New insights into frustrated lewis pairs: structural investigations of intramolecular phosphane-borane adducts by using modern solid-state NMR techniques and DFT calculations

Article abstract of DOI:10.1021/ja210160k

Covalent bonding interactions between the Lewis acid and Lewis base functionalities have been probed in a series of "frustrated Lewis pairs" (FLPs) (mainly substituted vinylene linked intramolecular phosphane-borane adducts), using solid-state nuclear mag

Full text of DOI:10.1021/ja210160k

Article
pubs.acs.org/JACS
New Insights into Frustrated Lewis Pairs: Structural Investigations of
Intramolecular Phosphane−Borane Adducts by Using Modern Solid-
State NMR Techniques and DFT Calculations
Thomas Wiegand,† Hellmut Eckert,*,† Olga Ekkert,‡ Roland Frohlich,‡ Gerald Kehr,‡ Gerhard Erker,*,‡
̈
and Stefan Grimme*,≠
†Institut fur Physikalische Chemie and Graduate School of Chemistry, WWU Munster, Corrensstrasse 30, D 48149 Munster,
̈
̈
̈
Germany
≠Mulliken Center for Theoretical Chemistry, Institut fur Physikalische und Theoretische Chemie, Universitat Bonn, Beringstrasse 4,
D 53115 Bonn, Germany
‡Organisch-Chemisches Institut, WWU Munster, Corrensstrasse 40, D 48149 Munster, Germany
̈
̈
̈
̈
S
* Supporting Information
ABSTRACT: Covalent bonding interactions between the Lewis acid and Lewis
base functionalities have been probed in a series of “frustrated Lewis pairs”
(FLPs) (mainly substituted vinylene linked intramolecular phosphane−borane
adducts), using solid-state nuclear magnetic resonance techniques and
accompanying DFT calculations. Both the 11B NMR isotropic chemical shifts
and nuclear electric quadrupolar coupling parameters turn out to be extremely
sensitive experimental probes for such interactions, revealing linear correlations
with boron−phosphorus internuclear distances. The principal component Vzz of
the 11B electric field gradient tensor is tilted slightly away (∼20°) from the
boron−phosphorus internuclear vector, leading to an improved understanding
of the remarkable reactivity of the FLPs. Complementary 31P{1H}-CPMAS
experiments reveal significant 31P−11B scalar spin−spin interactions (1J ≈ 50
Hz), evidencing covalent bonding interactions between the reaction centers.
Finally, 11B{31P} rotational echo double resonance (REDOR) experiments show systematic deviations from calculated curves
based on the internuclear distances from X-ray crystallography. These deviations suggest non-zero contributions from anisotropic
indirect spin−spin (J anisotropy) interactions, thereby offering additional evidence for covalent bonding.
most reactive examples is the four-membered heterocyclic
ethylene-bridged phosphane−borane adduct 9 (see Figure 1).6
Besides an extremely high reactivity toward dihydrogen, this
intramolecular adduct undergoes many other interesting
addition reactions.10 The class of intramolecular phosphane/
borane adducts has recently been extended to unsaturated C2-
INTRODUCTION
■
Frustrated Lewis acid/base pairs (FLPs), which feature
intramolecular Lewis acid and Lewis base functionalities in
close proximity but within constrained geometries, are of great
interest in the field of homogeneous catalysis.1 Their reactivity
stems from the presence of bulky, sterically demanding ligands
linked systems, which possess more moderated reactivities.16
A
t
(e.g., mesityl, Bu, and C6F5 groups) attached to the reactive
whole series of these molecules is synthesized by a 1,1-
carboboration reaction and therefore represents a new
application in the field of P−C bond activation chemistry.
Despite the large number of experimental reports in the field
of FLP chemistry, only a few deal with the theoretical
understanding of the chemical reaction mechanisms involving
centers, thereby inhibiting the anticipated Lewis acid/base
adduct formation.2 The close proximity of the Lewis acid and
the Lewis base moieties leads to a remarkable cooperative
reaction behavior that is in many cases comparable to that of
organometallic compounds. The most prominent example is
the activation of dihydrogen,3−7 which offers a completely new
line of research for the development of metal-free, environ-
mentally friendly hydrogenation catalysis. But also the
activation of CO2,8 carbonyl compounds,9,10 alkenes,11,12
dienes,13 and alkynes14 and the capture of NO15 have been
reported. A large variety of FLPs can be formed from a
phosphane Lewis base and a borane Lewis acid functionality
combined in an intermolecular4 or intramolecular manner.
Typical adducts (1−9) as investigated in the present manu-
script are shown in Figure 1. Of these, one of the earliest and
these molecules. Pap
́
ai17,18 and Grimme19 have proposed
dihydrogen activation mechanisms, in the latter case implicating
especially the polarization of dihydrogen in the electric field
generated by the donor/acceptor atoms of the intramolecular
adducts. A common structural characteristic of these com-
pounds enforcing cooperative effects is the presence of
Received: October 28, 2011
Published: January 20, 2012
© 2012 American Chemical Society
4236
dx.doi.org/10.1021/ja210160k | J. Am. Chem. Soc. 2012, 134, 4236−4249

Journal of the American Chemical Society
Article
Figure 1. Scheme showing the structures of the investigated compounds.
significant noncovalent interactions among the bulky ligands. In
addition the catalytic activity appears to be modulated by weak
interactions (of either the covalent or the noncovalent type)
between the two reactive centers. The present study introduces
modern solid-state NMR techniques, for the first time, as an
experimental probe of such interactions for a series of closely
related P/B pairs (mainly substituted vinylene-bridged
phosphane−borane adducts16), in which the internuclear
distance can be varied over a comparatively wide range by
suitable choice of ligands. We explore compositional trends
signifying weak 11B···31P covalent interactions by studying the
NMR parameters (such as nuclear electric quadrupolar
coupling constants, isotropic chemical shifts and indirect
spin−spin coupling constants) with the goal of obtaining a
deeper structural insight into the reactivity of those molecules.
11B MAS NMR spectroscopy has developed into a well-
established tool for the structural characterization of inorganic
boron compounds, as well as boranes.20 Both the 11B chemical
iso
CS
shifts, δ , and quadrupole coupling constants, CQ, are highly
Figure 2. Crystal structure of 6 acting as a representative model for the
sensitive to the boron coordination number, high-frequency
shifts and large CQ values being associated with trigonally
coordinated boron sites and low-frequency shifts and small CQ
values with four-coordinated boron species.20 Finally, the
electric field gradient (EFG) asymmetry parameter, η,
characterizes the degree to which the local symmetries of the
three- or four-coordinated boron species deviate from the axial
symmetries D3h or C3v, respectively. In the present study, we
establish 11B magic angle spinning (MAS) NMR in the field of
frustrated Lewis pairs and highlight the remarkable sensitivity of
its parameters toward even small changes in the surroundings
of the boron nuclei in local environments that might be
considered between three- and four-coordinated bonding
scenarios. As previous works have shown, the 11B EFG tensor
is also available with high accuracy by DFT calculations,
allowing a visualization of the individual tensor components in
the molecular axis frame.21,22 In a similar vein, the present study
examines 31P solid-state NMR parameters such as isotropic
chemical shifts and scalar 31P−11B spin−spin coupling
constants as probes for covalent boron−phosphorus inter-
actions. Finally, we explore the potential and limitations of
11B{31P} rotational echo double resonance (REDOR)23
experiments for internuclear distance measurements in these
compounds.
intramolecular B/P adducts.
extraordinarily rigid backbone of those molecules. In the
compounds 724 and 8,25 the Lewis acid and base functionalities
are linked via a trimethylene and cyclohexylene bridge,
respectively. In all molecules, the Lewis acidity of the boron
site is enhanced by electron-withdrawing C6F5 ligands, while
the phosphorus site is electron-rich due to ligands such as
mesityl or phenyl. Especially characteristic are significant
noncovalent π−π interactions between the electron-poor and
electron-rich arene ligands leading to a further stabilization of
these molecules.19 Compound 10 represents a typical Lewis
acid/base adduct,26,27 while compound 1128 acts as a reference
material without any interaction between the base and the acid
center. Compounds of type 11 are formed at the beginning of
the carboboration and are isolated as stable intermediates, while
compound 12 represents the borane starting material. In all
cases (except 11 and 12), the Lewis acid and Lewis base
moieties are in close contact exhibiting B···P internuclear
distances varying between 2.03 and 2.19 Å. Within the series of
substituted vinylene-bridged adducts, this distance can be
adjusted by choice of the ligands on the phosphorus Lewis base
site: the more bulky those ligands are, the larger is the boron−
phosphorus distance. While these adducts show some of the
typical FLP reactions, they do not activate dihydrogen under
standard reaction conditions (i.e., ambient temperature and 60
bar H2 pressure). Nevertheless, we note that the thermody-
namic window for the H2 activation is extremely small,
indicating that their reactivity is already moderated (relative
The substituted vinylene-bridged intramolecular adducts
(compounds 1−6 in Figure 1, for a characteristic structure
see Figure 2) consist of a heterocyclic, nearly planar four-
membered ring structure as present also in the most prominent
FLP, 9.6 The bridging olefinic functionality leads to an
4237
dx.doi.org/10.1021/ja210160k | J. Am. Chem. Soc. 2012, 134, 4236−4249

