Association and Dissociation of Grignard Reagents RMgCl and Their Turbo Variant RMgC?LiCl

Article abstract of DOI:10.1002/chem.201600699

Grignard reagents RMgCl and their so-called turbo variant, the highly reactive RMgC?LiCl, are of exceptional synthetic utility. Nevertheless, it is still not fully understood which species these compounds form in solution and, in particular, in which way LiCl exerts its reactivity-enhancing effect. A combination of electrospray-ionization mass spectrometry, electrical conductivity measurements, NMR spectroscopy (including diffusion-ordered spectroscopy), and quantum chemical calculations is used to analyze solutions of RMgCl (R=Me, Et, Bu, Hex, Oct, Dec, iPr, tBu, Ph) in tetrahydrofuran and other ethereal solvents in the absence and presence of stoichiometric amounts of LiCl. In tetrahydrofuran, RMgCl forms mononuclear species, which are converted into trinuclear anions as a result of the concentration increase experienced during the electrospray process. These trinuclear anions are theoretically predicted to adopt open cubic geometries, which remarkably resemble structural motifs previously found in the solid state. The molecular constituents of RMgCl and RMgC?LiCl are interrelated via Schlenk equilibria and fast intermolecular exchange processes. A small portion of the Grignard reagent also forms anionic ate complexes in solution. The abundance of these more electron-rich and hence supposedly more nucleophilic ate complexes strongly increases upon the addition of LiCl, thus rationalizing its beneficial effect on the reactivity of Grignard reagents.

Full text of DOI:10.1002/chem.201600699

DOI: 10.1002/chem.201600699
Full Paper
&
Grignard Reagents |Hot Paper|
Association and Dissociation of Grignard Reagents RMgCl and
Their Turbo Variant RMgCl·LiCl
Christoph Schnegelsberg,[a] Sebastian Bachmann,[b] Marlene Kolter,[a] Thomas Auth,[a]
Michael John,[a] Dietmar Stalke,[b] and Konrad Koszinowski*[a]
Chem. Eur. J. 2016, 22, 7752 – 7762
7752
ꢀ 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Paper
Abstract: Grignard reagents RMgCl and their so-called turbo
variant, the highly reactive RMgCl·LiCl, are of exceptional
synthetic utility. Nevertheless, it is still not fully understood
which species these compounds form in solution and, in
particular, in which way LiCl exerts its reactivity-enhancing
effect. A combination of electrospray-ionization mass spec-
trometry, electrical conductivity measurements, NMR spec-
troscopy (including diffusion-ordered spectroscopy), and
quantum chemical calculations is used to analyze solutions
of RMgCl (R=Me, Et, Bu, Hex, Oct, Dec, iPr, tBu, Ph) in tetra-
hydrofuran and other ethereal solvents in the absence and
presence of stoichiometric amounts of LiCl. In tetrahydrofur-
an, RMgCl forms mononuclear species, which are converted
into trinuclear anions as a result of the concentration in-
crease experienced during the electrospray process. These
trinuclear anions are theoretically predicted to adopt open
cubic geometries, which remarkably resemble structural
motifs previously found in the solid state. The molecular
constituents of RMgCl and RMgCl·LiCl are interrelated via
Schlenk equilibria and fast intermolecular exchange process-
es. A small portion of the Grignard reagent also forms anion-
ic ate complexes in solution. The abundance of these more
electron-rich and hence supposedly more nucleophilic ate
complexes strongly increases upon the addition of LiCl, thus
rationalizing its beneficial effect on the reactivity of Grignard
reagents.
Introduction
The outstanding synthetic utility of Grignard reagents has
always resulted in a keen interest in their speciation in solu-
tion. The classical studies by Schlenk and Schlenk showed that
Grignard reagents RMgX in ethereal solutions undergo dispro-
portionation to afford both R2Mg and MgX2 (the so-called
Schlenk equilibrium).[14] Later work of Ashby and others used
ebullioscopy and kinetic measurements as well as NMR, IR, and
X-ray absorption spectroscopy to determine the position of
the Schlenk equilibria and the aggregation state of RMgX.[15,16]
These studies pointed to the presence of monomeric and di-
meric RMgX in diethyl ether. In tetrahydrofuran (THF), only
monomeric species were found, but here the operation of the
Schlenk equilibrium is more pronounced and gives rise to the
formation of substantial amounts of R2Mg and MgX2, besides
RMgX. Electrochemical experiments also indicated the pres-
ence of ionic species, although only in low abundance.[17] The
nature of these ionic species remains unclear.
Since their discovery by Grignard in the year of 1900,[1] the or-
ganomagnesium halides RMgX that bear his name have been
known and cherished as the prototypical organometallic re-
agents.[2] Grignard reagents not only react readily with numer-
ous electrophiles, such as aldehydes, ketones, epoxides, ni-
triles, esters, and carbon dioxide,[3,4] but also efficiently under-
go transition metal-catalyzed conjugate addition[5] and cross-
coupling reactions.[6] Thus, they are indispensable tools for the
formation of new carbon–carbon bonds and continue to at-
tract a great deal of interest. Important milestones in the de-
velopment of Grignard reagents include the activation of the
magnesium metal for a more facile access to RMgX com-
pounds[7] and the synthesis and use of highly functionalized
Grignard reagents.[8] The latest pivotal innovation is marked by
the development of the so-called turbo-Grignard reagent
iPrMgCl·LiCl by Knochel and co-workers.[9] The presence of LiCl
strongly enhances the reactivity of this reagent and, thus, ena-
bles highly efficient magnesium–bromine exchange reactions.
Thereby, it further broadens the scope of accessible polyfunc-
tionalized Grignard reagents. Thanks to their superior per-
formance, iPrMgCl·LiCl and other LiCl-containing Grignard re-
agents are by now widely adopted in academia[10] and indus-
try.[11] At the same time, the addition of LiCl also significantly
accelerates the insertion of magnesium into organic halides.[12]
RMgCl·LiCl reagents, thus, represent a prime example of the
emergence of synergistic effects in the reactivity of bimetallic
main-group compounds.[13]
More recent studies have also addressed the microscopic
constitution of turbo-Grignard reagents. Krasovskiy and Kno-
chel suggested that the addition of LiCl to RMgCl in THF may
lead to ate complexes, which should exhibit an enhanced nu-
cleophilic reactivity (Scheme 1, top left).[9a] Quantum-chemical
calculations by Krasovskiy, Straub, and Knochel confirmed the
[a] C. Schnegelsberg, M. Kolter, T. Auth, Dr. M. John, Prof. Dr. K. Koszinowski
Institut für Organische und Biomolekulare Chemie
Georg-August-Universität Gçttingen
Tammannstraße 2, 37077 Gçttingen (Germany)
E-mail: konrad.koszinowski@chemie.uni-goettingen.de
[b] S. Bachmann, Prof. Dr. D. Stalke
Institut für Anorganische Chemie
Georg-August-Universität Gçttingen
Tammannstraße 4, 37077 Gçttingen (Germany)
Scheme 1. Constituents of iPrMgCl·LiCl assumed to be present in solution
(THF)[9a,18] and found in the solid state,[19] respectively. Coordinating solvent
molecules are omitted for clarity.
Supporting information for this article is available on the WWW under
http://dx.doi.org/10.1002/chem.201600699.
