Chlorobis(pentafluoroethyl)phosphane: Improved synthesis and molecular structure in the gas phase

Article abstract of DOI:10.1002/chem.201003048

(C₂F₅)₂PCl is now accessible through a significantly improved synthesis protocol starting from the technical product (C₂F₅)₃PF₂. (C₂F ₅)₃PF₂ was reduced in the first step with NaBH₄ in a solvent-free reaction at 120 °C. The product, P(C ₂F₅)₃, was treated with an excess of an aqueous sodium hydroxide solution to afford the corresponding phosphinite salt Na ⁺(C₂F₅)₂PO^- selectively under liberation of pentafluoroethane. Subsequent chlorination with PhPCl ₄ resulted in the selective formation of (C₂F ₅)₂PCl, which was isolated by fractional condensation in an overall yield of 66 %. The gas electron diffraction (GED) pattern for (C ₂F₅)₂PCl was recorded and found to be described by a two-conformer model. A quantum chemical investigation of the potential-energy surface revealed the possible existence of many low-energy conformers, each with a number of low-frequency vibrational modes and therefore large-amplitude motions. The conformer calculated to be most stable was also found to be most abundant by GED and comprised 61(5) % of the total. The molecular structure parameters determined by GED were in good agreement with those calculated at the MP2/TZVPP level of theory; the only significant difference was a discrepancy of about 3° in the C-P-C angle, which, for the lowest-energy conformer, was refined to 98.2(4)° and was calculated to be 94.9°.

Full text of DOI:10.1002/chem.201003048

DOI: 10.1002/chem.201003048
Chlorobis(pentafluoroethyl)phosphane: Improved Synthesis and Molecular
Structure in the Gas Phase
[
a]
[a]
[a]
[b]
Stuart A. Hayes, Raphael J. F. Berger, Norbert W. Mitzel,* Julia Bader, and
[
b]
Berthold Hoge*
Abstract:
A
C
H
T
U
N
G
T
R
E
N
N
U
N
G
(C F ) PCl is now accessible
yield of 66%. The gas electron diffrac-
tion (GED) pattern for (C F ) PCl was
fore large-amplitude motions. The con-
former calculated to be most stable
was also found to be most abundant by
GED and comprised 61(5)% of the
total. The molecular structure parame-
ters determined by GED were in good
agreement with those calculated at the
MP2/TZVPP level of theory; the only
significant difference was a discrepancy
of about 38 in the C-P-C angle, which,
for the lowest-energy conformer, was
refined to 98.2(4)8 and was calculated
to be 94.98.
2
5 2
through a significantly improved syn-
thesis protocol starting from the techni-
cal product (C F ) PF . (C F ) PF was
reduced in the first step with NaBH in
a solvent-free reaction at 1208C. The
product, P ACHTUNGTRENNUNG( C F ) , was treated with an
2 5 3
excess of an aqueous sodium hydroxide
solution to afford the corresponding
phosphinite salt Na ACHTNUGTRNEUNG( C F ) PO selec-
2 5 2
tively under liberation of pentafluoro-
ethane. Subsequent chlorination with
2
5 2
recorded and found to be described by
a two-conformer model. A quantum
chemical investigation of the potential-
energy surface revealed the possible
existence of many low-energy conform-
ers, each with a number of low-fre-
quency vibrational modes and there-
2
5
3
2
2
5
3
2
4
+
À
Keywords: density functional calcu-
lations · gas electron diffraction ·
molecular structures · fluorine ·
phosphorus
PhPCl resulted in the selective forma-
4
tion of (C F ) PCl, which was isolated
by fractional condensation in an overall
2
5 2
Introduction
result in stable and tunable chelating ligands. (CF ) PC H P-
3 2 2 4
AHCTUNGTNERNUG( CF ) was the first perfluoroalkylated bidentate phosphane
3 2
[1]
Bidentate ethylene bridged phosphane ligands are a versa-
tile class of chelating ligands for transition metals. Variation
of the substituents at the phosphorus atom allows a facile
modification of the steric and electronic properties of the
metal atom in complexes derived thereof. Although a wide
range of p-donor ligands exist, the number of electronically
tunable p-acceptor ligands is small. Common examples such
ligand. The synthesis reported by Burg and Street in 1963
proceeds by the reaction of tetrakis(trifluoromethyl)diphos-
phane with ethylene. Field and Wilkinson have developed
[2]
another approach based on the method reported by Rup-
[3]
pert and Volbach: Cl PC H PCl was treated with CF Br
2
2
4
2
3
in the presence of P ACHTUNGTRNEU(GN NEt ) . The yield of the unoptimized
2 3
reaction of 15% is very low, and further optimization has
not been reported so far.
as CO, NO, and PF lack the possibility of modification. Dis-
3
advantageously, the PF bond is quite susceptible towards
cleavage (e.g., by hydrolysis); this limits possible applica-
tions.
Therefore, the introduction of electron-withdrawing, yet
chemically relatively inert, perfluoroalkyl groups should
In 1989, Roddick and Ernst introduced the sterically more
[4]
demanding C F group into a phosphane derivative. The
2
5
reaction of Cl PC H PCl with C F Li at À908C resulted in
2
2
4
2
2 5
the selective formation of (C F ) PC H P ACHTUNGTRENNGNU( C F ) , which was
2 5 2 2 4 2 5 2
isolated in a 68% yield. Contrary to CF Li, which eliminates
3
[5]
LiF at temperatures below À1008C, C F Li can be handled
2
5
for up to 10 min, even at À788C. In complexes with chromi-
um, molybdenum, and tungsten, (C F ) PC H P(C F ) ex-
[
a] Dr. S. A. Hayes, Dr. R. J. F. Berger, Prof. Dr. N. W. Mitzel
Fakultꢀt fꢁr Chemie
ACHTUNGTRENNUNG
4
2
5
2
2
2
5 2
[6]
hibits strong p-acceptor behavior.