Journal of the American Chemical Society
Article
Experimentally determined chemical shift anisotropy (CSA) parame-
ters were included in the simulations (for details see Figure S1,
Supporting Information). All the simulations utilize magnetic B···P
dipole−dipole coupling constants calculated from the B···P
internuclear distances extracted from the crystallographic information
available.
Ab Initio Calculations. All calculations were carried out using the
program packages TURBOMOLE (version 6.0 and 6.3)41,42 and
GAUSSIAN (version GAUSSIAN09).43 The geometry optimizations
have been performed on a DFT meta-GGA (TPSS44) level of theory
(starting with the crystal structure if available) applying the recently
developed D3 dispersion correction45 and Ahlrich’s def2-TZVP46 basis
set. All geometry optimizations were performed within the
TURBOMOLE program suite. In all TURBOMOLE SCF calculations,
an energy convergence criterion of 10−7 Eh was chosen, and in all
geometry optimizations, an energy convergence criterion of 5 × 10−7
Eh was chosen. The integration grid was set to m4,47 and the RI
approximation48,49 was used.
For the calculations of nuclear electric quadrupole coupling
tensors,21,50 the positions of the heavy atoms were taken from the
crystal structure, whereas the positions of the hydrogen atoms were
optimized on the DFT TPSS level with Ahlrich’s def2-TZVP basis set.
The calculations of the electric field gradients were performed on a
GGA DFT level (functional B97-D51) using the program package
GAUSSIAN09. The def2-TZVP basis set obtained from the EMSL
database52,53 was modified in such a way that tighter basis functions on
the boron atom (extracted from the cc-pCVTZ basis set,54,55 for
details see Supporting Information section) were included for having a
more accurate description of the region near the boron nucleus. The
GAUSSIAN output files were analyzed by using the program
EFGShield,22 version 2.2, for determination of CQ and η values and
visualizing the orientation of the electric field gradient tensor in the
molecular geometry by using the DIAMOND software.56
to those of 8 and 9) by weak intramolecular interactions
between the centers. In the present study, quantitative
information to this effect will be presented based on combined
solid-state NMR experiments and DFT calculations.
EXPERIMENTAL SECTION
The synthesis and general characterization of compounds 1−12 have
been already reported,6,16,24−29 and relevant details are given in the
Supporting Information.
■
Solid-state NMR measurements were carried out on BRUKER
Avance III (300 MHz), Avance DSX 400, and BRUKER Avance DSX
500 spectrometers, corresponding to magnetic flux densities of 7.05,
9.4, and 11.74 T, respectively. All spectrometers were equipped with 4
mm NMR double and triple resonance probes operating at MAS
rotation frequencies between 2.4 and 14 kHz.
11B spectra were mainly measured on the 500 MHz spectrometer at
a Larmor frequency of 160.5 MHz. Signals were acquired following
excitation with 30° pulses about 0.6 μs in length and repetition times
ranging between 5 and 30 s. Proton decoupled 11B{1H} spectra were
1
obtained at 9.4 T with 90° pulses of about 1.5 μs length and a H
decoupling pulse length of 7.1 μs applied in a TPPM-1530 decoupling
scheme. 11B triple-quantum (TQ) MAS NMR spectra31 were obtained
at 11.74 T using a three-pulse z-filtering sequence.32 The optimized
lengths of the strong preparation and reconversion pulses were varied
between 3.8 and 5.5 μs and between 1.5 and 1.7 μs, respectively
(nutation frequency ν1 ≈ 130 kHz for a liquid sample). The single
quantum signal was detected by a soft pulse of 10 μs length in all cases
(nutation frequency ν1 ≈ 30 kHz for a liquid sample). The increment
for the evolution time was adjusted to 1/(14000 × 2n) s (with n value
of 1, 2, or 3), and a recycle delay of 5−30 s was used. A comparison of
the centers of gravity in the isotropic F1 and anisotropic F2
dimensions reveals the isotropic chemical shift, δCisSo, and the
“second-order quadrupolar effect (SOQE)” defined as SOQE =
CQ(1 + η2/3)1/2 with CQ and η representing the nuclear electric
quadrupolar coupling constant and the electric field gradient
asymmetry parameter, respectively. Those parameters were also
determined quantitatively by line shape fitting analysis of the MAS
spectra and MQMAS slices using the DMFIT software (version
2011).33 Chemical shifts are reported relative to a BF3·Et2O standard.
31P{1H} CPMAS spectra were measured at 9.4 T with 1H 90° pulse
lengths of 4−6 μs, a contact time of 5 ms, and a relaxation delay of 5 s.
Hartmann−Hahn conditions were adjusted on NH4H2PO4. An
efficient polarization transfer was achieved by a ramped-amplitude
CP step34 with νRF(1H) being swept from 54 to 27 kHz in 64 steps (in
the case of a 1H 90° pulse length of 4.6 μs). All spectra were acquired
with TPPM-15 proton decoupling during the data acquisition applying
decoupling pulses of 6.7−10 μs length (∼10/12 π pulses). Additional
31P{1H} CPMAS spectra using the SW-TPPM-15 decoupling
In case of model adduct calculations as a function of the boron−
phosphorus distance, the latter was fixed to characteristic values
ranging between 1.85 and 2.6 Å, while the rest of the structure was
optimized on the TPSS-D3 def2-TZVP level of theory; CQ and η
values were calculated at the B97-D/def2-TZVP level of theory.
The magnetic shielding calculations were performed within the
GIAO (gauge-independent atomic orbitals) framework.50,57 For 11B,
magnetic shieldings were calculated on the B3-LYP58,59 and BP-8660,61
level of theory with the def2-TZVP basis set using the TURBOMOLE
program package. Chemical shifts are referenced to BF3·Et2O by using
B2H6 (δ(B2H6) = 16.6 ppm vs BF3·Et2O) as an external standard
(σBP‑86(B2H6) = 81.04 ppm, σB3‑LYP(B2H6) = 84.23 ppm).62−65 31P
chemical shifts were calculated with the B3-LYP functional and the
def2-TZVP basis set. Chemical shifts were referenced to phosphoric
acid (σB3‑LYP = 274.31 ppm). 31P chemical shift anisotropy (CSA)
parameters were calculated on the B3-LYP TZVP66 level of theory
using the program package GAUSSIAN09.
scheme35 were recorded at 7.1 T with a H 90° pulse length of 3.5
1
31P/11B indirect spin−spin coupling constants were calculated on a
B3-LYP TZVP level of theory using the program package
GAUSSIAN09. The same geometries as in the calculations of nuclear
quadrupole couplings were used. In case of the trans model of 1, a
complete geometry optimization was carried out on the TPSS-D3/
def2-TZVP level of theory. The analysis of the distance-dependence of
μs, a contact time of 5 ms and a relaxation delay of 5 s. This
decoupling sequence changes the amplitude of the decoupling pulses
(optimized for each experiment, approximately 10/12 of the π pulse)
in a linear way between 75% and 125% of the initial value.36 Line
shape analysis was done by using the DMFIT software (version 2011).
Chemical shifts are reported relative to a 85% H3PO4 solution.
11B{31P} REDOR experiments were conducted at 9.4 T with the
compensated REDOR scheme using radio frequency power levels
corresponding to 180° pulses of 6.5−7.5 μs for 11B and 31P. The
phases of the 31P π pulses were alternated according to the XY4 phase
cycling scheme.37 Spinning speeds of 12−14 kHz were used. For
creating a reproducible magnetization in each experiment, a saturation
comb consisting of 32 90° pulses was applied. Constant-time-REDOR
(CT-REDOR) experiments38 (also with compensation) were
performed under similar experimental conditions with spinning speeds
of 5 and 10 kHz, respectively. The simulations of both CT-REDOR
and REDOR curves were carried out with the SIMPSON software.39
For an accurate description of the oscillatory part of the REDOR
curves, the REPULSION powder angle averaging scheme (rep2000)40
was applied and 36 equally spaced gamma angles were used.
1
the J spin−spin coupling constant for compound 10 was performed
by varying the boron−phosphorus distance within the framework of
the crystal structure.
Wiberg bond order indices67 in the Cartesian atomic orbital (CAO)
basis were calculated from the TPSS Kohn−Sham determinants by
using the TURBOMOLE program package and those in the natural
atomic orbitals (NAO) basis were calculated by using the NBO
program (version 3.1)68 as included in GAUSSIAN09.
RESULTS AND DISCUSSION
■
11B MAS NMR and DFT Calculations. Figure 3
summarizes the 11B MAS NMR data obtained on the
intramolecular adducts studied as well as on the reference
4238
dx.doi.org/10.1021/ja210160k | J. Am. Chem. Soc. 2012, 134, 4236−4249