Chem. Eur. J. 2016, 22, 7752 – 7762
www.chemeurj.org
7753
ꢀ 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Paper
assumed correlation between ate character and nucleophilic
reactivity.[18] In the same study, the authors also found evi-
dence for the operation of Schlenk equilibria and the forma-
tion of solvent-separated ion pairs from the ate complex and
its Li+ counter ion (Scheme 1, top right).[18] Such ion pairs were
not observed in crystals isolated from solutions of iPrMgCl·LiCl,
however. Instead, Lerner and collaborators found these crystals
to display neutral open-cube geometries of four Mg centers
without the incorporation of Li (Scheme 1, bottom).[19] The
same structure was obtained for crystals isolated from the cor-
responding conventional Grignard reagent iPrMgCl.[19,20]
ESI mass spectrometry
Before the analysis of solutions of Grignard reagents, the inlet line
of the mass spectrometer was extensively purged with dry THF.
Sample solutions (c=25 mm if not stated otherwise) were then
transferred into gas-tight syringes and injected into the ESI source
of an HCT quadrupole-ion trap mass spectrometer (Bruker Dalto-
nik) at a flow rate of 8 mLminÀ1. The ESI source was operated at
a voltage of Æ3000 V with N2 as nebulizer gas (0.7 bar) and drying
gas (5.0 LminÀ1, 373 K). To transfer the generated ions into the
quadrupole-ion trap, very mild conditions were applied to prevent
unwanted decomposition reactions.[21b,28] The helium-filled trap [es-
timated pressure p(He)ꢀ 0.3 Pa] was typically operated at a trap
drive of 40 to ensure efficient trapping and detection of ions of
medium m/z ratios (typically recorded m/z range: 50–1000). The
given assignments are based on the m/z ratios, the isotope pat-
terns (some of which were significantly broadened and, thus,
poorly resolved, presumably due to reactions with traces of back-
ground water during the m/z scan), and the gas-phase fragmenta-
tion behavior of the observed ions. In the gas-phase fragmentation
experiments, the ions of interest were mass-selected (with isolation
To elucidate the effect of LiCl upon Grignard reagents at the
molecular level, the direct identification of the in situ-formed
species by spectrometric and spectroscopic methods is essen-
tial. Herein, we analyze solutions of conventional Grignard re-
agents RMgCl and their turbo variant RMgCl·LiCl in THF by
a combination of electrospray-ionization (ESI) mass spectrome-
try, electrical conductivity measurements, and multinuclear
NMR spectroscopy, including diffusion-ordered spectroscopy
(DOSY). ESI mass spectrometry selectively probes charged spe-
cies and, thus, is very well suited to detect anionic ate com-
plexes, as we and others have shown previously.[21,22] Note,
however, that the notoriously high sensitivity of Grignard re-
agents toward hydrolysis and oxidation so far has prevented
the successful detection of intact organomagnesium species
by ESI mass spectrometry.[23,24] The analysis of these challeng-
ing analytes therefore also serves as a rigorous test for the ap-
plied methods. The qualitative insight afforded by ESI mass
spectrometry is ideally complemented by electrical conductivi-
ty measurements, which can provide more quantitative infor-
mation on the ionic constituents in solution.[21b,c,25] NMR spec-
troscopy, in turn, has the advantage of probing both neutral
and charged species. Moreover, its DOSY variant can be used
to determine the aggregation state of the sampled species,
which is particularly helpful for the characterization of organo-
metallics and metal bases in solution.[26] However, one draw-
back of NMR spectroscopic experiments is their relatively slow
time scale, which prevents the resolution of fast equilibria. To
facilitate the interpretation of the experimental results, we also
employ quantum chemical calculations.
widths of 8 u), subjected to excitation voltages of amplitude Vexc
,
and allowed to collide with the helium gas. Additional experiments
were carried out with a micrOTOF-Q II mass spectrometer (Bruker
Daltonik) under similar ESI conditions.[29] This instrument is
equipped with a time-of-flight mass analyzer and, thus, achieves
a higher mass resolution than the quadrupole-ion trap (typically,
deviations of 20 ppm were obtained when the instrument was ex-
ternally calibrated with a mixture of phosphazenes in H2O/acetoni-
trile). A drawback of the micrOTOF-Q II instrument is the imperfect
insulation of its ESI source against the surrounding atmosphere
and the resulting intake of trace amounts of water and oxygen.
Both instruments were controlled with the Compass software pack-
age (Bruker Daltonik), which was also used to calculate exact m/z
ratios and theoretical isotope patterns.
Electrical conductivity measurements
Electrical conductivity experiments were performed with a Seven
Multi instrument (Mettler Toledo) with a stainless steel electrode
cell (kcell =0.1 cmÀ1, calibrated against an aqueous solution of KCl
at 298 K) at T=298Æ1 K. Iodometric titration immediately after
each measurement proved that the sample solutions still had
ꢁ90% of the expected activity.
NMR spectroscopy
Samples of RMgCl or RMgCl·LiCl (c=25 mm) and equimolar
amounts of 1,2,3,4-tetraphenylnaphthalene (TPhN) were filled into
5 mm NMR tubes and kept inside the NMR spectrometer at 298 K
for 30 min (with an air flow of 400 LhÀ1) before the measurement,
if not stated otherwise. All experiments were performed with
a Bruker Avance 400 spectrometer equipped with an observe
broadband probe with a z-axis gradient coil with a maximum gra-
dient strength of 57 GcmÀ1 and without sample spinning during
measurements. The DOSY experiments employed a double stimu-
lated echo sequence with bipolar gradient pulses and three spoil
gradients with convection compensation (dstebpg3s).[30] The diffu-
sion time was set to D=0.1 s. The pulsed field gradients d/2 were
adjusted for every compound in a range of 2–3 ms. The delay for
gradient recovery was 0.2 ms and the eddy current delay 5 ms. For
each DOSY experiment, a series of 16 spectra on 32 K data points
was collected. The pulse gradients (g) were incremented from 2 to
98% of the maximum gradient strength in a linear ramp with
Experimental and Theoretical Methods
Materials and general methods
Standard Schlenk techniques were applied in all cases. THF and
[D8]THF were dried over sodium or potassium/benzophenone and
freshly distilled before use. Et2O was dried over sodium and freshly
distilled. Dioxane was freshly distilled and degassed. MeMgCl (in
THF, 3m), EtMgCl (in THF, 2m), and PhMgCl (in THF, 2m) were pur-
chased and used as received. BuMgCl, HexMgCl, OctMgCl,
DecMgCl, iPrMgCl, and tBuMgCl were synthesized according to
standard procedures (see the Supporting Information). Exact con-
centrations of the Grignard reagents were determined by iodomet-
ric titration.[27] Solutions of RMgCl·LiCl were prepared by combining
equimolar amounts of solutions of RMgCl and dry LiCl in THF.
Chem. Eur. J. 2016, 22, 7752 – 7762
www.chemeurj.org
7754
ꢀ 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Paper
a total time of the experiment of 51 min. After Fourier transforma-
tion and baseline correction, the diffusion dimension was pro-
cessed with the Topspin 3.1 software (Bruker Biospin). Diffusion co-
efficients processed with a line broadening of 2 Hz were calculated
by exponential fits with the T1/T2 software of Topspin.
sulted from the unavoidable mass discrimination inherent in
the employed mass analyzer. Note that we succeeded in sup-
pressing hydrolysis reactions essentially completely for some
systems (e.g., for BuMgCl; Figure 1), whereas in other cases the
observation of [RnMg3Cl6Àn(OH)]À and [RnMg3Cl5Àn(OH)2]À com-
plexes point to the partial occurrence of these unwanted pro-
cesses.