Although C F Li represents a very useful reagent for the
Lehrstuhl fꢁr Anorganische Chemie und Strukturchemie
Universitꢀt Bielefeld, Universitꢀtsstraße 25
2
5
alkylation of phosphorus halides, Roddick warned some
3
3615 Bielefeld (Germany)
Fax : (+49)521-106-6026
E-mail: mitzel@uni-bielefeld.de
years later not to warm a solution of C F Cl/nBuLi in dieth-
2
5
[7]
yl ether beyond À908C: after the addition of just a few
drops of the phosphorus compound, a violent, explosive de-
composition may occur. A safe and more efficient route to
[
b] Dr. J. Bader, Prof. Dr. B. Hoge
Fakultꢀt fꢁr Chemie, Anorganische Chemie II
Universitꢀt Bielefeld, Universitꢀtsstraße 25
(
CF ) PC H P
A
H
U
G
R
N
U
G
A
H
U
G
R
N
U
G
3
2
2
4
3
2
2
5
2
2
4
2
5 2
3
3615 Bielefeld (Germany)
quently
been
found
in
the
transformation
of
E-mail: b.hoge@uni-bielefeld.de
(
PhO) PCH CH P
A
H
U
G
R
N
U
(OPh) with Me SiCF and Me SiC F , re-
Supporting information for this article is available on the WWW
under http://dx.doi.org/10.1002/chem.201003048.
2
2
2
2
3
3
3
2 5
[8]
spectively.
3968
ꢂ 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2011, 17, 3968 – 3976

FULL PAPER
For efficient syntheses of chiral bidentate bis(pentafluoro-
ACHTUNGTRENNUNGe thyl)phosphane derivatives, the halides (C F ) PX (X=
2
5 2
Hal) represent well-suited starting materials. Kꢁndig and co-
workers have established chiral bidentate bis(pentafluoro-
phenyl)phosphinite ligands for the synthesis of chiral, Lewis
acidic transition-metal complexes for various applications in
[9]
asymmetric catalysis. The corresponding chiral ligand I has
been prepared in an almost quantitative yield through the
reaction of (R,R)- or (S,S)-hydrobenzoin with bromobis-
(
pentafluorophenyl)phosphane in the presence of triethyl-
[10]
ACHTUNGTRENNUNGa mine (Scheme 1).
2 5 2
Scheme 2. Reaction pathways for the synthesis of (C F ) PCl.
steps. In this protocol (C F ) PF was reduced with NaBH
4
2
5
3
2
in a solvent-free reaction at 1208C. The products, (C F ) P
2
5 3
Scheme 1.
and (C F ) PH, were separated by isothermal distillation
2
5 2
and (C F ) P was isolated in an overall yield of 83%.
2
5 3
Whereas the alkaline hydrolysis of triarylphosphane de-
rivatives with electron-withdrawing pentafluorophenyl or
Mono- and bis(pentafluoroethyl)phosphorus halides are
accessible through the reaction of C F I with white phospho-
tetrafluoropyridyl groups resulted in the formation of the
2
5
[
11]
[18]
rus, but the yields are too low to justify the drastic reac-
tion conditions in the search for an efficient preparative
access. The first synthesis of (C F ) PCl originates from 1972
corresponding phosphinous acid,
(C F ) P reacted in di-
2 5 3
ethyl ether with an excess aqueous solution of sodium hy-
droxide selectively under liberation of pentafluoroethane to
2
5 2
[12]
+
À
when Ang et al. successfully reacted (C F ) PI with AgCl.
give the corresponding phosphinite salt Na ACHTUNTGNERNUG( C F ) PO .
2 5 2
2
5
2
After 15 days, they obtained the product in 93% yield.
This phosphinite salt is very sensitive towards air and water.
It reacts with atmospheric oxygen to give bis(pentafluoro-
Herein, we present 1) an efficient synthesis and 2) the ex-
perimental gas-phase structure of (C F ) PCl, which can
+
À
A
H
U
G
R
N
U
G
ACHTUNGRTNNGE(U C F ) P(O)O , and hydrolyzes
2 5 2
2
5
2
serve as a model to describe the conformational behavior
and steric requirements of (C F ) P groups in larger mole-
+
À
thylphosphinate, Na C F (H)P(O)O . The freshly prepared
2
5
2
2
5
cules.
phosphinite salt was treated with PCl to yield the desired
3
(
C F ) PCl, but the separation of the product from an excess
2 5 2
of PCl was not successful by either fractional condensation
3
Results and Discussion
or by isothermal distillation. By making use of PhPCl as
4
the chlorinating reagent, this difficulty could be overcome.
+
À
Synthesis of (C F ) PCl: We developed a significantly im-
Treatment of the in situ generated Na ACHTUNGTERNNUG( C F ) PO with
2 5 2
2
5
2
proved synthesis of (C F ) PCl starting from the phosphor-
PhPCl in 1,6-dibromohexane selectively resulted in the for-
2
5
2
4
ane (C F ) PF . (C F ) PF is produced on a technical scale
mation of (C F ) PCl, which was isolated by fractional con-
2
5
3
2
2
5
3
2
2
5 2
by the electrochemical fluorination of triethylphosphane in
anhydrous HF and was generously provided by Merck
densation in an overall yield of 66%.
[13]
(
Darmstadt, Germany).
(C F ) PF adds aqueous HF to
First-principles investigations: Calculations were performed,
primarily, to supplement the GED investigation by way of
providing approximate amplitudes of vibration for each pair
2
5
3
2
À
+
[14]
form the strong acid H ACHTUNGTRENNNU[G (C F ) PF ] ·nH O, which hydro-
2 5 3 3 2
lyzes at 1308C to afford (C F ) P(O)OH. The chlorinating
[15]
2
5 2
[19]
reagent PhPCl transforms this acid into the corresponding
of atoms (in combination with the program SHRINK ) to
guide the construction of the molecular model, and for
values and uncertainties of flexible restraints to be intro-
duced into the refinement by the structure analysis restrain-
ed by ab initio calculations for electron diffraction (SARA-
4
phosphinic acid chloride, (C F ) P(O)Cl. In this way,
2
5 2
(
C F ) P(O)Cl is now conveniently accessible on a large
2
5
[
2
16]
scale.
[17]
As described in an earlier paper, (C F ) P(O)Cl reacted
2
5 2
[20]
with two equivalents of Bu SnH quantitatively with the lib-
CEN) method. The possibility of multiple conformers af-
forded by the two C F groups complicates the analysis of
3
eration of H to yield (C F ) POSnBu (Scheme 2). Subse-
2
2
5
2
3
2 5
quent reaction with PhPCl4 resulted in the formation of
C F ) PCl, which was separated from the reaction mixture
by fractional condensation in a yield of 60%.
the GED data and prior knowledge of the conformers likely
to be present is essential for a reliable interpretation of the
diffraction pattern. To this end, a two-dimensional potential-
energy surface (PES) for rotation around the two C-P-C-C
dihedral angles was calculated (Figure 1), which revealed
eight distinct local minima (labeled 1 to 8), each with a cor-
(
2
5 2
A further marked improvement of the synthesis of
(
C F ) PCl could be achieved by using an alternative reac-
2 5 2
tion pathway with a reduced number of required reaction
Chem. Eur. J. 2011, 17, 3968 – 3976
ꢂ 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
3969

N. W. Mitzel, B. Hoge et al.
Table 1. Relative energies of the low-energy conformers (energies in
À1
kJmol ).
