7Li, 31P, and1H pulsed Gradient Spin-Echo (PGSE) diffusion NMR spectroscopy and ion pairing: On the temperature dependence of the ion pairing in Li(CPh3), fluorenyllithium, and Li[N(SiMe 3)2] amongst other salts

Article abstract of DOI:10.1002/chem.200400867

⁷Li, ³¹P, and ¹H variable-temperature pulsed gradient spin-echo (PGSE) diffusion methods have been used to study ion pairing and aggregation states for a range of lithium salts such as lithium halides, lithium carbanions, and a lithium amide in THF solutions. For trityllithium (2) and fluorenyllithium (9), it is shown that ion pairing is favored at 299 K but the ions are well separated at 155 K. For 2-lithio-1,3-dithiane (13) and lithium hexamethyldisilazane (LiHMDS 16), low-temperature data show that the ions remain together. For the dithio anion 13, a mononuclear species has been established, whereas for the lithium amide 16, the PGSE results allow two different aggregation states to be readily recognized. For the lithium halides LiX (X = Br, Cl, I) in THF, the ⁷Li PGSE data show that all three salts can be described as well-separated ions at ambient temperature. The solid state structure of trityllithium (2) is described and reveals a solvent-separated ion pair formed by a [Li(thf)₄]⁺ ion and a bare triphenylmethide anion.

Full text of DOI:10.1002/chem.200400867

FULL PAPER
7
31
1
Li, P, and H Pulsed Gradient Spin-Echo (PGSE) Diffusion NMR
Spectroscopy and Ion Pairing: On the Temperature Dependence of the Ion
Pairing in Li(CPh ), Fluorenyllithium, and Li[N(SiMe ) ] amongst Other
3
3 2
Salts
[a]
Ignacio Fernꢀndez, Eloꢁsa Martꢁnez-Viviente, Frank Breher, and Paul S. Pregosin*
7
31
1
Abstract: Li, P, and H variable-tem-
lithio-1,3-dithiane (13) and lithium
hexamethyldisilazane (LiHMDS 16),
low-temperature data show that the
ions remain together. For the dithio
anion 13, a mononuclear species has
been established, whereas for the lithi-
um amide 16, the PGSE results allow
two different aggregation states to be
readily recognized. For the lithium hal-
ides LiX (X = Br, Cl, I) in THF, the
perature pulsed gradient spin-echo
PGSE) diffusion methods have been
(
used to study ion pairing and aggrega-
tion states for a range of lithium salts
such as lithium halides, lithium carb-
anions, and a lithium amide in THF
solutions. For trityllithium (2) and
fluorenyllithium (9), it is shown that
ion pairing is favored at 299 K but the
ions are well separated at 155 K. For 2-
7
Li PGSE data show that all three salts
can be described as well-separated ions
at ambient temperature. The solid state
structure of trityllithium (2) is de-
scribed and reveals a solvent-separated
Keywords: aggregation · diffusion ·
lithium · NMR spectroscopy · X-ray
diffraction
+
ion pair formed by a [Li(thf)4] ion
and a bare triphenylmethide anion.
Introduction
on molecular volumes and thus indirectly on molecular
[6]
weights. Several groups have used this approach to detect
higher aggregation in, for example, a zirconium-based poly-
Lithium reagents are widely used in both preparative inor-
[1]
[7]
I
[8]
ganic and organic chemistry. Regrettably, it is often the
case that differing degrees of solvation and aggregation of
the various lithium salts complicate the structural picture;
for example, tert-butyllithium can be a monomer, a dimer,
merization catalyst, a Cu cluster catalyst precursor, and
[
9]
an iron-based dendrimer, as well as in the characterization
of Pt molecular squares,
[
10]
[11,12]
among others.
For salts of
1
19
transition metals, H NMR (and F NMR) PGSE diffusion
measurements on the cation and anion can provide insight
into how these charged species interact. The diffusion con-
stants derived for the individual cations and anions directly
reflect interactions such as ion pairing between these spe-
[2]
or a tetramer. Since the chemistry of many lithium species
depends on the details of the interaction between the lithi-
[3]
um cation and the solvent, understanding the solution
structure of different lithium reagents is important for ra-
tionalizing and predicting their reactivity and selectivity.
The measurement of diffusion constants by pulsed gradi-
[
13–16]
[17]
cies.
Ion pairing can prove both helpful and detrimen-
[
18]
tal to the activity of different homogeneous catalysts. Nat-
[4,5]
ent spin-echo (PGSE) NMR methods
tracted increasing interest, as this technique provides data
has recently at-
urally, the diffusion constants D will be strongly solvent-de-
[
19]
pendent
dent.
and, to a lesser extent, concentration-depen-
[
20]
7
The combination of Li NMR and PGSE methods is still
relatively rare.
[
a] I. Fernꢀndez, E. Martꢁnez-Viviente, Dr. F. Breher,
Prof. Dr. P. S. Pregosin
[21,22]
[21a,b]
We note only two applications
in-
Laboratory of Inorganic Chemistry, ETHZ HCI Hçnggerberg 8093
Zurich (Switzerland)
E-mail: pregosin@inorg.ethz.ch
volving lithium diffusion studies in organometallic lithium
carbanion chemistry. One of these (on n-butyllithium) in-
[21b]
volved the DOSY
(diffusion-ordered NMR spectroscopy)
Supporting information for this article is available on the WWW
under http://www.chemeurj.org/ or from the author. Eight figures
and not PGSE methodology. We have recently communicat-
7
1
7
ed Li PGSE results in connection with structural problems
showing H and Li PGSE results (without a correction for the gyro-
magnetic ratio), and a gHMQC correlation for 13 at 173 K with one
figure.
[23]
in lithium phosphide chemistry. We report here a series of
7
31
1
ambient- and low-temperature Li, P, and H PGSE NMR
Chem. Eur. J. 2005, 11, 1495 – 1506
DOI: 10.1002/chem.200400867
ꢂ 2005 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
1495

results on both simple lithium salts and several organolithi-
um species in solution in THF. These new data, when com-
bined with X-ray crystallography, provide novel insight into
the question of ion pairing and, in general, on how solvated
lithium cations interact with selected inorganic and organo-
metallic anions.
Results and Discussion
Lithium halides: To begin with we consider the lithium ha-
lides, LiX. A vapor pressure osmometric study for LiCl sug-
[24]
gests a ratio MWapparent/MWreal of 1.83. A crystal structure
of [(LiCl) (hmpa) ], (hmpa = hexamethylphosphoramide,
4
4
[25]
P(O)(NMe ) ), has been reported; however, the reaction
2
3
of lithiated catechol with [W(O)Cl ] produced LiCl as a side
4
product that crystallizes as the dinuclear species, [(thf) Li(m-
2
[27]
[26]
Cl) Li(thf) ]. Furthermore, Reich and co-workers cate-
2
2
o
Figure 1. Plot of the ln(I/I ) versus arbitrary units proportional to the
square of the gradient amplitude for Li PGSE NMR diffusion measure-
gorize lithium chloride as a sturdy dimer according to
HMPA titrations. Consequently, there is some question with
respect to the structure of this simple salt.
7
ments on 60 mm LiX (X = Cl, Br, and I) samples at ambient tempera-
7
ture in THF ( Li: d = 4 ms; D = 70 ms).
LiBr is thought to contain about 80% monomeric species
and 20% higher aggregates in THF (MWapparent/MWreal
=
[24]
[30]
1
.23). LiI is considered to be monomeric in THF at room
tal D values together with the Stokes–Einstein equation,
[24]
temperature. Moreover, [LiI(thf) ], obtained from the re-
and suggest a relatively small solvated cation, such as
[Li(thf) ] . The viscosity of the (nondeuterated) THF at
3
+
action of LiH and iodine in THF, has a mononuclear struc-
4
[28]
[31]
ture in the solid state.
room temperature was taken from the literature.
To shed further light on the nature of these lithium halide
Although the local environment about the lithium center
is clear, the nature of the solvated halogen remains open. To
shed some light on this we have measured the Cl NMR
7
interactions, we measured Li PGSE NMR spectra for LiCl,
[29]
35
LiBr, and LiI as 60 mm solutions in THF.
Table 1 and
spectrum of a LiCl solution in THF (Figure 2), along with
1
0
2
ꢀ1
[a]
H
Table 1. D [ꢃ10 m s ] and r [ꢄ] values for LiX in THF at ambient
temperature.
[
b]
[c]
[d]
7
Nucleus
D
r
H
r
d ( Li)
Dn1/2
7
LiCl
LiBr
LiI
Li
Li
Li
11.1
11.1
11.2
15.3
15.3
4.3
4.3
4.3
3.1
3.1
4.8
3.2
3.2
4.2
4.3
4.4
4.7
4.7
4.7
3.0
0.48
0.59
0.48
2.2
2.4
2.2
7
7
3
1
7
1
3
7
7
7
1
HMPA
P
H
Li
H
LiCl + HMPA
9.87
0.36
3.8
14.9
14.9
11.4
11.1
10.8
3.0
1
P
LiCl (20 mm)
LiCl (60 mm)
LiCl (100 mm)
Li
Li
Li
[
a] 60 mm solutions, unless otherwise stated. [b] Experimental error is
ꢀ10
2 ꢀ1
about ꢁ2%; about ꢁ0.06 10 m s . [c] Standard deviation is about
0.1 ꢄ. [d] Estimated by using Chem3D, by averaging the distances be-
tween the centroid and the outer hydrogen. (THF, 299 K)
ꢁ
h
=
ꢀ
1
ꢀ1
0
.461 Kgs m .
Figure 1, which show the D values and diffusion data for
these, reveal that the cations of all three salts afford identi-
cal D values within the experimental error. The observed
7
+
Li chemical shifts are in agreement with ionic Li and the
7
+
narrow Li line widths suggest a symmetric environment
for this metal cation in all three salts. Clearly, the tetrasolvat-
ed cation is the most stable in THF. The hydrodynamic radii
rH, shown in Table 1, were obtained by using the experimen-
3
5
Figure 2. Cl NMR spectra (48.99 MHz) at 299 K for 60 mm solutions of
a) LiClO in D O, d = 962.5 ppm, b) LiCl in D
O, d = ꢀ45.0, and c)
LiCl in [D
]THF d = ꢀ93.2 ppm. General conditions: spectral width =
19596 Hz; repetition delay = 100 ms; 1024 scans; 12 min per spectrum.
4
2
2
8
1496
ꢂ 2005 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
Chem. Eur. J. 2005, 11, 1495 – 1506