Quantum chemical calculations
Electronic structure calculations were performed with the ORCA
program package (version 3.0.3).[31,32] The presented gas-phase
minimum energy structures were calculated without any con-
straints by the DFT method B3LYP[33] (as defined in the GAUSSIAN
program system) in combination with Grimme’s atom-pairwise dis-
persion correction with Becke–Johnson damping (DFT-D3bj).[34] For
all atoms, the def2-TZVP basis set[35] minimally augmented (ma) by
diffuse functions according to Truhlar and co-workers[36] was ap-
plied. To obtain minimum energy geometries in solution, the same
method was used together with the conductor-like screening
model COSMO.[37] All optimized structures were confirmed to corre-
spond to a minimum in energy by means of harmonic vibrational
frequency calculations (conducted at the same level of theory as
each geometry optimization).
Solutions of iPrMgCl and tBuMgCl also afforded prominent
trinuclear ate complexes of the type [RnMg3Cl7Àn]À, namely
[iPr5Mg3Cl2]À and [tBu3Mg3Cl4]À, respectively (see the Support-
ing Information, Figures S20–S24). Additional ions contained
alkoxy groups, which presumably resulted from reactions with
traces of dioxygen and the consecutive cleavage of the O-OR
bonds by the nucleophilic attack of metal-bound
R
groups.[15c,23c,38] Control experiments with sample solutions de-
liberately exposed to O2 indeed found a dramatic increase of
alkoxide complexes (see the Supporting Information, Figur-
es S25 and S26).
The negative ion-mode ESI mass spectra of solutions of
PhMgCl showed the trinuclear complex [Ph3Mg3Cl4]À as base
peak (see the Supporting Information, Figures S27–S29). In ad-
dition, the presence of di- and trinuclear ions with up to four
hydroxy groups indicated extensive hydrolysis reactions.
We also analyzed a 1:1 mixture of iPrMgCl and PhMgCl in
THF. The resulting ESI mass spectra indicated a series of trinu-
clear ate complexes, several of which contained both iPr and
Ph groups (see the Supporting Information, Figures S30–S32).
The distribution of the two organic substituents in the detect-
ed anions was not statistical, but showed a significant enrich-
ment in the iPr group.
Results
ESI mass spectrometry
RMgCl in the absence of LiCl
Upon analysis by negative ion-mode ESI mass spectrometry,
solutions of Grignard reagents RMgCl bearing primary alkyl
groups (R=Me, Et, Bu, Hex, Oct, Dec) showed a multitude of
organomagnesium ate complexes (Figure 1; Figures S1–S19
and Table S1 in the Supporting Information). In all cases,
The analysis of solutions of Grignard reagents RMgCl by pos-
itive ion-mode ESI mass spectrometry only led to the observa-
tion of [Mg3Cl3(OMe)2(THF)n]+ ions (n=3–5). As reported previ-
ously, these ions originated from reactions with surprisingly
long-lasting MeOH contaminations present in the ESI sour-
ce.[23c]
RMgCl·LiCl
Negative ion-mode ESI of solutions of turbo-type Grignard re-
agents RMgCl·LiCl (R=Me, Et, Bu, Hex, Oct, Dec, iPr, tBu, Ph)
gave mass spectra very similar to those obtained for the solu-
tions of the corresponding conventional Grignard reagents
RMgCl (see the Supporting Information, Figures S33–S43). Only
for DecMgCl and PhMgCl, a few Li+-containing ions, such as
[Dec6LiMg3Cl2]À and [PhMg2LiCl2(OH)3]À, could be detected (see
the Supporting Information, Figures S39, S40, and S43). Posi-
tive ion-mode ESI produced [Li(THF)n]+ (n=2, 3; see the Sup-
porting Information, Figure S44 for BuMgCl·LiCl).
Figure 1. Negative ion-mode ESI mass spectrum of a 25 mm solution of
BuMgCl in THF.
À
almost exclusively trinuclear anions of the type [RnMg3Cl7Àn
]
were observed. Variation of the concentration of the sample
solution did not change this result (see the Supporting Infor-
mation, Figure S7). For Grignard reagents with short and
Effects of solvents and additives
medium alkyl chains (R=Me, Et, Bu, Hex), alkyl-rich complexes
Negative ion-mode ESI mass spectrometry of a solution of
iPrMgCl in Et2O afforded [iPr4Mg3Cl3]À as base peak (see the
Supporting Information, Figures S45–S47). Upon addition of
À
[RnMg3Cl7Àn
]
(nꢁ4) predominated, whereas more balanced
distributions were found for Grignard reagents with long alkyl
substituents (R=Oct, Dec). It is not clear whether this trend re-
flects a genuine difference in behavior or whether it merely re-
10% dioxane to a solution of iPrMgCl in THF, [iPr5Mg3Cl2]À,
À
[iPr4Mg3(OiPr)nCl1Àn
]
(n=0 and 1), as well as [iPr11Mg5]À were
Chem. Eur. J. 2016, 22, 7752 – 7762
www.chemeurj.org
7755
ꢀ 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Paper
observed (see the Supporting Information, Figures S48–S51).
The latter ion is remarkable because it is the only pentanuclear
complex detected, which moreover exclusively contains organ-
ic substituents.
water present in the vacuum system of the mass spectrometer,
which led to the partial hydrolysis of the organomagnesium
ate complexes [Eq. (3) for the case of the trinuclear ions]. The
rates of these reactions apparently depended on the nature of
the organic substituent and were particularly high for the phe-
nylmagnesium ate complexes. For these species, multiple con-
secutive hydrolysis steps were observed (see the Supporting
Information, Figures S82–S84).
Reactivity in solution
We also used negative ion-mode ESI mass spectrometry to an-
alyze solutions of iPrMgCl·LiCl in THF/dioxane (9:1), to which 4-
bromoanisole (ArBr) had been added. From synthetic studies,
it is known that under these conditions bromine–magnesium
exchange reactions occur.[18] We could indeed observe the re-
sulting intermediates [Ar4Mg2Cl]À and [Ar5Mg2]À, which
showed the complete replacement of all exchangeable iPr
groups by aryl substituents (Figure 2 and Figures S52 and S53
in the Supporting Information). Unlike the vast majority of
other detected organomagnesium ate complexes, these spe-
cies were dinuclear. In addition, the trinuclear anion
[Ar4Mg3(OiPr)2Cl]À was present and again pointed to the occur-
rence of reactions with traces of O2.
½RnMg3Cl7ÀnŠÀ þ H2O ! ½RnÀ1Mg3Cl7ÀnðOHފÀ þ RH
ð3Þ
Electrical conductivity measurements
RMgCl+nLiCl
Solutions of Grignard reagents RMgCl (R=Bu and iPr) in THF
displayed relatively low electrical conductivities (Figure 3). In
contrast, the conductivity of a solution of tBuMgCl was mark-
edly higher. Upon addition of 0.5–2 equivalents of LiCl, the
conductivity significantly increased in all cases (Figure 3 and
Tables S3 and S4). With the exception of the iPrMgCl/nLiCl
system, the measured curves exhibited approximately linear
slopes without any apparent signs of saturation. Note that the
conductivity of LiCl in THF alone is negligibly low (see below)
and, thus, cannot account for the observed increase.[40]
Figure 2. Negative ion-mode ESI mass spectrum of a solution of the prod-
ucts formed in the reaction of iPrMgCl·LiCl (25 mm) with 4-bromoanisole
(ArBr, 25 mm) in THF/dioxane (9:1).
Figure 3. Specific electrical conductivities of solutions of RMgCl (approx.
250 mm) treated with different amounts of LiCl in THF at 298 K.
Gas-phase reactivity
In all cases examined,[39] the gas-phase fragmentation of the
organomagnesium ate complexes resulted in the loss of R2Mg
units [Eq. (1) and Table S2 and Figures S54–S75 in the Support-
ing Information]. In a few cases, the primary fragment ions un-
derwent a consecutive fragmentation and lost another R2Mg
unit [Eq. (2) and Figures S76–S78 in the Supporting Informa-
tion]. Analogous fragmentation reactions also occurred for the
pentanuclear [iPr11Mg5]À ion (see the Supporting Information,
Figure S79) and the lithium-containing complexes (see the
Supporting Information, Figures S80 and S81).