Conformer
BP86
B3PW91
MP2
DE
DE
DG
DE
1
2
3
4
0.0
2.1
2.0
5.0
1.5
0.0
1.0
1.4
2.5
0.5
0.0
1.7
2.6
4.5
6.2
0.0
2.0
2.5
2.8
1.0
[
s
a]
C
[
a] Transition state. The mode corresponding to the imaginary frequency
was ignored in the free energy calculation.
obtain the free energy disfavor the C transition state. How-
s
ever, the presence of relatively low frequencies of vibration
may render the calculated free energies less reliable than
the electronic energies for predicting the relative stabilities
of the conformers.
Figure 1. PES (BP86/SVP) for rotation around the two PÀC bonds.
Minima are numbered roughly according to the energy ordering and
labels 1 to 4 relate the local minima on the PES to the conformers shown
in Figure 2.
Considerations for the modeling of GED data: A number of
problems are encountered when attempting to construct a
GED model for nonrigid systems, such as the compound de-
scribed herein, namely, the likelihood of multiple conform-
ers, possible large-amplitude motions, uncertainties in the
amplitudes of vibration and distance corrections, and the
correlation between these three factors.
The first of these, the possibility of multiple conformers,
perhaps requires the most effort to resolve. This is because
the presence of additional con-
responding enantiomer. Of these eight local minima, confor-
mer 1 was calculated to have the lowest energy, while con-
formers 2 to 4 (Figure 2) were also calculated to have low
enough energies that they might also contribute significantly
to the diffraction pattern. The relative energies calculated at
formers not only increases the
number of parameters to be re-
fined, but introduces problems
in resolving the individual dis-
tances. As in the case of single
conformers with similar distan-
ces or pseudosymmetry, the
average distance may be well
resolved, but the differences be-
tween distances will be correlat-
ed to various other parameters,
such as the amplitudes of vibra-
tion, distance corrections, and
the conformer ratios themselves
(
conversely, if these differences
are fixed, they may influence
the refined conformer ratios).
As an illustration of this, the PÀ
Cl distance in conformer 1 was
calculated
(B3PW91/6-
Figure 2. Calculated structures of the four lowest-energy conformers, showing the atom numbering used in the
GED refinement (Figure 1).
3
11G(2d)) to be 2.055 ꢃ,
whereas in conformer 4 it was
calculated to be 2.051 ꢃ. Since
the BP86/SVP, B3PW91/6-311G(2d), and MP2/TZVPP
levels are shown in Table 1. In addition to these minima, the
P and Cl are both strong scatterers for electrons, the average
PÀCl distance for all conformers will be well resolved, but
transition state (C ) for the interconversion of conformer 1
the difference between conformers will be strongly correlat-
ed to the amplitude of vibration, u, so the refined value of
rP-Cl in 1 will depend on the fraction of 1 in the sample
under investigation. In this case the difference is small and
it is arguable whether it is necessary to take into account
s
with its enantiomer was predicted to be low and the calcu-
lated energies for this conformer are included in Table 1.
While the electronic energies calculated by the two meth-
ods are in good agreement, the calculated corrections to
3970
www.chemeurj.org
ꢂ 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2011, 17, 3968 – 3976

Chlorobis(pentafluoroethyl)phosphane
FULL PAPER
such a small difference. However, angles are more easily de-
formed than bonds, therefore, differences in distances sepa-
rated by two bonds may be markedly different. For example,
the C-P-C angle was calculated to be 95.38 in conformer 1
and 100.48 in conformer 4, leading to a difference of 10 pm
in the nonbonded C(3)···C(4) distance. If both of these con-
formers were present in significant quantities, the choice of
a single-conformer model would significantly distort the re-
fined structure, yielding a systematic error that would not be
expressed in the standard deviations of the refined parame-
ter values. On the other hand, the differences between the
conformers for some local internal coordinates (such as
those for functional groups remote from the site of varia-
tion) may be negligible and can therefore be considered to
be identical for each conformer. In principle, the SARA-
CEN method can be used to solve these problems, howev-
er, a full implementation of this method for medium-sized
molecules with more than a single conformer becomes ex-
tremely cumbersome as well as increasing the likelihood of
errors during coding. Thus, the final model, for which de-
tailed results are presented, is a pseudo-SARACEN type
using a number of assumptions, details for which are given
in the following two sections.
Given the results of the calculations, (C F ) PCl should be
2
5 2
expected to exhibit large-amplitude motions according to
À1
both definitions. In fact, if a calculated value of 100 cm is
used as the criteria for separating small- and large-ampli-
tude motion, each possible conformer would have at least
four large-amplitude modes: two of these corresponding to
rotations around the two PÀC bonds and two corresponding
to rotations of the two CF groups. However, further calcu-
3
lations of the potential energy for the CF rotation reveal
3
À1
that there is a substantial barrier to rotation (13 kJmol
when using the B3PW91 functional), almost an order of
magnitude greater than the barrier to interconversion of the
conformers arising from rotation around the PÀC bonds.
Therefore, any large-amplitude-motion effects will be most
significant for the rotations of the two CF groups, which
3
[20]
were thus identified as the appropriate starting point for in-
vestigating the appropriateness of
model.
a pseudoconformer
Molecular structure models for the GED analysis: Inspec-
tion of the PES (Figure 1) reveals some interesting features
that can be used to guide the construction of an appropriate
GED model. The global minimum, 1, is located in a very
flat region of the PES and can be described as distorted
The remaining problems mentioned above are related to
that of large-amplitude motion; a phenomenon often en-
countered in GED. This can be defined in various ways, but
is often associated with at least one low-frequency oscilla-
from C symmetry, but with a low barrier to interconversion
s
of enantiomers. This conformer is likely to be dominant in
the contribution to the diffraction image, not just because it
is the predicted global minimum, but also because of the rel-
atively broad nature of this minimum. In addition, due to
the low energy barrier between the enantiomers, it is con-
ceivable that the compound could be well described by Cs
symmetry, either because of problems predicting the true
PES arising from theoretical approximations, or because the
vibrational wave function may straddle the potential-energy
barrier leading to a large amplitude of the wave function at
the barrier. Thus, initial attempts to fit the data were based
on a C -symmetric model, first, by using full C symmetry,
À1
tion (generally less than 50 to 100 cm ) around a broad,
shallow minimum in the PES. Alternatively, it can be con-
sidered to be a property of nonrigid molecules that possess
several minima on the PES, separated by barriers that are
[21]
surmountable within the timescale of the experiment.