Journal of the American Chemical Society
Article
Figure 4. 11B MAS NMR spectrum of the educt of the 1,1-
carboboration, B(C6F5)3 (12), acquired at 11.7 T with a rotation
frequency of 14 kHz (a) and simulated spectrum using the SIMPSON
program package (b) assuming a CQ of 4.3 MHz, an η value of 0.02,
and Δσ = 300 ppm. The inset on top shows the DFT calculated (B97-
D, def2-TZVP (modified), see Experimental Section) orientation of
the EFG in the molecular axis frame. Vzz is oriented perpendicular to
the boron coordination plane.
Figure 3. 11B MAS NMR spectra for the investigated P/B compounds
(straight lines) and their corresponding simulations (dashed lines): (a)
1 (R = mes, R′ = ph), (b) 2 (R = ph, R′ = ph), (c) 3 (R = ph, R′ = n-
propyl), (d) 4 (R = ph, R′ = PPh2), (e) 5 (R = mes, R′ = tolyl), (f) 6
(R = ph, R′ = me), (g) 7, (h) 8, (i) 9, (j) 10, and (k) 11. All spectra
were acquired at 11.7 T, except g and i, which were measured at 9.4 T
under TPPM-15 proton decoupling. The + sign marks impurities.
al. for related compounds (e.g., trimesitylborane, BMes3, and
triphenyl borate, B(OPh)3), the second-order quadrupolar
perturbation line shape of the MAS center band is significantly
affected by a large chemical shift anisotropy.20 While the crystal
structure of B(C6F5)3 is not known, the characteristic 11B MAS
NMR line shape is consistent with a local D3h symmetry at the
boron center, as expected from the fact that all three ligands are
identical. The DFT calculation of the 11B electric field gradient
in the gas phase yields a CQ value of 4.04 MHz and an η value
of zero, which is in good agreement with the experimental
results obtained in the solid state. The principal EFG
component, Vzz, is aligned perpendicular to the molecular
plane. The simulation that best reproduces the experimental
spectrum indicates that the electric field gradient tensor at the
11B site and the 11B magnetic shielding tensors are coincident.
The four-coordinated boron atom of compound 11 exhibits the
materials 10 and 11. Isotropic chemical shifts, as well as CQ and
η values determined from these data, either by simulating the
central transition line shapes using DMFIT33 and SIMPSON39
codes or by analyzing the relevant F1 cross sections along the
F2 dimensions of 2-D triple-quantum (TQ) MAS NMR
spectra,31,32 are summarized in Table 1. TQMAS experiments
are particularly useful for the analysis of spectra with
overlapping signals. The reference compound data in particular
indicate the large variations in the 11B NMR line shapes that are
in principle possible in these systems. Trispentafluorophenyl-
borane (B(C6F5)3), 12, shows a line shape that is strongly
affected by second-order quadrupolar effects characterized by a
large quadrupolar coupling constant (4.3 MHz) and an
asymmetry parameter close to zero (see Figure 4 and Figure
S2, Supporting Information). As previously noted by Bryce et
a
iso
Table 1. Experimentally and Quantum-Chemically Determined 11B δ , CQ, and η Values
CS
f
iso
CS
iso
δ
(expt, ppm, 0.5)
CQ (expt, MHz, 3%)
η (expt, 0.1)
δ
(calcd, ppm)
CS
CQ (calcd, MHz)
η (calcd)
angle (Vzz, B, P; deg)
b
c
b
b
1
0.3
1.54 /1.55
0.19
−0.6
−7.4
1.58
0.12
22.4
b
b
b
b
b
b
e
e
e
e
2
−7.6 ; −5.9
1.25 ; 1.36
0.18 ; 0.15
1.43
0.05
21.6
b
c
b
b
3
−6.6
−4.7
0.3
1.34 /1.35
0.15
−7.6
−6.6
−1.9
−9.2
−10.2
7.3
1.37
1.36
1.51
1.33
1.49
2.14
0.04
0.05
0.13
0.07
0.53
0.39
23.2
24.2
23.7
22.8
8.5
b
c
b
b
4
1.31 /1.31
0.15
b
c
b
b
b
5
1.55 /1.57
0.17
b
c
b
6
−7.5
−9.1
8.6
1.25 /1.27
0.16
c
c
c
7
1.43
0.55
c
c
c
8
2.10
0.43
20.0
c
c
c
e
e
e
e
9
3.3
1.8
0.6
14.6
2.62
0.26
19.2
c
c
b
b
10
11
12
−7.4
−16.7
58.7
1.59 /1.63
0.15
−8.8
−19.5
54.5
1.60
0.47
0.02
0.4
0.5
c
d
d
0.33
0.3
c
c
c
e
e
e
4.26
0.02
4.04
0.00
a
Chemical shifts were calculated on a DFT B3-LYP/def2-TZVP, electric field gradients on a DFT B97-D/def2-TZVP (modified) level of theory.
b
c
d
Determined from line shape analysis of slices from MQMAS experiments. Determined from line shape analysis of 11B MAS spectra. Determined
e
from 11B SATRAS spectrum (see Figure S3, Supporting Information). Fully geometry-optimized structures are used within the calculations.
f
Orientation of the DFT calculated (B97-D, def2-TZVP (modified)) main principal component of the EFG tensor, Vzz, expressed by the angle
between this parameter and the B···P distance vector.
4239
dx.doi.org/10.1021/ja210160k | J. Am. Chem. Soc. 2012, 134, 4236−4249

Journal of the American Chemical Society
Article
Table 2. Comparison of Crystallographic B···P Distances and Bond Angle Sums around B of the Investigated Compounds with
Theoretical Values from a Full DFT Geometry Optimization on the TPSS Level of Theory (Basis def2-TZVP) Using the
Recently Developed D3 Dispersion Correction
dcryst(B···P), Å
2.115(2)
dcalc(B···P), Å
angle sum B(cryst), deg
angle sum B(calcd), deg
1
2.130
2.046
2.046
2.047
2.131
2.046
2.079
2.174
2.261
2.221
3.234
344.1(2)
a
346
352
351
350
346
351
345
344
352
343
334
2
a
3
2.038(3)
2.038(7)
2.094(2)
2.026(2)
2.060(2)
349.2(2)
348.1(5)
342.1(1)
349.0(2)
342.4(2)
344.2(4)/343.8(4)
a
4
5
6
7
b
8
2.188(5) /2.206(5)
9
a
10
11
2.180(6)
339.9(4)
326.5(4)/329.0(4)
b
3.257(6) /3.232(5)
a
b
No crystal structure available. Used within the DFT calculations.
Figure 5. 11B TQ-MAS spectrum (left) of the phosphane−borane adduct 2 (R = ph, R′ = ph) and corresponding 1D slices along F2 (right)
performed at 11.7 T with a spinning frequency of 14 kHz. The spectrum shows two slightly different resonances with isotropic chemical shifts of
−7.6 ppm and −5.9 ppm, respectively, with a SOQE of 1.24 and 1.31 MHz, respectively. The corresponding experimental F1 slices and their
simulations are shown on the right.
smallest quadrupolar coupling constant within this series (0.33
MHz, η = 0.3), reflecting a boron site with nearly tetrahedral
symmetry. These values were determined by fitting the
intensity distribution within the MAS side band pattern,
which arises from the effect of magic-angle spinning on the
anisotropically broadened first-order quadrupolar satellites
(SATRAS, see Figure S3, Supporting Information). Finally,
the spectrum of the PPh3−B(C6F5)3 adduct 10 is characterized
by an intermediate quadrupolar coupling constant (1.6 MHz)
and an asymmetry parameter close to zero, consistent with the
local C3v symmetry of the central boron atom. The
trimethylene-bridged phosphane−borane adduct 7 shows a
characteristic 11B MAS NMR line shape reflecting a
quadrupolar coupling constant of 1.4 MHz and an asymmetry
parameter of 0.55. In this case, the large asymmetry parameter
can be attributed to the unsymmetric ligation pattern (C6F5
groups and the trimethylene bridge) and the significant
interaction with the phosphorus Lewis base center evident
from the crystal structure.
The remaining spectra shown in Figure 3 reveal the wide
range of chemical shifts and quadrupolar interaction parameters
measured for the different intramolecular adducts investigated
in this study. The phosphane−borane systems 1−6 and the
PPh3−B(C6F5)3 Lewis acid/base adduct 10 give rise to
intermediate quadrupolar coupling strengths (CQ values near
1.5 MHz) thus clearly reflecting the local distortions from the
ideal trigonal local geometry present in 12. The latter
distortions are also readily apparent from the angle sums,
obtained from both the DFT calculations and the crystal
structures (see Table 2). Notably, the distortions appear to
maintain the local C3 axis, as reflected by the calculated and
measured asymmetry parameters, which are found close to
zero. The isotropic chemical shift values range within the
interval 0.3 to −7.6 ppm. The agreement with the theoretically
calculated values is excellent (R2 = 0.998, see Table 1 and
4240
dx.doi.org/10.1021/ja210160k | J. Am. Chem. Soc. 2012, 134, 4236−4249

Journal of the American Chemical Society
Article
Figure S4, Supporting Information). For compound 2, the
TQMAS spectrum clearly shows two slightly different boron
positions (CQ = 1.25 and 1.36 MHz, δCisoS = −7.6 and −5.9 ppm)
in an approximate intensity ratio of 2:1 (see Figure 5 and
Figure S5, Supporting Information). We believe that these
different sites might arise from packing effects or disorder in
the solid state, which may explain why up to now the crystal
structure could not be solved. Our unconstrained DFT
geometry optimizations (TPSS-D3,44,45 def2-TZVP,46 see
Experimental Section) yield a boron−phosphorus distance of
2.05 Å, which is in good agreement with the experimentally
determined distances within the series of intramolecular
adducts (see Table 2). Finally, relatively large CQ values are
measured for the extremely reactive model compounds 8 and 9.
In these cases, the asymmetry parameters of 0.43 and 0.6 reflect
the considerable distortion of the electric field gradient caused
by the nonsymmetric substitution pattern at the three-
coordinated boron site. In the case of compound 9, the
experimental 11B chemical shifts and quadrupolar coupling
parameters show sizable deviations from the calculated values.
This might reflect the more fluxional character of this
compound in the solid state. As a result, a correct calculation
would have to include an averaging process over multiple
configurations, which was not done in the present study.
The interesting question is whether the 11B chemical shifts
and quadrupolar coupling parameters can be related to special
structural characteristics that define the high reactivity of FLPs
and therefore may allow further insights into the reaction
behavior of those molecules. As previous calculations have
shown,19 the electric field generated by the Lewis acid and
Lewis base functionality is responsible for the extremely fast
heterolytic cleavage of dihydrogen. Moreover, these calculations
have revealed that the dissociation of the H2 molecule takes
place nearly barrier-free after the molecule is oriented within
the reactive pocket. Therefore, it is reasonable to assume that
the boron−phosphorus distance is a very important structural
characteristic having a significant influence on the reactivity of
these molecules. Our investigations show that the 11B NMR
parameters for compounds 1−8 are strongly correlated with
this parameter. Figure 6 illustrates the high sensitivity of 11B
isotropic chemical shifts to the B···P distances for the series of
these intramolecular adducts. The general trend that larger
distances result in a high-frequency shift of the 11B resonance is
clearly apparent. The data are compared with chemical shift
calculations carried out for the model compound Ph2P−C2H4−
B(C6F5)2, where the boron−phosphorus distance was changed
from 1.8 to 2.6 Å and full relaxation of all other degrees of
freedom was guaranteed. The closely parallel chemical shift
trends observed for the experimental and theoretical data
clearly confirm these correlations.
Figure 6. Correlation between experimentally determined 11B
chemical shifts in case of the bridged B/P adducts ( ) and in case
■
of DFT calculations (chemical shifts on the BP-86/def-TZVP level of
theory) for a model compound (see inset) with different B···P
distances (★, for technical details see Experimental Section). Larger
B···P distances result in a stronger deshielding of the boron atom. The
straight line represents a linear regression for the experimental values
(R2 = 0.85) and the dashed line for the DFT calculated values using
the model system (R2 = 0.99). Compound 9 is omitted from the
correlation because of its fluxional character and because no
experimental B···P distance is known, and compound 10 is excluded,
because the computational model is inadequate for this molecule,
where both B and P are bound to three aromatic ligands.
■
Figure 7. Correlation between experimentally determined ( , from
line shape analysis of 11B MAS experiments) and DFT-calculated
(B97-D/def2-TZVP (modified), ★) 11B CQ values and boron−
phosphorus distances: larger B···P distances result in larger CQ values
due to less distortion from trigonal geometry at the boron site. The
error bars indicate an experimental uncertainty of about 3%. The
straight line shows a linear regression for the experimental values (R2 =
0.94) and the dashed line a linear regression for the DFT-calculated
values (R2 = 0.91). Compound 9 is omitted from the correlation,
because of its fluxional character (see text) and because no
experimental internuclear distance is available. Compound 10 is
omitted from the correlation, because it does not represent an
intramolecular adduct.
Besides the chemical shifts, especially, the 11B CQ values are
modified by the intramolecular interactions between the Lewis
centers. These effects are summarized in Figure 7 in which the
CQ values are plotted against the experimental B···P distances
for the compounds 1−8. The smallest quadrupolar coupling
constants are observed for the intramolecular adduct 6, while
the largest CQ results in case of compound 8. Figure 7 points
out that the larger the B···P distances are, the larger are the CQ
values indicating lesser degrees of distortion of the boron site
from trigonal geometry. These results suggest that the 11B
quadrupolar coupling constant can be interpreted in terms of
the interaction between the Lewis acid and Lewis base
functionality. Again, theoretical investigations on the model
system Ph2P−C2H4−B(C6F5)2 over the distance range 1.8−2.6
Å confirm these findings: larger quadrupolar coupling constants
result for larger B···P distances due to reduced Lewis acid/
Lewis base interactions (Figure 8b). These calculations also
show that the asymmetry parameter tends to decrease strongly
as a function of B···P distances (Figure 8a).
4241
dx.doi.org/10.1021/ja210160k | J. Am. Chem. Soc. 2012, 134, 4236−4249

Journal of the American Chemical Society
Article
Figure 8. DFT calculations of quadrupolar coupling parameters (B97-D, def2-TZVP) for a model compound (see inset of Figure 6) with varying
boron−phosphorus distances from 1.8 to 2.6 Å (for details, see Experimental Section) illustrating the correlations between the B···P distance and the
asymmetry parameter η (a) and the quadrupolar coupling constant CQ (b). The data highlight the fact that larger B···P distances result in a smaller
distortion of the boron site from trigonal geometry leading to larger CQ and smaller η values. The inset in panel a shows the potential energy curve
with a minimum at 2.06 Å for the TPSS-D3/def2-TZVP level of theory.
A more profound understanding of the experimentally
observed correlations of the EFG with the boron−phosphorus
distance is obtained by visualizing the DFT-calculated EFG
principal axis tensor elements in the molecular frame. In case of
main group elements such as boron, the EFG is dominated by
the outer core and valence shell electronic distribution.21
Therefore an accurate description of both the valence and the
core shell electrons is required. To this end, we used Ahlrich’s
def2-TZVP46 basis set, which is modified at the boron atom by
using tighter basis functions extracted out of Dunning’s
correlation consistent basis set54,55 (for more details, see
Experimental Section and Figure S6, Supporting Information).
Our theoretical studies are based on calculations in the gas
phase, thus neglecting the interactions between the discrete
molecular units in the solid state. As indicated in Figure 9, the
DFT B97-D51 level of theory allows a very accurate
determination of 11B CQ values resulting in a linear correlation
between experimentally and theoretically determined values
(the R2 value is determined as 0.997). The theoretical studies
reveal in case of the substituted vinylene-linked intramolecular
adducts asymmetry parameters ranging between 0.04 and 0.13.
These values are consistent with a local boron geometry that is
still close to trigonal as is expected from the crystallographically
determined bond angle sum for the boron site of about 340−
350°.
The strong distance-dependence of the 11B quadrupolar
coupling constant can be understood by analysis of the
orientation of the largest principal component of the electric
field gradient tensor in the molecular frame. In all cases, the Vzz
component points only slightly away from the B···P vector (for
examples, see Figure S7, Supporting Information, exact values
are given in Table 1) and is therefore strongly influenced by the
interaction among both Lewis centers. The herein developed
combination of experimental and theoretical NMR work allows
an analysis of the electric field described in the theoretical work
by Grimme and co-workers in terms of the electric field
gradient from the viewpoint of the boron site. Since the
calculated principal component of the electric field gradient
points slightly away from the boron−phosphorus axis (see
Table 1), it is reasonable to assume that a completely linear
geometry of the P−H−H−B unit (as, for example, proposed by
Pap
́
ai and co-workers17) is not necessarily an essential
requirement for an efficient activation of H2 within the
FLP.19 These considerations hold also for the adducts 8 and
9, which are extremely sensitive toward a reaction with H2.
31P{1H} CPMAS NMR and DFT Calculations. Figure 10
summarizes the 31P{1H} CPMAS NMR spectra of the
intramolecular adducts of our present study, and Table 3
summarizes the spectroscopic parameters determined from
them. The 31P isotropic chemical shifts fall within a narrow
characteristic chemical shift window between approximately 25
and 5 ppm directly reflecting the extent of the intramolecular
adduct formation. In case of the system 4, two signals are
observed at 24.5 and −8.3 ppm because there are two
crystallographically distinct phosphorus positions: one interact-
ing with the boron Lewis acid site and the other one standing
Figure 9. Correlation between experimentally (from 11B MAS NMR
experiments) and theoretically (DFT, B97-D, def2-TZVP (modified))
determined CQ values with an R2 value of 0.997 and a slope of 0.93.
The deviation from unity is partly attributed to uncertainties in the
experimental values (estimated at 3%, see error bars). Compound 9
is not included in this correlation because of the complications arising
from its fluxional character (see text).
4242
dx.doi.org/10.1021/ja210160k | J. Am. Chem. Soc. 2012, 134, 4236−4249

Journal of the American Chemical Society
Article
In case of the intramolecular adducts 1, 3, 5, and 6, the 31P
spectra resolve a clear multiplet structure. As illustrated in
Figure 11 for compound 1, the peak splittings are nearly field
independent on a hertz scale and therefore suggest the
influence of indirect 31P···10B/11B spin−spin couplings. In
addition, the asymmetric peak separations exhibit the clear
signature of dipolar interactions with a quadrupolar nucleus. It
is well-known that heteronuclear dipole−dipole interactions of
spin-1/2 with nuclei experiencing strong quadrupolar coupling
constants are not completely eliminated under MAS con-
ditions.69 In the context of first-order perturbation theory, the
31P MAS NMR transitions in the presence of this interaction
can be written as:
2
S(S + 1) − 3m
υ
= − m|J| −
d
m
S(2S − 1)
(1)
wherein J are the isotropic 10/11B, 31P indirect spin−spin
coupling constants, S are the nuclear spin quantum numbers for
the quadrupolar nuclei (I = 3/2 and 3 for 11B and 10B,
respectively), m are the orientational quantum numbers, and d
is the residual dipolar coupling.70,71 The latter value is given by:
Figure 10. 31P{1H} CPMAS NMR spectra acquired at 9.4 T with a
spinning frequency of 10 kHz applying the TPPM-15 decoupling
scheme for the following phosphane−borane adducts: (a) 1 (R = mes,
R′ = ph), (b) 2 (R = ph, R′ = ph), (c) 3 (R = ph, R′ = n-propyl), (d) 4
(R = ph, R′ = PPh2), (e) 5 (R = mes, R′ = tolyl), (f) 6 (R = ph, R′ =
me), (g) 7, (h) 8, (i) 9, and (j) 10. All the spectra were acquired at 9.4
T under TPPM-15 proton decoupling. The + sign marks impurities.
ΔJ
3
−3C D −
Q
(
)
2
D
2
D
D
d =
[3 cos β − 1 + η sin β cos 2α ]
20ν
(2)
s
with D describing the B,P direct dipolar coupling constant, ΔJ
the anisotropy of the indirect spin−spin coupling tensor, νs the
10B or 11B Larmor frequency, CQ the 10B or 11B quadrupolar
coupling constant, η the asymmetry parameter, and αD and βD
the Euler angles defining the orientation of the dipolar vector in
the principal axis system of the EFG.72 To the best of our
knowledge, we report the first fully resolved asymmetric
splitting for a 11B,31P spin system, although many other
examples of scalar couplings between spin 1/2 and quadrupolar
nuclei have been published including the 31P/69,71Ga,72
31P/55Mn,73 31P/63,65Cu,74−76 31P/99,101Ru,77 and 31P/95,97Mo78
spin systems. Partially resolved J-multiplets in 31P,11B spin
systems have been previously published.79,80 The line shape
analysis is achieved by using the DMFIT software (see Table 3
and Figure S9, Supporting Information). Our simulations take
into account the natural abundances of 10B and 11B, 19.9% and
apart. As illustrated by Figure S8, Supporting Information, both
phosphorus species can be unambiguously differentiated by
31P{11B} REDOR measurements because the B···P internuclear
distances (2.04 vs 4.25 Å) differ significantly: at the chosen
dipolar evolution time (0.4 ms), only the resonance at 24.5
ppm is attenuated upon 11B dipolar recoupling. The assignment
of this resonance to the adduct-forming moiety is supported by
DFT magnetic shielding calculations on the B3-LYP level of
theory leading to chemical shifts of 23.6 ppm and −10.1 ppm
for both respective resonances.
In the case of the P/B pair 2, two different peaks with slightly
different chemical shifts are observed. Consistent with the
conclusions from 11B MAS NMR, this observation confirms the
assumed disorder phenomenon suspected in the solid state.
Table 3. 31P Isotropic Chemical Shifts, Indirect 10/11B−31P Spin−spin Coupling Parameters, Residual Dipolar Couplings, and
Chemical Shift Anisotropy Parameters (Determined by Line Shape Analysis of Slow-Spinning MAS Experiments) for the
a
Investigated Compounds
J(31P,11B)
5, Hz
J(31P,10B), J(31P,11B), d(31P,11B) 3,
d(31P,10B) 3,
δiso, ppm
|Δσ|, ppm (ησ)
Hz
Hz, calcd
Hz
Hz
δiso, ppm, calcd
|Δσ|, ppm (ησ), calcd
b
c
b
c
1
17.8
∼15.3/10.7
14.5
54.5
52.0
83.8 (0.49)
18.0
51.1
47.1
51.4
52.4
56.1
53.2
45.5
28.7
10.9
49.9
−5.5 (−4.1 )
−11.5 (−8.6 )
24.1
17.0
90.0 (0.40)
60.6 (0.19)
2
b
c
b
c
3
54.0 (0.56)
68.2/87.5 (0.77/0.95)
94.4 (0.56)
17.2
−6.0 (−4.5 )
−12.5 (−9.4 )
16.5
59.7 (0.32)
4
24.5/−8.3
14.4
23.6/−10.1
19.3
71.1/72.9 (0.66/0.89)
104.9 (0.48)
66.7 (0.51)
b
c
b
c
5
52.9
52.0
17.5
17.2
−6.0 (−4.5 )
−12.5 (−9.4 )
b
c
b
c
6
14.7
55.6 (0.65)
−6.0 (−4.5 )
−12.5 (−9.4 )
11.8
7
21.1
56.6 (0.85)
28.8
60.6 (1.00)
8
23.1
78.2 (0.40)
32.0
80.0 (0.37)
9
21.7
69.9 (0.92)
38.0
85.6 (0.61)
10
11
18.6
30.4 (0.83)
28.1
5.4 (0.28)
−130.5
450.3 (0.05)
−161.1
478.2 (0.16)
a
The calculated J coupling constants, Δσ, and η values result from DFT calculations (B3-LYP, TZVP), in the latter case using the convention Δσ =
σ33 − 1/2(σ11 + σ22), ησ = (σ22 − σ11)/(σ33 − σiso) and |σ33 − σiso| > |σ11 − σiso| > |σ22 − σiso|. Calculated chemical shifts (DFT, B3-LYP, def2-TZVP)
are referenced to phosphoric acid. Measured at 7.1 T. Measured at 9.4 T.
b
c
4243
dx.doi.org/10.1021/ja210160k | J. Am. Chem. Soc. 2012, 134, 4236−4249

Journal of the American Chemical Society
Article
Figure 11. (left) 31P{1H} CPMAS spectra of compound 1 (R = mes, R′ = ph) acquired at different field strengths. (top) Spectrum measured at 7.1 T
using the SW-TPPM-15 decoupling scheme (b) and simulated spectrum (a) showing also the contributions due to 31P···11B (dotted line) and
31P···10B (dashed line) indirect spin−spin coupling. (bottom) Spectrum measured at 9.4 T using the TPPM-15 decoupling scheme (b) and simulated
spectrum (a) showing also the contributions due to 31P···11B (dotted line) and 31P···10B (dashed line) indirect dipolar coupling. Simulation
parameters are given in Table 3. (right) 31P{1H} CPMAS spectra acquired with the SW-TPPM-15 decoupling scheme at 7.1 T of compounds 1 (a),
3 (b), 5 (c), and 6 (d) where the J coupling multiplet could be fully or partially resolved.
80.1%, respectively, and the fact that the J and d values for 10B
and 11B must be scaled according to the magnetogyric ratios
(γ(10B)/γ(11B) = 0.33) and nuclear quadrupole moments
(Q(10B)/ Q(11B) = 2.084), respectively.81 Figure 11 shows the
31P{1H} CP-MAS spectra of 1 at two different field strengths
and their corresponding simulations comprising both the
contributions from dipolar interactions with 10B and 11B. The
simulations reveal that the indirect J-coupling between 31P and
11B dominates the line shape splitting and 1J(31P···11B) is
determined to be 54.5 Hz. This relatively large coupling
constant characterizes the covalent interactions between the
Lewis acid and Lewis base through a weak but non-negligible
covalent bond between boron and phosphorus. It is reasonable
Table 4. 31P Full Widths at Half Maximum (fhwm) at Two
Different Field Strengths for Those Molecules for Which the
Spin−Spin Coupling Multiplets Could Not Be Resolved
a
fhwm (7.0 T), Hz
fhwm (9.4 T), Hz
4 (24.5 ppm)
230
118
201
174
904
236
209
119
4 (−8.3 ppm)
7
186
8
152
9
1100
223
10
a
Nearly identical peak widths (except for 9) suggest that spin−spin
couplings exert a dominant influence on the line shapes, whereas the
broad peak in the case of 9 can be attributed to chemical shift
distribution or possibly also dynamic effects. In case of 4, the
remarkable differences in the peak widths directly reflect the influence
1
to assume that this coupling belongs to a J coupling rather
than a 3J coupling via the olefinic backbone based on additional
evidence discussed in more detail below. Similar splittings are
observed for compounds 3, 5, and 6 (see Figure 11), whereas in
some of the adducts measured in this study, the multiplet
splittings are not clearly resolved. We attribute this to degraded
resolution owing to a lower degree of sample crystallinity or
rapid fluctuations of the spins among the various 10B and 11B
Zeeman levels leading to significant relaxation broadening.
Nevertheless, even in those cases, field-dependent line width
analyses suggest that these interactions contribute to the 31P
NMR line shapes observed (see Table 4), except for compound
9, the line shape of which is dominated by chemical shift
distribution effects indicating considerable disorder in the solid
state. Additionally the large line width observed for 9 may
reflect dynamic processes on the NMR spectra, consistent with
its suspected fluxional character in the solid state.
1
3
of J and J 31P···10B/11B spin−spin couplings, respectively, on the
MAS NMR line shapes.
The alternative possibility that the observed peak splitting
3
1
actually arises from J, rather than J, scalar coupling has been
experimentally addressed by analyzing the linewidths of the 31P
resonances of 4. The resonance of the cis-phosphorus site
interacting with the boron Lewis acid site exhibits a much larger
line width than the resonance of the phosphorus site oriented
trans to the boron unit (209 vs 119 Hz at 9.4 T, see Table 4).
This difference is attributed to a much larger 1J coupling
3
constant in comparison to that of a J coupling. Additionally,
we perfomed theoretical calculations of scalar coupling
constants for answering this question satisfactorily. All the J
coupling analysis described within this work has been
4244
dx.doi.org/10.1021/ja210160k | J. Am. Chem. Soc. 2012, 134, 4236−4249

Journal of the American Chemical Society
Article
performed on a hybrid DFT level of theory using the functional
B3-LYP.58,59 For 1, these calculations lead to a 11B···31P J
coupling constant of 51.1 Hz, in very good agreement with the
experimental value (54.5 Hz). Such values are comparable in
inversion pulse trains (S0). Under such conditions, the
normalized difference signal ΔS = (S0 − S)/S0 changes
periodically (like cos ΔΦ) as a function of the dipolar evolution
time, NTr, where
1
magnitude to J (11B···31P) values measured in solution for the
ΔΦ = 4 2 NT D sin β cos β sin α
adduct between PPh3 and BH379,82 (approximately 60 Hz) and
also calculated for the classical Lewis acid/base adduct 10.
Calculations on this model compound under systematic
variation of the B···P distance reveal a strong distance
dependence, as illustrated in Figure S11, Supporting
Information. Finally, the 3J coupling constant of the trans
isomer of 1 (see Figure S12, Supporting Information) is
calculated as 18.3 Hz. Because this value corresponds to the
maximum value based on the Karplus curve83,84 (dihedral angle
of ∼180°), a significantly smaller 3J value would be expected for
the cis-isomer. Based on all of these considerations, it is safe to
conclude that the observed peak splittings arise from 1J and not
(4)
r
In this expression, N is the number of rotor cycles, Tr is the
rotor period, and the angles α and β are Euler angles describing
the orientation of the dipolar vector in the MAS rotor axis
system. For a polycrystalline sample, a powder average must be
taken by appropriate integration over all the Euler angles.
2π
π
ΔS
0
1
4π
= 1 −
dα
sin β cos(ΔΦ) dβ
∫
∫
S
0
0
(5)
Equations 4 and 5 yield the so-called REDOR curve, which is a
plot of ΔS/S0 as a function of the dipolar evolution time, NTr.
For a two-spin system, simulation of this curve yields the
dipolar coupling constant D from which the internuclear
distance rIS can be extracted via eq 3.
3
from J couplings between 31P and the boron nuclides. This
result is in accordance with the computed Wiberg bond-order
indices.67 A bond-order index of about one indicates the
presence of a mostly covalent single bond, while values close to
zero point to ionic or van der Waals interactions. In the case of
the unsaturated intramolecular adducts the Wiberg bond-order
indices range between 0.73 and 0.91 in the CAO basis and 0.78
and 0.82 in the NAO basis (see Table S10, Supporting
Information). While the WBIs in the CAO basis do not
correlate well with the B···P internuclear distances, a much
more consistent trend for them is found in the NAO basis, even
though individual differences are rather small. The covalent
bonding interaction is further probed by a natural bond orbitals
(NBO) analysis,85,86 which reveals a real bond between the
phosphorus and boron moieties (see Table S10, Supporting
Information) in all of the compounds investigated. The
percentage of the NBO on the natural atomic hybrid localized
at B correlates well with the B···P internuclear distances, with
the highest numbers being observed for compounds 3, 4, 6, and
7 (d(B···P) = 2.02−2.06 Å), intermediate values for
compounds 1 and 5 (d(B···P) ≈ 2.09−2.12 Å), and much
lower numbers being observed for compounds 8 and 9 (longest
distances). Only compound 10 presents a somewhat larger
deviation from this trend.
The intramolecular adducts of the present study can be
considered as ideal 31P−11B two-spin systems, for which the
internuclear distance should be easily measurable using the
REDOR experiment. In the present work, we will demonstrate
the calibration of this approach for molecules with known
boron−phosphorus distances, which can then be extended to
adducts with unknown structures.
Figure 12 shows the 11B{31P} REDOR curve for 3. Data have
been corrected for potential limitations in the 31P excitation
Distance Measurements by 11B{31P} REDOR Experi-
ments. Against the background of obtaining deeper insight
into the interactions between the Lewis acid and base
functionalities, it is desirable to determine boron−phosphorus
distances even in those compounds for which no crystal
structure data are available. Solid-state NMR offers an
important opportunity in such cases by conducting rotational
echo double resonance (REDOR) experiments. Such REDOR
experiments are designed to measure the heteronuclear
magnetic dipole coupling constant, D, between two nuclei I
and S
31
Figure 12. 11B{ P} compensated REDOR curve ( , calibration factor
■
●
▲
a = 1.4), REDOR curve ( ), and compensation ( ) for the B/P
adduct 3 (spinning frequency 14 kHz). SIMPSON simulations
assuming a B···P distance of 1.9 Å (orange dashed line), 2.038 Å
(crystallographic value, blue dashed line), 2.06 Å (red straight line),
2.1 Å (green dashed line), and 2.2 Å (brown dashed line) are shown.
Optimum agreement between the experimental and simulated
REDOR curves is achieved by assuming a B···P distance of 2.06 Å.
All simulations include the experimental 31P CSA parameters (Δσ = 54
ppm, ησ = 0.56).
⎛
⎞
μ
8π
0
−3
D =
hγ γ r
IS
I S
⎜
⎝
⎟
⎠
2
(3)
efficiencies, using a compensation scheme developed in our
laboratory,87 applied with a calibration factor of 1.4. As Figure
12 reveals, the data set shows the oscillatory behavior
characteristic of an isolated two-spin system as predicted by
eqs 4 and 5. The experimental REDOR curve is compared with
various simulated curves for different internuclear distances.
These REDOR curves are further influenced by the 31P
under MAS conditions, where γI and γS are the corresponding
gyromagnetic ratios and rIS is the internuclear distance. While
the magnetic dipole−dipole coupling is averaged out over the
MAS rotor cycle, it can be reintroduced by applying additional
inversion pulse trains to one (or both) of the nuclei during the
rotor cycle. This recoupling process diminishes the signal
amplitude S, in relation to that observed in the absence of the
4245
dx.doi.org/10.1021/ja210160k | J. Am. Chem. Soc. 2012, 134, 4236−4249

Journal of the American Chemical Society
Article
chemical shift anisotropy, and the experimentally determined
values (Δσ = 54 ppm, ησ = 0.56) have been included in these
simulations. The oscillating part of the experimental curve is
best reproduced by a simulation based on an internuclear
distance of 2.06 Å, which differs slightly from the crystallo-
graphic value of 2.038 Å. Analogous measurements on the
other adducts of the present study show that these under-
estimations of B···P distances are rather general. The REDOR
data tend to underestimate the 31P−11B dipolar coupling
strengths systematically, leading to an overestimation of the
internuclear distances. Despite this limitation, it will be possible
to use REDOR data for obtaining boron−phosphorus
internuclear distances for intramolecular adducts with unknown
crystal structures or in amorphous systems by calibration
against reference data.
As a matter of fact, the experimental and simulated REDOR
curves based on the crystallographic B···P distance can be made
to agree with each other by including a contribution arising
from the J coupling anisotropy (ΔJ) in the simulations (see
Figure 13a,b for two representative adducts). As expressed by
eq 2, the combination of the direct dipolar coupling and the J
anisotropy leads to a reduced effective dipole−dipole coupling,
which ultimately determines the frequency of the dipolar
oscillations. The determined ΔJ values for the intramolecular
adducts range between 60 and 150 Hz; for the adduct 10 a
slightly larger value of 200 Hz is observed (see Table 5).
Additionally, we performed 11B{31P} constant-time REDOR
(CT-REDOR) experiments in which the evolution time (i.e.,
the number of rotor periods) is held constant and the positions
of the dephasing π-pulses are stepped through the rotation
period.38 Again, these studies were conducted with a
compensation scheme. Figure 14 shows compensated CT-
REDOR curves for 3 with three different evolution times
probing characteristic parts of the REDOR curve. The results
support the conclusion drawn for the standard REDOR
experiments. The boron−phosphorus distances extracted
from the oscillatory part of the CT-REDOR curve tend to be
overestimated, and the experimental CT-REDOR curves can be
correctly reproduced based on the crystallographic B···P
distance if non-zero anisotropies of the J coupling tensor
(with ΔJ values ranging between 60 and 219 Hz) are included
in the simulations (see Table 5). We emphasize that especially
in systems with long 11B relaxation times and low boron
contents (by mass), CT-REDOR experiments lead to
significant savings in experimental time compared with
conventional REDOR measurements.
Figure 13. (a) 11B{31P} compensated REDOR curve for the B/P
adduct 3 (spinning frequency 14 kHz). The SIMPSON simulation
assuming a B···P distance of 2.038 Å (crystallographic value, straight
line) is shown. Perfect agreement between the experimental and
simulated REDOR curve based on the crystallographic B···P distance
can only be achieved if a contribution arising from the J coupling
anisotropy (ΔJ) is included in the simulations. The simulation includes
the experimental 31P CSA parameters (Δσ = 54 ppm, ησ = 0.56) and
an assumed J coupling anisotropy of 60 Hz (coincident dipolar and J
coupling tensors are assumed). (b) 11B{31P} compensated REDOR
curve ( , calibration factor a = 1.3) and REDOR curve ( ) for the B/
P adduct 4 (spinning frequency 13 kHz). The SIMPSON simulations
assume a three-spin system with B···P distances of 2.038 Å (P1) and
4.25 Å (P2) and include the 31P CSA parameters (P1 Δσ = 68 ppm, ησ
= 0.77 and P2 Δσ = 88 ppm, ησ = 0.95) and an assumed J coupling
anisotropy of 60 Hz.
■
▲
Finally, for all of the adducts investigated in the present
study, we have noticed a subtle distortion of the experimental
REDOR and CT-REDOR curves in the sense that the extent of
the initial dephasing in the limit of small dipolar evolution
times falls slightly below the extent of dephasing predicted from
the oscillatory part (see Figure 13). That is, both parts of the
REDOR curve would actually result in slightly different values
of effective dipolar coupling constants. We speculate that these
distortions result mainly from heteronuclear interactions with
the homonuclearly coupled proton spin reservoir88 or other
systematic errors as already described in the literature.89−91 The
origin of these deviations can be explored by conducting these
Table 5. B···P Distances from Crystallography and “Best-Fit”
Values Obtained from the Oscillatory Part of the 11B{31P}
REDOR Curves, Neglecting the Influence of Anisotropic J-
Coupling Interactions (ΔJ = 0) and “Best-Fit” Values of ΔJ
a
dREDOR(B···P)
(dCT‑REDOR(B···P)), Å
ΔJREDOR
dcryst(B···P), Å
(ΔJCT‑REDOR), Hz
1
2.115(2)
2.038(3)
2.038(7)
2.026(2)
2.180(6)
2.18 (2.14)
2.06 (2.13)
2.06
150 (60)
60 (219)
60
3
4
6
2.06 (2.06)
2.29 (2.29)
80 (80)
200 (200)
10
a
1
Numbers given in parentheses are best-fit values from CT-REDOR
REDOR experiments with H multipulse decoupling or on
data.
perdeuterated compounds, to be examined in future work.
The non-zero contributions of the anisotropy of the J
coupling tensor suggested by our REDOR experiments lend
further support to the presence of covalent interactions
between the acid and base centers in the intramolecular
adducts investigated.
4246
dx.doi.org/10.1021/ja210160k | J. Am. Chem. Soc. 2012, 134, 4236−4249

Journal of the American Chemical Society
Article
31
Figure 14. 11B{ P} compensated CT-REDOR curve ( ) and CT-REDOR curve ( ) for the B/P adduct 3 with evolution times of (a) 0.2, (b) 0.8,
and (c) 1.6 ms for representing the initial region of a REDOR curve (a) and the oscillatory region (c). Perfect agreement between the experimental
and simulated CT-REDOR curves based on the crystallographic B···P distance can only be achieved if a contribution arising from the J coupling
anisotropy (ΔJ) is included in the simulations. The straight line in panels a and b belongs to SIMPSON simulations assuming a B···P distance of
2.038 Å and using the experimental 31P CSA parameters (Δσ = 54 ppm, ησ = 0.56) and a J coupling anisotropy of 334 Hz (coincidence of the dipolar
and J coupling tensors is assumed). In panel c, the best agreement between simulation and experiment is obtained by assuming a B···P distance of
2.038 Å, the experimentally determined 31P CSA parameters, and a J coupling anisotropy of 219 Hz.
■
▲
of solid-state NMR techniques characterizing the weak bonding
interactions within boron/phosphorus-based FLPs.
CONCLUSIONS
■
Solid-state NMR techniques are established for the structural
characterization of frustrated Lewis pairs (FLPs), especially for
investigating the interactions between the Lewis acid and base
functionalities. 11B chemical shifts and quadrupolar coupling
parameters can be interpreted in terms of the strength of the
interaction between the Lewis acid and base functionality: the
stronger this interaction, the lower are the 11B isotropic
chemical shifts and the smaller is the 11B quadrupolar coupling
constant. The latter result is supported by DFT calculations of
electric field gradients on a GGA level for single molecules in
the gas phase, which are in excellent agreement with
experimentally obtained values. These DFT calculations reveal
that the principal electric field gradient component, Vzz, is tilted
about 20° away from the boron−phosphorus director. These
results explain the strong B···P distance dependence of the
quadrupolar coupling constant. They may also relate to the fact
that it is not necessary to assume a linear orientation of H2 in
the reactive pocket of the FLP for the understanding of the
mechanism of the dihydrogen activation.
In summary, we conclude that “frustration” in FLP chemistry
does not mean the complete suppression of covalent
interactions between the Lewis acid and base centers. As a
matter of fact, the residual electron density between the
reaction centers may be a necessary requirement for the typical
FLP behavior. To explore whether and how the communication
between these functionalities may explain the remarkable
cooperative reaction behavior, ongoing NMR and DFT studies
are directed toward characterizing the intermediate steps of the
H2 activation process.
ASSOCIATED CONTENT
■
S
* Supporting Information
Complete ref 43, synthesis and general characterization of
compounds 1−12, and additional solid-state NMR and DFT
details. This material is available free of charge via the Internet
at http://pubs.acs.org.
31P{1H} CPMAS NMR spectra are influenced by indirect
31P−11B spin−spin coupling, manifesting itself by asymmetric
peak splittings independent of magnetic field strength. For the
substituted vinylene-bridged intramolecular adducts, the
experimental 1J(11B...31P) spin−spin coupling constants ex-
tracted from these spectra are on the order of 50 Hz, in close
agreement with DFT calculations. These results suggest that a
significant covalent interaction is still apparent in intra-
molecular phosphane−borane adducts. This is also demon-
strated by 11B{31P} REDOR and CT-REDOR experiments,
which suggest the influence of non-zero contributions of the J
coupling tensor anisotropy leading to a slightly reduced
effective dipolar coupling constant. Analysis of REDOR curves
assuming ΔJ = 0 leads to a systematic overestimation of the
boron−phosphorus distance (by up to 0.1 Å in some cases). If J
anisotropies on the order of 100 Hz are taken into
consideration, however, the REDOR curves are well repro-
duced based on the crystallographic boron−phosphorus
distances. With these considerations in mind, the method
lends itself to distance measurements in intramolecular adducts
with unknown crystal structures or amorphous materials.
Overall, the present study illustrates the power and potential
AUTHOR INFORMATION
■
Corresponding Author
eckerth@uni-muenster.de; erker@uni-muenster.de; grimme@
thch.uni-bonn.de
ACKNOWLEDGMENTS
■
This work was supported by the SFB 858 “Cooperative
Systems in Chemistry”. T.W. thanks the Fonds der Chemischen
Industrie for a research fellowship. He acknowledges additional
support by the NRW Forschungsschule “Molecules and
Materials” and also thanks Dipl.-Chem. Birgitta Schirmer and
Dr. Thomas Echelmeyer for helpful discussions. We cordially
thank Dr. M. Sajid for providing us with a sample of compound
8.
REFERENCES
■
(1) Stephan, D. W.; Erker, G. Angew. Chem., Int. Ed. 2010, 49, 46.
(2) Lewis, G. N. Valence and the Structure of Atoms and Molecules;
Chemical Catalogue Company: New York, 1923.
(3) Stephan, D. W. Dalton Trans. 2009, 3129.
(4) Welch, G. C.; Stephan, D. W. J. Am. Chem. Soc. 2007, 129, 1880.
4247
dx.doi.org/10.1021/ja210160k | J. Am. Chem. Soc. 2012, 134, 4236−4249

Journal of the American Chemical Society
Article
(5) Welch, G. C.; Juan, R. R. S.; Masuda, J. D.; Stephan, D. W. Science
2006, 314, 1124.
(40) Bak, M.; Nielsen, N. C. J. Magn. Reson. 1997, 125, 132.
(41) Ahlrichs, R.; Furche, F.; Hattig, C. TURBOMOLE, version 6.0/
̈
(6) Spies, P.; Erker, G.; Kehr, G.; Bergander, K.; Frohlich, R.;
6.3, Universitat Karlsruhe, 2009.
̈
̈
(42) Ahlrichs, R.; Bar, M.; Haser, M.; Horn, H.; Kolmel, C. Chem.
Grimme, S.; Stephan, D. W. Chem. Commun. 2007, 5072.
̈
̈
̈
(7) Spies, P.; Kehr, G.; Bergander, K.; Wibbeling, B.; Frohlich, R.;
Phys. Lett. 1989, 162, 165.
̈
Erker, G. Dalton Trans. 2009, 1534.
(43) Frisch, M. J. et al. Gaussian 09; Gaussian, Inc., Wallingford, CT,
(8) Momming, C. M.; Otten, E.; Kehr, G.; Frohlich, R.; Grimme, S.;
2009.
̈
̈
Stephan, D. W.; Erker, G. Angew. Chem., Int. Ed. 2009, 48, 6643.
(9) Moebs-Sanchez, S.; Bouhadir, G.; Saffon, N.; Maron, L.;
Bourissou, D. Chem. Commun. 2008, 3435.
(44) Tao, J.; Perdew, J. P.; Staroverov, V. N.; Scuseria, G. E. Phys.
Rev. Lett. 2003, 91, No. 146401.
(45) Grimme, S.; Antony, J.; Ehrlich, S.; Krieg, H. J. Chem. Phys.
(10) Momming, C. M.; Kehr, G.; Wibbeling, B.; Frohlich, R.; Erker,
G. Dalton Trans. 2010, 39, 7556.
(11) McCahill, J. S. J.; Welch, G. C.; Stephan, D. W. Angew. Chem.,
Int. Ed. 2007, 46, 4968.
̈
̈
2010, 132, No. 154104.
(46) Weigend, F.; Ahlrichs, R. Phys. Chem. Chem. Phys. 2005, 7, 3297.
(47) Treutler, O.; Ahlrichs, R. J. Chem. Phys. 1995, 102, 346.
(48) Eichkorn, K.; Treutler, O.; Ohm, H.; Haser, M.; Ahlrichs, R.
Chem. Phys. Lett. 1995, 240, 283.
̈
̈
(12) Momming, C. M.; Fromel, S.; Kehr, G.; Frohlich, R.; Grimme,
̈
̈
̈
S.; Erker, G. J. Am. Chem. Soc. 2009, 131, 12280.
(49) Eichkorn, K.; Weigend, F.; Treutler, O.; Ahlrichs, R. Theor.
(13) Ullrich, M.; Seto, K. S. H.; Lough, A. J.; Stephan, D. W. Chem.
Chim. Acta 1997, 97, 119.
Commun. 2009, 2335.
(14) Dureen, M. A.; Stephan, D. W. J. Am. Chem. Soc. 2009, 131,
8396.
(50) Kaupp, M.; Buhl, M.; Malkin, V. G. Calculation of NMR and EPR
̈
parameters; Wiley-VCH: Weinheim, Germany, 2004.
(51) Grimme, S. J. Comput. Chem. 2006, 27, 1787.
(52) Feller, D. J. Comput. Chem. 1996, 17, 1571.
(53) Schuchardt, K. L.; Didier, B. T.; Elsethagen, T.; Sun, L.;
Gurumoorthi, V.; Chase, J.; Li, J.; Windus, T. L. J. Chem. Inf. Model.
2007, 47, 1045.
(54) Woon, D. E.; Dunning, T. H. J. Chem. Phys. 1995, 103, 4572.
(55) Peterson, K. A.; Dunning, T. H. J. Chem. Phys. 2002, 117, 10548.
(56) Pennington, W. T. J. Appl. Crystallogr. 1999, 32, 1028.
(57) Ditchfield, R. Mol. Phys. 1974, 27, 789.
(15) Cardenas, A. J. P.; Culotta, B. J.; Warren, T. H.; Grimme, S.;
Stute, A.; Frohlich, R.; Kehr, G.; Erker, G. Angew. Chem., Int. Ed. 2011,
̈
50, 7567.
(16) Ekkert, O.; Kehr, G.; Frohlich, R.; Erker, G. J. Am. Chem. Soc.
2011, 133, 4610.
(17) Rokob, T. A.; Hamza, A.; Stirling, A.; Soos
Chem., Int. Ed. 2008, 47, 2435.
(18) Hamza, A.; Stirling, A.; Andras Rokob, T.; Papai, I. Int. J.
Quantum Chem. 2009, 109, 2416.
(19) Grimme, S.; Kruse, H.; Goerigk, L.; Erker, G. Angew. Chem., Int.
Ed. 2010, 49, 1402.
̈
́
, T.; Pap
́
ai, I. Angew.
́
́
(58) Becke, A. D. J. Chem. Phys. 1993, 98, 5648.
(59) Stephens, P. J.; Devlin, F. J.; Chabalowski, C. F.; Frisch, M. J. J.
Phys. Chem. 1994, 98, 11623.
(60) Becke, A. D. Phys. Rev. A 1988, 38, 3098.
(61) Perdew, J. P. Phys. Rev. B 1986, 34, 7406.
(62) Antol, I.; Glasovac, Z.; Eckert-Maksic, M. New J. Chem. 2004,
28, 880.
(63) Onak, T.; Landesman, H.; Williams, R.; Shapiro, I. J. Phys. Chem.
1959, 63, 1533.
(20) Bryce, D. L.; Wasylishen, R. E.; Gee, M. J. Phys. Chem. A 2001,
105, 3633.
(21) Autschbach, J.; Zheng, S.; Schurko, R. W. Concepts Magn. Reson.,
Part A 2010, 36A, 84.
(22) Adiga, S.; Aebi, D.; Bryce, D. L. Can. J. Chem. 2007, 85, 496.
(23) Gullion, T.; Schaefer, J. J. Magn. Reson. 1989, 81, 196.
(64) Buehl, M.; Schleyer, P. v. R. J. Am. Chem. Soc. 1992, 114, 477.
(65) Onak, T.; Diaz, M.; Barfield, M. J. Am. Chem. Soc. 1995, 117,
1403.
(24) Spies, P.; Frohlich, R.; Kehr, G.; Erker, G.; Grimme, S. Chem.
̈
Eur. J. 2008, 14, 333.
(25) Axenov, K. V.; Momming, C. M.; Kehr, G.; Frohlich, R.; Erker,
̈
̈
(66) Schafer, A.; Huber, C.; Ahlrichs, R. J. Chem. Phys. 1994, 100,
̈
G. Chem.Eur. J. 2010, 16, 14069.
5829.
(26) Jacobsen, H.; Berke, H.; Doring, S.; Kehr, G.; Erker, G.;
̈
(67) Wiberg, K. B. Tetrahedron 1968, 24, 1083.
(68) Glendering, E. D.; Reed, A. E.; Carpenter, J. E.; Weinhold, F.
NBO, version 3.1.
(69) Massiot, D.; Fayon, F.; Deschamps, M.; Cadars, S.; Florian, P.;
Montouillout, V.; Pellerin, N.; Hiet, J.; Rakhmatullin, A.; Bessada, C.
C. R. Chim. 2010, 13, 117.
Frohlich, R.; Meyer, O. Organometallics 1999, 18, 1724.
̈
(27) Massey, A. G.; Park, A. J. J. Organomet. Chem. 1964, 2, 245.
(28) Ekkert, O.; Kehr, G.; Frohlich, R.; Erker, G. Chem. Commun.
̈
2011, 47, 10482.
(29) Massey, A. G.; Park, A. J.; Stone, F. G. A. Proc. Chem. Soc. 1963,
212.
(70) Olivieri, A. J. Am. Chem. Soc. 1992, 114, 5758.
(71) Harris, R. K.; Olivieri, A. C. Prog. Nucl. Magn. Reson. Spectrosc.
1992, 24, 435.
(72) Moran, K. L.; Gier, T. E.; Harrison, W. T. A.; Stucky, G. D.;
Eckert, H.; Eichele, K.; Wasylishen, R. E. J. Am. Chem. Soc. 1993, 115,
10553.
(30) Bennett, A. E.; Rienstra, C. M.; Auger, M.; Lakshmi, K. V.;
Griffin, R. G. J. Chem. Phys. 1995, 103, 6951.
(31) Medek, A.; Harwood, J. S.; Frydman, L. J. Am. Chem. Soc. 1995,
117, 12779.
(32) Amoureux, J.-P.; Fernandez, C.; Steuernagel, S. J. Magn. Reson.,
Ser. A 1996, 123, 116.
(33) Massiot, D.; Fayon, F.; Capron, M.; King, I.; Le Calve,
Alonso, B.; Durand, J.-O.; Bujoli, B.; Gan, Z.; Hoatson, G. Magn.
Reson. Chem. 2002, 40, 70.
(34) Peersen, O. B.; Wu, X. L.; Kustanovich, I.; Smith, S. O. J. Magn.
Reson., Ser. A 1993, 104, 334.
́
S.;
(73) Lindner, E.; Fawzi, R.; Mayer, H. A.; Eichele, K.; Pohmer, K.
Inorg. Chem. 1991, 30, 1102.
(74) Brunklaus, G.; Chan, J. C. C.; Eckert, H.; Reiser, S.; Nilges, T.;
Pfitzner, A. Phys. Chem. Chem. Phys. 2003, 5, 3768.
(75) Scheer, M.; Gregoriades, L.; Bai, J.; Sierka, M.; Brunklaus, G.;
Eckert, H. Chem.Eur. J. 2005, 11, 2163.
(35) Thakur, R. S.; Kurur, N. D.; Madhu, P. K. Chem. Phys. Lett.
2006, 426, 459.
(76) Tang, J. A.; Ellis, B. D.; Warren, T. H.; Hanna, J. V.; Macdonald,
C. L. B.; Schurko, R. W. J. Am. Chem. Soc. 2007, 129, 13049.
(77) Eichele, K.; Wasylishen, R. E.; Corrigan, J. F.; Doherty, S.;
Carty, A. J.; Sun, Y. Inorg. Chem. 1993, 32, 121.
(78) Eichele, K.; Wasylishen, R. E.; Maitra, K.; Nelson, J. H.; Britten,
J. F. Inorg. Chem. 1997, 36, 3539.
(36) Vinod Chandran, C.; Madhu, P. K.; Kurur, N. D.; Brauniger, T.
̈
Magn. Reson. Chem. 2008, 46, 943.
(37) Gullion, T.; Baker, D. B.; Conradi, M. S. J. Magn. Reson. 1990,
89, 479.
(38) Echelmeyer, T.; van Wullen, L.; Wegner, S. Solid State Nucl.
̈
Magn. Reson. 2008, 34, 14.
(79) Power, W. P. J. Am. Chem. Soc. 1995, 117, 1800.
(80) Ashbrook, S. E.; Dowell, N. G.; Prokes, I.; Wimperis, S. J. Am.
Chem. Soc. 2006, 128, 6782.
(39) Bak, M.; Rasmussen, J. T.; Nielsen, N. C. J. Magn. Reson. 2000,
147, 296.
4248
dx.doi.org/10.1021/ja210160k | J. Am. Chem. Soc. 2012, 134, 4236−4249

Journal of the American Chemical Society
Article
(81) Harris, R. K.; Becker, E. D.; Cabral de Menezes, S. M.;
Goodfellow, R.; Granger, P. Pure Appl. Chem. 2001, 73, 1795.
(82) Huffman, J. C.; Skupinski, W. A.; Caulton, K. G. Cryst. Struct.
Commun. 1982, 11, 1435.
(83) Karplus, M. J. Am. Chem. Soc. 1963, 85, 2870.
(84) Karplus, M. J. Chem. Phys. 1959, 30, 11.
(85) Foster, J. P.; Weinhold, F. J. Am. Chem. Soc. 1980, 102, 7211.
(86) Reed, A. E.; Weinhold, F. J. Chem. Phys. 1983, 78, 4066.
(87) Chan, J. C. C.; Eckert, H. J. Magn. Reson. 2000, 147, 170.
(88) Mitchell, D. J.; Evans, J. N. S. Chem. Phys. Lett. 1998, 292, 656.
(89) Elbers, S.; Strojek, W.; Koudelka, L.; Eckert, H. Solid State Nucl.
Magn. Reson. 2005, 27, 65.
(90) Strojek, W.; Fehse, C. M.; Eckert, H.; Ewald, B.; Kniep, R. Solid
State Nucl. Magn. Reson. 2007, 32, 89.
(91) Raskar, D. B.; Eckert, H.; Ewald, B.; Kniep, R. Solid State Nucl.
Magn. Reson. 2008, 34, 20.
4249
dx.doi.org/10.1021/ja210160k | J. Am. Chem. Soc. 2012, 134, 4236−4249