BuMgCl/LiCl: Job plot
To obtain complementary information, we also employed the
method of continuous variation, commonly known as a Job
plot.[41] To this end, we considered different binary mixtures of
BuMgCl and LiCl in THF with a constant total concentration
ctotal =c(BuMgCl)+c(LiCl)=250 mm. Plotting the electrical con-
ductivities of these mixtures against the molar fraction of
BuMgCl resulted in a curve with a pronounced maximum at
x(BuMgCl)=0.5 (Figure 4). The curve showed a slightly asym-
metric shape due to the different conductivities of the two
pure components: that of LiCl was negligibly low,[40] unlike
that of BuMgCl (see above).
À
½RnMg3Cl7Àn
Š
! ½RnÀ2Mg2Cl7ÀnŠÀ þ R2Mg
ð1Þ
ð2Þ
À
½RnÀ2Mg2Cl7Àn
Š
! ½RnÀ4MgCl7ÀnŠÀ þ R2Mg
These unimolecular reactions were partly obscured by the
parallel occurrence of bimolecular reactions with traces of
Chem. Eur. J. 2016, 22, 7752 – 7762
www.chemeurj.org
7756
ꢀ 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Paper
Figure 4. Job plot for a mixture of BuMgCl and LiCl in THF at 298 K,
total =250 mm.
c
Figure 6. Comparison of experimentally derived molecular weights (red)
with values expected for [RMgCl(THF)2] (black) and [R2Mg2Cl2(THF)2] (blue).
NMR Spectroscopy
alkyl chain of the Grignard reagents (Figure 6). This behavior
strongly suggested that the probed compounds RMgCl be-
longed to a common series of homologous species. From the
observed increase of the molecular weights per additional
C2H4 unit in the alkyl chain, it can be directly deduced that the
probed organomagnesium complexes contained exactly one
alkyl group. The obtained molecular weights match those cal-
culated for [RMgCl(THF)2] very well (Figure 6). In contrast, as-
suming the presence of, for example, dinuclear
[R2Mg2Cl2(THF)2] would not reproduce the derived molecular
weights at all. In the case of BuMgCl, we also investigated the
concentration dependence of the diffusion coefficient. A ten-
fold concentration increase from 25 mm to 250 mm resulted in
no significant increase of the diffusion coefficient. At c=1.0m,
the highest concentration that could be probed, the diffusion
coefficient slightly increased. This behavior could indicate the
partial formation of a higher aggregate interrelated to the
[RMgCl(THF)2] species via a fast equilibrium; the validity of the
ECC DOSY method has, however, not been rigorously demon-
strated for highly concentrated sample solutions.[26g]
RMgX in the absence of LiX
1H and 13C NMR spectra of Grignard reagents RMgCl (R=Et, Bu,
Hex, Oct, Dec, iPr) and of BuMgBr in [D8]THF showed the up-
field-shifted signals characteristic of methylene or methine
groups, respectively, directly bound to magnesium (Figure 5
and Figures S85–S113 in the Supporting Information). In all
cases, a single set of signals was observed, suggesting the
presence of only a single distinct species of BuMgCl or the op-
eration of fast equilibria. Cooling down the sample to 193 K
did not result in additional signals.
BuMgCl+LiCl
The addition of 1 equivalent of LiCl to solutions of BuMgCl had
no discernible effects on the H and 13C NMR spectra (see the
1
Figure 5. 1H DOSY spectrum of a 25 mm solution of BuMgCl in [D8]THF at
298 K. TPhN=1,2,3,4-tetraphenylnaphthalene added as internal reference;
the signal marked with an asterisk results from traces of grease.
7
Supporting Information, Figures S115–S117). The Li NMR spec-
tra of the BuMgCl·LiCl solutions exhibited a signal at d=
0.28 ppm, which slightly differed from the chemical shift of
0.49 ppm measured for solutions of LiCl (see the Supporting
Information, Figure S118).
1
Moreover, we used H DOSY experiments to determine the
1
diffusion coefficients of the organomagnesium species
(Figure 5 and Figures S88, S92, S96, S100, S104, S108, and S113
in the Supporting Information). We assigned a fixed value to
the diffusion coefficient of an internal reference (1,2,3,4-tetra-
phenylnaphthalene; logDref,fix =À9.1054) for all measurements
and, thus, normalized all diffusion coefficients of the Grignard
reagents. By using a carefully chosen external calibration curve
(ECC; see the Supporting Information, Figure S114 and Ta-
bles S5–S9), we could convert these diffusion coefficients into
molecular weights. These molecular weights showed a monoto-
nous and evenly spaced increase with growing length of the
The H DOSY experiments determined diffusion coefficients
for BuMgCl·LiCl, which were consistently higher than those for
simple BuMgCl (see the Supporting Information, Figure S119
and Table S10). This finding indicated that the addition of LiCl
led to the (partial) association of the two components. The
molecular weight deduced from the experiment (MWdet
ꢁ
299.6 gmol
)
À1 [42] appears to agree with that calculated for a 1:1
adduct of BuMgCl·2THF and LiCl (MWcalc =303.5 gmolÀ1) and
to suggest that such species form almost quantitatively. If, in-
stead, it is assumed that the adduct of the two components
binds 3 molecules of THF, thus affording BuMgCl·LiCl·3THF
Chem. Eur. J. 2016, 22, 7752 – 7762
www.chemeurj.org
7757
ꢀ 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Paper
(MWcalc =375.6 gmolÀ1), the determined molecular weight im-
plies its incomplete formation. In addition, we also studied sol-
utions of BuMgBr·LiBr, whose NMR spectra closely resembled
those of BuMgBr as well (see the Supporting Information, Fig-
ures S120–S124).
Discussion
Aggregation state of RMgCl
The observation of trinuclear complexes by ESI mass spectrom-
etry disagrees with the presence of mononuclear species infer-
red from the DOSY measurements. DOSY has recently been es-
tablished as a standard method for determining the aggrega-
tion state of analytes in solution.[26] Furthermore, the outcome
of the DOSY experiments is in accordance with the current
consensus on the behavior of Grignard reagents in
THF.[15,16a,b,d,e] Thus, it appears beyond any doubt that the
probed RMgCl compounds form mononuclear species in THF.
The ECC DOSY measurements afford further information in
that they point to the attachment of two molecules of THF to
each RMgCl molecule by relating diffusion coefficients to mo-
lecular weights. This finding fully matches expectations, be-
cause the magnesium center thus reaches its preferred coordi-
nation number of 4.
Quantum chemical calculations
Trinuclear ate complexes
For a better understanding of the almost exclusive formation
of trinuclear ate complexes upon ESI of solutions of RMgCl and
RMgCl·LiCl, we calculated the energetically most stable gas-
phase structures of the model complexes [Mg3Cl7]À and
[Me7Mg3]À. Considering 12 different starting geometries for
each species (see the Supporting Information, Figure S125), 4
distinct minima for [Mg3Cl7]À and 2 for [Me7Mg3]À could be
identified. In both cases, the global minimum corresponded to
an open cubic structure, which was predicted to be significant-
ly lower in energy than the other isomer(s) (Figure 7 and Fig-
ure S126 and Table S11 in the Supporting Information).