From the point of view of GED, the first case poses a great-
er difficulty for the refinement of a molecular structure be-
cause the approximation of harmonic motion, routinely used
to calculate the amplitudes of vibration and utilized by pro-
[22]
[19]
grams such as ASYM and SHRINK, is no longer ap-
propriate. It is worth mentioning here that cubic force con-
s
s
then by breaking the symmetry by allowing independent
movement of the two perfluoroethyl groups. The R factors
for these models were 14 and 9.5%, respectively, so a struc-
[23]
stants have been incorporated into the SHRINK program,
but these are only considered for the calculation of distance
corrections, k, rather than amplitudes of vibration, u; in ad-
dition, since a perturbation approach is used, the ka3,1 dis-
tance corrections can be less reliable than the harmonic kh1
corrections when large amplitudes are involved. These prob-
lems are often tackled by using a “dynamic” model that ac-
counts for the low-frequency mode with a set of pseudocon-
formers, each weighted according to a Boltzmann popula-
tion of a parameterized potential-energy function (as point-
ture with full C symmetry was rejected on the basis of the
s
poorer fit of the data. It was noticed, however, that the re-
fined parameter values were similar for the two models,
showing that some structural information could be extracted
from the diffraction data, even if the exact conformers and
conformer ratios were unknown. The natural extension of
this approach was a pseudoconformer model, and two such
models investigated were those fitting potential-energy func-
[24]
2
2
4
ed out by Zewail et al., this name can be misleading as
the model itself is static, though accounting for vibrational
motion). This approach has met with considerable success;
however, it is only practical for compounds with one, or at
most two, large-amplitude modes. An additional concern is
that this type of model is most appropriate when the motion
involved is independent of the structural and vibrational
properties of the remainder of the molecule, which is never
completely true.
tions of the form V(f)=A
A
H
U
G
E
N
N
(fÀB) and V(f)=Af +Bf (f
is the difference between the two C-P-C-C angles). Howev-
er, neither of these models gave better agreement between
the theoretical and experimental intensity curves than the
single-conformer model. With the exception of the C sym-
s
metric model, all models tested resulted in similar refined
parameters.
While a reasonable fit to the experimental data could be
achieved by using these simplified models, a small disagree-
Chem. Eur. J. 2011, 17, 3968 – 3976
ꢂ 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
3971

N. W. Mitzel, B. Hoge et al.
ment at about 4.5 ꢃ in the radial distribution curve could
not be accounted for. Theoretical radial distribution func-
tions were generated for the calculated geometries of the
four low-energy conformers, 1 to 4, as well as the transition
ing Information. The full number of degrees of freedom for
such a model would be 84 (=2
A
H
U
G
R
N
U
G
simplified compared with the full number of degrees of free-
dom, but the assumptions adopted were mostly in the de-
state for enantiomer inversion, C , as shown in Figure 3. The
scriptions of the terminal CF groups. More specifically, the
s
3
CF groups were assumed to be identical, with local C sym-
3
3v
metry, so that all CÀF distances and F-C-F angles in these
groups were considered to be equal, as were the CÀC dis-
tances that joined each CF group to the rest of the mole-
3
cule. However, only the CÀC distance was included explicit-
ly as an independent parameter, whereas the CÀF distance
and F-C-F angle were related to the remaining distances and
angles of the same type by way of weighted averages and
differences (p , p , p , and p ). These assumptions can be
6
7
42
43
justified for two reasons: First, because these groups are lo-
cated at the termini of the molecule, any errors in the pa-
rameters describing these groups will have a minimal impact
on the nonbonded distances in the rest of the molecule.
Second, because the CF groups are reasonably remote from
3
the PÀC bonds (rotations around which define the various
Figure 3. Comparison of the experimental (exptl) radial distribution
curve (RDC) with those predicted from calculated geometries by using
one of two methods: a) generation of Fourier-transformed theoretical in-
tensity curves or b) a superposition of Gaussian functions for each pair of
atoms. The experimental curve and curve 1(a) were obtained by multiply-
conformers) these groups will not be affected to a large
extent by conformational differences, and therefore, should
be similar in structure. The three remaining assumptions
were based on the results of ab initio calculations, namely,
equal PÀCl bond lengths in conformers 1 and 2, an equal
average CÀF distance in 1 and 2, and a single F-C-F angle
2
À1
À1
F
ing the molecular intensity curves by sexp
A
H
U
T
N
N
[À0.002s
A
H
U
G
R
N
N
(ZClÀfCl
)
A
T
N
T
E
N
N
(Z
F
Àf
) ]
prior to Fourier transformation.
for the CF bridges.
2
upper two curves show the Fourier transformation of molec-
ular intensity curves obtained from experimental data
In each conformer, the geometry around the central phos-
phorus atom was defined by using the bonded distances, the
two Cl-P-C angles, and the C-P-C angle. The fluorine atom
(
exptl) and the calculated (B3PW91/6-311G(2d)) geometry
of conformer 1 by using amplitudes of vibration calculated
using SHRINK. As in the simple model mentioned above,
there is a small disagreement between the calculated struc-
ture and experiment at 4.5 ꢃ in the radial distribution curve
positions in the CF bridge were defined by using a riding
model, with a rock (rk), twist (tw) and wag (wg) angle for
2
each group (p22 to p ), to allow full flexibility of orientation
33
of these groups. These were defined as rotations around
three perpendicular vectors, all of which pass through the
central C atom: rock is the rotation around a vector perpen-
dicular to the P-C-C plane, twist is the rotation around the
bisector of the P-C-C angle, and wag is the rotation around
the remaining orthogonal vector.
(
Figure 3), thus the assumption of pseudo-C symmetry was
s
not a critical problem. The lower curves, 1(b) to 4(b) and Cs,
were calculated probability distributions, also using the cal-
culated geometries and amplitudes in combination with
SHRINK, thus, the effect of the Fourier transform can be
seen by comparing plots 1(a) and 1(b) (Figure 3), which
shows that the breadth of the peaks corresponding to short
distances are most drastically affected. Curves 1(b) and 2(b)
are very similar and most closely match the experimental
curve, whereas the remaining curves exhibit significant dif-
ferences to experiment in the region 2.5 to 3.5 ꢃ and on the
basis of these observations, conformers 3 and 4 were exclud-
ed from subsequent analysis. It can also be seen that super-
imposing the RDCs for conformers 1 and 2 may relieve the
problem at 4.5 ꢃ associated with a single-conformer model,
thus a two-conformer model was identified as the most ap-
propriate for this system.