PGSE Diffusion NMR Spectroscopy and Ion Pairing in Lithium Salts
FULL PAPER
two model measurements for 60 mm solutions of Li(ClO4)
For the LiCl sample with 12 equivalents of HMPA at
7
and LiCl in water. The marked increase, relative to the
155 K, Figure 3 shows two Li signals, d
=
0.14 and
3
5
aqueous model salts, of the Cl line width in THF to 362 Hz
suggests that the chloride is no longer in a symmetric envi-
ronment, perhaps owing to a small percentage of ion pair-
ꢀ0.18 ppm. The low-frequency resonance appears as a re-
[
32]
ing.
Table 1 also shows the concentration dependence of the D
values for samples of LiCl in the 20–100 mm range. The de-
crease in the diffusion coefficient when the concentration is
increased (ca. 7.7%) arises from both aggregation and sol-
vent viscosity effects and is in the range observed by us pre-
[20,33]
viously.
HMPA is a highly polar, aprotic solvent that has been em-
[34,35]
ployed as an additive in lithium chemistry.
Its usefulness
stems from its ability to strongly coordinate the amide
+
oxygen atom to lithium to afford [Li(hmpa) ] . The various
n
7
1
31
Li, H, and P PGSE data for 60 mm solutions of lithium
chloride containing 12 equivalents of HMPA at ambient
temperature in THF are also given in Table 1. The agree-
1
31
ment between the H and P PGSE data provides a check
7
on the reliability of the Li results in that these nuclei reside
within the same cation, assuming HMPA complexation.
The coordination of the larger HMPA, in place of THF,
to the lithium center slows the translation of the lithium
7
Figure 3. Variable-temperature Li NMR of a 60 mm LiCl solution in
THF with 12 equivalents of HMPA added (155.45 MHz).
7
cation and produces a decrease in the Li D value (DD =
ꢀ
10
2
ꢀ1
2
ꢀ
1.23ꢃ10 m s ). Because of equilibrium effects, the
solved quintet ( J(P, Li) = 7.6 Hz) owing to the spin-spin
coupling of the four P atoms of the coordinated HMPA,
and thus can be assigned to the [Li(hmpa)4] ion 1. The re-
3
1
movement of the HMPA is slowed as well. As expected, the
H, Li, and P NMR spectra all show single resonances,
1
7
31
+
owing to facile ligand exchange at ambient temperature.
To suppress the dynamics, a 60 mm solution of LiCl in
THF (without HMPA) was cooled to 155 K and studied. De-
creasing the temperature had only a small effect on the ap-
maining signal arises from solvated LiCl. Below 173 K, the
H NMR spectrum of this sample reveals two methyl reso-
1
nances at d = 2.74 and 2.68 ppm, with the higher frequency
3
1
signal corresponding to coordinated HMPA. The P NMR
7
2
pearance of the 1D Li NMR spectra for the lithium chlo-
spectrum shows a new signal at d = 27.36 ppm ( J
=
P, Li
7
+
ride/THF sample. The observed Li line width at 155 K,
7.6 Hz), which is assigned to the [Li(hmpa) ] ion 1. The
4
3
1
31
Dn1/2 = 4.1 Hz, is only slightly larger than that found at
corresponding P, P 2D exchange spectrum (Figure 4a)
clearly shows that, despite the observed multiplicity on the
high-frequency signal, the solvent HMPA is still exchanging
with the complexed HMPA at 155 K. Figure 4b shows an
299 K, Dn1/2 = 2.2 Hz. The D value at 155 K (see Table 2) is
much smaller, owing to the increased solution viscosity;
however, as this viscosity value has been estimated previous-
[
23]
1
31
ly,
equation,
changed relative to the ambient-temperature diffusion data.
the radius r , calculated from the Stokes–Einstein
H, P gHMQC spectrum, which permits the assignment of
the methyl groups in the complexed HMPA.
H
[30]
indicates that the lithium environment is un-
1
7
31
Table 2 also provides data from the H, Li, and P PGSE
measurements at 155 K for the HMPA sample, as well as
data for the 60 mm reference solution of HMPA. From these
1
0
2
ꢀ1
[a]
H
Table 2. D [ꢃ10 m s ] and r [ꢄ] values in THF at 155 K.
low-temperature data we calculate an r value of about
H
[b]
[c]
[d]
7
+
Nucleus
D
r
H
r
d ( Li)
Dv1/2
7.7 ꢄ in THF for the [Li(hmpa) ] ion. The solid-state struc-
4
7
+
[36]
LiCl
HMPA
Li
H
P
0.252
0.351
0.350
4.3
3.1
3.1
4.7
3.0
0.55
4.1
ture of [Li(hmpa)4] has been described. From these X-
ray data one can estimate the rotational radius
1
3
[37]
to be
1
about 6.7 ꢄ. This discrepancy is to be expected, as the solid-
state r value does not involve a solvent shell. In any case the
PGSE results confirm the simple mononuclear cationic
structure in solution.
LiCl + HMPA
7
1
3
7
1
3
[e]
Solvated LiCl
Li
0.14
14.6
17.6
[f]
H
P
0.351
0.352
0.142
0.142
0.140
3.1
1
[f]
3.1
7.7
7.7
7.8
+
[
Li(hmpa)
4
] , 1
Li
H
6.7
ꢀ0.18
1
Lithium triphenylmethane (2): Lithium triphenylmethane,
P
LiCPh 2, has been studied previously by UV spectrosco-
3
[
a] 60 mm solutions. [b] Experimental error is about ꢁ2%. [c] Standard
[38]
[39]
[40]
py, various NMR methods, and X-ray crystallography.
deviation is about ꢁ0.1 ꢄ. [d] Estimated by using Chem3D, by averaging
These studies demonstrate that the LiCPh salt should be
the distances between the centroid and the outer hydrogen. [e] Not ob-
3
ꢀ3
ꢀ1
ꢀ1
tained. [f] Free HMPA. h (THF, 155 K) = 10.431ꢃ10 Kgs
m
.
considered as a solvent-separated salt in solution, and that
Chem. Eur. J. 2005, 11, 1495 – 1506
www.chemeurj.org
ꢂ 2005 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
1497

P. S. Pregosin et al.
[
39b]
13
et al.
of the anionic carbon, d
0.1 ppm, is found at higher fre-
The C chemical shift
=
9
3
quency than that of the sp -hy-
bridized precursor,
7.3 ppm, in analogy with the
known C data for the ipso-
carbon atom of phenyllithium
and other metalated phenyl de-
d
=
5
1
3
[43]
[44]
rivatives.
The low-frequency
shift of C-4 is consistent with
delocalization of some negative
charge into the phenyl rings.
Tables 4 and
5 show the
PGSE results for 60 mm solu-
3
1
31
Figure 4. a) 2D P, P EXSY NMR (161.92 MHz) and b) 2D gHMQC (400.13 MHz) of the 60 mm sample LiCl tions of LiCPh and HCPh in
3
3
+
4
plus HMPA in THF at 155 K. The [Li(hmpa) ] ion is indicated as 1.
THF at both ambient tempera-
ture and 155 K. The equiva-
lence of the experimental D
decreasing the temperature generally favors the separated
values for both the cation and anion in LiCPh at 299 K in-
3
[41]
[42]
7
1
ions.
However, Bauer and Lochmann
report a Li, H
dicates strong ion pairing. Moreover, the calculated r value
H
HOESY contact for this salt (0.35m) in THF at room tem-
of 6.4 ꢄ is much larger than that expected for either the iso-
lated [Li(thf)4] ion or the triphenylmethide anion. Our
PGSE data do not allow us to distinguish between a contact
ion pair (CIP) with, perhaps, fewer than four THF solvent
molecules, and a solvent-separated ion pair (SSIP). Howev-
+
perature, which implies some (perhaps only brief) contact.
1
13
Table 3 shows our H and C NMR data for LiCPh at
3
2
99 K in THF. These data are in good agreement with those
[39a]
reported by Sandel and Freedam,
and later by Jackman
1
13
[a]
3 9 9 7 4 7 2
Table 3. H and C NMR data for LiCPh , LiC13H , Li(2-Ph)C H , and LiC H S in THF at 299 K.
1
13
Site
H
J
H,H
C
3
LiCPh 2
1
2
3
4
5
–
–
–
–
90.1 (57.3)
149.7 (144.6)
123.8 (129.7)
127.5 (128.4)
112.5 (126.4)
7.40 (7.20)
6.60 (7.34)
6.04 (7.26)
8.6, 1.2
8.6, 7.0
7.0, 1.2
9
LiC13H 9
1
2
3
4
4
8
9
, 8
, 7
, 6
, 5
a
7.37 (7.62)
6.87 (7.35)
6.50 (7.43)
7.98 (7.90)
–
8.0, 0.9
8.0, 6.6, 1.1
7.7, 6.6, 0.9
7.7, 1.1
–
116.1 (125.2)
119.0 (126.9)
108.1 (126.9)
118.6 (120.0)
123.2 (143.6)
137.7 (142.2)
65.8 (36.9)
a
–
–
5.98 (3.97)
9 7
Li(2-Ph)C H 10
1
2
4
5
3
8
9
1
1
, 3
6.38 (3.91, 7.40)
–
7.30 (7.44, 7.55)
6.44 (7.31, 7.23)
–
88.2 (39.0, 126.6)
128.6 (136.5)
117.2 (121.0, 123.7)
111.58 (126.7, 124.9)
129.5 (145.8, 143.6)
141.2 (146.8)
128.6 (125.8)
129.5 (128.8)
120.5 (127.6)
–
, 7
, 6
a, 7a
–
–
–
7.79 (7.80)
7.22 (7.45)
6.93 (7.33)
8.4, 1.4
8.4, 7.2
7.2, 1.4
0
1
4 7 2
LiC H S 13
2
4
5
2.75 (3.79)
2.33 (2.81)
2.07 (2.03)
25.5 (27.1)
33.6 (31.7)
30.8 (29.9)
,6
[
a] The spectral data for the neutral precursor are given in parentheses.
1498
ꢂ 2005 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
Chem. Eur. J. 2005, 11, 1495 – 1506

PGSE Diffusion NMR Spectroscopy and Ion Pairing in Lithium Salts
FULL PAPER
1
0
2
ꢀ1
[a]
Table 4. D [ꢃ10 m s ] and r
H
[ꢄ] values in THF at 299 K.
changes by less than 1 ppm be-
tween 298 and 155 K. Further,
the absence of a detectable
[
b]
[c]
[d]
7
Nucleus
D
r
H
r
d ( Li)
Dn1/2
7
LiCPh
3
2
Li
7.40
7.38
6.4
6.4
ꢀ0.45
10.2
1
13
7
H
C, Li coupling constant over
the whole range of tempera-
tures strongly suggests an ionic,
as opposed to a covalent, inter-
1
HCPh
LiFlu 9
3
H
12.5
3.8
5.4
5.4
2.7
5.3
5.2
3.0
2.4
4.2
4.2
5.0
5.0
2.8
4.3
4.0
3.2
7
Li
8.85
8.81
17.44
8.89
9.16
15.96
20.17
11.4
11.3
9.51
9.55
16.7
10.9
ꢀ1.44
ꢀ3.24
13.8
5.0
1
H
1
HFlu
LiInd 10
H
action between these two
7
7
Li
atoms. The Li line width, Dn1
/2
1
H
=
2.6 Hz, is even smaller than
1
HInd
HDith
LiHMDS 16
H
3.9
2.6
1
that observed at ambient tem-
perature (Dn1/2 = 10.2 Hz), de-
spite the more viscous solvent,
which suggests tetrahedral sym-
metry at the lithium atom.
H
7
Li
0.78
1.10
11.0
32.6
1
6
0 mm
LiHMDS 16
00 mm
H
[
e]
7
Li
1
6
H
HMDS
TMSS
3.2
4.4
1
7
The H and Li PGSE data at
55 K show differing rates of
1
[
a] 60 mm solutions unless otherwise noted. [b] Experimental error is about ꢁ2%. [c] Standard deviation is
about ꢁ0.1 ꢄ. [d] Estimated by using Chem3D, by averaging the distances between the centroid and the outer
hydrogen. [e] Measurements obtained in a 0.6m sample using a coaxial NMR tube (ID = 1.96 mm; OD = triphenylmethide moieties and
translation for the lithium and
ꢀ
3
ꢀ1
ꢀ1
2
.97 mm) separated by a spacer. h (THF, 299 K) = 0.461ꢃ10 Kgs
m
.
thus smaller and different rH
values of 4.9 and 4.2 ꢄ, respec-
tively. These results imply little
or no interaction between these
ions at this temperature. The
cation and anion are translating
independently (thus explaining
the smaller line width of the
lithium signal) with the 4.9 ꢄ
value in good agreement with
1
0
2
ꢀ1
[a]
H
Table 5. D [ꢃ10 m s ] and r [ꢄ] values in THF at 155 K.
[
b]
[c]
[d]
7
Nucleus
D
r
H
r
d ( Li)
Dn1/2
7
LiCPh
HCPh
3
2
Li
H
H
Li
H
H
Li
H
0.224
0.258
0.282
0.223
0.253
0.396
5.23
4.9
4.2
3.9
4.9
4.3
2.8
4.4
4.4
2.4
ꢀ0.41
ꢀ1.13
0.11
2.6
1
1
3
4.0
7
LiFlu 9
2.8
3.6
1
our
expectation
for
the
1
HFlu
+
[
e]
7
[Li(thf)4] ion. It would appear
that, at low temperature in
THF solution, mobile ions are
favored, whereas at ambient
temperature the ions pair
strongly. There is no tempera-
ture at which covalent bonding
between the Li and the anionic
carbon is preferred. We shall
offer an explanation for this
temperature dependence after
discussing all of the remaining
salts.
LiDith 13
52 K
1
2
5.22
9.31
[
e]
1
HDith
52 K
H
2.6
2.6
2
7
[f]
LiDith 13
Li
H
H
0.16
19.5
1
0.245
0.447
4.5
2.4
1
HDith
LiHMDS 16
mononuclear
[
e,g]
7
Li
H
Li
H
5.14
5.17
4.32
4.26
7.83
4.3
4.3
5.3
5.3
2.8
0.66
1.33
8.5
9.8
1
2
50 K
dinuclear
50 K
7
1
2
[
e,g]
1
HMDS
50 K
TMSS
H
3.2
4.4
2
1
H
0.233
4.3
[
a] 60 mm solutions unless otherwise noted. [b] Experimental error is about ꢁ2%. [c] Standard deviation is
We have found three X-ray
about ꢁ0.1 ꢄ. [d] Estimated by using Chem3D, by averaging the distances between the centroid and the outer
structures in which a Li(CPh)
3
hydrogen. [e] A coaxial NMR tube (ID = 1.96 mm; OD = 2.97 mm) separated by a spacer was used. [f] Not
[40]
ꢀ
3
ꢀ1
ꢀ1
ꢀ3
ꢀ1
ꢀ1
moiety is involved.
Com-
obtained. [g] 0.60m; h (THF, 155 K) = 10.431ꢃ10 Kgs
THF, 252 K) = 0.801ꢃ10 Kgs
m ; h (THF, 250 K) = 0.825ꢃ10 Kgs m ; h
ꢀ
3
ꢀ1
ꢀ1
(
m
.
pounds 3 and 4 were crystal-
lized from Et O and n-hexane
2
solution, respectively. Complex
5
was prepared from THF using
7
er, we note that the Li line width is only slightly broader
(
two moles of crown ether per mole of triphenyl derivative.
The solid-state structure of LiCPh prepared from THF is
+
ꢀ
Dn1/2 = 10.2 Hz) than that of the [Li(thf)4] ion as a Cl
3
salt (Dn1/2 = 2.2 Hz), suggesting that the local symmetry at
the lithium ion has not changed drastically.
not known.
To complement the diffusion data for LiCPh , we allowed
3
The low-temperature NMR measurements for 2 were
both informative and surprising. In contrast to many organo-
HCPh to react with nBuLi in THF at ꢀ308C for 60 min.
3
Addition of n-pentane to the red solution, followed by stor-
1
3
lithium compounds, the methanide
C
chemical shift
age overnight at RT, afforded air-sensitive crystals of
Chem. Eur. J. 2005, 11, 1495 – 1506
www.chemeurj.org
ꢂ 2005 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
1499

P. S. Pregosin et al.
allows delocalization of the negative charge into the adja-
[46]
cent rings. This feature is observed in almost all carban-
ionic moieties attached to heterocycles or phenyl rings re-
ꢀ
[40b,47]
ꢀ
[48]
ported so far in the literature (e.g., Ph C ,
Ph pyC ,
3
2
ꢀ
[49]
or py CH , with py = pyridyl). The planes of the phenyl
2
rings deviate from the ideal planar alignment (~22.28 for
Ph1 and Ph1’ and ~26.48 for Ph2). Consequently, the carb-
anion reveals a propeller-like structure. The average of the
three C1ꢀC
bond lengths, about 1.45 ꢄ, is in excellent
agreement with those values found in the structure of 5,
ipso
1
.45 ꢄ, in which all three separations are identical.
[
{Li(THF) }(CPh )] 2 suitable for X-ray diffraction (see
The structure of [(Li([12]crown-4) ](CPh ) (5), can also be
4
3
2
3
[
Eq. (1)]). Compound 2 is quite stable for several weeks in
thought of as a separated ion pair in which the average
angle of the propeller ring conformation is 31.28, similar to
that found for 2. Consequently, the diffusion data at 155 K,
together with the X-ray structure for 2, support separated
ions.
the absence of air or moisture.
Fluorenyllithium (9): Fluorenyllithium has been extensively
[3a,50]
[3a,51]
studied. Ultraviolet-visible
and conductivity
mea-
surements suggest that this salt forms a solvent-separated
ion pair (SSIP) in THF at temperatures below ꢀ308C, and is
predominantly ion paired when the temperature is above
1
3
2
58C. These findings were corroborated by Edlund using C
[52]
[53]
NMR spectroscopy and later on by Schleyer and Bauer
by using a Li, H HOESY NMR approach. Only five X-ray
examples of fluorenyllithium (9) have been reported. Of
[45]
6
1
The X-ray crystal structure determination
of
2
[54]
(
Figure 5) supports the description of the proposed solvent-
separated ion pair (SSIP) in which one half of the “free”
these, the most relevant to the present discussion are the di-
ꢀ
+
[54c]
Ph C carbanion, as well as the cationic [Li(thf)4] moiety,
ethyl ether complex, [C H Li(Et O) ]
(6), the quinucli-
3
13
9
2
2
[54a]
are generated by crystallographic twofold symmetry. The
dine salt, [C H Li(NC H ) ]
(7), and the diglyme com-
1
3
9
7
13 2
ꢀ
[54e]
main structural feature in the carbanion Ph C is the pres-
plex [Li(C H )(diglyme) ] (8).
Salts 6 and 7 are thought
3
13
9
2
ence of a central trigonal-planar carbon atom (C1), which
to be CIPs, whereas 8 has been characterized as an SSIP.
Fluorenyllithium 9 in THF solution could be prepared in
1
3
1
an analogous fashion to that used for 2. C and H NMR
1
3
data for 9 are also shown in Table 3. In the C NMR spec-
trum the anionic carbon resonance is found at d
5.8 ppm. This represents a high-frequency shift relative to
=
6
1
3
1
the neutral precursor at d = 36.9 ppm. These C and H
Figure 5. Molecular
structure
and
numbering
scheme
for
[
{Li(thf) }(CPh )] (2). The thermal ellipsoids are drawn at a 30% proba-
4
3
NMR data are in complete agreement with those described
[52,55]
bility level. Hydrogen atoms of the THF molecules have been omitted
for clarity. Selected bond lengths [pm] and angles [8]: C1ꢀC10 144.2(2),
C1ꢀC20 147.2(3), Li1ꢀO1 189.7(3), Li1ꢀO2 190.3(4); C11-C10-C15
in the literature.
1
7
The H and Li PGSE results for a 60 mm solution of 9 in
THF at ambient temperature are given in Table 4. The
equivalent D values for both the anion and cation point to
1
1
2
13.9(2), C21-C20-C21’ 115.3(2), C10-C1-C10’ 123.7(2), C10-C1-C20
18.1(1); (equivalent atoms generated by ꢀx+1, ꢀy+2, z and ꢀx, ꢀy+
7
, z).
an ion pair in THF solution. The Li line width of 13.8 Hz is,
1500
ꢂ 2005 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
Chem. Eur. J. 2005, 11, 1495 – 1506

PGSE Diffusion NMR Spectroscopy and Ion Pairing in Lithium Salts
FULL PAPER
once again, slightly broader, in agreement with our observa-
tions for 2. From the X-ray structure of [C H Li(Et O) ] 6,
1
3
9
2
2
one can estimate a rotational radius of 4.9 ꢄ. Therefore, the
.4 ꢄ rH value based on the measured D value is satisfacto-
5
ry. Consequently, we suggest that the fluorenyllithium (9)
exists as a mononuclear ion pair at 299 K in THF. The D
and rH values for a 60 mm sample of the neutral precursor
(
HFlu) in THF are also shown. Diffusion data from a so-
lution of 9 at 155 K gave rH values of 4.9 ꢄ and 4.3 ꢄ for
the cation and anion respectively. These data point again to
well-separated ions (Dn1/2 = 2.8 Hz) under these conditions.
We assume that the increase in the size of the anion relative
to the rH value for fluorene (HFlu) is related to a solvent
shell.
values estimated from the x-ray structures of 11 and 12 (4.6
and 5.0 ꢄ, respectively). Consequently, we suggest that the
(2-phenyl)indenyllithium species 10 exists as a mononuclear
ion pair in THF. The D and r values for a 60 mm sample of
H
the neutral precursor (HInd) in THF are also shown in
Table 4.
Crystals of fluorenyllithium were obtained by treatment
of fluorene with nBuLi in THF at ꢀ608C for 10 min and
then subsequent layering of the orange solution with n-pen-
tane at ꢀ408C for 12 h (see [Eq. (2)]). Diffraction studies on
these air-sensitive, low-melting (m.p. ca. ꢀ308C) crystals
support the formulation [C H Li(thf) ]; however, apart
2-Lithio-1,3-dithiane (13): There is considerable evidence
3
that the anionic carbon of 2-lithio-1,3-dithiane (13) is sp -hy-
bridized and very localized. This conclusion is supported by
1
3
the low-frequency C chemical shift of the lithiated carbon
1
3
9
4
from determining the unit cell and recognizing the
(d = 25.8 ppm) relative to the neutral precursor (d =
+
[60,61]
13
[Li(thf)4] ion, the data could not be satisfactorily refined
31.7 ppm).
Low-temperature (173 K) C NMR mea-
6
13
owing to disorder.
surements on 2-( Li)lithio-2-( C)-dithiane, reported by See-
bach and coworkers reveal a triplet, J( Li, C) = 10.1 Hz,
[61]
6
13
6
for C-2. As this multiplicity implies a single Li atom, 13
must be either a mononuclear salt or a dinuclear aggregate
with the two components held together by an SꢀLiꢀC
[62]
bridge. Cryoscopic measurements
in THF confirm that
the 2-methyl derivative of 13 is mononuclear. The HMPA
titration technique has been applied to 13 and reveals a con-
[63]
tact ion pair (CIP).
X-ray crystal structures for the lithium-bridged dinuclear
[
64]
2
-methyl-2-lithiodithiane TMEDA (14) and the mononu-
[65]
(
2-Phenyl)indenyllithium (10): The indenyl anion represents
clear salt, [(2-phenyl-2-lithiodithiane)(tmeda)(thf)] (15)
yet another interesting model system for solvation and ion-
pairing studies because of its rigidity and thermal stability.
Its lithium salts have been extensively investigated by UV-
have been described previously.
[56]
[39b,52a,57]
[58]
visible,
NMR,
and X-ray methods.
Table 3
13
1
shows our C and H NMR data for the 2-phenyl derivative
1
13
10. The H and the C spectra show six and nine individual
resonances, respectively, as would be expected from a struc-
1
3
ture possessing a mirror plane. In the C spectrum the ob-
served C-2 (d = 128.6 ppm) and C-1,3 (d = 88.2 ppm) reso-
nances (see Table 3) are in accord with what is found for
3
[59]
other h -C H derivatives.
9
7
Only two X-ray structures of indenyllithium derivatives
have been reported. Both were described as contact ion
To the best of our knowledge the THF solution structure
of 13 is as yet unknown. Salt 13 is not stable in THF at
[58]
[58a]
13
pairs.
For the TMEDA salt [C H Li(N C H )] (11),
299 K but is stable at 252 K. Our C data (d anionic CH =
9
7
2
6
16
1
13
1
the lithium atom is coordinated to one bidentate TMEDA
molecule and to the indenyl group with an equal distance of
25.5 ppm, J( C, H) at 173 K = 132.5 Hz) are in agreement
[61]
with those reported by Seebach et al.. Variable-tempera-
[58b]
7
2.377 ꢄ. On the other hand, the salt 12,
crystallized in
ture Li spectra reveal a significant temperature dependence
THF, shows a lithium cation coordinated by three molecules
of THF.
Table 4 shows D and rH values for 10 at 299 K. The rH
values for cation and anion, 5.3 ꢄ and 5.2 ꢄ, respectively,
are almost identical. The rH value of 5.3 ꢄ, based on the
measured D value, is satisfactory when compared to the r
of the line width (Figure 6) with Dn at 173 K = 15.3 Hz
1
/2
[66]
relative to 3.6 Hz value at 253 K.
Table 5 gives diffusion data for 13 in THF at 252 K and re-
veals that ion pairing persists at this temperature (r
=
H
[30]
7
4.4 ꢄ for both ions ). At 155 K the Li resonance is even
broader than at 173 K, Dn1/2 = 19.5 Hz, and the longitudinal
Chem. Eur. J. 2005, 11, 1495 – 1506
www.chemeurj.org
ꢂ 2005 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
1501

P. S. Pregosin et al.
0
.060m and 0.60m were prepared. At room temperature the
1
H NMR of a 0.060m solution of Li[N(SiMe ) ] in THF con-
3
2
7
sists of a sharp singlet at d = ꢀ0.09 ppm, whereas in the Li
NMR spectrum a single resonance with a line width of Dn1
/2
=
11.0 Hz at d=1.10 ppm is found. For the 0.6m sample,
the observed lithium resonance was considerably broader,
1
7
Dn1/2 = 32.6 Hz (see Table 4). H and Li PGSE diffusion
data for the two concentrations, 0.060m and 0.60m, at 299 K
are given in Table 4.
When the temperature for the 0.6m sample is decreased
1
(
see Figure 7), two distinct species are observed in both H
7
Figure 6. Variable-temperature Li NMR spectra of a 60 mm solution of
7
lithium-1,3-dithiane 13 in THF (the Li line widths Dn1/2 are indicated).
relaxation time, T1 = 32 ms, is rather short. Consequently,
1
only the H D value was obtained. Nevertheless, the rH
value of 4.5 ꢄ at 155 K, is similar to that found at 252 K and
suggests that the two ions remain close. This r of 4.5 ꢄ, al-
H
[30]
though smaller than the 5.0 ꢄ estimated for 15, suggests
that, in THF solution, the 2-lithio-1,3-dithiane 13 exists as a
mononuclear species. The D and rH values for a 60 mm
sample of the neutral precursor (HDith) in THF are also
shown in Tables 4 and 5.
7
Figure 7. Variable-temperature Li NMR spectra of a 0.6m solution of
LiHMDS 16 in THF (155.45 MHz).
Lithium hexamethyldisilazide Li[N(SiMe ) ]: It was report-
3
2
ed early on that Li[N(SiMe ) ] (16) demonstrates a mononu-
3
2
[67]
7
1
clear/dinuclear equilibrium in THF.
using Li and N-labeled Li[N(SiMe ) ],
Subsequent studies
and Li NMR spectra. Integration of the H NMR spectrum
6
15
[68]
1
7
confirmed the
afforded a mononuclear/dinuclear ratio of 44:56. H and Li
PGSE measurements at 250 K,
3
2
[71]
earlier conclusions. Solid-state structures of Li[N(SiMe ) ]
a temperature at which
3
2
aggregates containing oxygen or nitrogen polydentate li-
gands reveal the existence of both mononuclear and dinu-
clear solvated species. The most relevant for our discussion
are the mononuclear structures 17 and 18, containing
TMEDA and N,N,N’,N’’,N’’-pentamethyldiethylenetriamine,
the two lithium resonances are relatively sharp, gave the dif-
fusion data shown in Table 5. Representative Li diffusion
data for the 0.6m sample are shown in Figure 8. From the
X-ray structure of model 19, we can estimate a radius of
7
about 5.6 ꢄ. We find a D value for the mononuclear salt
[69]
[31]
respectively, and the bridged mono-solvated dinuclear salt
that corresponds to a fairly small r value of 4.3 ꢄ. The D
H
[
70]
19.
value for the presumed dinuclear species affords an r value
H
of 5.3 ꢄ, which is much closer to the 5.6 ꢄ suggested from
the solid-state study on 19. To the best of our knowledge
7
this represents the first application of Li PGSE measure-
ments to the recognition of different lithium aggregates in
solution, in this case mononuclear and dinuclear lithium
amides.
Conclusion
The position of the equilibrium for [Li{N(SiMe )} ] in
THF is recognized to be a function of concentration. The
3
2
[67]
These various lithium salts nicely show that PGSE methods
are quite useful for recognizing ion pairing (or the lack
thereof) and aggregation in THF solution. For several of the
organolithium species it is now clear that ion pairing is fa-
observed mononuclear/dinuclear ratio for a 0.69m sample is
[67]
5
1:49; for a 0.069m solution, this ratio changes to 88:12.
For this reason, two samples of different concentration,
1502
ꢂ 2005 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
Chem. Eur. J. 2005, 11, 1495 – 1506

PGSE Diffusion NMR Spectroscopy and Ion Pairing in Lithium Salts
FULL PAPER
PGSE results allow the two different aggregation states to
be readily recognized and, where available, the solid-state
structures support the PGSE data. The D values for the
simple LiX salts reflect the relative stability of the
+
[
Li(thf)4] ion. Clearly, THF represents a “special” solvent
[6,20]
in that, in contrast to dichloromethane,
where partial ion
pairing is normal, THF shows a tendency to separate the
ions at low temperature.
Experimental Section
General: Glassware was dried overnight in a 1108C oven to remove
moisture. THF was freshly distilled from potassium before use. All chem-
icals used for sample preparation were obtained from Merck, Sigma–Al-
drich, or Fluka and were of reagent grade. HMPA was distilled from
2 2
CaH , and stored under N over molecular sieves. HMPA is a carcinogen-
ic agent, so adequate precautions were taken to avoid all forms of expo-
sure. All reactions and sample manipulations were carried out using stan-
dard Schlenk techniques under nitrogen. The NMR samples were pre-
pared in standard 5-mm NMR tubes, which were flushed with nitrogen,
oven dried, and fitted with a plastic cap. The outside top portion of the
tube was held securely by paraffin film and grease.
Lithium halides: [D
tubes containing the corresponding salt. The addition of HMPA (31.5 mL,
60 mmol) was performed on a 60 mm sample of LiCl (0.64 mg, 60 mmol)
placed in an N flushed Schlenk tube. To get the HMPA to dissolve, the
tube had to be repeatedly warmed slightly and shaken.
Organolithiums: LiCPh , LiC13 , Li(2-Ph)C , LiC
SiMe ] were obtained as air-sensitive species by adding 1.1 equivalent
of nBuLi (1.6m in n-hexane) to a cooled solution (N liquid/acetone) of
the neutral precursors HCPh , H(2-Ph)C , HC , and
3
HN(SiMe in [D ]THF solution. Lithium triphenylmethane (LiCPh )
8
]THF (0.5 mL) was added to oven-dried 5-mm NMR
3
2
3
H
9
9
H
7
4 7 2
H S , and Li[N-
(
3 2
)
2
3
, HC13H
9
9
H
7
4 7 2
H S
3
)
2
8
was generated by the addition of nBuLi (21 mL, 1.6m in n-hexane) to
oven-dried NMR tubes containing triphenylmethane (60 mmol, 7.33 mg)
and freshly distilled [D
generated a dark red solution upon mixing.
8
]THF (0.5 mL) cooled to ꢀ788C. The addition
The same protocol was applied for the remaining lithium species, which
became dark orange, green, colorless, and light yellow upon shaking, for
fluorenyllithium (LiC13
lithio-1,3-dithiane (LiC
LiHMDS), respectively.
H
9
), (2-phenyl)indenyl lithium (Li(2-Ph)C
9 7
H ), 2-
4
H S), and lithium bis(trimethylsilane) amide
7
(
NMR spectroscopy: All multinuclear room- and low-temperature experi-
ments were performed on a 400 MHz Bruker AVANCE spectrometer
equipped with a microprocessor-controlled gradient unit and an inverse
Figure 8. a) Plot of the ln(I/I
square of the gradient amplitude for H (anion) and Li (cation) PGSE
diffusion measurements on a 0.6m sample of LiHMDS (HMDS
o
) versus arbitrary units proportional to the
1
multinuclear probe with an actively shielded z axis gradient coil.
NMR spectra were referenced to the residual signal of [D
H
1
7
7
]THF at d =
.76 ppm as an internal standard. P and Li NMR chemical shifts are re-
=
31
7
1
1
7
N(SiMe
3
)
2
) in THF solution at 250 K. H (d = 3 ms; D = 69 ms); Li (d
31
ferred to external 85% H
3
PO
O for Li (155.454 MHz).
The PGSE NMR diffusion measurements were carried out using the
4
for P (161.923 MHz) and LiCl 9.7m in
=
11 ms; D = 28 ms); b) After correcting for the gyromagnetic ratio of
7
D
2
7
1
7
Li, D and d, the slopes observed for the H and Li measurements are
identical within the experimental error.
[
5,6]
stimulated echo pulse sequence.
The shape of the gradient pulse was
rectangular, and its strength varied automatically in the course of the ex-
periments. The D values were determined from the slope of the regres-
2
vored at 299 K but the ions are well separated at 155 K;
however for the dithianyl salt 13, the low-temperature data
show that the ions remain together. The explanation for the
observed temperature dependence of the ion pairing is not
completely clear; however, it is known that the dielectric
constant of THF increases markedly (more than doubles!)
on cooling from ambient temperature to 155 K. Consequent-
ly, we assign the increased stability of the separated ions at
sion line ln(I/I
o
) versus G , according to Equation (a). I/I
o
= observed
spin echo intensity/intensity without gradients, G = gradient strength, D
delay between the midpoints of the gradients, D = diffusion coeffi-
cient, d = gradient length.
=
[72]
ꢀ
ꢁ
ꢀ
ꢁ
I
d
2
2
ln
¼ ꢀðgdÞ G Dꢀ 3
D
ðaÞ
I
0
The measurements were carried out without spinning. The sample tem-
perature was calibrated before the PGSE measurements by introducing a
thermocouple inside the bore of the magnet. The calibration of the gradi-
155 K to a marked increase in the stability of the separated
ions due to increased solvation. For the lithium amide, the
Chem. Eur. J. 2005, 11, 1495 – 1506
www.chemeurj.org
ꢂ 2005 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
1503

P. S. Pregosin et al.
ents was carried out by means of a diffusion measurement of HDO in
ganometallics 1986, 5, 1851; d) W. Bauer, W. R. Winchester, P.
von R. Schleyer, Organometallics 1987, 6, 2371.
ꢀ
9
2
ꢀ1 [73]
ꢀ4
D
2
O (DHDO = 1.9ꢃ10 m s
We believe that our previously reported experimental error in D values,
2%, can be extended to the measurements performed here. All of the
)
, which afforded a slope of 2.022 10 ).
[3] a) Ions and Ion Pairs in Organic Reactions, Vol. 1 (Ed.: M. Szwarc),
Wiley, New York, 1972, Chapter 2; b) D. Seebach, Angew. Chem.
1988, 100, 1685; Angew. Chem. Int. Ed. Engl. 1988, 27, 1624; c) ion
pair structures have been proposed as a key feature controlling 1,2-
and 1,4- addition to enones: a) T. Cohen, W. D. Abraham, M.
Myers, J. Am. Chem. Soc. 1987, 109, 7923; b) W. H. Sikorsky, H. J.
Reich, J. Am. Chem. Soc. 2001, 123, 6527.
[4] C. S. Johnson Jr., Prog. Nucl. Magn. Reson. Spectrosc. 1999, 35,
203.
[5] P. Stilbs, Prog. Nucl. Magn. Reson. Spectrosc. 1987, 23, 1.
ꢁ
data leading to the reported D values afforded lines whose correlation
coefficients were above 0.999. To check the reproducibility, three differ-
ent measurements with different diffusion parameters (D and/or d) were
carried out. The gradient strength was incremented in 8% steps from
1
0% to 90%, so that 8–10 points could be used for regression analysis.
1
7
31
A measurement of H, Li, and P T
sion experiment, and the recovery delay set to (3–5)T
gives the Li T
1
was carried out before each diffu-
. Table 6 below
(ms) values for all the lithiated species of interest.
1
7
1
[
6] P. S. Pregosin, E. Martꢁnez-Viviente, P. G. A. Kumar, Dalton Trans.
003, 4007.
2
[
7] a) C. Zuccaccia, N. G. Stahl, A. Macchioni, M. C. Chen, J. A. Rob-
erts, T. J. Marks, J. Am. Chem. Soc. 2004, 126, 1448; b) D. E. Ba-
bushkin, H. H. Brintzinger, J. Am. Chem. Soc. 2002, 124, 12869.
8] A. Pichota, P. S. Pregosin, M. Valentini, M. Wçrle, D. Seebach,
Angew. Chem. 2000, 112, 153; Angew. Chem. Int. Ed. 2000, 39, 157.
9] C. B. Gorman, J. C. Smith, M. W. Hager, B. L. Parkhurst, H. Sierzpu-
towska-Gracz, C. A. Haneys, J. Am. Chem. Soc. 1999, 121, 9958.
7
Table 6. Li longitudinal relaxation times T
1
[ms] of lithiated species.
1
2
9
10
13
16
16
[
[
(60 mm) (600 mm)
T
T
T
1
1
1
(299 K)
(250 K)
–
–
710
–
621 1200
–
150
112
–
–
118
–
–
79
30
49
16
(155 K) 498 1680 1590 1630 32
[
[
10] B. Olenyuk, M. D. Lovin, J. A. Whiteford, P. J. J. Stang, J. Am.
Chem. Soc. 1999, 121, 10434.
11] D. L. Reger, J. R. Gardinier, P. J. Pellechia, M. D. Smith, K. J.
Brown, Inorg. Chem. 2003, 42, 7635.
Details for the PGSE measurements:
1
H PGSE diffusion measurements: Room-temperature diffusion parame-
ters: D = 19–68 ms, d = 2–3 ms; at low temperature: D = 27–68 ms, d
3–12 ms. The number of scans varied between 8 and 16 per increment.
[12] K. Hayamizu, W. S. Price, J. Magn. Reson. 2004, 167, 328.
[13] a) M. Valentini, P. S. Pregosin, H. Rꢆegger, Organometallics 2000,
19, 2551; b) D. Drago, P. S. Pregosin, A. Pfaltz, Chem. Commun.
2002, 286; c) P. S. Pregosin, E. Martꢁnez-Viviente, P. G. A. Kumar,
Dalton Trans. 2003, 4007; d) E. Martꢁnez-Viviente, P. S. Pregosin,
Inorg. Chem. 2003, 42, 2209; e) P. G. A. Kumar, P. S. Pregosin, J. M.
Goicoechea, M. K. Whittlesey, Organometallics 2003, 22, 2956.
[14] a) M. Valentini, P. S. Pregosin, H. Rꢆegger, J. Chem. Soc. Dalton
Trans. 2000, 4507; b) E. Martꢁnez-Viviente, H. Rꢆegger, P. S. Prego-
sin, J. Lꢅpez-Serrano, Organometallics 2002, 21, 5841; c) Y. Chen,
M. Valentini, P. S. Pregosin, A. Albinati, Inorg. Chim. Acta 2002,
=
Typical experimental times were 0.5–1 h.
7
31
Li and P PGSE diffusion measurements: To obtain a suitable attenua-
tion for these nuclei, the diffusion delay D and/or the gradient strength d
were increased. With D values higher than 200 ms no signal could be ac-
quired. The most suitable option available was then a considerable incre-
ment of the gradient lengths (d):
3
1
P PGSE: At room temperature: D = 16–33 ms, d = 3–8 ms; at low
temperature: D = 38–60 ms, d = 19–24 ms. The number of scans varied
between 64 and 512 scans per increment. Typical experimental times
were 1–4 h.
3
27, 4; d) T. J. Geldbach, F. Breher, V. Gramlich, P. G. A. Kumar,
P. S. Pregosin, Inorg. Chem. 2004, 43, 1920.
7
[15] a) C. Zuccaccia, G. Bellachioma, G. Cardaci, A. Macchioni, Organo-
metallics 2000, 19, 4663; b) A. Burini, J. P. Fackler, R. Galassi, A.
Macchioni, M. A. Omary, M. A. Rawashdeh-Omary, B. R. Pietroni,
S. Sabatini, C. Zuccaccia, J. Am. Chem. Soc. 2002, 124, 4570;
c) N. G. Stahl, C. Zuccaccia, T. R. Jensen, T. J. Marks, J. Am. Chem.
Soc. 2003, 125, 5256.
Li PGSE: At room temperature: D = 18–95 ms, d = 4–6 ms; at low
temperature: d = 28–77 ms, d = 10–26 ms. The number of scans varied
between 64 and 200 scans per increment. Typical experimental times
were 1–4 h.
To avoid convection, the PGSE diffusion measurements at 250 K were
carried out using a commercial coaxial insert (ID = 1.96 mm, OD =
2
[
16] a) H. P. Mo, T. C. Pochapsky, J. Phys. Chem. B 1997, 101, 4485; b) S.
Beck, A. Geyer, H. H. Brintzinger, Chem. Commun. 1999, 2477;
c) Q. Jaing, H. Rꢆegger, L. M. Venanzi, Inorg. Chim. Acta 1999, 290,
.97 mm) inserted into a standard 5 mm NMR tube and held in a concen-
[
74]
tric manner with a spacer.
6
4; d) R. M. Stoop, S. Bachmann, M. Valentini, A. Mezzetti, Orga-
nometallics 2000, 19, 4117; e) N. E. Schlçrer, E. J. Cabrita, S. Berger,
Angew. Chem. 2002, 114, 114; Angew. Chem. Int. Ed. 2002, 41, 107.
17] a) A. Pfaltz, J. Blankenstein, R. Hilgraf, E. Hormann, S. McIntyre,
F. Menges, M. Schonleber, S. P. Smidt, B. Wꢆstenberg, N. Zimmer-
manns, Adv. Synth. Catal. 2003, 5, 33; b) D. Drago, P. S. Pregosin, A.
Pfaltz, Chem. Commun. 2002, 286.
18] E. P. Kꢆndig, C. M. Saudan, G. H. Bernardinelli, Angew. Chem.
1999, 111, 1298; Angew. Chem. Int. Ed. 1999, 38, 1220.
[19] D. Zuccaccia, S. Sabatini, G. Bellachiom, G. Cardaci, E. Clot, A.
[
[
Acknowledgement
P.S.P. thanks the Swiss National Science Foundation and the ETH Zurich
for financial support. I. F. thanks the Junta de Andalucꢁa for a research
contract, and E.M.-V. thanks the Seneca Foundation (Comunidad Autꢅn-
oma de la Regiꢅn de Murcia, Spain) for a grant. We thank Dr. Heinz
Rꢆegger for his experimental assistance and advice.
Macchionis, Inorg. Chem. 2003, 42, 5465.
[
20] M. Valentini, H. Rꢆegger, P. S. Pregosin, Helv. Chim. Acta 2001, 84,
833. If chemical equilibria are involved, there may be a marked
concentration dependence.
2
[
[
1] a) B. J. Wakefield, The Chemistry of Organolithium Compounds,
Pergamon, Oxford, 1974; b) W. Bauer, P. von R. Schleyer, Adv.
Carbanion Chem. 1992, 1, 89; c) A.-M. Sapse, P. von R. Schleyer,
[21] a) F. Breher, J. Grunenberg, S. C. Lawrence, P. Mountford, H. Rꢆeg-
ger, Angew. Chem. 2004, 116, 2775; Angew. Chem. Int. Ed. 2004, 43,
2521–2524; b) I. Keresztes, P. G. Williard, J. Am. Chem. Soc. 2000,
122, 10228–10229; c) M. Ginley, U. Henriksson, J. Colloid Interface
Sci. 1992, 150, 281; d) L. v. Dam, B. Andreasson, L. Nordenskiçld,
Chem. Phys. Lett. 1996, 262, 737; e) F. Hallwass, M. Engelsberg,
A. M. Simas, J. Phys. Chem. A 2002, 106, 589.
Lithium Chemistry:
Wiley, New York, 1995.
2] a) C. Strohmann, T. Seibel, K. Strohfeldts, Angew. Chem. 2003, 115,
669; Angew. Chem. Int. Ed. 2003, 42, 4531; b) T. Kottke, D. Stalke,
Angew. Chem. 1993, 105, 619; Angew. Chem. Int. Ed. Engl. 1993, 32,
80; c) R. D. Thomas, M. T. Clarke, R. M. Jensen, T. C. Young, Or-
A Theoretical and Experimental Overview,
4
5
1504
ꢂ 2005 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
Chem. Eur. J. 2005, 11, 1495 – 1506

PGSE Diffusion NMR Spectroscopy and Ion Pairing in Lithium Salts
FULL PAPER
[
[
22] a) K. Hayamizu, Y. Aihara, S. Arai, C. G. Martinez, J. Phys. Chem.
B 1999, 103, 519; b) K. Hayamizu, E. Akiba, J. Chem. Phys. 2002,
[40] a) J. J. Brooks, G. D. Stucky, J. Am. Chem. Soc. 1972, 94, 7333;
b) M. M. Olmstead, P. P. Power, J. Am. Chem. Soc. 1985, 107, 2174;
c) R. A. Barlett, H. V. R. Dias, P. P. Power, J. Organomet. Chem.
1988, 341, 1.
[41] a) T. E. Hogen-Esch, Adv. Phys. Org. Chem. 1977, 15, 153; b) T. E.
Hogen-Esch, J. Smid, J. Am. Chem. Soc. 1966, 88, 318.
[42] W. Bauer, L. Lochmann, J. Am. Chem. Soc. 1992, 114, 7482.
[43] D. Seebach, R. Hꢈssig, J. Gabriel, Helv. Chim. Acta 1983 66, 308; S.
Berger; U. Fleischer, C. Geletneky, J. C. W. Lohrenz, Chem. Ber.
1995 128, 1183–1186.
1
17, 5929.
23] I. Fernꢀndez, E. Martꢁnez-Viviente, P. S. Pregosin, Inorg. Chem.
004, 43, 4555.
2
[
[
24] M. K. Wong, A. Popov, I. J. Inorg. Nucl. Chem. 1972, 34, 3615.
25] D. Barr, W. Clegg, R. E. Mulvey, R. Snaith, J. Chem. Soc. Chem.
Commun. 1984, 79.
[
[
26] F. E. Hahn, S. Rupprecht, Z. Naturforsch B 1991, 46, 143.
27] H. J. Reich, J. P. Borst, R. R. Dykstra, D. P. Green, J. Am. Chem.
Soc. 1993, 115, 8728.
[44] M. Tschoerner, P. W. Kunz, P. S. Pregosin, Magn. Reson. Chem.
[
[
28] H. Nçth, R. Waldçr, Z. Naturforsch B 1998, 53, 1525.
1
999, 37, 91; E. Martꢁnez-Viviente, P. S. Pregosin, M. Tschoerners,
7
29] The sensitivity of Li is not sufficient to allow measurements on the
Magn. Reson. Chem. 2000, 38, 23; M. Tschoerner, P. S. Pregosin,
Inorg. Chim. Acta 1999, 290, 95.
1
1
–2 mm solutions used in our H NMR diffusion studies. The existing
7
Li PGSE studies were made on 60 mm solutions. The literature
often reports concentrations of >100 mm.
[
45] Crystal structure of 2: single crystals were obtained from THF at
4
room temperature; C35H47LiO , orthorhombic, space group Pnn2; a
=
=
[
30] The hydrodynamic radii r
stein equation: D = (k T)/(6phr), in which D is the diffusion coeffi-
cient, k is the Boltzman constant, T is the temperature in Kelvin, h
H
were calculated from the Stokes–Ein-
3
11.469(1), b = 13.747(1), c = 10.093(1) ꢄ; V = 1591.3(1) ꢄ ; Z
ꢀ3
B
2; 1calcd = 1.124 Mgm ; crystal dimensions 0.92ꢃ0.65ꢃ0.55 mm;
B
diffractometer Bruker SMART Apex; MoKa radiation, 200 K, 2Vmax
49.428; 12619 reflections, 2700 independent (Rint 0.0611),
direct methods; refinement against full-matrix (versus F ) with
SHELXTL (ver. 6.12) and SHELXL-97, 184 parameters, R1
.0498 and wR2 (all data) = 0.1426 max./min residual electron den-
is the viscosity of the solution. It has been suggested that the factor
c (=6 in the equation) is not valid for small species whose van der
Waals radii are <5 ꢄ (J. T. Edward, J. Chem. Educ. 1970, 47, 261).
This factor can be adjusted by using a semiempirical approach (see:
H.-C. Chen, S.-H. Chen, J. Phys. Chem. 1984, 88, 5118; P. J. Espino-
sa, J. G. de la Torre, J. Phys. Chem. 1987, 91, 3612) derived from the
microfriction theory proposed by Wirtz and co-workers (A. Gierer,
K. Wirtz, Z. Naturforsch. A 1953, 8, 522; A. Spernol, K. Wirtz, Z.
Naturforsch. 1953, 8, 532), in which c is expressed as a function of
=
=
2
=
0
ꢀ3
sity 0.196/ꢀ0.194 eꢄ . The hydrogen atoms were placed in ideal-
ized positions and included as riding atoms. CCDC-247992 contains
the supplementary crystallographic data for this paper. These data
can be obtained free of charge from The Cambridge Crystallograph-
ic Data Centre via www.ccdc.cam.ac.uk/data_request/cif. For fluore-
nyllithium: at 200 K: a = 9.236(2), b = 9.674(2), c = 17.239(3) ꢄ,
2
.234
the solute-to-solvent ratio of radii: c = 6/[1 + {0.695 (rsolv/r
To be consistent and facilitate comparisons we have used the
Stokes–Einstein equation as shown (c = 6), although we recognize
H
)
}].
3
a = 93.863(4), b = 103.927(5), g = 112.588(4)8, V = 1358 ꢄ .
[
46] Only one example of an unassociated alkyl lithium compound con-
taining a “free” pyramidal carbanion in solution and in the solid
state has been characterized so far: F. Breher, J. Grunenberg, S. C.
Lawrence, P. Mountford, H. Rꢆegger, Angew. Chem. 2004, 116,
that, for example, for the organic precursors such as HCPh
NH(SiMe , perhaps a smaller value would be better. For LiX in
THF at 299 K a correction (c = 5.3) is necessary.
3
or
3 2
)
[
[
31] C. L. Yaws, Chemical Properties Handbook, McGraw–Hill, New
York, 1999 (online: http//www.knovel.com).
32] A few per cent contribution from, for example, a 10 KHz Cl line
width, Dn1/2, due to a structure with strong ion pairing, would be suf-
ficient to account for the observed change in Dn1/2. Typical line
widths for covalently bound Cl are of the order of thousands of Hz,
2
575; Angew. Chem. Int. Ed. 2004, 43, 2521.
ꢀ
3
5
[47] Triphenylmethanide salts containing the “free” [Ph C] anion have
3
been reported in a) P. P. Power, Acc. Chem. Res. 1988, 21, 147;
b) J. S. Alexander, K. Ruhland-Senge, Angew. Chem. 2001, 113,
2
732; Angew. Chem. Int. Ed. 2001, 40, 2658; c) S. Harder, Chem.
Eur. J. 2002, 8, 3229, and references therein.
for example, 20 KHz for CCl
4
: see NMR and the Periodic Table
[
[
48] U. Pieper, D. Stalke, Organometallics 1993, 12, 1201.
49] a) H. Gornitzka, D. Stalke, Angew. Chem. 1994, 106, 695; Angew.
Chem. Int. Ed. Engl. 1994, 33, 693; b) H. Gornitzka, D. Stalke, Or-
ganometallics 1994, 13, 4398.
50] a) T. T. Hogen-Esch, J. Smid, J. Am. Chem. Soc. 1966, 88, 307;
b) H. W. Vos, C. MacLean, N. H. Velthorst, J. Chem. Soc. 1976, 72,
(Eds.: R. K. Harris, B. E. Man), Academic Press, London, 1978,
4
page 421; or 13 KHz for CCl dissolved in ether: see M. A. Fedotov,
O. L. Malkina, V. G. Malkin, Chem. Phys. Lett. 1996, 258, 330. The
presence of some ion pairing would not change the average volume
of the solvated salt very much.
[
[
33] The variation in D is probably due to a change in aggregation as
well as solvent viscosity. See: a) A. Chandra, B. Bagchi, J. Chem.
Phys. 2000, 113, 3226; b) H. Falkenhagen, Z. Phys. 1931, 32, 745.
34] W. H. Sikorski, H. J. Reich, J. Am. Chem. Soc. 2001, 123, 6527. We
noted that N,N’-dimethylpropylene urea (DMPU), as additive, is
preferable to HMPA on toxicity grounds: T. Mukhopadhyay, D. See-
bach, Helv. Chim. Acta 1982, 65, 385.
35] a) D. Barr, M. J. Doyle, R. E. Mulvey, P. R. Raithby, D. Reed, R.
Snaith, D. S. Wright, J. Chem. Soc. Chem. Commun. 1989, 318;
b) F. E. Romesburg, J. H. Gilchrist, A. T. Harrison, D. J. Fuller, D. B.
Collum, J. Am. Chem. Soc. 1991, 113, 5751.
36] D. Barr, W. Clegg, R. E. Mulvey, R. Snaith, J. Chem. Soc. Chem.
Commun. 1984, 226.
37] The rotational radius is the radius of the sphere determined by ro-
tating a molecular model or X-ray structure around its geometric
center: see: K. W. Mattison, U. Nobbmann, D. Dolak, American
Biotechnology Laboratory 2001, 19, 66.
6
3.
[
[
51] T. T. Hogen-Esch, J. Smid, J. Am. Chem. Soc. 1966, 88, 318.
52] a) U. Edlung, Org. Magn. Reson. 1979, 12, 661; b) D. Johnels, U.
Edlund, J. Am. Chem. Soc. 1990, 112, 1647; c) I. Sethson, B. Elias-
son, U. Edlund, Magn. Reson. Chem. 1991, 29, 1012.
53] D. Hoffmann, W. Bauer, P. von R. Schleyer, J. Chem. Soc. Chem.
Commun. 1990, 208.
[
[
[
[
54] a) J. J. Brooks, W. Rhine, G. D. Stucky, J. Am. Chem. Soc. 1972, 94,
7
339; b) M. Kçnemann, G. Erker, R. Frçhlich, E. U. Wꢆrthwein, J.
Am. Chem. Soc. 1997, 119, 11155; c) M. Hakasson, C. H. Ottosson,
A. Boman, D. Johnels, Organometallics 1998, 17, 1208; d) C. Uffing,
R. Kçppe, H. Schnçckel, Organometallics 1998, 17, 3512; e) S. Nean-
der, Kçrnich, F. Olbrich, J. Organomet. Chem. 2002, 656, 89.
[
[
[55] a) R. H. Cox, H. W. Terry, Jr., L. W. Harrison, J. Am. Chem. Soc.
1971, 93, 3297; b) D. H. OꢇBrien, C. R. Russel, A. J. Hart, J. Am.
Chem. Soc. 1979, 101, 633.
[
[
38] E. Buncel, B. Menon, J. Org. Chem. 1979, 44, 317.
39] a) V. R. Sandel, H. H. Freedam, J. Am. Chem. Soc. 1966, 88, 1272;
b) J. B. Grutzner, J. M. Lawlor, L. M. Jackman, J. Am. Chem. Soc.
[56] a) H. W. Vos, H. H. Blom, N. H. Velthorst, C. MacLean, J. Chem.
Soc. Perkin Trans. 2 1972, 635; b) H. W. Vos, C. MacLean, N. H.
Velthorst, J. Chem. Soc. Faraday Trans. 2 1976, 63.
1
972, 94, 2306; c) R. Waack, M. A. Doran, E. B. Baker, G. A. Olah,
[57] a) U. Edlund, Org. Magn. Reson. 1977, 9, 593; b) J. Giessen, C.
Goojier, C. MacLean, N. H. Velthorst, Chem. Phys. Lett. 1978, 55,
33; c) C. Gooijer, N. H. Velthorst, Org. Magn. Reson. 1979, 12, 1979;
J. Am. Chem. Soc. 1966, 88, 1272; d) D. H. OꢇBrien, C. R. Russell,
A. Hart, J. Am. Chem. Soc. 1979, 101, 633.
Chem. Eur. J. 2005, 11, 1495 – 1506
www.chemeurj.org
ꢂ 2005 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
1505

P. S. Pregosin et al.
c) B. Eliasson, T. Lejon, U. Edlund, R. L. Mueller, S. W. Staley, J.
Phys. Org. Chem. 1990, 3, 9.
58] a) W. E. Rhine, G. D. Stucky, J. Am. Chem. Soc. 1975, 97, 737; b) I.
Hoppe, M. Marsch, K. Harms, G. Boche, D. Hoppe, Angew. Chem.
overlap with the THF resonance. We believe that the increased line
broadening at low temperature stems from the spin-spin coupling.
[67] B. Y. Kimura, T. L. Brown, J. Organomet. Chem. 1970, 26, 57.
[68] a) F. E. Romesberg, M. P. Bernstein, J. H. Gilchrist, A. T. Harrison,
D. J. Fuller, D. B. Collum, J. Am. Chem. Soc. 1993, 115, 3475;
b) B. L. Lucht, D. B. Collum, J. Am. Chem. Soc. 1994, 116, 6009;
c) B. L. Lucht, D. B. Collum, Acc. Chem. Res. 1999, 32, 1035, and
references therein.
[
[
1
995, 107, 2328; Angew. Chem. Int. Ed. Engl. 1995, 34, 2158.
59] a) M. C. Chisholm, M. J. Hampden-Smith, J. C. Huffman, J. D.
Martin, K. A. Stahl, Polyhedron 1988, 7, 1991; b) M. L. H. Green,
P. K. Konidaris, D. M. Michaellidou, P. Mountford, J. Chem. Soc.
Dalton Trans. 1995, 155; c) J. R. Ascenso, C. G. Azevendo, I. S. Gon-
calves, E. Herdtweck, D. S. Moreno, M. Pessanha, C. C. Romao, Or-
ganometallics 1995, 14, 3901; d) J. R. Ascenso, I. S. Goncalves, E.
Herdtweck, C. C. Romao, J. Organomet. Chem. 1996, 508, 169.
60] S. E. Browne, S. E. Asher, E. H. Cornwall, J. K. Frisoli, L. J. Harris,
E. A. Salot, E. A. Sauter, M. A. Trecoske, P. S. Veale, J. Am. Chem.
Soc. 1984, 106, 1432.
[69] K. W. Henderson, A. E. Dorigo, Q. Y. Liu, P. G. Williard, J. Am.
Chem. Soc. 1997, 119, 11855.
[70] a) L. M. Engelhardt, B. S. Jolly, P. C. Punk, C. L. Raston, B. W. Skel-
ton, A. H. White, Aust. J. Chem. 1986, 39, 1337; b) H. Mack, G.
Frenzen, M. Bendikov, M. S. Eisen, J. Organomet. Chem. 1997, 549,
39.
[
7
7
1 1
[71] Li T (250 K, mononuclear) = 79.4 ms; Li T (250 K, dinuclear) =
[
[
[
61] D. Seebach, J. Gabriel, R. Hꢈssig, Helv. Chim. Acta 1984, 67, 1083.
62] W. Bauer, D. Seebach, Helv. Chim. Acta 1984, 67, 1972.
63] a) H. J. Reich, J. P. Borst, J. Am. Chem. Soc. 1991, 113, 1835; b) H. J.
Reich, J. P. Borst, R. R. Dykstra, Tetrahedron 1994, 50, 5869.
64] R. Amstutz, D. Seebach, P. Seiler, B. Schweizer, J. D. Dunitz,
Angew. Chem. 1980, 92, 59; Angew. Chem. Int. Ed. Engl. 1980, 19,
48.6 ms. At temperatures below 250 K the T values of both lithium
aggregates were <30 ms.
[72] D. J. Metz and A. Glines, J. Phys. Chem. 1967, 71, 1158.
[73] H. J. V. Tyrrell, K. R. Harris, Diffusion in Liquids, Butterworths,
1
[
London, 1984.
[74] E. Martꢁnez-Viviente, P. S. Pregosin, Helv. Chim. Acta 2003, 86,
5
3.
2364.
[
65] R. Amstutz, J. D. Dunitz, D. Seebach, Angew. Chem. 1981, 93, 487;
Angew. Chem. Int. Ed. Engl. 1981, 20, 465.
[
66] A DEPT spectrum, together with a gHMQC correlation at 173 K,
Received: August 23, 2004
Published online: January 18, 2005
7
13
suggest a Li, C coupling of about 22 Hz, although there is partial
1506
ꢂ 2005 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
Chem. Eur. J. 2005, 11, 1495 – 1506