It remains to be clarified why the ESI mass spectrometry ex-
periments arrive at a different result. The ESI process produces
small charged droplets, from which the finally observed bare
analyte ions are emitted into the gas phase. As the charged
droplets rapidly lose solvent molecules due to evaporation, the
effective concentration of the analyte increases. This increase
may result in a shift of the aggregation equilibria,[21,43] which
can rationalize the observed formation of higher aggregates.
Nevertheless, the high specificity of this process and the
strong preference for the generation of trinuclear complexes
are remarkable. According to the theoretical calculations, the
trinuclear anions adopt open cubic structures. This type of
structure is closely related to that found for crystals of iPrMgCl
and iPrMgCl·LiCl (Scheme 1)[19] and can be converted into the
latter by the addition of one [MgX]+ moiety. It thus appears
that the ESI process does not form random higher aggregates,
but anionic analogues of the species present in the solid state.
One might even argue that in this case the ESI process emu-
lates the crystallization process at the molecular level.
Figure 7. Calculated gas-phase minimum-energy structures for [Mg3Cl7]À
(left) and [Me7Mg3]À (right).
MeMgCl+LiCl
We also used quantum chemical calculations to predict the
structures of different minima for the system MeMgCl+LiCl. In
the gas phase, the association of the two components results
in a heterobimetallic adduct, in which two chlorine atoms
bridge the two metal centers. Heterolytic dissociation of this
adduct affords the corresponding anionic Mg complex
(Figure 8). Essentially the same species are found when solva-
tion by THF is taken into account by using COSMO[37] and ex-
plicitly considering THF molecules to complete the assumed
tetrahedral first coordination spheres of the metal ions (see
the Supporting Information, Figure S127).
For reagents bearing phenyl or 4-anisyl substituents, the ESI-
mass spectrometric experiments detected not only trinuclear,
but also dinuclear complexes. This finding indicates a special
situation for aryl systems.[44]
Effect of LiCl
The present results consistently indicate that Grignard reagents
RMgCl and LiCl react with each other to form ate complexes
(Scheme 2).[45] This reaction involves a 1:1 stoichiometry of
both components, as is clearly seen from the electrical conduc-
tivity measurements and the Job plot with its maximum at
Figure 8. Calculated gas-phase minimum-energy structures for the adduct
formed from MeMgCl and LiCl (left) and the anionic complex resulting from
its heterolytic dissociation (right).
Scheme 2. Formation of ate complexes from RMgCl and LiCl and their heter-
olytic dissociation.[45] Coordinating solvent molecules are omitted for clarity.
Chem. Eur. J. 2016, 22, 7752 – 7762
www.chemeurj.org
7758
ꢀ 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Paper
À
2 RMgCl Ð ½RMgðsolvފþ þ ½RMgCl2Š
ð4Þ
x=0.5 for the case of BuMgCl. The quantum-chemical calcula-
tions directly show the ate character of the 1:1 adduct, in
which the Lewis-acidic Mg center has added the chloro ligand
provided by LiCl. Heterolytic dissociation of the contact-ion
pair furnishes the free ions detectable by the ESI-mass spectro-
metric and conductometric experiments (Scheme 2). Thus, our
findings qualitatively agree very well with the hypothesis of
Krasovskiy, Straub, and Knochel.[18] Apart from the fact that ate
complexes form, it is also of interest to which extent they do
so. The formation of trinuclear anions during the ESI process
implies that the initially present [RMgCl2]À species find plenty
of neutral RMgCl to react with. This apparent abundance of
neutral RMgCl can be considered as a first indication for the in-
complete formation of ate complexes from RMgCl and LiCl in
the sample solutions. Further evidence pointing into this direc-
tion comes from the conductivity measurements. The lack of
saturation effects, except for the case of iPrMgCl, in the titra-
tion experiments clearly indicates that the equilibrium be-
tween the separated components and the ate complex must
lie on the left side for concentrations of cꢀ250 mm. Likewise,
the relatively broad maximum in the Job plot indicates that
the association between BuMgCl and LiCl does not occur
quantitatively. As outlined above, the interpretation of the
DOSY results depends on the number of THF molecules bind-
ing to the Li+[BuMgCl2]À complex. As suggested by the quan-
tum chemical calculations, the ate complex binds three THF
molecules such that both metal centers are tetrahedrally coor-
dinated. Under this assumption, the DOSY results are also in
accordance with an incomplete formation of ate complexes
(for c=25 mm).
The addition of LiCl strongly increases the concentration of
free ions in solution because the resulting ate complexes
Li+[RMgCl2]À can undergo heterolytic dissociation much more
easily than simple RMgCl. A very similar situation is found for
the related case of RZnCl and Li+[RZnCl2]À.[25b] As the conduc-
tivity measurements show, the tendency toward dissociation
also depends on the nature of the organic substituent. The
higher conductivities recorded for the tert-butyl system indi-
cate that sterically demanding substituents facilitate heterolytic
dissociation. We noted a similar behavior for the cyanocuprates
LiCuR2·LiCN.[25a] An important difference between RMgCl·LiCl
and LiCuR2·LiCN lies in the higher molar conductivities of the
latter, which exceed those of the former by one order of mag-
nitude.[25a] On the basis of the degree of heterolytic dissocia-
tion estimated for the cyanocuprates,[25a] we thus infer that, at
synthetically relevant concentrations in THF, at most a few per-
cent of the turbo-Grignard reagents undergo heterolytic disso-
ciation.
Schlenk equilibria
À
The simultaneous observation of several different [RnMg3Cl7Àn
]
ate complexes by ESI mass spectrometry can be viewed as
direct evidence for the operation of Schlenk-type equilibria.
This finding is in line with previous results.[15,16a,b,d,e] Probably,
the Schlenk equilibria operate not only in solution, but also in
the nanodroplets produced during the ESI process, where the
increased analyte concentrations make encounters between in-
dividual RMgCl molecules more likely. In comparison to the
Grignard reagents bearing primary alkyl groups, those with
a secondary or tertiary alkyl substituent apparently had a some-
The present findings can also be compared with our previ-
ous results on the related system RZnCl·LiCl.[25b] For the latter,
ate complexes Li+[RZnCl2]À form, which are completely analo-
gous to the magnesates observed in the present work. Inter-
estingly, the organozinc halides show a higher tendency to
add LiCl than the Grignard reagents (Kass ꢀ100 LmolÀ1 for
BuZnCl).[25b] This difference apparently does not reflect simple
trends in Lewis acidity, which is higher for MgII than for ZnII.
Possibly, the higher chemical hardness and oxophilicity of the
MgII center favor its interaction with THF as the harder Lewis
base, whereas the less hard ZnII center supposedly can better
bind chloride as the less hard Lewis base.[46]
what lower tendency to undergo exchange reactions and form
À
different [RnMg3Cl7Àn
]
ate complexes. Possibly, this deviating
behavior is caused by steric effects. In the case of iPrMgCl, the
addition of dioxane was found to shift the Schlenk equilibrium
strongly, thus corroborating earlier reports.[18]
The failure of the NMR spectroscopic experiments to detect
different organomagnesium species even at low temperatures
points to high efficiencies of the intermolecular exchange reac-
tions. The occurrence of such exchange reactions was directly
proven by the ESI-mass spectrometric detection of mixed com-
plexes upon combination of iPrMgCl and PhMgCl. The fact
that the detected ate complexes do not show a statistical dis-
tribution of the two different organic substituents possibly re-
flects different ESI activities of the isopropyl- and phenyl-con-
taining magnesium ate complexes.
Heterolytic dissociation of RMgCl and Li+[RMgCl2]À
Both the ESI mass spectrometry experiments and the conduc-
tivity measurements show that solutions of Grignard reagents
in THF contain ionic species even in the absence of added LiCl.