Because there was more than one angle or distance of a
particular type (details are provided in the Supporting Infor-
mation in Table S5), they were introduced by means of aver-
age and difference parameters. This is because the average
values are often well resolved by GED, but the differences
are usually poorly resolved. By using this implementation,
the differences can be fixed or restrained to calculated
[20]
values. In this sense the SARACEN method was adopted
to allow all 43 geometric parameters to be refined, 8 of
which were refined unrestrained. In addition, the ratio of
conformers 1 and 2 was refined unrestrained, and seven
groups of amplitudes of vibration were refined. These seven
groups of amplitudes were of two types, the first type con-
sisted of six groups of amplitudes for interatomic distances
less than about 3.7 ꢃ. The second type was a single group
consisting of all distances larger than 3.7 ꢃ, which were sus-
pected to undergo large-amplitude motion. For each group
in the first category, amplitudes for similar interatomic dis-
GED refinement and molecular structure of (C F ) PCl:
2
5 2
The two-conformer model (comprising conformers 1 and 2)
with which the following results were determined was con-
structed by using 43 independent geometrical parameters
and 1 conformer ratio, as listed in Table S5 in the Support-
3972
www.chemeurj.org
ꢂ 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2011, 17, 3968 – 3976

Chlorobis(pentafluoroethyl)phosphane
FULL PAPER
tances were tied together and an overall scale-factor refined,
which was restrained to the calculated value with an uncer-
tainty of 10%. Amplitudes of the second type were also tied
together with fixed ratios, while a single scale factor was re-
fined, but in this case no restraint was applied.
The refined radial distribution curve (RDC) and molecu-
lar intensity curves for the two-conformer model are shown
in Figure 4, and a selection of the refined GED parameter
vibrational model, indicating that the uncertainty in this
value was larger than the standard deviations obtained from
the least-squares fit [this dependence on the model type was
not due to distance corrections applied to the CÀC distan-
ces, which were in the range of 0.001 ꢃ, rather it was due to
those applied to the nonbonded distances, which indirectly
determined this distance]. The calculated values were closer
to the rh1 structure type than the r structure type, but, in
g
both cases, were calculated to be longer than determined by
GED. In the present study there is also a strong influence of
the structure type (0.016 ꢃ), however, in this case, the value
calculated by MP2 theory is very close to the experimental
rh1 model.
In contrast to the CÀC distances, the experimental C-P-C
and P-C-C angles were found to be relatively independent
of the structure type and for both P ACHTUNGTRENUN(NG C F ) and (C F ) PCl
2 5 3 2 5 2
the values of these parameters were determined to be larger
than those predicted by MP2 theory. In the study on
P ACHTUNGTRNEUN(G C F ) , a parallel determination of the crystal structure by
2 5 3
X-ray diffraction at 100 K revealed that in the solid state
the P-C-C angles were close to those predicted by theory,
whereas the C-P-C angle was intermediate between the
GED and ab inito values. It is possible to propose a number
of explanations for these results, including the influence of
packing effects in the crystal, deficiencies in MP2 theory,
and the influence of temperature and vibrational motion on
the GED structural parameters. An estimate of the reliabili-
ty of calculated values can be obtained by looking at the
basis-set convergence and comparing correlated methods
with those that omit electron correlation; thus, the accuracy
of MP2 theory can be assessed by comparing the calculated
values with those calculated by HF, and both of these are
shown for (C F ) PCl in Table 2. For the C-P-C angle, the
2
5 2
differences between the methods are about 28, and for the
P-C-C angle, around 18. These values are comparable to the
differences between the GED and MP2 values of about 3
and 18 for the respective angles. Curiously, HF theory seems
to be better than MP2 for predicting the values of these pa-
rameters and, in general, predicts the GED structure re-
markably accurately, arguably outperforming the B3PW91
method, which is thought to be a higher-level theory. How-
ever, the drawback of the HF method can be clearly seen by
the poor predictions for the CÀF distances, which are sys-
Figure 4. Top: experimental and difference GED molecular intensity
curves for the two-conformer refinement of (C
mental and difference radial distribution curves for the two-conformer
refinement of (C PCl.
2 5 2
F ) PCl. Bottom: experi-
F
5
)
2
tematically underestimated by about 0.02–0.03 ꢃ.
2
It is also interesting to compare the structures of P
ACHTUNGTRENNUNG( C F )
2 5 3
and (C F ) PCl in the context of correlation between the
2
5 2
values for 1 and 2 are listed in Table 2 and compared with
nonbonded F···F distances on adjacent perfluoroethyl
groups and the identification of stable minima on the PES.
For all four of the most stable conformers of (C F ) PCl, 1
calculated values. As found for the related compound
[25]
P ACHTUNGTRENNUNG( C F ) , the experimental GED structure and the struc-
2 5 3
2
5 2
ture calculated at the MP2 level of theory were, in general,
in good agreement. In this previous study the largest differ-
ences between the experimental and theoretical structures
were in the CÀC distances, the C-P-C angles, and the P-C-C
angles; the CÀC distances were determined to be
to 4 (Figure 2), these nonbonded distances appear to be
maximized and is consistent with the idea that a significant
factor influencing the preferred conformations in these
types of compounds is repulsion between fluorine atoms on
[25]
neighboring branches.
0
.01–0.03 ꢃ shorter by GED and the angles about 28 larger.
The study on P(C F ) also revealed that the experimental
CÀC distance was strongly influenced by the selection of the
ACHTUNGTRENNUNG
2
5 3
Chem. Eur. J. 2011, 17, 3968 – 3976
ꢂ 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
3973

N. W. Mitzel, B. Hoge et al.
Table 2. Values of critical geometric parameters as determined by GED and calculated at various levels of first time, a description of the
[
a]
theory.
steric
requirements
of
a
[
b]
L/R
Conformer
GED
B3PW91
HF
(r
MP2
(r
(
C F ) P group and its confor-
2
5 2
[
c]
[d]
[e]
(
r
g
)
A
H
U
G
R
N
U
G
)
(r
e
)
e
)
e
)
mational behavior. These are of
interest when such a group is
incorporated into new ligand
systems.
distances
PÀCl
1
2
1
2
1
2
1
2
1
2
1
1
2
1
2
1
2
1
2
2.015(4)
2.015(4)
1.904(3)
1.911(3)
1.894(3)
1.900(3)
1.528(3)
1.528(3)
1.528(3)
1.528(3)
1.332(1)
1.346(2)
1.346(2)
1.357(2)
1.356(2)
1.346(2)
1.354(2)
1.356(2)
1.349(2)
2.018(3)
2.018(3)
1.907(3)
1.914(3)
1.897(3)
1.904(3)
1.544(3)
1.544(3)
1.544(3)
1.544(3)
1.333(1)
1.346(2)
1.347(2)
1.357(2)
1.357(2)
1.347(2)
1.354(2)
1.357(2)
1.349(2)
2.055
2.052
1.926
1.932
1.915
1.920
1.543
1.544
1.547
1.543
1.331
1.343
1.341
1.356
1.352
1.341
1.348
1.352
1.344
2.028
2.026
1.913
1.918
1.906
1.910
1.542
1.543
1.544
1.542
1.306
1.321
1.319
1.330
1.328
1.319
1.325
1.328
1.321
2.026
2.023
1.913
1.918
1.903
1.909
1.542
1.544
1.546
1.543
1.330
1.346
1.343
1.356
1.353
1.343
1.350
1.353
1.346
PÀC
CÀC
L
R
L
R
First-principles calculations
revealed an unexpected com-
plexity of the potential energy
hypersurface with respect to the
two torsional angles C-P-C-C
(t , t , where for C symmetry,
1
2
s
CÀF (CF
CÀF(13)
av)
3
t1 + t =0), giving rise to at
least eight different minima po-
tentially contributing to the
2
R
R
R
R
L
L
L
L
CÀF(14)
CÀF(15)
CÀF(16)
electron
scattering
pattern
under experimental conditions.
However, a thorough analysis
of the molecular scattering in-
tensity curves together with a
critical inspection of calculated
relative conformer energies
showed that only two main con-
formers dominated the molecu-
lar ensemble in the gas jet. The
angles
C-P-C
C-P-C
Cl-P-C
1
2
1
2
1
2
1
2
1
2
98.6(4)
101.0(4)
99.8(3)
99.0(3)
101.4(3)
99.7(3)
112.0(3)
112.7(3)
114.8(3)
112.4(3)
98.2(4)
95.3
97.5
99.0
98.3
100.8
99.1
111.0
110.8
113.9
111.3
96.9
99.0
99.8
99.0
101.3
99.8
112.1
112.6
114.6
112.3
94.9
97.2
98.8
98.1
100.5
98.8
110.9
111.5
113.6
111.2
100.5(4)
100.1(3)
99.3(3)
101.7(3)
100.0(3)
111.8(3)
112.5(3)
114.6(3)
112.2(3)
L
L
R
R
L
most
abundant
conformer,
P-C-C
6
1(5)% (GED), has a structure
that is close to Cs symmetry
with the magnitude of t and t2
R
1
differing
by
approximately
torsional angles
2
5(1)8. The C-P-C angle in this
C-P-C-C(9), t
1
L
1
2
1
2
À169.0(10)
À102.4(9)
À164.6(9)
171.9(10)
À169.1(10)
À102.6(9)
À165.3(9)
171.7(10)
À167.9
À103.3
À165.8
171.3
À168.4
À100.7
À169.1
170.7
À167.7
À102.3
À165.0
171.5
conformer was determined to
be 98.2(4)8.
In this way, a large fraction
of the potential hypersurface
C-P-C-C(5), t
2
R
[
a] All distances are given in ꢃngstroms and angles are given in degrees. [b] L and R indicate the left- or
right-hand side of the molecule, as shown in Figure 2. [c] Average distance between atoms <r>. [d] Average describing the conformational
interatomic distances determined by using distance corrections for nonbonded atom pairs based on harmonic possibilities within the (C F ) P
2
5 2
curvilinear motion, see reference [19]. [e] Equilibrium structure distance.
group has been scanned by
quantum chemical methods. In
essence, the results show that
Conclusion
repulsive F···F interactions dominate this conformational
space; a fact that has also been found for (C F ) P in a relat-
2
5 3
A
C
H
T
U
N
G
T
R
E
N
N
U
N
G
(C F ) PCl has great potential to be used as a starting mate-
ed, but slightly different manner [P
cone angle of 193 (solid) and 1918 (gas), whereas the P-
(CF ) ligand exhibits a significantly smaller Tolman cone
ACHTUNGTRNEN(GU C F ) has a Tolman
2
5
2
2 5 3
[26]
rial for the synthesis of new electronically and sterically
fine-tuned ligands. With this work, we established a signifi-
cantly improved synthesis of (C F ) PCl starting from the
technical product (C F ) PF with yields of up to 66%. As
previously demonstrated,
AHCTUNGTRENNUNG
3
3
angle of only 1288 in the coordination sphere of the iron
2
5 2
[27]
complex [Fe(CO) {P ACHTUNGTRNENG(U CF ) } ] ]. The mutual dependence of
2 3 3 3
2
5
3
2
26]
[
(C F ) PCl reacts selectively
the conformations of the two C F groups can be metaphori-
2 5
2
5
2
with alcohol derivatives, ROH, in the presence of K CO as
cally described by two cogwheels gripping each other.
A gas-phase and solid-state structural investigation on
P ACHTUNGTNERNU(G C F ) demonstrated the significant difference in the steric
2 5 3
2
3
a base to give the corresponding phosphinite derivatives
C F ) POR. Further investigations to synthesize bis(penta-
(
2
5
2
fluoroethyl)phosphinite ligands (cf. Scheme 1) and corre-
sponding Lewis acidic transition-metal complexes for appli-
cations in catalysis are in progress.
bulk in comparison to the P
ACHTUNGTRENNUNG( CF ) ligand. Similarly, the
3 3
(C F ) P group also has a substantially higher steric shield-
2
5 2
ing of the phosphorus atom than the P ACHTUNGTRENUNG( CF ) group.
3 2
With the experimental determination of the molecular
structure of (C F ) PCl in the gas phase, we provide, for the
2
5 2
3974
www.chemeurj.org
ꢂ 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2011, 17, 3968 – 3976

Chlorobis(pentafluoroethyl)phosphane
FULL PAPER
support. We want to thank Dr. N. Ignatiev (Merck KGaA, Darmstadt,
Germany) for helpful discussions.
Experimental Section
Synthesis of (C
7 mmol) in 1,6-dibromohexane (20 mL) was degassed for 15 min in
vacuum before the addition of (C P(O)Cl (2.5 g, 7.8 mmol). After
stirring the mixture for 1 h, all volatile compounds were removed in
vacuum. PhPCl (1.24 g, 5.0 mmol) was added and the solution was
2 5 2 2 5 2 3
F ) PCl via (C F ) P(O)Cl: A solution of Bu SnH (4.9 g,
1
2
F
5
)
2
[1] A. B. Burg, G. B. Street, J. Am. Chem. Soc. 1963, 85, 3522.
[2] L. D. Field, M. P. Wilkinson, Tetrahedron Lett. 1992, 33, 7601.
[3] W. Volbach, I. Ruppert, Tetrahedron Lett. 1983, 24, 5509.
[4] M. F. Ernst. D. M. Roddick, Inorg. Chem. 1989, 28, 1624.
[5] R. D. Chambers, Fluorine in Organic Chemistry, Wiley, New York
1973, p. 345.
4
stirred for 20 min. The volatile compounds were removed and separated
under dynamic vacuum conditions by using three traps at À30, À78, and
À1968C to separate minor amounts of the solvent and (C
2
F
5
)
2
PCl. The
À1968C trap contained a colorless liquid identified as (C
0%).
5 2 2 5 3 2 5 3
Synthesis of (C F ) PCl via P ACTHNGUTRNEN(UG C F ) : A solution of P ACHUTNGNNEURG( C F ) (12.4 mmol)
2
F
5
)
2
PCl (1.3 g,
[6] M. F. Ernst, D. M. Roddick, Organometallics 1990, 9, 1586.
[7] D. M. Roddick, Chem. Eng. News 1997, 75, 6.
6
[
8] M. B. Murphy-Jolly, L. C. Lewis, A. J. M. Caffyn, Chem. Commun.
005, 4479.
2
2
in diethyl ether was treated with an excess of a 1.5m aqueous solution of
NaOH (25 mmol) and stirred for 30 min. 1,6-Dibromohexane was added
to the separated organic phase and all volatile compounds were removed
[
9] For example, see: A. Badoiu, G. Bernardinelli, C. Besnard, E. P.
Kꢁndig, Org. Biomol. Chem. 2010, 8, 193, and references therein.
10] M. E. Bruin, E. P. Kꢁndig, Chem. Commun. 1998, 2635.
11] a) F. W. Bennett, H. J. Emelꢄus, R. N. Haszeldine, J. Chem. Soc.
[
[
by applying vacuum overnight. After the addition of PhPCl
4
(3.5 g,
1
3 mmol) dissolved in degassed 1,6-dibromohexane (5 mL) and addition-
1
953, 1565; b) H. J. Emelꢄus, J. D. Smith, J. Chem. Soc. 1959, 375;
al stirring for 10 min at room temperature, the product was isolated by
c) A. H. Cowley, T. A. Furtsch, D. S. Dierdorf, J. Chem. Soc. Chem.
Commun. 1970, 523.
12] H. G. Ang, M. E. Redwood, B. O. West, Aust. J. Chem. 1972, 25,
fractional condensation. The À1968C trap contained a colorless liquid
1
9
identified as (C
2
F
5
)
2
PCl (2.5 g, 66%). F NMR (Et
2
O, RT): d=À80.8
3
2
31
[
(
d, J
A
C
H
T
U
N
G
T
R
E
N
N
U
N
G
(P,F)=14 Hz), À116.5 ppm (d,
J
A
H
U
T
E
N
N
(P,F)=61 Hz); P NMR (Et
2
O,
[28]
19
3
4
93.
RT): d=60.4 ppm (m); (Lit.
:
F NMR: d=À81.63 (d,
J AHCTUNGTRENNUN(G P,F)=
2
31
[
13] N. V. Ignatꢅev, P. Sartori, J. Fluorine Chem. 2000, 103, 57; U. Heider,
1
(
4.6 Hz), À117.2 ppm (d,
septet of quintets)).
First-principles calculations: All calculations were spin restricted and
J
A
H
U
T
E
N
N
(P,F)=58.6 Hz); P NMR: d=61.17 ppm
V. Hilarius, P. Sartori, N. Ignatiev (Merck Patent GmbH), WO 2000/
2
1969, US 6,264,818, EP 1037896, 2000.
[14] N. V. Ignatꢅev, H. Willner, P. Sartori, J. Fluorine Chem. 2009, 130,
183; N. Ignatyev, M. Schmidt, A. Kuehner, V. Hilarius, U. Heider,
[
29,30]
those with the hybrid density functional B3PW91,
which was used
1
for the calculation of geometries and to provide a force-field for subse-
quent calculations of the amplitudes of vibration by using SHRINK,
A. Kucheryna, P. Sartori, H. Willner (Merck Patent GmbH), WO
2003/002579, US 7,094,328, EP 1399453 B1, 2003.
[
31]
were performed by using the Gaussian 03 suite of programs. The re-
[
32]
[15] U. Welz-Biermann, N. Ignatyev, M. Weiden, U. Heider, A. Kuchery-
na, H. Willner, P. Sartori (Merck Patent GmbH), WO 2003/0087110,
US 7,202,379, 2003.
16] N. Ignatyev, E. Aust, D. Bejan, H. Willner (Merck Patent GmbH),
WO 2010/009791A1, 2010.
17] B. Hoge, J. Bader, H. Beckers, Y. S. Kim, R. Eujen, H. Willner, N.
Ignatiev, Chem. Eur. J. 2009, 15 , 3567.
18] B. Hoge, W. Wiebe, Angew. Chem. 2008, 120, 8236; Angew. Chem.
Int. Ed. 2008, 47, 8116.
19] V. A. Sipachev, J. Mol. Struct. 1985, 126–132, 143.
20] a) A. J. Blake, P. T. Brain, H. McNab, J. Miller, C. A. Morrison, S.
Parsons, D. W. H. Rankin, H. E. Robertson, B. A. Smart, J. Phys.
Chem. 1996, 100, 12280; b) N. W. Mitzel, D. W. H. Rankin, Dalton
Trans. 2003, 3650.
21] Stereochemical Applications of Gas-Phase Electron Diffraction, Part
A: The Diffraction Technique (Eds.: I. Hargittai, M. Hargittai),
VCH, Weinheim, 1988, and references therein.
maining calculations were performed by using the Turbomole program.
In all cases, in which the method is stated as MP2, the resolution-of-the-
identity (RI) method was used, in conjunction with the def2-TZVPP
basis set, with the default partitioning scheme for defining the frozen
[
[
[
[
33]
core. For HF calculations, the def2-TZVPP basis set was also used to
enable a direct comparison of the MP2 and HF geometries, but those de-
[
33]
[33]
noted BP86 were performed by using the SVP basis set. Calculations
at the B3PW91 level were performed in conjunction with the 6-311G(2d)
basis set. Geometries and energies of 1, 2, 3, and 4 calculated at the RI-
MP2/def2-TZVPP level of theory are provided in the Supporting Infor-
mation.
[
[
Gas electron diffraction: Electron scattering intensities for (C
2 5 2
F ) PCl
were recorded at room temperature on a combination of reusable Fuji
[
34]
and Kodak imaging plates by using a Balzers KD-G2 Gas Eldigraph
equipped with a new electron source (STAIB Instruments) operating at
around 60 kV. These data were reduced to total intensities by using the
[
[
35]
program PIMAG by Strand et al. (version 040827) and further data re-
duction (yielding molecular intensity curves), the molecular structure re-
finement, and the electron wavelength determination (from benzene
[
[
[
22] L. Hedberg, I. M. Mills, J. Mol. Spectrosc. 2000, 203, 82.
23] V. A. Sipachev, Struct. Chem. 2000, 11, 167.
24] M. M. Lin, D. Shorokhov, A. H. Zewail, J. Phys. Chem. A 2009, 113,
[
36]
data) were performed by using version 3.0 of the ed@ed program. Fur-
ther details about the Bielefeld GED apparatus, the data handling, and
approaches to the refinement are given in reference [37]. Full informa-
tion relating to the refinement is provided in the Supporting Information:
The data analysis parameters for each data set (R factors, scale factors,
data ranges, weighting points, nozzle-to-plate distances, and electron
wavelengths), a list of interatomic distances, amplitudes of vibration, u,
4
075.
[
[
[
25] S. A. Hayes, R. J. F. Berger, B. Neumann, N. W. Mitzel, J. Bader, B.
Hoge, Dalton Trans. 2010, 39, 5630.
26] B. Hoge, J. Bader, B. Kurscheid, N. Ignatyev, E. Aust (Merck Patent
GmbH), WO 2010/009818A1, 2010.
27] a) B. J. Dunne, R. B. Morris, A. G. Orpen, J. Chem. Soc. Dalton
Trans. 1991, 653; b) A. B. Burg, Inorg. Chem. 1986, 25, 4751.
28] T. Mahmood, J. M. Shreeve, Inorg. Chem. 1986, 25, 3128.
29] A. D. Becke, J. Chem. Phys. 1993, 98, 5648.
30] J. P. Perdew, K. Burke, Y. Wang, Phys. Rev. B 1996, 54, 16533.
31] Gaussian 03, Revision B03, M. J. Frisch, G. W. Trucks, H. B. Schle-
gel, G. E. Scuseria, M. A. Robb, J. R. Cheeseman, J. A. Montgom-
ery, Jr., T. Vreven, K. N. Kudin, J. C. Burant, J. M. Millam, S. S.
Iyengar, J. Tomasi, V. Barone, B. Mennucci, M. Cossi, G. Scalmani,
N. Rega, G. A. Petersson, H. Nakatsuji, M. Hada, M. Ehara, K.
Toyota, R. Fukuda, J. Hasegawa, M. Ishida, T. Nakajima, Y. Honda,
O. Kitao, H. Nakai, M. Klene, X. Li, J. E. Knox, H. P. Hratchian,
J. B. Cross, V. Bakken, C. Adamo, J. Jaramillo, R. Gomperts, R. E.
Stratmann, O. Yazyev, A. J. Austin, R. Cammi, C. Pomelli, J. W.
and distance corrections for the curvilinear perpendicular motion, kh1
,
[
[
[
[
and refined molecular coordinates for 1 and 2 from the rh1 refinement
are also provided. Only three pairs of parameters were correlated by
more than 50%, namely, p
1
/u
2
=À0.51, p11/p42 =0.70 and k
1
/k
2
=À0.75.
Acknowledgements
This work was supported by the Deutsche Forschungsgemeinschaft.
Merck KGaA, Darmstadt, Germany, is also acknowledged for financial
Chem. Eur. J. 2011, 17, 3968 – 3976
ꢂ 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
3975

N. W. Mitzel, B. Hoge et al.
Ochterski, P. Y. Ayala, K. Morokuma, G. A. Voth, P. Salvador, J. J.
Dannenberg, V. G. Zakrzewski, S. Dapprich, A. D. Daniels, M. C.
Strain, O. Farkas, D. K. Malick, A. D. Rabuck, K. Raghavachari,
J. B. Foresman, J. V. Ortiz, Q. Cui, A. G. Baboul, S. Clifford, J. Cio-
slowski, B. B. Stefanov, G. Liu, A. Liashenko, P. Piskorz, I. Komaro-
mi, R. L. Martin, D. J. Fox, T. Keith, M. A. Al-Laham, C. Y. Peng,
A. Nanayakkara, M. Challacombe, P. M. W. Gill, B. Johnson, W.
Chen, M. W. Wong, C. Gonzalez, J. A. Pople, Gaussian, Inc., Wall-
ingford CT, 2004.
Ahlrichs, J. Comp. Chem. 1993, 14, 907; e) A. Schꢀfer, C. Huber, R.
Ahlrichs, J. Chem. Phys. 1994, 100, 5829.
[34] W. Zeil, J. Haase, L. Wegmann, Z. Instr. 1966, 74, 84.
[35] S. Gundersen, S. Samdal, T. G. Strand, H.-V. Volden, J. Mol. Struct.
2007, 832, 164.
[36] a) S. L. Hinchley, H. E. Robertson, K. B. Borisenko, A. R. Turner,
B. F. Johnston, D. W. H. Rankin, M. Ahmadian, J. N. Jones, A. H.
Cowley, Dalton Trans. 2004, 2469; b) A. M. Reilly, D. A. Wann,
D. W. H. Rankin, J. Phys. Chem. A 2009, 113, 938.
[
[
32] R. Ahlrichs, M. Bꢀr, M. Hꢀser, H. Horn, C. Kçlmel, Chem. Phys.
Lett. 1989, 154–164, 165.
33] a) R. Ahlrichs, Phys. Chem. Chem. Phys. 2004, 6, 5119; b) F. Wei-
gend, M. Hꢀser, Theor. Chem. Acc. 1997, 97, 331; c) F. Weigend, M.
Hꢀser, H. Patzelt, Chem. Phys. Lett. 1998, 294, 143; d) F. Haase, R.
[37] R. J. F. Berger, M. Hoffmann, S. A. Hayes, N. W. Mitzel, Z. Natur-
forsch. B 2009, 64, 1259.
Received: October 21, 2010
Published online: March 7, 2011
3976
www.chemeurj.org
ꢂ 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2011, 17, 3968 – 3976