Fluorinated boroxin-based anion receptors for lithium ion batteries: Fluoride anion binding, ab initio calculations, and ionic conductivity studies

Article abstract of DOI:10.1021/jp901952t

Novel fluorinated boroxines, tris(2,6-difluorophenyl)boroxin (DF), tris(2,4,6-trifluorophenyl)boroxin (TF), and tris(pentafluorophenyl)boroxin (PF), have been investigated for potential applications in lithium ion batteries through fluoride anion binding,

Full text of DOI:10.1021/jp901952t

5918
J. Phys. Chem. A 2009, 113, 5918–5926
Fluorinated Boroxin-Based Anion Receptors for Lithium Ion Batteries: Fluoride Anion
Binding, Ab Initio Calculations, and Ionic Conductivity Studies
Nanditha G. Nair,† Mario Blanco,*,‡ William West,*,§ F. Christoph Weise,⊥ Steve Greenbaum,⊥
and V. Prakash Reddy*,†
Department of Chemistry, Missouri UniVersity of Science and Technology, Rolla, Missouri 65409, Department
of Chemistry, California Institute of Technology, Pasadena, California 91109, Jet Propulsion Laboratory,
California Institute of Technology, Pasadena, California 91109, and Hunter College,
City UniVersity of New York, New York 10021
ReceiVed: March 3, 2009; ReVised Manuscript ReceiVed: April 3, 2009
Novel fluorinated boroxines, tris(2,6-difluorophenyl)boroxin (DF), tris(2,4,6-trifluorophenyl)boroxin (TF), and
tris(pentafluorophenyl)boroxin (PF), have been investigated for potential applications in lithium ion batteries
through fluoride anion binding, ab initio calculations, and ionic conductivity measurements. Structures of the
fluorinated boroxines and boroxin-fluoride complexes have been confirmed by comparing their 19F and 11
B
NMR chemical shifts with those obtained by the DFT-GIAO method. The stoichiometry of the fluoride anion
binding to these boroxines has been shown to be 1:1 using 19F NMR and UV-vis spectroscopy. UV-vis
spectroscopic studies show the coexistence of more than one complex, in addition to the 1:1 complex, for
perfluorinated boroxin, PF. DFT calculations (B3LYP/6-311G**) show that the fluoride ion complex of DF
prefers unsymmetrical, covalently bound structure (7) over the symmetrically bridged species (10) by 12.5
kcal/mol. Rapid equilibration of the fluoride anion among the three borons in these boroxines results in a
single 19F NMR absorption for all of the aromatic ortho- or para-fluorines at ambient temperature. The effect
of these anion receptors on lithium ion conductivities was also explored for potential applications in dual ion
intercalating lithium batteries.
Introduction
benzodioxaborole (3) as an anion receptor for fluoride anions
and demonstrated its effect on the dissociation of LiF.2
The state of the art electrolytes in the conventional lithium
ion batteries are based on a variety of carbonate-based solvents
and LiPF6. Coordination of Lewis-basic carbonate solvents with
lithium salts results in a decrease in lithium ion transference
numbers. Further, reaction of LiPF6 or related salts with traces
of water in the electrolyte solvents releases Bronsted acids such
as HF, which serve as initiators for the ring-opening polymer-
ization of cyclic carbonate solvents and other undesirable effects
including corrosion of cathode materials, which limits the
specific energies obtainable from these batteries.
A rechargeable C-F cathode would be very desirable in terms
of specific energy relative to conventional lithium ion transition
metal oxide cathodes. However, there are numerous difficulties
associated with reversible fluorination of carbon which relate
to, to a large extent, the nature of the C-F bond. To date, all
C-F cathodes, regardless of preparation conditions, are strictly
nonrechargeable. Recently, a novel dual ion (Li+, F-) intercalat-
ing lithium fluoride battery has been developed using LiF as a
low-molecular-weight electrolyte dissolved in nonaqueous
solvents aided by boron-based anion receptors such as tris(pen-
tafluorophenyl)borate (TPFB; 1) and tris(hexafluoroisopropy-
l)borate (THFIPB; 2).1 This new battery cell chemistry is based
upon the dual intercalation of lithium and fluoride ions into
graphite anodes and cathodes, respectively.1 Amine and co-
workers have developed 2-(pentafluorophenyl)tetrafluoro-1,3,2-
Yang and McBreen and co-workers have shown that with
TPFB (1) and THFIPB (2) and tris(pentafluorophenyl)borane
(TPFPB), when used as anion receptors in LiF/carbonate
electrolytes, lithium ion transference numbers could be realized
as high as 0.7, and acceptable room-temperature ionic conduc-
tivity in the range of 10-3 Scm-1 was observed.3-5 Solubility
of the otherwise insoluble LiF also dramatically increased in
the presence of these anion receptors.2,3,6 The fluoride adducts
of the anion receptors are also expected to provide a protective
solid electrolyte interface (SEI) on the graphite surfaces, which
should lead to improved capacity retention and power capabi-
lity.2,7,8 TPFB at relatively higher concentrations, however,
significantly raises the interfacial impedance of lithium ion
batteries.9
The optimal characteristics of the anion receptors for applications
in high-specific-energy lithium ion batteries include their ability
to sustain sufficiently high electrochemical windows, high solubili-
ties, and diffusivities of their fluoride complexes and, more
importantly, their fluoride anion affinities. The low affinity of
fluoride anions to the receptors would not result in the sufficient
* To whom correspondence should be addressed. E-mail: preddy@mst.edu.
Tel: (573)341-4768. Fax: (573)341-6033 (V.P.R.).
† Missouri University of Science and Technology.
‡ Department of Chemistry, California Institute of Technology.
§ Jet Propulsion Laboratory, California Institute of Technology.
⊥ City University of New York.
10.1021/jp901952t CCC: $40.75  2009 American Chemical Society
Published on Web 04/27/2009

Boroxin-Based Anion Receptors for Lithium Ion Batteries
J. Phys. Chem. A, Vol. 113, No. 20, 2009 5919
dissolution of LiF into the electrolyte solvents, while at the same
time, too strong fluoride anion binding to the anion receptor would
not efficiently release the bound fluoride anions during charging
with resultant co-intercalation into the graphite cathodes.1
TF: δ19F, -99.9 (apparent triplet, J ) 7.5 Hz, ortho-F), 106.8
(t, J ) 8 Hz; para-F); δ11B, 28.0.
PF: δ19F, -132.4 (m, ortho-F), -153.7 (t, J ) 19 Hz, para-
F), -163.5 (m, meta-F). δ11B, 27.3 (broad s), 20.4 (broad s).
Complex 7: δ19F, -110.4 (broad s, ortho-F); (additional small
peaks, δ19F, -102 to -105, were observed, presumably corre-
sponding to multiple fluoride exchange equilibria); δ11B, 28.5
(broad s), 20.4 (broad s), 3.0 (m).
Complex 8: δ19F, -107.9 (m, ortho- and para-F) (additional
small peaks, δ19F, -98 to -115, were observed, presumably
corresponding to multiple fluoride exchange equilibria); δ11B,
28.0 (broad s), 20.1 (broad s), 2.8 (m).
Complex 9: PF: δ19F, -139.3 (m, ortho-F), -155.3 (t, para-F),
-162.9 (m, meta-F). δ11B, 20.4(broad s), 1.95 (broad s), 1.2 (m).
Addition of 2 mol equiv of TBAF simplified the 19F NMR
spectra of the complexes 7 and 8. There was no apparent change
in the spectra of complex 9. The δ19F remained constant upon
further addition of up to 3 mol equiv of TBAF to each of these
complexes (7, 8, and 9). The 19F and 11B NMR spectra of these
complexes with 2 mol equiv of fluoride anion are as follows
Complex 7 (with excess TBAF): δ19F, -110.4 (broad s); δ11B,
20.4 (broad s), 1.2 (broad s).
Here, we disclose our results on the effect of fluorinated
boroxines, tris(2,6-difluorophenyl)boroxin (DF; 4), tris-
(2,4,6)trifluorophenylboroxin (TF; 5), and tris(pentafluo-
rophenyl)boroxin (PF; 6) as novel anion receptors for
potential applications in lithium ion batteries, through ab
initio theory, fluoride anion binding studies, and ionic
conductivities. These cyclic fluorinated boroxines have
additional advantages as their Lewis acidity could be readily
modulated by varying the number of fluorines on the aromatic
rings. Our density functional theory (DFT) calculations
substantiate this hypothesis and serve as a predictive tool
for identifying the optimal anion receptors.
Complex 8 (with excess TBAF): δ19F, -107.9 (m, ortho-
and para-F); δ11B, 20.4, 1.2.
Complex 9 (with excess TBAF): δ19F PF: δ19F, -139.3 (m,
ortho-F), -155.3 (t, J ) 19 Hz, para-F), -162.9 (m, meta-F);
δ11B, 20.4 (broad s), 1.2 (broad s).
UV-Vis Spectroscopic Studies. UV-vis spectroscopic
studies were carried out on a Cary-50 UV-vis spectrophotom-
eter equipped with a 1 cm path length quartz cell by monitoring
the absorbances between 200 and 300 nm. Micropipets used
were capable of delivering 1 mL solutions, having adjustable
ranges between 100 and 1000 µL. All UV-vis measurements
were performed at ambient temperature. Stock solutions (1 ×
10-5 M) of DF (4), PF (6), and tetrabutylammonium fluoride
(TBAF) were prepared in acetonitrile. The stoichiometry of the
DF-fluoride (PF-fluoride) complexes was determined using
Job’s continuous variation method as follows.
A series of solutions consisting of DF (PF) and TBAF were
prepared by mixing the above stock solutions so that the total
concentration of the combined solutions was maintained constant
at 1 × 10-5 M, and the UV absorbances were recorded
separately for each solution. The stoichiometry of the complexes
was determined by a plot of the absorbances (multiplied by the
mol fraction of boroxines) versus mol fractions of the TBAF.
Ab Initio Calculations. We conducted thermodynamic
calculations, in the form of electronic energies, solvation free
energies in propylene carbonate, and binding energetics using
quantum mechanical density functional theory (B3LYP)12 for
reactions of the fluoride anion binding to novel fluorinated
phenyl boroxin anion receptors. We performed full geometry
optimizations with solvation effects obtained through a con-
tinuum dielectric model.13 The molecular solvation surface of
the anion receptor was defined with the molecular radius (2.7Å)
and dielectric constant of propylene carbonate (ε ) 64.5). All
wave functions are open shell (UDFT), calculated with a high-
level basis set (LACVP**) which translates to the 6-311 G**
basis for main group elements. Ab initio calculations were
carried out on a Dell Linux Cluster (80 node/160 core, Single
Core Xeon, 3.06 GHz) running Linux/OSCAR, with 160 GB
of RAM and 8.6 TB of disk space. 11B and 19F NMR chemical
shifts were calculated using density functional theory, at the
B3LYP/6-311G** level, for the various fluorinated receptors
and fluoride complexed anions. The geometries were fully
Experimental Methods Section
Materials. Acetonitrile (anhydrous, >99.8%), dimethylsul-
foxide (>99.6%), dichloromethane (>99.8%), ethyl acetate
(>99.5%), and tetrabutylammonium fluoride hydrate (>98%)
were obtained from Aldrich and used as received. The fluori-
nated boroxines DF (4), TF (5), and PF (6) were obtained by
thermal dehydration of the corresponding arylboronic acids at
100 °C for about one hour.10 The products were obtained
essentially pure as confirmed by their 19F NMR and 11B NMR
spectra (vide infra). Further confirmation of the structures was
provided by matching the observed chemical shifts with those
obtained at the GIAO/B3LYP/6-31G** level.11
NMR Spectra. 19F NMR spectra (at 376 MHz) and 11B NMR
spectra (128 Hz) were obtained on a Varian Inova 400 MHz
spectrometer in anhydrous CH3CN (in unlocked mode) or CD3CN
solutions. 19F NMR spectra were referenced to CFCl3 (δ19F ) 0),
and 11B NMR spectra were referenced to BF3-OEt2 (δ11B ) 0).
A series of solutions (about 200 mM) of boroxines (DF, TF,
and PF) were prepared by dissolving each of the boroxines in
various solvents (acetonitrile, DMSO, ethyl acetate, and dichlo-
romethane), and appropriate amounts of tetrabutylammonium
fluoride (TBAF) were added to them so that the resulting
solutions had mole ratios of boroxine/fluoride of 1:0, 1:1, 1:2,
and 1:3. These homogeneous solutions were transferred into 5
mm NMR tubes, CFCl3 was added as an internal reference, and
19F NMR spectra were recorded at ambient temperature. 11B
NMR spectra were obtained for the same solutions using BF3-
etherate as the external reference.
The reaction of boroxines (DF, TF, and PF) with the fluoride
anion was exothermic. A clear homogeneous solution resulted
immediately after the addition of TBAF to DF and TF. However,
reaction of PF with TBAF gave homogeneous solutions only in
DMSO, acetonitrile, and dichloromethane, but not in ethyl acetate.
The 11B and 19F NMR spectra for the boroxines and their fluoride
complexes 7, 8, and 9 (1:1 mol ratio) in acetontrile are as follows:
DF: δ19F, -103.0 (broad s); δ11B, 28.5.

5920 J. Phys. Chem. A, Vol. 113, No. 20, 2009
Nair et al.
Figure 1. Experimental (acetonitrile solvent; in parentheses) and GIAO/B3LYP/6-31G** calculated δ19F (δCFCl3 ) 0) and 11B NMR spectra for
DF, TF, and PF; δ11B are shown using arrows (δBF3 ·OEt2 ) 0).
optimized (basis set 6-311G**) in a continuum dielectric
medium appropriate for propylene carbonate (PC). The 19F and
11B NMR chemical shifts were computed with respect to CFCl3
Structural Characterization. The boroxines 4-6 were
characterized by 19F and 11B NMR spectroscopy, and the
structures were confirmed through comparison of the chemical
shifts with those obtained by the ab initio GIAO/B3LYP/6-
31G**11 method (Figure 1). The 19F NMR spectrum for DF (4)
showed a single absorption at -103.0 ppm, which is very close
to the GIAO/B3LYP/6-31G** of -101.7 (∆δ ) 1.3). TF (5)
showed two 19F absorptions at δ19F -99.9 (ortho-F) and -106.8
(para-F), agreeing with those of DFT-GIAO calculations,
-104.8 (ortho-F, ∆δ ) 4.82) and -105.5 (para-F, ∆δ ) 2.34).
PF (6) showed three absorptions at -132.4 (ortho-F), -153.7
(para-F), and -163.0 (meta-F). The ortho-, meta-, and para-
fluorine assignments are supported by the ab initio calculated
chemical shifts, although the theoretically obtained chemical
shifts deviate from the experimental values by 6-10 ppm.
Further, the 11B NMR spectra closely match the experimental
values for DF (∆δ ) 1.7), TF (∆δ ) 0.2), and PF (∆δ ) 0.5),
showing that the GIAO 11B NMR chemical shifts are reliable
indicators of the structural characterization of the boroxines.17,18
Solvation Effects. It has been established through X-ray
diffraction studies that tris(pentafluorophenylborane) can form
1:1 adducts with solvents such as acetonitrile.19 In order to see
similar solvation effects on fluorinated boroxines, we have
explored the effect of solvents on the stabilization of the boron
centers through 19F NMR spectroscopy (Figure 2). In the
relatively non-nucleophilic solvent, dichloromethane, DF shows
a single absorption at δ19F -104.5, whereas in acetonitrile,
dimethylsulfoxide, and ethyl acetate, the absorption has shifted
downfield by 1 ppm to -103. It shows a significant solvation
effect from these solvents. Similarly, for TF, δ19F(CH2Cl2)
signals at -101.1(ortho-F) and -103.2 (para-F) are shifted
downfield by 1 ppm for the ortho-fluorine, and para-fluorine is
shifted by 4 ppm upfield in DMSO, acetonitrile, and ethyl
acetate.
-
(δ19F ) 0.0) and BF4 (δ11B ) 0.0), respectively. The latter
-
δ11B(BF4-) were reconverted to δ11B(BF3 ·Et2O) (δ11B(BF4
)
) -2.2 with respect to δ11B(BF3 ·OEt2; δ11B ) 0).
Lithium Ion Conductivities. A series of 0.1 M boroxine-
based anion receptors dissolved along with LiF in propylene
carbonate (PC) were prepared using an excess of LiF relative
to the anion receptor (0.1 M anion receptor + 0.2 M LiF) in
order to probe for the fluoride anion binding. The ionic
conductivities of these solutions were measured using a Pt/Pt
conductivity cell through electrochemical impedance spectros-
copy, calibrated with standard commercially available samples.
Results and Discussion
Boron-based anion receptors such as tris(pentafluorophenyl)bo-
rane have recently been used for selective binding of fluoride anions
in lithium ion batteries.2,5,8,14 In order to systematically fine-tune
the binding efficiencies of the boron-based anion receptors, we have
now synthesized a series of fluorinated arylboroxinessDF (4), TF
(5), and PF (6)sby thermal dehydrative cyclizations of the
corresponding arylboronic acids. These boroxines are novel
compounds and have not been used in lithium ion battery
applications. The structures of these compounds were readiy
assigned based on 19F and 11B NMR spectra, which closely match
with those calculated by ab initio theory (Figure 1). 11B NMR are
especially diagnostic of the structures as deviations of the experi-
mental chemical shifts from GIAO/B3LYP-6-31G** derived values
are too small, ∆δ ) 1.7 (DF), 0.2 (TF), 0.5 (PF).
The synthetic method adapted is comparatively simple and
allows synthesis of a wide variety of fluorinated boroxines from
their corresonding arylboronic acids.10,15,16 Further, the binding
efficiencies of these boroxines can be readily optimized by tailoring
the degree of fluorination on the aryl rings or by incorporating
additional electron-withdrawing groups. These anion receptors
could be readily monitored for their anion binding efficiency using
19F and 11B NMR spectroscopy. We have explored the ionic
conductivities, fluoride anion binding efficiencies, and binding
stoichiometries of these boroxines. These results are further
substantiated by ab initio theoretical calculations.
On the other hand, PF (6) shows relatively more sensitivity
to solvation effects. In the more coordinating DMSO solvent,
the δ19F (CH2Cl2) signals at -132.8 (ortho-F), -149.6 (para-
F), and -161.5 (meta-F) are split into multiple peaks from -131
to -164 ppm in DMSO, indicating the formation of stable
PF-DMSO adducts. In acetonitrile and ethyl acetate solvents,
the δ19F(CH2Cl2) absorptions are shifted downfield by 0.5 ppm
for ortho-fluorines, upfield by 5 ppm for para-fluorines, and
upfield by 2 ppm for the meta-fluorines.
The 19F NMR of PF, TF, and DF in various solvents thus
clearly show the importance of solvation. In the cases of TF
and DF, there is rapid exchange of the solvent molecules among
the three boron centers of the complex, so that only a single
19F NMR absorption is observed for all of the aromatic ortho-
or para-fluorines. However, in relatively more polar DMSO
solvent, additional distinct absorptions for PF-DMSO adducts

Boroxin-Based Anion Receptors for Lithium Ion Batteries
J. Phys. Chem. A, Vol. 113, No. 20, 2009 5921
lithium fluoride electrolyte in nonaqueous organic solvents but also
dramatically increases the lithium ion conductivities. Reversibility
of fluoride anion binding is crucial in the next generation dual ion
intercalating lithium ion batteries. In order to understand the
structural effects on the fluoride binding to anion receptors, we
carried out DFT calculations on a series of fluorinated boroxines,
and the results are shown in Table 1 and Figure 3. As can be seen
from Figure 3, fluoride anion binding is correlated with the number
of the fluorines on the aryl rings; the more fluorines in the aryl
rings, the higher the binding efficiency of the boronic anhydrides
(boroxines). Because there is more than one boron capable of anion
capture, we estimated the binding energies for the second and third
fluoride and report the data in Table 1. These calculations are less
reliable because they involved highly charged species. Nonetheless,
it appears that multiple bindings are favorable (exothermic). Second
and third anion binding energies do not follow the same trend as
a function of fluorination that the first anion binding follows.
To understand the binding abilities of the fluorinated borox-
ines with the fluoride anion, we have carried out 19F NMR
spectroscopic studies for solutions of boroxines with varied
amounts of TBAF. A series of samples of boroxines (4-6) were
prepared in acetonitrile, dichloromethane, and DMSO and
contained 0, 1, 2, and 3 equiv of TBAF and CFCl3 (δCFCl3 )
0) as the internal reference, and the spectra were recorded in
the unlocked mode. The 19F and 11B NMR spectra of the
boroxin-fluoride complexes (7-9) in acetonitrile are sum-
marized in Figure 5. The DF-fluoride complex, 7 (with 2 equiv
of TBAF), shows a single absorption, δ19F -110.4 (ortho-F).
The TF-fluoride complex, 8 (with 2 equiv of TBAF), also
shows only a single 19F absorption at -107.9 ppm due to the
coincidental merging of the ortho- and para-fluorines; that is,
the ortho-fluorines are shielded much farther than the para-
fluorine with respect to the corresponding neutral boroxin. The
PF-fluoride complex, 9 (with 1 or 2 equiv of TBAF), shows
three 19F absorptions at -139.3 (ortho-F), -155.3 (meta-F), and
-162.9 (para-F).
The assignments of the chemical shifts of these boroxin-
fluoride complexes, 7, 8, and 9, could be readily confirmed
through their DFT-GIAO chemical shifts (Figure 4). The DFT-
GIAO derived δ19F for the ortho-fluorines in these complexes
vary slightly from each other, whereas they show single
absorptions in solution NMR, indicating rapid rotation across
the B-aryl bonds in the solution phase. Further, the DFT values
also reflect different chemical environments for the two sets of
aryl groups. In solution, due to the rapid intramolecular exchange
of the fluoride anion among the three borons, a single δ19F
absorption is observed (vide infra). The DFT-GIAO predicts a
δ19F of -166.7, -168.8, and -164.6 for the B-F fluorine in
structures 7, 8, and 9, respectively. These absorptions could not
be observed in solution, perhaps due to the rapid intramolecular
fluoride anion exchange (vide infra).
Upon addition of 1 mol equiv of TBAF to PF (6), in
acetonitrile solvent, complete disappearance of the 19F NMR
peaks for the starting material was observed, and a three line
19F NMR spectrum for the PF-fluoride complex, 9, was
observed; δ19F -139.3, -155.3, -162.9 (Figure 5). However,
for DF (4) and TF (5), upon addition of 1 mol equiv of TBAF,
a complex spectrum resulted with multiple absorptions corre-
sponding to the boroxin-fluoride complexes (7 and 8), presum-
ably because of their equilibration with unreacted boroxines.
Addition of 2 mol equiv of TBAF to these solutions resulted in
surprisingly simplified spectra showing only one absorption (less
than 0.1% of the starting material could be observed by 19F
NMR). These results show that PF has relatively stronger
Figure 2. 19F NMR spectra of DF (4), TF (5), and PF (6) in CH3CN
(A), EtOAc (B), DMSO (C), and CH2Cl2 (D).
could be observed. This indicates relatively slow exchange of
the DMSO with PF. Unlike the starting boroxines, the
boroxine-fluoride adducts (in the presence of excess fluoride
anion) have similar δ19F in all of the solvents studied (CH2Cl2,
acetonitrile, and DMSO), showing that the solvent participation
is minimal in the case of the fluoride adducts.
Fluoride Anion Binding. Fluoride anion binding to anion
receptors not only helps the solubility of the otherwise insoluble

5922 J. Phys. Chem. A, Vol. 113, No. 20, 2009
Nair et al.
TABLE 1: Fluoride (F-) Anion Binding Energies (B3LYP, LACVP**), Solvation Energies, and Degree of Fluorination for
Various Fluorinated Arylboroxinesa
compound
degree of fluorination
binding energy (kcal/mol)
tris(pentafluorophenyl)boroxin (6)
15
12
9
9
9
74.1
73.1
71.0
70.0
69.3
67.5
61.6
(2,6-difluorophenyl)-bis(pentafluorophenyl*)boroxin
bis(2,6-difluorophenyl)-(pentafluorophenyl*)boroxin
bis(2,6-difluorophenyl*)-(pentafluorophenyl)boroxin
tris(2,4,6-trifluorophenyl)boroxin (5)
tris(2,6-difluorophenyl)boroxin (4)
tris(4-fluorophenyl)boroxin
6
3
Second Anion Captured
(2,6-difluorophenyl)-bis(pentafluorophenyl**)boroxin
bis(2,6-difluorophenyl*)-(pentafluorophenyl*)boroxin
13
10
49.1
45.9
Third Anion Captured
(2,6-difluorophenyl*)-bis(pentafluorophenyl**)boroxin
bis(2,6-difluorophenyl*)-(pentafluorophenyl**)boroxin
14
11
16.6
11.2
a The * indicates the location of fluoride anion binding.
for each of these complexes (with excess fluoride in case of
DF and TF) shows that there is a rapid intramolecular fluoride
exchange between the three borons (eq 2) with an upper limit
of free energy of activation of 12.5, 12.4, and 12.0 kcal/mol
for boroxin-fluoride complexes 7, 8, and 9, respectively
(corresonding to exchange rate constants of 4400, 4800, and
9200 s-1, respectively).20 Presumably due to this low barrier
for exchange, the δ19F for B-F fluorine could not be observed
in solution (vide supra).
Anion receptor PF (6), in acetonitrile, in the presence of 1
mol equiv of fluoride showed three new absorptions at -139.3
(ortho-F), -155.3 (para-F), and -162.9 (meta-F) corresponding
to the PF-fluoride complex, and the absorptions for PF
completely disappeared (Figure 5). Further addition of TBAF
(2-3 mol equiv) resulted in no change of the spectra, showing
that the stoichiometry of the PF-fluoride complex is 1:1. The
formation of the fluoride complex is essentially irreversible as
the absorptions corresonding to the starting PF could not be
observed even after addition of 1 mol equiv of TBAF. The
ortho-fluorines are relatively more shielded in the PF-fluoride
complex (9) as compared to the meta- and para-fluorines, as
expected due to the proximity of the negatively charged boron.
11B NMR Studies. The 11B NMR for the fluoride complexes
(7, 8, and 9) are shown in Figure 5. The DF-fluoride complex
(7) shows two 11B absorptions at δ11B 19.8 (sp2-B) and 1.2 (sp3-
B). The TF-fluoride and PF-fluoride complexes similarly show
two 11B absorptions at δ11B (TF), 20.1 (sp2-B) and 1.2 (sp3-B),
and δ11B (PF), 20.4 (sp2-B) and 1.2 (sp3-B). The DFT-GIAO
derived chemical shifts are in reasonably close agreement (∆δ
) 2.8-5.9) with experimental values; therefore, the assignments
of the 11B absorptions in the complexes can be confirmed. In
the presence of 1 mol equiv of TBAF, about 20% of the
unreacted starting materials could be observed in the case of
DF and TF, in addition to the above two 11B absorptions for
the fluoride complexes, 7 and 8, showng their equilibration with
residual boroxines. Further, the chemical shifts of the sp3-
hybridized boron in complexes 7 and 8 is shielded by 2 ppm in
the presence of 2 equiv of TBAF, presumably due to the weak
binding of the second fluoride anion to these complexes.
Figure 3. Effect of the degree of aryl fluorination on the fluoride anion
binding energies (kcal/mol) from DFT (B3LYP).
fluoride affinity as compared to DF and TF. The data in Table
1 are also in accordance with these results; that is, the higher
the aryl-fluorination, the stronger the fluoride binding.
Addition of 1 mol equiv of TBAF to PF, however, resulted
in instantaneous complete disappearance of the starting material
as monitored by 11B NMR spectroscopy. The PF-fluoride
complex, 9, showed three relatively broad signals at δ11B 20.4
(sp2-B), 2.2 (sp3-B), and 1.4 (sp3-B) (Figure 5). Further addition
of TBAF to this solution resulted in gradual broadening of the
peak at 2.2 ppm. Although we cannot clearly rationalize this
At first glance, we would have expected distinct 19F absorp-
tions for the two sets of aromatic rings for complexes 7-9;
that is, the aryl rings bound to sp3-hybridized boron are
chemically different than those attached to sp2-hybridized boron.
The observation of a single ortho-19F (or para-19F) absoption

Boroxin-Based Anion Receptors for Lithium Ion Batteries
J. Phys. Chem. A, Vol. 113, No. 20, 2009 5923
Figure 4. Experimental (acetonitrile solvent; δ19F in parentheses) and GIAO/B3LYP/6-31G** calculated δ19F (δCFCl3 ) 0) for the boroxine-fluoride
complexes (7-9); δ11B are shown using arrows (δBF3 ·OEt2 ) 0).
UV-Vis Studies. UV-vis spectroscopy is commonly used
to identify the stoichiometry of host-guest complexes.21 In order
to further confirm the 19F NMR derived stoichiometry for the
boroxine-fluoride complexes, we have carried out UV-vis
studies for these complexes using continuous variation method
(Job’s plots).
UV-vis spectra were recorded for a series of DF/TBAF (PF/
TBAF) solutions; the mole ratios of DF to fluoride (PF to
fluoride) were continually varied while keeping the total
concentration of each of their solutions constant. The measured
UV-vis absorbances at 265, 265, 260, and 254 nm (multiplied
by the mole fraction of boroxines) were plotted against mole
fractions of the TBAF (Figure 6).22,23 Through these Job’s plots,
the stoichiometry of the DF-fluoride complex was determined
as 1:1. In the case of DF, the 1:1 stoichiometry is consistently
observed at all wavelengths, 269, 265, 260, and 254 nm (Figure
6B). The stoichiometry of host-guest complexation is inde-
pendent of the UV-vis wavelength in Job’s plots if a single
complex is formed. On the other hand, the dependence of Job’s
maxima on wavelength is indicative of the coexistence of more
than one complex.21,24,25 The Job’s plots for PF, on the other
hand, show deviations from 1:1 stoichiometry at different
wavelengths, indicating the possibility of coexistence of more
than one complex for PF (Figure 6C) (vide supra).
Ab Intio Structures of Boroxin-Fluoride Complexes. We
have optimized structures of symmetrically bound and unsym-
metrically bound fluoride anion complexes of DF at the B3LYP/
6-31G** level in order to clarify the nature of the fluoride anion
binding to the cyclic boroxines. Aldridge and co-workers have
earlier shown that a related triboramacrocyclic compound, 1,4,7-
trifloro-1,4,7-triboracyclononane (11) and its perfluorinated
analogue have symmetrical fluoride anion binding to all three
boron atoms at the HF/6-31+G* level.26
Figure 5. 19F (top) and 11B (bottom) NMR spectra of DF (4), TF (5),
and PF (6) in acetonitrile at various molar equivalents of TBAF;
boroxin/TBAF ) 1:3 (A); boroxin/TBAF ) 1:2 (B); boroxin/TBAF
) 1:1 (C); boroxin without added TBAF (D).
unexpected observation for PF, perhaps a second fluoride anion
binding may be involved in this case. In accordance with this
expectation, UV-vis studies also show possible existence of
multiple equilibria, that is, coexistence of more than one
PF-fluoride complex (vide infra). Upon further addition of up
to 3 mol equiv of TBAF, the spectra for boroxin-fluoride
complexes 7, 8, and 9 remained unchanged.
However, we have found at higher DFT levels of calculations
at B3LYP/6-311G** that the symmetyrical C3V structure for the

5924 J. Phys. Chem. A, Vol. 113, No. 20, 2009
Nair et al.
Figure 7. B3LYP/6-311G** structures of asymmetic and sym-
metrically bridged [DF-F]- complexes, corresonding to structures 7
(left) and 10 (right).
becomes more tetragonal, and the propeller angle increases
significantly (79.7°).
Lithium Ion Conductivities. The lithium ion conductivities
of the fluorinated boroxin (4, 5, and 6) solutions are estimated
as near 0.5 mS/cm (Figure 9). The relative conductivities of
these solutions are in the following order: PF (6) > DF (4) >
TF (5). As a means of comparison, the conductivity of a solution
using the well-known tris(pentafluorophenyl)borane (TPFPB)5,8,14
yields virtually similar ionic conductivity as the PF solution
with the same molarity. While the fluorinated boroxin/LiF ratio
of the samples was 1:2, visual examination of the solution
indicated that not all of the LiF was dissolved in the solution,
and some fluorinated boroxin may also be insoluble at these
molarities. On the basis of the observation that the ionic
conductivity of the fluorinated boroxin/LiF did not exceed that
of TPFPB/LiF, which is known to form 1:1 complexes, we
conclude that the fluorinated boroxin/LiF ratio did not exceed
1:1. The solubility of all of the fluorinated boroxines in
propylene carbonate is qualitatively lower than TPFPB, which
necessarily limits the conductivity of the solutions.
It should be noted that the conductivity of perfluorinated
boroxin (6) is significantly higher than that the difluorinated or
trifluorinated boroxines (4 and 5). Although there is no clear-
cut correlation of conductivities with electron-withdrawing
substituents, these studies show that the highly fluorinated
boroxines show significant effect on the dissociation of LiF and
thereby their increased conductivities. However, the lithium ion
conductivities of these solutions are nearly 10 times smaller
than those for 1 M LiPF6 in PC, which signifies that even better
anion receptors are needed to achieve relatively higher
conductivities.
Figure 6. UV-vis spectra (in acetonitrile) obtained from the continu-
ous variation method (A) and Job’s plots for DF (B) and PF (C); B
and C are the overlay plots for absorbances at 269, 265, 260, and 255
nm.
complex, 10, is relatively 12.5 kcal/mol higher in energy as
compared to that of the unsymmetrically bound species, 7.
Structures 7 and 10 are energy minima on the potential energy
surface at this level of theory (Figure 7).
The relatively low energy for the unsymmetric structure (7)
implies its entropic advantage over the symmetric structure, 10.
The fluoride bound complex, 7, shows significantly upfield
shifted 19F signals for the fluorine atoms attached to aromatic
rings. However, we were not able to observe the absorptions
corresponding to the fluoride anion attached to boron. The lack
of detection of the boron-bound fluoride absorption indicates
that there is rapid equilibrium between three unsymmetrical
structures relative to the NMR time scale at ambient temperature
with an estimated free energy of activation of 12.0-12.5 kcal/
mol (vide supra). Geometries for the PF and [PF-F]- are shown
in Figure 8. In the neutral form, the PF anion receptor is a
propeller-shaped molecule with a small angle (36.4°) between
the boroxine ring and the phenyl groups. In the charged state
[PF-F]-, the boron atom that captures the fluoride anion

Boroxin-Based Anion Receptors for Lithium Ion Batteries
J. Phys. Chem. A, Vol. 113, No. 20, 2009 5925
Figure 8. B3LYP/6-311G** solution (propylene carbonate) geometry-optimized structures of neutral PF (left) and the corresponding charged
[PF-F]- complex (right).
atoms in these boroxines, with an estimated upper limit of
about 12 kcal/mol.
Acknowledgment. The work described here was carried out,
in part, at the Jet Propulsion Laboratory, California Institute of
Technology, under contract with the National Aeronautics and
Space Administration. Authors W.C.W. and M.B. gratefully
acknowledge the support of this work by the Army Research
Office. N.G.N. gratefully acknowledges graduate fellowship
from Missouri S&T.
References and Notes
(1) West, W. C.; Whitacre, J. F.; Leifer, N.; Greenbaum, S.; Smart,
M.; Bugga, R.; Blanco, M.; Narayanan, S. R. J. Electrochem. Soc. 2007,
154, A929–A936.
(2) Chen, Z.; Amine, K. Electrochem. Commun. 2007, 9, 703–707.
(3) McBreen, J.; Lee, H. S.; Yang, X. Q.; Sun, X. J. Power Sources
2000, 89, 163–167.
(4) Li, L. F.; Lee, H. S.; Li, H.; Yang, X. Q.; Nam, K. W.; Yoon,
W. S.; McBreen, J.; Huang, X. J. J. Power Sources 2008, 184, 517–521.
(5) Xie, B.; Lee, H. S.; Li, H.; Yang, X. Q.; McBreen, J.; Chen, L. Q.
Electrochem. Commun. 2008, 10, 1195–1197.
(6) Choi, N.-S.; Ryu, S.-W.; Park, J.-K. Electrochim. Acta 2008, 53,
6575–6579.
Figure 9. Conductivity of anion receptor-salt solutions in propylene
carbonate (PC) as a function of temperature.
(7) Chen, Z.; Liu, J.; Amine, K. Electrochim. Acta 2008, 53, 3267–
3270.
(8) Sun, X.; Lee, H. S.; Yang, X.-Q.; McBreen, J. Electrochem. Solid-
State Lett. 2003, 6, A43–A46.
(9) Chen, Z.; Amine, K. J. Electrochem. Soc. 2006, 153, A1221–A1225.
(10) Frohn, H. J.; Adonin, N. Y.; Bardin, V. V.; Starichenko, V. F. Z.
Anorg. Allg. Chem. 2002, 628, 2827–2833.
(11) Cao, Y.; Beachy, M. D.; Braden, D. A.; Morrill, L.; Ringnalda,
M. N.; Friesner, R. A. J. Chem. Phys. 2005, 122, 224116/224111–224116/
224110.
(12) Becke, A. D. J. Chem. Phys. 1993, 98, 5648–5652.
(13) Tannor, D. J.; Marten, B.; Murphy, R.; Friesner, R. A.; Sitkoff,
D.; Nicholls, A.; Honig, B.; Ringnalda, M.; Goddard, W. A. J. Am. Chem.
Soc. 1994, 116, 11875–11882.
(14) Sun, X.; Lee, H. S.; Lee, S.; Yang, X. Q.; McBreen, J. Electrochem.
Solid-State Lett. 1998, 1, 239–240.
We have also measured lithium diffusivities by 7Li NMR for
the solutions of LiF in boroxines, 4-6, and PFPB, which ranged
from 1.88 to 2.12 × 10-6 cm2/sec. The latter diffusivity is
relatively higher than the reported value for Li+/propylene
carbonate (1.5 × 10-6 cm2/sec).27 The discrepancy shows the
relatively reduced association of Li+ with the boroxines and
PFPB, which is also in accord with high lithium ion conductivi-
ties in the presence of boroxines (Figure 9).
Conclusion
In summary, fluorinated arylboroxines are efficient anion
receptors, comparable to the state of the art PFPB anion
receptor.8,14,19 The structures of the fluoride complexes DF, TF,
and PF are confirmed by matching the experimental 19F and
11B NMR chemical shifts with those obtained from DFT
calculations.
Our ab initio calculations show that the symmetrically
bound fluoride complex, 10, is, unexpectedly, less favorable
as compared to the unsymmetrically bound species, 7, by
12.5 kcal/mol. The coalescence of the 19F NMR signals of
these boroxines at ambient temperature shows relatively rapid
equilibration of the fluoride anion among the three boron
(15) Zeng, H.; Hua, R. J. Org. Chem. 2008, 73, 558–562.
(16) Rambo, B. M.; Lavigne, J. J. Chem. Mater. 2007, 19, 3732–3739.
(17) Hong, D.; Carroll, P. J.; Sneddon, L. G. J. Organomet. Chem. 2003,
680, 61–65.
(18) Bluhm, M. E.; Bradley, M. G.; Butterick, R., III; Kusari, U.;
Sneddon, L. G. J. Am. Chem. Soc. 2006, 128, 7748–7749.
(19) Jacobsen, H.; Berke, H.; Doering, S.; Kehr, G.; Erker, G.; Froehlich,
R.; Meyer, O. Organometallics 1999, 18, 1724–1735.
(20) We have estimated the rate constant (k) of the fluoride exchange
at ambient temperature using the equation k ) (π/sqrt2)∆δ, where ∆δ is
the separation of the corresponding 19F signals at slow exchange limit (in
hertz), obtained from the DFT-GIAO calculations, since the signal
coalescence occurs well below ambient temperature. Symmetry-related DFT
chemical shifts are averaged, and the difference in δ19F of sp3B-aryl and
sp2B-aryl ortho-fluorines is taken as ∆δ. Through this estimated exchange

5926 J. Phys. Chem. A, Vol. 113, No. 20, 2009
Nair et al.
rate constant, an upper limit of the activation barrier was calcaulted using
the Eyring equation, k ) (kBT/h)e-∆G*/RT; (a) Sandstrom, J. Dynamic NMR
Spectroscopy; Academic Press: New York, 1982; (b) Raban, M.; Kost, D.;
Carlson, E. H. J. Chem. Soc., Chem. Commun. 1971, 1913, 1656.
(21) Connors, K. A. Binding Constants: The Measurements of Molecular
Complex Stability; John Wiley: New York, 1987.
(24) Atkinson, G. F. J. Chem. Educ. 1974, 51, 792–798.
(25) Vosburgh, W. C.; Cooper, G. R. J. Am. Chem. Soc. 1941, 63, 437–
442.
(26) Aldridge, S.; Fallis, I. A.; Howard, S. T. Chem. Commun. 2001,
231–232.
(27) Hayamizu, K.; Aihara, Y. Electrochim. Acta 2004, 49, 3397–3402.
(22) Hirose, K. J. Inclusion Phenom. Macrocyclic Chem. 2001, 39, 193–
209.
(23) Karikari, A. S.; Mather, B. D.; Long, T. E. Biomacromolecules
2007, 8, 302–308.
JP901952T