These ions presumably form in an ionic disproportionation
[Eq. (4)]. As reported previously, the inferred organomagnesi-
um cations[20] have so far eluded detection by ESI mass spec-
trometry because they apparently undergo extremely fast pro-
tolysis reactions with traces of protic solvents remaining in the
ionization source.[23c] We have, however, previously observed
analogous or similar intact organometallic cations in solutions
of organozinc and -indium halides.[21b,25b,47]
Reactivity
Taken together, our present results and the previous findings
of Knochel and co-workers[9,18] point to the in situ-formed mag-
nesates as the reactive component of turbo-Grignard reagents.
The present study does not differentiate between contact ion
pairs Li+[RMgCl2]À and free [RMgCl2]À anions as active species,
but the previously reported small effect of added Li+-selective
Chem. Eur. J. 2016, 22, 7752 – 7762
www.chemeurj.org
7759
ꢀ 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Paper
[12]crown-4[18] suggests that Li+[RMgCl2]À and [RMgCl2]À exhib-
it comparable reactivities in bromine–magnesium exchange re-
actions. The intermediates formed in these reactions could be
directly observed by ESI mass spectrometry. The detected spe-
cies resemble those found for turbo-Grignard reagents
RMgCl·LiCl prepared by insertion reactions.
NMR spectroscopy has the advantage of probing analytes ir-
respective of their charge state. For the problem under investi-
gation, simple one-dimensional NMR spectroscopy proved to
be of limited value, however. In contrast, the ECC DOSY
method was much more instructive. This technique employs
only one internal reference. By using this method and system-
atically varying the length of the organic substituents in the
probed systems, we were able to determine the aggregation
states of these systems unambiguously. The same approach
also holds promise for the characterization of other analytes
by DOSY.
Apart from the synthetically valuable bromine–magnesium
exchange reactions, the investigated Grignard reagents and
their turbo variants also undergo unwanted decomposition re-
actions with traces of moisture or oxygen. The propensity
toward both reactions strongly depends on the organic sub-
stituent of the Grignard reagent. Hydrolysis proceeds particu-
larly quickly for PhMgCl and PhMgCl·LiCl, but the reason for
their enhanced reactivities is unclear. In contrast, oxidation re-
actions only occur for iPrMgCl, tBuMgCl, and their turbo coun-
terparts. The higher reactivities of secondary and tertiary alkyl
groups indicate the involvement of radicals in these processes.
Finally, the gas-phase fragmentation experiments deter-
mined the unimolecular reactivity of the polynuclear organo-
magnesate anions. The observed loss of organyl-rich neutral
fragments resembles the reactions found for other polynuclear
organometallic ate complexes.[21a,b] This fragmentation pattern
maximizes the number of electron-withdrawing halogen atoms
in the anionic fragments and, thus, helps to stabilize the nega-
tive charge.
Conclusion
By using a combination of different experimental methods and
quantum-chemical calculations, we were able to identify and
characterize the molecular constituents of Grignard reagents
RMgCl and their turbo variant RMgCl·LiCl in ethereal solvents.
DOSY measurements of solutions of both reagents in THF
showed the presence of complexes containing a single magne-
sium center. Negative ion-mode ESI mass spectrometry result-
ed in the detection of trinuclear complexes, which apparently
formed due to the concentration increase associated with the
ESI process. Quantum chemical calculations suggested that
these trinuclear complexes adopt open cubic geometries and,
thus, markedly resemble structures previously found for
iPrMgCl and iPrMgCl·LiCl in the solid state. The simultaneous
observation of organyl-rich complexes [RnMg3Cl7Àn]À (nꢁ4) and
organyl-poor ones (nꢂ3), moreover directly points to the oc-
currence of pronounced Schlenk equilibria. The highly dynamic
behavior of the organomagnesium species is also reflected in
the formation of exchange products upon mixing two different
Grignard reagents. ESI mass spectrometry is ideally suited to
analyze such processes, as it also unambiguously demonstrat-
ed the occurrence of magnesium–bromine exchange reactions
upon treatment of a Grignard reagent with an aryl bromide.
The remarkable similarity of the negative ion-mode ESI mass
spectra recorded for solutions of conventional Grignard re-
agents RMgCl on the one hand and for their turbo variant
RMgCl·LiCl on the other indicates that they do not exhibit sig-
nificant qualitative differences. Both reagents form ate com-
plexes to some extent, which heterolytically dissociate to
afford the detected free ions. The degree of magnesate forma-
tion and heterolytic dissociation is low for RMgCl and strongly
increases upon the addition of LiCl, as revealed by the electri-
cal conductivity measurements. Thus, the two reagents differ
at a quantitative level. The larger fraction of the more electron-
rich and, thus, putatively more nucleophilic ate complexes in
the case of the turbo-Grignard reagents explains their en-
hanced reactivity.
Comparison of analytical methods
Among the different analytical methods applied in the present
work, ESI mass spectrometry provides the most detailed infor-
mation. This technique affords insight at the strictly molecular
level and can, for example, resolve fast Schlenk equilibria. The
present work comprises the first successful analysis of intact
Grignard reagents by ESI mass spectrometry and thus demon-
strates that it can also be successfully applied to the analysis
of extremely reactive and sensitive species. However, draw-
backs of this method include its limitation to charged analytes
and the perturbation introduced by the very nature of the ESI
process. As the present findings clearly show, this perturbation
can result in drastic shifts of the aggregation equilibria. Fur-
thermore, very weakly bound complexes, such as THF adducts
of the magnesium ate anions, do not survive the ESI process
(in marked contrast to their more stable cationic ana-
logues).[23c]
Like ESI mass spectrometry, electrical conductivity measure-
ments only probe the charged components of the sample so-
lution. These measurements do not perturb the analyzed
system and, thus, provide direct information on the equilibria
operating in solution. A rigorous quantitative analysis of the
obtained data is difficult, however, due to the complicating
effect of inter-ionic interactions. These interactions are very
pronounced for the highly concentrated sample solutions in- Acknowledgements
vestigated in the present work. Experiments with less concen-
trated solutions, in turn, would necessarily suffer from an in-
creased degree of unwanted hydrolysis and/or oxidation reac-
tions.
We thank Nils Gehrmann, Nico Graw, Hanna Hubrich, Vanessa
Reusche, Lucas Weber, and Rahel Ziemer for carrying out some
of the conductometric experiments and gratefully acknowl-
Chem. Eur. J. 2016, 22, 7752 – 7762
www.chemeurj.org
7760
ꢀ 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Paper
[17] a) P. Jolibois, Compt. Rend. 1912, 155, 353–355; b) W. V. Evans, R. Pear-
son, J. Am. Chem. Soc. 1942, 64, 2865–2871; c) R. E. Dessy, R. M. Jones,
J. Org. Chem. 1959, 24, 1685–1689.
edge financial support by the Deutsche Forschungsgmein-
schaft (KO 2875/4-1).
[18] A. Krasovskiy, B. F. Straub, P. Knochel, Angew. Chem. Int. Ed. 2006, 45,
159–162; Angew. Chem. 2006, 118, 165–169.
[19] F. Blasberg, M. Bolte, M. Wagner, H.-W. Lerner, Organometallics 2012, 31,
1001–1005.
Keywords: Grignard reagents · mass spectrometry · NMR
spectroscopy · reactive intermediates · salt effect
[20] For examples of a structurally characterized anionic trisorganyl magne-
sium ate complex and an organomagnesium cation, see: C. Lichten-
berg, T. P. Spaniol, I. Peckermann, T. P. Hanusa, J. Okuda, J. Am. Chem.
Soc. 2013, 135, 811 –821.
[1] V. Grignard, C. R. Acad. Sci. 1900, 130, 1322–1324.
[2] D. Seyferth, Organometallics 2009, 28, 1598–1605.
[3] M. S. Kharasch, O. Reinmuth, Grignard Reactions of Nonmetallic Substan-
ces, Prentice-Hall, New York, 1954.
[4] K. Oshima in Science of Synthesis, Vol. 7 (Ed.: H. Yamamoto), Thieme,
Stuttgart, 2004, pp. 573–596.
[21] a) K. Koszinowski, P. Bçhrer, Organometallics 2009, 28, 100–110; b) K.
Koszinowski, J. Am. Chem. Soc. 2010, 132, 6032–6040; c) A. Putau, H.
Brand, K. Koszinowski, J. Am. Chem. Soc. 2012, 134, 613–622; d) T. D.
Blümke, T. Klatt, K. Koszinowski, P. Knochel, Angew. Chem. Int. Ed. 2012,
51, 9926–9930; Angew. Chem. 2012, 124, 10064–10068.
[5] a) G. Fouquet, M. Schlosser, Angew. Chem. Int. Ed. Engl. 1974, 13, 82–83;
Angew. Chem. 1974, 86, 50–51; b) D. Martin, S. Kehrli, M. d’Augustin, H.
Clavier, M. Mauduit, A. Alexakis, J. Am. Chem. Soc. 2006, 128, 8416–
8417; c) S. R. Harutyunyan, T. den Hartog, K. Geurts, A. J. Minnaard, B. L.
Feringa, Chem. Rev. 2008, 108, 2824–2852.
[6] a) M. Tamura, J. K. Kochi, J. Am. Chem. Soc. 1971, 93, 1487–1489;
b) R. J. P. Corriu, J. P. Masse, J. Chem. Soc. Chem. Commun. 1972, 144a;
c) K. Tamao, K. Sumitani, M. Kumada, J. Am. Chem. Soc. 1972, 94, 4374–
4376; d) K. Tamao, K. Sumitani, Y. Kiso, M. Zembayashi, A. Fujioka, S.-i.
Kodama, I. Nakajima, A. Minato, M. Kumada, Bull. Chem. Soc. J. 1976, 49,
1958–1969; e) C. E. I. Knappke, A. Jacobi von Wangelin, Chem. Soc. Rev.
2011, 40, 4948–4962.
[7] a) R. D. Rieke, P. M. Hudnall, J. Am. Chem. Soc. 1972, 94, 7178–7179;
b) R. D. Rieke, S. E. Bales, J. Am. Chem. Soc. 1974, 96, 1775–1781; c) T. P.
Burns, R. D. Rieke, J. Org. Chem. 1987, 52, 3674–3680; d) R. D. Rieke, Sci-
ence 1989, 246, 1260–1264.
[8] a) L. Boymond, M. Rottländer, G. Cahiez, P. Knochel, Angew. Chem. Int.
Ed. 1998, 37, 1701–1703; Angew. Chem. 1998, 110, 1801–1803; b) I.
Sapountzis, P. Knochel, Angew. Chem. Int. Ed. 2002, 41, 1610–1611;
Angew. Chem. 2002, 114, 1680–1681; c) A. E. Jensen, W. Dohle, I. Sap-
ountzis, D. M. Lindsay, V. A. Vu, P. Knochel, Synthesis 2002, 565–569;
d) P. Knochel, W. Dohle, N. Gommermann, F. F. Kneisel, F. Kopp, T. Korn,
I. Sapountzis, V. A. Vu, Angew. Chem. Int. Ed. 2003, 42, 4302–4320;
Angew. Chem. 2003, 115, 4438–4456.
[9] a) A. Krasovskiy, P. Knochel, Angew. Chem. Int. Ed. 2004, 43, 3333–3336;
Angew. Chem. 2004, 116, 3396–3399; b) F. Kopp, A. Krasovskiy, P. Kno-
chel, Chem. Commun. 2004, 2288–2289; c) H. Ren, A. Krasovskiy, P. Kno-
chel, Org. Lett. 2004, 6, 4215–4217; d) C. B. Rauhut, V. A. Vu, F. F. Flem-
ing, P. Knochel, Org. Lett. 2008, 10, 1187–1189; e) N. M. Barl, V. Werner,
C. Sämann, P. Knochel, Heterocycles 2014, 88, 827–844.
[10] R. L.-Y. Bao, R. Zhao, L. Shi, Chem. Commun. 2015, 51, 6884–6900. Cor-
rection: R. L.-Y. Bao, R. Zhao, L. Shi, Chem. Commun. 2015, 51, 9744.
[11] D. Hauk, S. Lang, A. Murso, Org. Process Res. Dev. 2006, 10, 733–738.
[12] a) F. M. Piller, P. Appukkuttan, A. Gavryushin, M. Helm, P. Knochel,
Angew. Chem. Int. Ed. 2008, 47, 6802–6806; Angew. Chem. 2008, 120,
6907–6911; b) F. M. Piller, A. Metzger, M. A. Schade, B. A. Haag, A. Gav-
ryushin, P. Knochel, Chem. Eur. J. 2009, 15, 7192–7202.
[13] a) R. E. Mulvey, F. Mongin, M. Uchiyama, Y. Kondo, Angew. Chem. Int. Ed.
2007, 46, 3802–3824; Angew. Chem. 2007, 119, 3876–3899; b) E. Hevia,
R. E. Mulvey, Angew. Chem. Int. Ed. 2011, 50, 6448–6450; Angew. Chem.
2011, 123, 6576–6578; c) B. Haag, M. Mosrin, H. Ila, V. Malakhov, P. Kno-
chel, Angew. Chem. Int. Ed. 2011, 50, 9794–9824; Angew. Chem. 2011,
123, 9968–9999; d) A. Harrison-Marchand, F. Mongin, Chem. Rev. 2013,
113, 7470–7562.
[22] a) B. H. Lipshutz, K. L. Stevens, B. James, J. G. Pavlovich, J. Am. Chem.
Soc. 1996, 118, 6796–6797; b) B. H. Lipshutz, J. Keith, D. J. Buzard, Orga-
nometallics 1999, 18, 1571–1574; c) M. A. Henderson, T. K. Trefz, S. Col-
lins, M. Y. Wang, J. S. McIndoe, Organometallics 2013, 32, 2079–2083;
d) T. K. Trefz, M. A. Henderson, M. Y. Wang, S. Collins, J. S. McIndoe, Or-
ganometallics 2013, 32, 3149–3152; e) T. K. Trefz, M. A. Henderson, M.
Linnolahti, S. Collins, J. S. McIndoe, Chem. Eur. J. 2015, 21, 2980–2991.
[23] a) S. Sakamoto, T. Imamoto, K. Yamaguchi, Org. Lett. 2001, 3, 1793–
1795; b) M. A. Schade, J. E. Fleckenstein, P. Knochel, K. Koszinowski, J.
Org. Chem. 2010, 75, 6848–6857; c) C. Schnegelsberg, T. D. Blümke, K.
Koszinowski, J. Mass Spectrom. 2015, 50, 1393–1395.
[24] To avoid the difficulties associated with the direct analysis of organo-
magnesium ions by ESI mass spectrometry, O’Hair and co-workers de-
veloped an elegant gas-phase synthesis of these species: a) R. A. J.
O’Hair, A. K. Vrkic, P. F. James, J. Am. Chem. Soc. 2004, 126, 12173–
12183; b) G. N. Khairallah, C. Thum, R. A. J. O’Hair, Organometallics 2009,
28, 5002–5011; c) G. N. Khairallah, E. J. H. Yoo, R. A. J. O’Hair, Organome-
tallics 2010, 29, 1238–1245; d) M. G. Leeming, G. N. Khairallah, G. da Sil-
va, R. A. J. O’Hair, Organometallics 2011, 30, 4297–4307; e) G. N. Khairal-
lah, C. C. L. Thum, D. Lesage, J.-C. Tabet, R. A. J. O’Hair, Organometallics
2013, 32, 2319–2328.
[25] a) A. Putau, K. Koszinowski, Organometallics 2011, 30, 4771–4778;
b) J. E. Fleckenstein, K. Koszinowski, Organometallics 2011, 30, 5018–
5026.
[26] a) D. Li, I. Keresztes, R. Hopson, P. G. Williard, Acc. Chem. Res. 2009, 42,
270–280; b) G. Kagan, W. Li, R. Hopson, P. G. Williard, Org. Lett. 2010,
12, 520–523; c) D. R. Armstrong, P. García-lvarez, A. R. Kennedy, R. E.
Mulvey, J. A. Parkinson, Angew. Chem. Int. Ed. 2010, 49, 3185–3188;
Angew. Chem. 2010, 122, 3253–3256; d) D. R. Armstrong, W. Clegg, P.
García-lvarez, M. D. McCall, L. Nuttall, A. R. Kennedy, L. Russo, E. Hevia,
Chem. Eur. J. 2011, 17, 4470–4479; e) P. García-lvarez, R. E. Mulvey,
J. A. Parkinson, Angew. Chem. Int. Ed. 2011, 50, 9668–9671; Angew.
Chem. 2011, 123, 9842–9845; f) A.-C. Pçppler, M. M. Meinholz, H.
Faßhuber, A. Lange, M. John, D. Stalke, Organometallics 2012, 31, 42–
45; g) A. Hernµn-Gómez, E. Herd, E. Hevia, A. R. Kennedy, P. Knochel, K.
Koszinowski, S. M. Manolikakes, R. E. Mulvey, C. Schnegelsberg, Angew.
Chem. Int. Ed. 2014, 53, 2706–2710; Angew. Chem. 2014, 126, 2744–
2748; h) R. Neufeld, D. Stalke, Chem. Sci. 2015, 6, 3354–3364; i) R. Neu-
feld, M. John, D. Stalke, Angew. Chem. Int. Ed. 2015, 54, 6994–6998;
Angew. Chem. 2015, 127, 7100–7104.
[27] A. Krasovskiy, P. Knochel, Synthesis 2006, 5, 890–891.
[28] A. Putau, K. Koszinowski, Organometallics 2010, 29, 3593–3601; Addi-
tion/Correction: A. Putau, K. Koszinowski, Organometallics 2010, 29,
6841–6842.
[14] W. Schlenk, W. Schlenk, Jr., Ber. 1929, 62, 920–924.
[29] K. Koszinowski, F. Lissy, Int. J. Mass Spectrom. 2013, 354/355, 219–228.
[30] a) A. Jerschow, N. Müller, J. Magn. Reson. 1997, 125, 372–375; b) A. Jer-
schow, N. Müller, J. Magn. Reson. Ser. A 1996, 123, 222–225.
[31] F. Neese, WIREs Comput. Mol. Sci. 2012, 2, 73–78.
[15] a) E. C. Ashby, Q. Rev. Chem. Soc. 1967, 21, 259–285; b) F. W. Walker,
E. C. J. Ashby, J. Am. Chem. Soc. 1969, 91, 3845–3850; c) G. E. Parris,
E. C. J. Ashby, J. Am. Chem. Soc. 1971, 93, 1206–1213; d) E. C. Ashby,
Pure Appl. Chem. 1980, 52, 545–569.
[32] F. Neese, ORCA—An ab initio, DFT and semiempirical SCF-MO Package,
Version 3.0.3, Max-Planck-Institute for Chemical Energy Conversion,
2013.
[16] a) R. M. Salinger, H. S. Mosher, J. Am. Chem. Soc. 1964, 86, 1782–1786;
b) M. B. Smith, W. E. Becker, Tetrahedron 1967, 23, 4215–4227; c) T.
Holm, Acta Chem. Scand. 1969, 23, 579–586; d) R. Benn, H. Lehmkuhl,
´
[33] a) A. D. Becke, J. Chem. Phys. 1993, 98, 5648–5652; b) P. J. Stephens, F. J.
Devlin, C. F. Chabalowski, M. J. Frisch, J. Phys. Chem. 1994, 98, 11623–
11627.
K. Mehler, A. Rufinska, Angew. Chem. Int. Ed. Engl. 1984, 23, 534–535;
Angew. Chem. 1984, 96, 521–523; e) T. S. Ertel, H. Bertagnolli, Poly-
hedron 1993, 12, 2175–2184.
Chem. Eur. J. 2016, 22, 7752 – 7762
www.chemeurj.org
7761
ꢀ 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Paper
[34] a) S. Grimme, J. Antony, S. Ehrlich, H. Krieg, J. Chem. Phys. 2010, 132,
154104; b) S. Grimme, S. Ehrlich, L. Goerigk, J. Comput. Chem. 2011, 32,
1456–1465.
the recommended limit for the determination of exact MWdet values
(see the Supporting Information).
ˇ
ˇ
[43] N. G. Tsierkezos, J. Roithovµ, D. Schrçder, M. Oncµk, P. Slavícek, Inorg.
[35] F. Weigend, R. Ahlrichs, Phys. Chem. Chem. Phys. 2005, 7, 3297–3305.
[36] J. Zheng, X. Xu, D. G. Truhlar, Theor. Chem. Acc. 2011, 128, 295–305.
[37] a) A. Klamt, G. Schüürmann, J. Chem. Soc. Perkin Trans. 2 1993, 799–
805; b) S. Sinnecker, A. Rajendran, A. Klamt, M. Diedenhofen, F. Neese, J.
Phys. Chem. A 2006, 110, 2235–2245.
Chem. 2009, 48, 6287–6296.
[44] In the case of PhMgCl, their partial hydrolysis and the ability of the re-
sulting OH groups to form hydrogen bonds could, however, possibly
also favor the formation of dinuclear complexes.
[45] LiCl is known to be present as dimeric Li2Cl2 in THF: a) M. K. Wong, A. I.
Popov, J. Inorg. Nucl. Chem. 1972, 34, 3615–3622; b) H. J. Reich, J. P.
Borst, R. R. Dykstra, P. D. Green, J. Am. Chem. Soc. 1993, 115, 8728–
8741.
[38] G. Sosnovsky, J. H. Brown, Chem. Rev. 1966, 66, 529–566.
[39] No gas-phase fragmentation experiments were performed for
À
[EtnMg3Cl7Àn
]
because extensive overlaps between the isotopologues
of different ions prevented a meaningful mass selection and analysis.
[40] a) M. Hojo, T. Ueda, Z. Chen, M. Nishimura, J. Electroanal. Chem. 1999,
468, 110–116; b) D. Das, J. Solution Chem. 2008, 37, 947–955.
[46] R. G. Pearson, J. Am. Chem. Soc. 1963, 85, 3533–3539.
[47] J. E. Fleckenstein, K. Koszinowski, Chem. Eur. J. 2009, 15, 12745–12753.
[41] J. S. Renny, L. L. Tomasevich, E. H. Tallmadge, D. B. Collum, Angew. Chem.
Int. Ed. 2013, 52, 11998–12013; Angew. Chem. 2013, 125, 12218–12234.
[42] The determined value only corresponds to a lower value because the
associated molar van der Waals density of BuMgCl·LiCl·2THF exceeds
Received: February 15, 2016
Published online on May 6, 2016
Chem. Eur. J. 2016, 22, 7752 – 7762
www.chemeurj.org
7762
ꢀ 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim