residues[27]
=
(whose C C units are thus involved) and slow oxido–
hydroxido tautomerization.[28]
Even more remarkably, it turned out that upon treatment
of THF solutions of 2 either with 18O2 or an excess of PhI18O
again the incorporation of several 18O atoms into 2 occurred,
Angew. Chem. Int. Ed. 2014, 53, 12741 –12745
ꢀ 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim