Model compounds based on cyclotriphosphazene and hexaphenylbenzene with tethered Li+-solvents and their ion-conducting properties

Article abstract of DOI:10.1021/cm103559d

To improve the lithium ion conductivities of currently used electrolytes, it is critical to understand how the transport of the lithium ions within the matrix is influenced by their interactions with solvating moieties. Therefore, well-defined model compo

Full text of DOI:10.1021/cm103559d

ARTICLE
pubs.acs.org/cm
Model Compounds Based on Cyclotriphosphazene and
Hexaphenylbenzene with Tethered Liþ-Solvents and Their
Ion-Conducting Properties
J€org Thielen, Wolfgang H. Meyer,* and Katharina Landfester
Max Planck Institute for Polymer Research, Ackermannweg 10, D-55128 Mainz, Germany
ABSTRACT: To improve the lithium ion conductivities of currently used electrolytes,
it is critical to understand how the transport of the lithium ions within the matrix is
influenced by their interactions with solvating moieties. Therefore, well-defined model
compounds based on cyclotriphosphazene (CTP) and hexaphenylbenzene (HPB)
cores were prepared, bearing side groups containing the structural element of ethylene
carbonate, which is the common solvent for lithium salts used as electrolytes in Li-ion
batteries. All model compounds were highly pure and thermally stable up to at least
250 °C, covering a broad range of glass transition temperatures from ꢀ79 °C up to
þ3.5 °C. The temperature-dependent ionic conductivities of the blends follow a
WilliamꢀLandelꢀFerry (WLF) type behavior with the corresponding glass transition
temperatures as reference. Though the glass transition temperatures of the blends are
low, their conductivities are only in the range of typical polymer electrolytes, implying that the coordination between the cyclic
carbonate functionality and the Li-ion is apparently too tight to allow for fast Li-ion dynamics.
KEYWORDS: cyclotriphosphazene, hexaphenylbenzene, cyclic carbonate, lithium ion conductor, model compounds
’ INTRODUCTION
safely, technical measures such as hermetic metal encasements or
relief valves are necessary, leading to a decrease in the effective
energy density of the battery. Therefore, over the last decades,
major efforts have been made to develop solvent-free ion con-
ducting polymer electrolytes. Starting with Wright’s discovery in
1973, showing that alkali metal salt complexes of poly(ethylene
oxide) (PEO) show substantial ionic conductivity,8 several main
routes of polymer electrolytes have been developed: (i) dry solid
polymer electrolytes (SPEs), (ii) gel polymer electrolytes, (iii)
compositepolymer electrolytes, and(iv) ionic liquids.9ꢀ11 Despite
great efforts and testing of various systems as replacements, the
ideal electrolyte has not yet been identified.
Previous studies in our group based on our general approach
of “immobilizing” ion solvents revealed, that tethering the struc-
tural element of EC to a poly(meth)acrylate backbone yielded
rather surprising high conductivities, despite the comparably high
glass transition temperatures (Tg).12 Those did not serve as
reference temperature for the conductivity in the corresponding
modified WilliamꢀLandelꢀFerry (WLF) plot, thus suggesting
that the ion mobility is controlled by side chain dipole relaxation
modes rather than by the segmental motion of the polymer
backbone.
Efficient energy storage and conversion is playing a key role in
overcoming the present and future challenges in energy supply.
Batteries provide portable, electrochemicalstorageof green energy
sources such as solar, wind, or water power, and potentially allow
for a reduction of the dependence on fossil fuels, which is of great
importance with respect to the issue of global warming. In view of
energy density and energy drain, the development of rechargeable
lithium ion batteries has to be considered as a milestone for the
internal power supply of the tremendously increasing amount of
mobile and portable electric and electronic devices.1,2 In such
batteries, lithium ions are solvated by an organic solvent and
diffuse freely between the anode and cathode which are physically
isolated by a separator membrane.3,4 Most commonly used elec-
trolytes comprise blends of highly polar carbonates which are
reasonably good solvents for Li-salts,5 e.g. ethylene carbonate
(EC), propylene carbonate (PC), diethyl carbonate (DEC), or
dimethyl carbonate (DMC).
In case of small battery packages, these electrolytes have proven
to fulfill safety and market concerns in terms of high ionic con-
ductivity at ambient temperatures (in the order of 10ꢀ3 S cmꢀ1
)
combined with a high boiling and a low melting point as well as
safe performance over a wide temperature range.6 However, the
usage of organic liquids in large batteries as in electric vehicles
poses potential safety hazards. Due to the high flammability and
the present vapor pressure of the organic liquids, abrupt leakage
when mechanical forces are applied (e.g., in car accidents) or
overheating as a course of sudden and uncontrolled discharge
(“short circuits”) may lead to ignition or in the last resort even to
explosion of the battery.7 As a consequence, to use the electrolytes
Considering these results, which are indicating that the ion
mobility is decoupled from the other membrane relevant proper-
ties of the polymer, model compounds were prepared to study
the influence of both the glass transition temperatures and the
spacer length on the ionic conductivity in more detail. Perfectly
Received: December 14, 2010
Revised:
February 11, 2011
Published: March 22, 2011
r
2011 American Chemical Society
2120
dx.doi.org/10.1021/cm103559d Chem. Mater. 2011, 23, 2120–2129
|

Chemistry of Materials
ARTICLE
defined model compounds are highly suitable for reliable analysis
of the molecular ion transport, governed by the complex inter-
play of numerous parameters within an electrolyte, such as
viscosity, salt dissociation, ionꢀsolvent interaction, or ion asso-
ciation, hence affecting the bulk conductivity. It is anticipated
that this study not only provides suggestions for optimization
directions, but also reveals further insights into the possible ion
conducting mechanism. The latter is fundamental for tailored
improvement of the current electrolytes and to provide materials
for low-cost secondary batteries of high voltage, capacity, and
rate-capability, in strong demand for the potential mass market of
electric transport and the urgent need to reduce CO2 emissions.
In agreement with these considerations, we have chosen two
series of model compounds with the objective to cover a preferably
broadrange ofglass transition temperatures. One seriesisbased on
the relatively flexible cyclotriphosphazene (CTP). The polypho-
sphazene itself, which is well-known for low glass transition
temperatures,14ꢀ16 was not chosen, since building up the cyclic
carbonate group from an alkenole functionalized polyphospha-
zene via epoxidation with 3-chloroperoxybenzoic acid potentially
leads to degradation and chain cleavage due to the rather harsh and
oxidative reaction conditions. Indeed, degradation and rearrange-
ment of a variety of different functionalized polyphosphazenes is
well documented,14,16ꢀ18 but imperfectly defined substances are
not suitable for the outlined investigations.
Thermogravimetric analysis (TGA) was measured under nitrogen at
a heating rate of 10 K minꢀ1 using a TGA/SDTA-851 (Mettler-Toledo).
Differential scanning calorimetry (DSC) was carried out on a Mettler-
Toledo DSC-30 under nitrogen at a heating rate of 10 K minꢀ1
.
Impedance spectroscopy was recorded as a function of temperature
using a Solartron SI 1260 impedance/gain phase analyzer with a high
resolution dielectric converter (Alpha high-resolution dielectric analy-
zer, Novocontrol) in the range from 10ꢀ2 to 107 Hz. The measurements
were performed using plain, polished stainless steel or platinum electro-
des with a temperature controlled cryostat under N2 (Novocontrol).
Direct current (dc) conductivities were usually obtained from the lower-
frequency plateau values of the real part of the alternating current (ac)
conductivities (Bode plot).
Preparation of the Blends. The model compounds and the
corresponding amount of the respective lithium salt were dissolved in a
minimum amount of dry THF (H20 e 50 ppm, dry DMF was used in
case of 13) in a nitrogen filled glovebox. The homogeneous solutions
were concentrated in vacuum (10ꢀ3 mbar) at 80 °C for several days.
Complete evaporation of the organic solvent was checked by NMR
spectroscopy. Finally the viscous, honeylike electrolytes were placed
between polished stainless steel electrodes keeping a fixed distance of
the electrodes of about 0.1 mm by means of Teflon spacers.
Synthesis. 1. CTP-Based Model Compounds
2,2,4,4,6,6-Hexakis-(5-hexenyl-1-oxy)-cyclotriphosphazene (1). 5-
Hexen-1-ol (4.61 g, 46.02 mmol) was added to a suspension of NaH
(95%, 1.16 g, 45.97 mmol) in 20 mL of dioxane. The mixture was stirred
at 50 °C for 24 h under argon atmosphere to form the corresponding
alkoxide. N3P3Cl6 (1.32 g, 3.80 mmol) was dissolved in 60 mL freshly
distilled dioxane and added dropwise to the alkoxide suspension. The
resultant mixture was stirred at 60 °C for 48 h under an atmosphere of
dry argon. Completion of the reaction was checked by 31P NMR. The
mixture was then filtrated and the solvent evaporated in vacuum. The
crude product was purified by column chromatography (CHCl3/
MeOH, 100/1, v/v, retardation factor Rf = 0.80) leading to a colorless,
viscous liquid 1 (2.13 g, 2.92 mmol, 77% yield).1H NMR (DMSO, 250
MHz): δ 1.33ꢀ1.44 (m, 12H, —OCH2CH2—), 1.53ꢀ1.63 (m, 12H,
—CH2CH2CHd), 1.96ꢀ2.05 (m, 12H, —CH2—CHd), 3.77ꢀ3.85
(m, 12H, —OCH2—), 4.91ꢀ5.04 (m, 12H, dCH2), 5.69ꢀ5.86 (m,
6H, —CHdCH2).13C NMR (DMSO, 250 MHz): δ 24.4, 29.1, 32.7,
64.9, 114.9, 138.4. 31P NMR (DMSO, 700 MHz): δ 18.03 (s). MALDI-
TOF m/z (%) 729.9 (100).
The second series is based on the rather stiff hexaphenylben-
zene (HPB), possibly allowing self-assembly and local packing.19
In both cases, the tethered cyclic carbonate (2-oxo-1,3-diox-
olane) groups serve as ion solvating moieties. The spacer length
between the cyclic carbonate moieties and the different cores as
well as the position of the cyclic carbonate was varied, mainly to
reveal the impact of the local mobility of the solvating units on
the performance as lithium ion conductor.
’ EXPERIMENTAL SECTION
Materials. 5-hexen-1-ol, 5-cis-octen-1-ol, 9-decen-1-ol, and tetra-
ethylene glycol monomethyl ether were dried over calcium hydride and
distilled under argon before use. Tetrahydrofuran (THF, 99.9%, Acros)
and dioxane (99.9%, Acros) were dried over sodium and distilled under
argon before use. Acetonitrile and methylene chloride were dried with
activated molecular sieves (4 Å). Hexachlorocyclotriphosphazene
(99.99%, resublimed, Aldrich) and all other chemicals were used as
received without further purification unless indicated otherwise. Air or
moisture-sensitive reactions were performed in thoroughly flame-dried
glass vessels in an inert atmosphere of dry argon using standard Schlenk
techniques. Analytical thin layer chromatography (TLC) was carried out
on F-254 percoated silica gel 60 plates (Machery Nagel). Visualization
was performed with UV light (254 nm), potassium permanganate
solution, or iodine stain. Column chromatography on all compounds
was conducted with silica gel 60 (0.063ꢀ0.2 mm pore size) from Fluka.
Measurements. 1H and13C NMR spectra were recorded on a Bruker
AC 300 MHz and a Bruker Avance III 250 MHz spectrometer; 31P NMR
spectra were measured with either a Bruker Avance III 500 MHz or a
Bruker Avance III 700 MHz spectrometer. 1H and 13C chemical shifts are
quotedonthe δ-scale in unitsof parts per million(ppm) usingtheresidual
solvent protons as internal standard. 31P signal shifts were referred to
triphenylphosphine (TPP) (δ = ꢀ6.00 ppm) as external reference.
Coupling constants (J) are reported in hertz (Hz), and splitting patterns
are designated as s (singlet), d (doublet), dd (double doublet), t (triplet),
m (multiplet), and br (broad). MALDI mass spectra were obtained on a
Bruker-Daltonics Reflex-Tof.
2,2,4,4,6,6-Hexakis-(4-(oxiran-2-yl)-but-1-oxy)-cyclotriphospha-
zene (2). 2,6-Ditert-butyl-4-methylphenol (BHT) (7.67 g, 34.81 mmol)
and 1 (1.50 g, 2.06 mmol) were dissolved in 50 mL methylene chloride.
The solution was cooled to 0 °C and 3-chloroperoxybenzoic acid
(MCPBA) (77%, 7.81 g, 34.85 mmol) was added portionwise. The
mixture was stirred vigorously for 48 h. After complete epoxidation, the
mixture was washed three times with 10 wt % sodium hydroxide solution
and water. The organic phase was dried over MgSO4 and then
concentrated to dryness. After purification by column chromatography
(gradient, n-hexane/ethyl acetate, 5/1, v/v, passing over to ethyl acetate
and finally MeOH, Rf,n-hexane/ethylacetate,5/1 = 0.11) compound 2 was
obtained as a slightly orange, viscous oil (1.34 g, 1.62 mmol, 79% yield).
1H NMR (CDCl3, 250 MHz): δ 1.35ꢀ1.80 (m, 36H, —OCH2—
(CH2)3—), 2.42 (dd, 6H, 2J = 4.9, 3J = 2.7, —CH—(—O—)—CH2),
2.70 (dd, 6H, 2J = 4.8, 3J = 3.6, —CH—(—O—)—CH2), 2.80ꢀ2.92
(m, 6H, —CH—(—O—)—CH2), 3.80ꢀ3.95 (m, 12H, —OCH2—).
13C NMR (CDCl3, 250 MHz): δ 22.5, 30.1, 32.2, 47.1, 52.3, 65.7. 31
P
NMR (DMSO, 700 MHz): δ 17.24 (s). MALDI-TOF m/z (%) 825.9
(100).
2,2,4,4,6,6-Hexakis-(4-(4-butoxy)-1,3-dioxolan-2-one)-cyclotripho-
sphazene (3). Tetrabutylammonium iodide (0.18 g, 0.49 mmol), tribu-
tyltin iodide (0.20 g, 0.49 mmol), and 2 (1.34 g, 1.62 mmol) were
homogenously mixed in 25 mL of THF. Under N2 atmosphere, CO2 was
2121
dx.doi.org/10.1021/cm103559d |Chem. Mater. 2011, 23, 2120–2129

Chemistry of Materials
ARTICLE
passed continuously through the clear solution for 24 h at 50 °C. The
mixture was then concentrated and the crude product was purified by
column chromatography (ethyl acetate/THF, 2/1, v/v, Rf = 0.48).
Product 3 was obtained as a colorless, highly viscous oil (1.01 g, 0.93
2,2,4,4,6,6-Hexakis-(4-ethyl-5-(4-butoxy)-1,3-dioxolan-2-one)-cy-
clotriphosphazene (9). Chlorocobalt tetraphenylporphyrin ((TTP)
CoIIICl) catalyst was synthesized following a reported procedure.20ꢀ22
A stainless steel autoclave (250 mL) was charged with a solution of 8
(0.50 g, 0.50 mmol), (TPP)CoIIICl (2 mol %) and N,N-dimethylami-
nopyridine (DMAP, 4 mol %) in 15 mL methylene chloride. After
purging with nitrogen, the reaction was started by pressurization of the
solution with CO2 up to 50 bar. The mixture was then heated to 100 °C
accompanied by a pressure increase to 80 bar and stirred for 72 h at this
temperature. The autoclave was allowed to cool to room temperature
and excess CO2 was discharged. The mixture was concentrated, and the
crude product was purified by column chromatography (ethyl acetate/n-
hexane, 5/1, v/v, Rf = 0.25). Product 9 was obtained as slightly orange,
highly viscous oil (0.42 g, 3.34 mmol, 67% yield). 1H NMR (CDCl3, 250
MHz): δ 1.03 (t, 18H, 3J = 7.3, —CH3), 1.36ꢀ1.83 (m, 48H, —CH2CH3,
—OCH2(CH2)3—), 3.90 (br, 12H, —OCH2—), 4.13ꢀ4.30 (m, 0.72 H,
trans-CH2—CH—(O(CdO)O)—CH—C2H5), 4.48ꢀ4.72 (m, 11.28
H, cis-CH2—CH—(O(CdO)O)—CH—C2H5). 13C NMR (CDCl3,
250 MHz): δ 10.3, 22.4, 22.5, 28.7, 29.9, 65.7, 80.0, 81.5, 154.9. 31P
NMR (CDCl3, 700 MHz): δ 18.57 (s). MALDI-TOF m/z (%) 1258
(100).
1
mmol, 57% yield). H NMR (DMSO, 250 MHz): δ 1.23ꢀ1.51 (m,
12H, —OCH2CH2CH2—), 1.52ꢀ1.83 (m, 24H, —OCH2CH2—, —
CH2CH—(O(CdO)O)—CH2), 3.84 (m, 12H, —OCH2—), 4.11
2
3
(dd, 6H, J = 8.0, J = 7.2, —CH—(O(CdO)O)—CH2), 4.56 (dd,
6H, 2J = 8.1, 3J = 7.2, —CH—(O(CdO)O)—CH2), 4.73ꢀ4.84 (m,
6H, —CH—(O(CdO)O)—CH2). 13C NMR (DMSO, 250 MHz):
δ 20.5, 29.2, 32.4, 64.9, 69.2, 76.9, 154.9. 31P NMR (DMSO, 500
MHz): δ 17.21 (s). MALDI-TOF m/z (%) 1090 (100).
2,2,4,4,6,6-Hexakis-(9-decenyl-1-oxy)-cyclotriphosphazene (4).
Synthetic procedure followed that for compound 1 to afford 4 as
colorless liquid (column chromatography, n-hexane/ethyl acetate, 4/1,
v/v, Rf = 0.88, 82% yield). 1H NMR (CDCl3, 250 MHz): δ 1.17ꢀ1.44
(m, 60H, —OCH2CH2—(CH2)5—), 1.52ꢀ1.70 (m, 12H, —OCH2
CH2—), 1.94ꢀ2.06 (m, 12H, —CH2—CHd), 3.82ꢀ3.92 (m, 12H, —
OCH2—), 4.85ꢀ5.01 (m, 12H, dCH2), 5.69ꢀ5.85 (m, 6H, —
CHdCH2). 13C NMR (CDCl3, 250 MHz): δ 25.9, 29.1, 29.3, 29.5,
29.6, 30.5, 34.0, 66.0, 114.4, 139.3. 31P NMR (CDCl3, 700 MHz):
δ 17.85 (s). MALDI-TOF m/z (%) 1067 (100).
2,2,4,4,6,6-Hexakis-(2-(2-(2-(2-methoxyethoxy)ethoxy)ethoxy)ethoxy)-
cyclotriphosphazene (10).23. Synthetic procedure followed that for
compound 1 to afford 10 as colorless oil (column chromatography, n-
2,2,4,4,6,6-Hexakis-(8-(oxiran-2-yl)-oct-1-oxy)-cyclotriphospha-
zene (5). Synthetic procedure followed that for compound 2 to afford 5
as slightly orange, viscous oil in 78% yield (column chromatography,
n-hexane/ethyl acetate, 10/1, v/v, passing over to pure ethyl acetate,
Rf,n-hexane/ethylacetate,5/1 = 0.2). 1H NMR (CDCl3, 250 MHz): δ 1.18ꢀ1.71
(m, 84H, ꢀOCH2ꢀ(CH2)7ꢀ), 2.44 (dd, 6H, 2J = 5.2, 3J = 2.6, ꢀCHꢀ-
(ꢀOꢀ)ꢀCH2), 2.73 (dd, 6H, 2J = 5.1, 3J = 4.0, ꢀCHꢀ(ꢀOꢀ)ꢀCH2),
2.82ꢀ2.92 (m, 6H, ꢀCHꢀ(ꢀOꢀ)ꢀCH2), 3.79ꢀ3.95 (m, 12H,
ꢀOCH2ꢀ). 13C NMR (CDCl3, 250 MHz): δ 25.9, 26.2, 29.4, 29.6,
29.7, 30.5, 32.7, 47.3, 52.6, 66.0. 31P NMR (DMSO, 700 MHz): δ17.25 (s).
2,2,4,4,6,6-Hexakis-(4-(8-octoxy)-1,3-dioxolan-2-one)-cyclotri-
phosphazene (6). Synthetic procedure followed that for compound 3 to
afford 6asslightlyyellow, viscousoil(columnchromatography,n-hexane/
1
hexane/THF, 1/3, v/v, Rf = 0.28, 90% yield). H NMR (CDCl3, 250
MHz): δ 3.34 (s, 18H, —CH3), 3.48ꢀ3.54 (m, 12H, P—OCH2CH2—),
3.55ꢀ3.68 (m, 72H, —O—(CH2CH2O)3—CH3), 4.00 (br, 12H, P—
OCH2—). 13C NMR (CDCl3, 250 MHz): δ 59.3, 65.2, 70.2, 70.7, 70.8,
72.1. 31P NMR (CDCl3, 700 MHz): δ 17.11 (s). MALDI-TOF m/z (%)
1379 (100).
2. HPB-Based Model Compounds
Hexakis-(4-(hept-4-enoate)-phenyl)-benzene (11). 6-Heptenoyl
chloride was synthesized following a reported procedure.24 Hexakis-
(4-hydroxyphenyl)-benzene25 (0.20 g, 0.32 mmol) and dry pyridine
(125 mL) were mixed at 60 °C in a Schlenk flask. Under an N2
atmosphere, a solution of 6-heptenoyl chloride (0.56 g, 3.81 mmol) in
5 mL dry acetonitrile was added dropwise at 60 °C under stirring. The
temperature was increased to 85 °C, and the mixture was stirred at this
temperature for 24 h. Subsequently, 15 mL acetonitrile, 15 mL ethyl
acetate, and 15 mL water were added, and the organic phase was washed
three times with 10 mL of 15 wt % aqueous sodium carbonate and once
with 15 mL water. The combined organic layer was dried over MgSO4,
concentrated, and the residue was subjected to column chromatography
with n-hexane/ethyl acetate (3/1, v/v, Rf = 0.65) as eluent. Product 11
was obtained as a white powder (0.35 g, 0.27 mmol, 84% yield). 1H NMR
(CDCl3, 250 MHz): δ 1.35ꢀ1.51 (m, 12H, —CH2CH2—CHdCH2),
1.57ꢀ1.75 (m, 12H, —O—C(dO)—CH2CH2—), 1.96ꢀ2.13 (m,
1
ethyl acetate, 4/1, v/v, Rf = 0.55, 75% yield). H NMR (CDCl3, 250
MHz): δ 1.17ꢀ1.53 (m, 60H, —OCH2CH2—(CH2)5—), 1.54ꢀ1.82
(m, 24H, —OCH2CH2—, —CH2CH—(O(CdO)O)—CH2), 3.87
(m, 12H, —OCH2—), 4.05 (dd, 6H, 2J = 8.3, 3J = 7.1, —CH
—(O(CdO)O)—CH2), 4.52 (dd, 6H, 2J = 8.2, 3J = 7.1, —CH
—(O(CdO)O)—CH2), 4.64ꢀ4.75 (m, 6H, —CH—(O(CdO)O)
—CH2). 13C NMR (CDCl3, 250 MHz): δ 24.6, 25.9, 29.3, 29.5, 30.4,
34.1, 66.0, 69.6, 77.3, 155.4. 31P NMR (CDCl3, 700 MHz): δ 17.22 (s).
MALDI-TOF m/z (%) 1426 (100).
2,2,4,4,6,6-Hexakis-(cis-5-octenyl-1-oxy)-cyclotriphosphazene (7).
Synthetic procedure followed that for compound 1 to afford 7 as
colorless, viscous liquid (column chromatography, methylene chloride,
Rf = 0.85, 89% yield). 1H NMR (CDCl3, 250 MHz): δ 0.91 (t, 18H, 3J =
7.3, —CH3), 1.31ꢀ1.46 (m, 12H, —OCH2CH2—), 1.54ꢀ1.70 (m,
12H, —OCH2CH2CH2—), 1.92ꢀ2.06 (m, 24H, —CH2—CHdCH
—CH2—), 3.83ꢀ3.92 (m, 12H, -OCH2-), 5.20ꢀ5.40 (m, 12H,
ꢀCH2ꢀCH=CH-CH2ꢀ). 13C NMR (CDCl3, 250 MHz): δ 14.5,
20.7, 26.0, 26.8, 30.0, 65.8, 128.8, 132.2. 31P NMR (CDCl3, 700 MHz):
δ 18.56 (s). MALDI-TOF m/z (%) 898 (100).
3
12H, —CH2—CHdCH2), 2.42 (t, 12H, J = 7.4, —O—C(dO)—
CH2—), 4.88ꢀ5.06 (m, 12H, —CH2—CHdCH2), 5.67ꢀ5.88 (m, 6H,
—CH2—CHdCH2), 6.63 (d, 12H, 3J = 8.7, ortho-ArH), 6.74 (d, 12H,
3J = 8.7, meta-ArH). 13C NMR (CDCl3, 250 MHz): δ 24.5, 28.5, 33.6,
34.4, 115.0, 120.3, 132.4, 137.7, 138.6, 140.2, 148.8, 171.8. MALDI-TOF
m/z (%) 1315 (Naþ-Peak, 100).
Hexakis-(4-(5-oxiran-2-yl-pentanoate)-phenyl)-benzene (12). 11
(0.35 g, 0.27 mmol) was dissolved in 80 mL methylene chloride. The
solution was cooled to 0 °C, and MCPBA (77%, 0.61 g, 2.72 mmol) was
added portionwise. The mixture was stirred vigorously for 96 h. After
complete epoxidation, the mixture was washed three times with 2 wt %
sodium hydroxide solution and water. The organic phase was dried over
MgSO4 and then concentrated to dryness. The crude product 12 was
obtained as white powder and used without further purification (0.36 g,
0.26 mmol, 96% yield). 1H NMR (CDCl3, 250 MHz): δ 1.40ꢀ1.80 (m,
36H, —O—C(dO)—CH2—(CH2)3—), 2.36ꢀ2.50 (m, 18H, —O—
C(dO)—CH2—, —CH—(—O—)—CH2), 2.72 (dd, 6H, 2J = 5.0, 3J =
2,2,4,4,6,6-Hexakis-(4-(3-ethyloxiran-2-yl)-but-1-oxy)-cyclotripho-
sphazene (8). Synthetic procedure followed that for compound 2 to
afford 8 after column chromatography (methylene chloride/ethyl acetate,
1
Rf = 0.85) as slightly yellow, highly viscous oil in 82% yield. H NMR
(DMSO, 250 MHz): δ 0.89ꢀ1.02 (t, 18H, 3J = 7.3, ꢀCH3), 1.35ꢀ1.73
(m, 48H, ꢀCH2CH3, ꢀOCH2ꢀ(CH2)3ꢀ), 2.74ꢀ2.90 (m, 12H, ꢀCHꢀ
(ꢀOꢀ)ꢀCHꢀC2H5), 3.74ꢀ3.90 (m, 12H, ꢀOCH2ꢀ). 13C NMR
(DMSO, 250 MHz): δ 10.5, 20.6, 22.5, 26.6, 29.5, 56.0, 57.1, 64.9. 31P
NMR (DMSO, 700 MHz): δ 18.81 (s). MALDI-TOF m/z (%) 994 (32),
1016 (Naþ-Peak, 32), 1032 (Kþ-Peak, 100).
2122
dx.doi.org/10.1021/cm103559d |Chem. Mater. 2011, 23, 2120–2129

Chemistry of Materials
ARTICLE
4.0, —CH—(—O—)—CH2), 2.83ꢀ2.96 (m, 6H, —CH—(—O
reaction of 17 provides 1,2-bis(4-(2-(2-(2-(2-methoxyethoxy)ethoxy)
ethoxy)-ethoxy)phenyl)ethyne (18).28 After subsequent one-pot cyclo-
trimerization of 18 product 19 is obtained as colorless oil (overall yield
32%).25,26 1H NMR (CDCl3, 300 MHz): δ 3.33 (s, 18H, ꢀCH3),
3.49ꢀ3.55 (m, 12H, CH2ꢀOCH3), 3.58ꢀ3.68 (m, 60H, ꢀO(CH2-
CH2O)2ꢀOCH2CH2ꢀOCH3), 3.69ꢀ3.75 (m, 12H, ArꢀOꢀ
CH2CH2ꢀ), 3.86ꢀ3.93 (m, 12H, ArꢀOꢀCH2ꢀ), 6.37 (d, 12H, 3J =
8.7, ortho-ArH), 6.60 (d, 12H, 3J = 8.7, meta-ArH). 13C NMR (CDCl3,
300 MHz): δ 59.2, 67.1, 70.0, 70.7, 70.9, 113.1, 132.5, 133.8, 140.4,
156.1. MALDI-TOF m/z (%) 1171 (100), 1793 (Naþ-Peak, 14), 1809
(Kþ-Peak, 14).
3
3
—)—CH2), 6.63 (d, 12H, J = 8.6, ortho-ArH), 6.74 (d, 12H, J =
8.6, meta-ArH). 13C NMR (CDCl3, 250 MHz): δ 24.8, 25.7, 32.3, 34.4,
47.9, 54.3, 120.3, 132.4, 137.7, 140.2, 148.7, 171.6.
Hexakis-(4-(5-(2-oxo-1,3-dioxolan-4-yl)-pentanoate)-phenyl)-ben-
zene (13). A stainless steel autoclave (250 mL) was charged with a solution
of 12 (0.36 g, 0.26 mmol), tetrabutylammonium iodide (0.06 g, 0.16
mmol), and tributyltin iodide (0.07 g, 0.16 mmol) in 50 mL of freshly
distilled THF. After purging with nitrogen, the reaction was started by
pressurization of the solution with CO2 up to 50 bar. The mixture was
then heated to 80 °C accompanied by a pressure increase to 80 bar and
stirred for 120 h at this temperature. The autoclave was allowed to cool to
room temperature and excess CO2 was discharged. The product was
precipitated in a minimum amount of ethyl acetate, filtered, and washed
with 100 mLchilled methanol, 100 mLchilledhexane, and a small amount
of chilled acetone to obtain 13 as a white powder (0.34 g, 0.21 mmol, 79%
’ RESULTS AND DISCUSSION
Model Compound Synthesis. The synthetic route employed
to prepare 2-oxo-1,3-dioxolane substituted model compounds
based on CTP is following a three step reaction (Scheme 1a):
First, N3P3Cl6 is reacted with the appropriate sodium alkoxide of
ω-hexen-1-ol (1) or ω-decen-1-ol (4), respectively, followed by
epoxidation with 3-chloroperoxybenzoic acid (MCPBA) to yield
hexakis-(4-(oxiran-2-yl)but-1-oxy)-cyclotriphosphazene (2) and
hexakis-(8-(oxiran-2-yl)oct-1-oxy)-cyclotriphosphazene (5). It
is essential to use equimolar rather than catalytic amounts of
the antioxidant 2,6-ditert-butyl-4-methylphenol (BHT) referred
to the amount of MCPBA to prevent serious degradation of the
CTP.29 In the final step, 2 and 5 are converted into hexakis-(4-
(4-butoxy)-1,3-dioxolan-2-one)-cyclotriphosphazene (3) and
hexakis-(4-(8-octoxy)-1,3-dioxolan-2-one)-cyclotriphosphazene
(6) by transition metal catalyzed insertion of CO2 into the
epoxide ring.30
1
yield). H NMR (DMSO, 250 MHz): δ 1.23ꢀ1.81 (m, 36H, —O—
C(dO)—CH2—(CH2)3—), 2.41ꢀ2.53 (m, 12H, —O—C(dO)—
CH2—), 4.11 (dd, 6H, 2J = 8.0, 3J = 7.3, —CH—(O(CdO)O)—CH2),
4.55 (dd, 6H, 2J = 8.0, 3J = 7.2, —CH—(O(CdO)O)—CH2),
3
4.68ꢀ4.88 (m, 6H, —CH—(O(CdO)O)—CH2), 6.66 (d, 12H, J =
8.3, ortho-ArH), 6.87 (d, 12H, 3J = 8.3, meta-ArH). 13C NMR (DMSO,
250 MHz): δ 23.6, 23.8, 32.5, 33.2, 69.2, 76.9, 120.0, 131.8, 137.1, 139.7,
148.1, 154.9, 171.0.
Hexakis-(4-(undec-10-enoate)-phenyl)-benzene (14). Synthetic
procedure followed that for compound 11 to afford product 14 as
colorless oil (column chromatography, n-hexane/THF, 4/1, v/v, Rf =
0.53, 78% yield). 1H NMR (CDCl3, 250 MHz): δ 1.18ꢀ1.41 (m, 60H, —
O—C(dO)—(CH2)2—(CH2)2—, —(CH2)3—CH2—CHdCH2),
1.56ꢀ1.72 (m, 12H, —O—C(dO)—CH2CH2—), 1.97ꢀ2.05 (m,
3
12H, —CH2—CHdCH2), 2.40 (t, 12H, J = 7.1, —O—C(dO)—
CH2—), 4.86ꢀ5.03 (m, 12H, —CH2—CHdCH2), 5.70ꢀ5.87 (m, 6H,
—CH2—CHdCH2), 6.63 (d, 12H, 3J = 8.7, ortho-ArH), 6.74 (d, 12H, 3J
= 8.7, meta-ArH). 13C NMR (CDCl3, 250 MHz): δ 25.0, 29.1, 29.3, 29.4,
29.5, 34.0, 34.6, 114.4, 120.3, 132.4, 137.7, 139.4, 140.2, 148.8, 172.0.
MALDI-TOF m/z (%) 1628 (10), 1651 (Naþ-Peak, 100).
In order to obtain an ethyl-substituted cyclocarbonate, alco-
holate 7 was used. The CO2 insertion into the internal cis-
epoxide 8 using tributyltin iodide/tetrabutylammonium iodide
as catalyst system was not exceeding 80% of conversion, as
evidenced by 1H NMR, despite a higher reaction temperature of
100 °C compared to the terminal epoxides and an applied CO2
partial pressure of 95 bar. Alternatively, lithium bromide applied
in combination with n-methyl pyrrolidone under similar reaction
conditions than described before resulted also only in a partial
conversion of about 90%.31 However, a rather rarely reported
full conversion of CO2 insertion into the internal epoxide was
achieved in an autoclave using the more efficient catalyst system
consisting of 2 mol % chlorocobalt tetraphenylporphyrin ((TPP)
4CoIIICl) and 4 mol % dimethylaminopyridine (DMAP).20ꢀ22
The commercially available CoII-tetraphenylporphyrin was not
sufficient due to the limited catalytic activity compared to
oxidized (TPP)CoIIICl.22 In our case, the product cis:trans ratio
for (TPP)CoIIICl/DMAP amounts to 94:6, supporting the
proposed mechanistic pathway from Paddock et al., which
suggests retention of the configuration due to double inversion
of the stereochemistry at the attacked carbon.22 In contrast, for
the partially converted product using the catalyst system Bu4NI/
Bu3SnI mainly inversion of the configuration occurs as the
product cis:trans ratio was 20:80.
Hexakis-(4-(9-oxiran-2-yl-nonanoate)-phenyl)-benzene (15).
Synthetic procedure followed that for compound 12 to afford product
1
15 as white powder (95% yield). H NMR (CDCl3, 250 MHz): δ
1.18ꢀ1.71 (m, 84H, —O—C(dO)—CH2—(CH2)7—), 2.35ꢀ2.45
(m, 18H, —O—C(dO)—CH2—, —CH—(—O—)—CH2),
2.68ꢀ2.74 (dd, 6H, 2J = 5.1, 3J = 3.9, —CH—(—O—)—CH2),
2.82ꢀ2.91 (m, 6H, —CH—(—O—)—CH2), 6.61 (d, 12H, 3J = 8.6,
ortho-ArH), 6.73 (d, 12H, 3J = 8.6, meta-ArH). 13C NMR (CDCl3, 250
MHz): δ 25.0, 26.1, 29.3, 29.4, 29.5, 29.6, 32.7, 34.5, 47.3, 52.7, 120.3,
132.4, 137.7, 140.2, 148.8, 172.0. MALDI-TOF m/z (%) 1746 (Naþ-
Peak, 100), 1762 (Kþ-Peak, 60), 1786 (Naþ/Kþ-Peak, 33).
Hexakis-(4-(9-(2-oxo-1,3-dioxolan-4-yl)nonanoate)-phenyl)-
benzene (16). Synthetic procedure followed that for compound 13
to afford product 16 aswhite powder(77% yield). 1H NMR(CDCl3, 250
MHz): δ 1.21ꢀ1.86 (m, 84H, —O—C(dO)—CH2—(CH2)7—), 2.41
(t, 12H, 3J = 7.4, —O—C(dO)—CH2—), 4.05 (dd, 6H, 2J = 8.0, 3J =
7.2, —CH—(O(CdO)O)—CH2), 4.50 (dd, 6H, 2J = 8.1, 3J = 7.2, —
CH—(O(CdO)O)—CH2),
4.60ꢀ4.75
(m,
6H,
—CH
3
—(O(CdO)O)—CH2), 6.62 (d, 12H, J = 8.6, ortho-ArH), 6.74 (d,
12H, 3J = 8.6, meta-ArH). 13C NMR (CDCl3, 250 MHz): δ 24.6, 24.9,
29.1, 29.2, 29.3, 34.1, 34.5, 69.6, 77.4, 120.3, 132.4, 137.8, 140.4, 148.8,
155.3, 171.9. EA (%): calc. C 68.86; H 6.99. Found: C 68.78; H 6.91.
Hexakis-(4-(2-(2-(2-(2-methoxyethoxy)ethoxy)ethoxy)ethoxy)-phe-
nyl)-benzene (19). Compound 19 was prepared following a reported
procedure: 1-(2-(2-(2-(2-methoxyethoxy)ethoxy)ethoxy)ethoxy)-4-
bromobenzene (17) is formed by reacting 2-(2-(2-(2-methoxyethox-
y)ethoxy)ethoxy)ethyl-4-methylbenzensulfonate with 4-bromophenol
in presence of anhydrous K2CO3 in DMF.26,27 A one-pot Sonogashira
CTP-based model compounds with alternative side groups
containing less oxygen atoms, e.g. lactone side groups, were not
accessible starting from the epoxy functionalized substances 2
and 5. Rather, full conversion was not achieved following a
modified procedure applying 1-morpholino-2-trimethylsilyl acet-
ylene and BF3 OEt2,32 or the reaction conditions were too harsh
3
in terms of the elevated temperature as for e.g. the reaction of the
epoxides 2 and 5 with diethylmalonate.33 Consequently,
2123
dx.doi.org/10.1021/cm103559d |Chem. Mater. 2011, 23, 2120–2129

Chemistry of Materials
ARTICLE
Scheme 1. (a) Synthesis of Cyclic Carbonate and Oligo(oxyethylene) Functionalized Model Compounds Based on a CTP Core
and (b) Synthesis of Cyclic Carbonate and Oligo(oxyethylene) Functionalized Model Compounds Based on an HPB Core
significant degradation of the phosphazene ring was observed,
notable by several peaks around 0 ppm in the corresponding 31P
NMR spectra.
compounds. It is essential that the esterification is performed prior
to epoxidation, since otherwise the basic reaction conditions of the
esterification lead to epoxide ring-opening and subsequent forma-
tion of trans-1,2-dioles. In both cases, moreover, complete transi-
tion metal catalyzed CO2 insertion was successful only at a CO2
partial pressure of about 80 bar at 80 °C.
Attempts to synthesize an HPB-based model compound with
a shorter alkyl spacer, e.g. in order to allow for a higher stacking
probability of the HPB moieties, proved difficult due to limited
solubility. Since the resulting product out of the reaction of
hexakis-(4-hydroxyphenyl)-benzene and epichlorhydrin was
The synthesis route for the HPB-based model compounds
started from hexakis-(4-hydroxyphenyl)-benzene (Scheme 1b).25
Hexakis-(4-(hept-4-enoate)-phenyl)-benzene (11) and hexa-
kis-(4-(undec-10-enoate)-phenyl)-benzene (14) were synthesized
via esterification of 6-heptenoyl chloride and 10-undecenoyl chlor-
ide respectively with hexakis-(4-hydroxyphenyl)-benzene in the
presence of pyridine. The build-up of the cyclic carbonate side
group followed a similar way described above for the CTP model
2124
dx.doi.org/10.1021/cm103559d |Chem. Mater. 2011, 23, 2120–2129

Chemistry of Materials
ARTICLE
Scheme 2. Proposed Mechanism for the Thermal Induced
Rearrangement of the Cyclotriphosphazene Side Groups35
Figure 1. (a) 31P NMR (left) and 1H NMR spectra (right) of CTP model
compound 6. (b) 1H NMR spectrum of HPB model compound 16.
Figure 3. 31Pꢀ P COSY NMR of model compound 6 after heating for
31
72 h at 100 °C in DMSO.
1H and 31P NMR spectra of hexakis-(4-(8-octoxy)-1,3-dioxolan-
2-one)-cyclotriphosphazene (6). Similar to all CTP-based model
compounds, the 31P NMR of 6 exhibits a sharp singlet at þ17.22
ppm, proving that only equivalent phosphorus atoms exist in the
CTP core in agreement with a fully substituted phosphazene ring.
The 1H NMR spectra confirm the chemical structure, which were
further approved by 13C NMR and MALDI mass spectrometry.
Thermal Properties. All model compounds were stable up to
at least 250 °C when investigated by TGA under N2 applying a
heating rate of 10 K minꢀ1 (Figure 2). The onset of decomposi-
tion in case of the CTP model compounds bearing cyclic carbonate
units is at around 250 °C ( 5 °C, whereas the oligo(oxyethylene)
functionalized equivalent 10 is even stable up to 300 °C, indicating,
that the cyclic carbonate decomposes most likely before the
phosphazene ring does. Model compounds based on HPB cores
(13 and 16) appear to be slightly more stable compared to the
CTPs. Similar to the CTP analogues, the oligo(oxyethylene)
functionalized HPB model compound 19 is less sensitive to heat.
The thermal stability of the considered blends with lithium bis-
(trifluoromethanesulfone)imide (LiTFSI) for the particular com-
pounds was comparable.
Figure 2. TGA traces of the pure model compounds under N2 at a
constant heating rate of 10 K minꢀ1
.
hardly soluble in any common solvent, for the pressurized CO2
insertion, only a limited degree of conversion (≈67%, verified by
elemental analysis) was achieved.34 This also applies for the
reaction pathway starting from hexakis-(4-hydroxyphenyl)-ben-
zene and 4-pentenoyl chloride.
It is important to note that for the CTP model compounds, a
rearrangement of the side groups could be observed after
extensive heating at temperatures above 80 °C. The rearrange-
ment follows the proposed mechanism shown in Scheme 2 as
evidenced by 31P NMR.35
For both model compound series, an oligo(oxyethylene)
functionalized reference was synthesized (10 and 19).
Model Compound Characterization. All model compounds
are soluble in THF, dioxane or DMF. Characterization after
31
After heat treatment, the 31Pꢀ PCOSYNMR(Figure3) shows
1
purification and drying in vacuum by H, 13C, and 31P NMR
three multiplets at around þ15, þ7, and 0 ppm which are not
present in the initial sample. Strong 2J and 4J coupling between the
three phosphorus nuclei can be observed. The splitting pattern is
spectroscopy revealed well-defined, completely functionalized,
and highly pure products. Figure 1a shows the representative
2125
dx.doi.org/10.1021/cm103559d |Chem. Mater. 2011, 23, 2120–2129

Chemistry of Materials
ARTICLE
Figure 4. DSC traces (second heating) of the pure model compounds
under N2 at a constant heating rate of 10 K minꢀ1
.
Figure 5. Arrhenius plot of the temperature dependent conductivity of
the model compounds blended with LiTFSI (molar ratio Li:O = 1:50,
except for PDOA: Li:O = 1:16.6). The lines through the measured data
points represent the WLF fits.
in each case a double doublet due to the coupling of the particular
phosphorus nucleus with the remaining two magnetic nonequi-
valent phosphorus nuclei in the cycle. The NMR data indicate that
the CTP core remains intact, while the substitution pattern may be
a matter of the samples thermal history.
analysis of the nature of the glass transition and more important the
molecular basis of lithium ion transport by solid state NMR is
currently in progress. These studies potentially allow for a more
rational understanding of molecular dynamics within the matrix and
qualify for structureꢀproperty correlation in such kind of electro-
lytes. The results will be discussed in a forthcoming publication.
The oligo(oxyethylene) functionalized model compounds 10
and 19 exhibit lower glass transition temperatures than the cyclic
carbonate analogues, probably due to the higher mobility of the
oligo(oxyethylene) side groups and their weaker electrostatic
interactions compared to the more polar 2-oxo-1,3-dioxolane
functionalities.
Differential scanning calorimetry (DSC) was used to study
thermal transitions of the model compounds, whereby second
heating profiles were compared. At a heating rate of 10 K minꢀ1
,
apparent glass transitions (Tg) were observed for all CTP-based
substances (Figure 4).
The thermal behavior of the HPB-based model compounds is
more difficult to address, since the thermal transitions are very
much kinetically controlled. Whereas pure 13 shows a sharp,
reproducible melting transition at 152 °C, the equivalent model
compound 16 bearing a longer alkyl spacer shows a melting peak at
58 °Cjustinthefirst heating scan. Further heating and cooling scans
exhibit a glass transition at 3.5 °C only, where even prolonged
annealing for several hours at 15 °C did not lead to partial
crystallization.
The Tg’s of the model compounds cover a broad temperature
range. The Tg of the pure CTP model compound 3 is ꢀ28.9 °C.
An increase of the side chain length may induce a greater free
volume; hence the onset of segmental motion and the glass
transition is shifted to lower temperatures. Consequently, com-
pound 6 exhibits a slightly lower Tg as compared to compound 3.
The Tg of the model compound 9 with the internal 2-oxo-1,3-
dioxolane group is higher, probably because the mobility of the
side chain is decreased as compared to the mobility of the side
chains in compounds 3 and 6.
Notably, all blends discussed in this work have been investi-
gated in the amorphous state, which was confirmed based on
X-ray diffraction data.
Ionic Conductivity. At 40 °C, the ionic conductivities of the
homogeneous blends of the model compounds with LiTFSI reach
4.3 ꢁ 10ꢀ6 S cmꢀ1 in the case of hexakis-(4-(8-octoxy)-1,3-
dioxolan-2-one)-cyclotriphosphazene (6) and 8.9 ꢁ 10ꢀ8 S cmꢀ1
for the respective HPB model 16, each at a molar Li:O ratio of
1:50. The observable bulk conductivity of the model compounds
apparently scales with the spacer length, since the conductivity
decreases from models 6 to 3 and 16 to 13, respectively, when the
appropriate model compounds with longer and shorter alkyl
spacers between core and the lithium ion solvating 2-oxo-1,3-
dioxolane groups are compared (Figure 5).
Adding LiTFSI to the CTP-based model compounds increases
the glass transition temperatures with increasing lithium salt
concentration. In contrast, the glass transition temperature of the
HPB-based model compound 16 decreases slightly upon addi-
tion of LiTFSI. Besides, the blend of 13 with LiTFSI does not
show a melting peak as pure 13 does but instead a glass transition
well below the melting peak of pure compound 13.
This raises the question of the nature of the observed glass
transitions. While in the CTP-based models both core and side
chains are rather flexible and may be involved in the glass transition,
in the HPB-based models the core is rather stiffand merely the onset
of the side chain motion appears to fit the relatively low glass
transition temperatures of the blends. In both cases, a thorough
The oligo(oxyethylene) functionalized references 10 and 19
exhibit conductivities which exceed the ionic conductivities of
the 2-oxo-1,3-dioxolane functionalized analogues by about 2
orders of magnitude. This corresponds to the clearly lower glass
transition temperatures of these models (Table 1).
Interestingly, the conductivity of the CTP-based model com-
pound 9 with an ethyl-substituted cyclic carbonate exhibits an
even lower conductivity as its appropriate counterpart, model
compound 3 with unsubstituted cyclic carbonate. Obviously, the
ethyl side group at the carbonate cycle does not cause a weaker
interaction between the solvating moieties and the Li-ions.
In relation to the polymer which we studied earlier (poly(2-
oxo-1,3-dioxolan-4-yl)methyl acrylate, PDOA, Tg = 327 K),12
2126
dx.doi.org/10.1021/cm103559d |Chem. Mater. 2011, 23, 2120–2129

Chemistry of Materials
ARTICLE
Table 1. Apparent Glass Transition Temperatures, log σ
(at 40 °C), and the Calculated WLF Parameters C1 and C2 for
the Model Compounds Blended with LiTFSI in a Molar Ratio
Li:O = 1:50
Tref = Tg log σ (40 °C)
model compounda
(°C)
(S cmꢀ1
)
C1
C2 (K)
R2b
3
ꢀ18.4
ꢀ34.1
ꢀ12.1
ꢀ61.6
þ12.7
þ2.3
ꢀ5.89
ꢀ5.37
ꢀ6.38
ꢀ3.69
ꢀ8.30
ꢀ7.05
ꢀ5.18
24.02
24.29
27.10
20.96
26.56
22.30
25.02
49.35
52.63
48.27
58.52
39.46
44.44
50.76
0.999
0.999
0.999
0.997
0.998
0.999
0.999
6
9
10
13
16
19
ꢀ39.2
a See the Experimental Section or Scheme 1a and b for sample number-
ing. b R2 is the correlation coefficient obtained between linear fit and
experimental data by plotting (T ꢀ Tg)/log(σ/σ(Tg)) vs T ꢀ Tg.
Figure 6. WLF master curve of the temperature dependent conductiv-
ity of the model compounds blended with LiTFSI (molar ratio Li:O =
1:50) vs the reduced temperature T ꢀ Tg.
the phosphazene model 9 is showing only slightly higher ionic
conductivity despite the much lower Tg (Table 1). This is
consistent with the fact that, for PDOA, the reference tempera-
ture in a WLF-treatment is found to have values more than 100 K
below the actual Tg. The EO-functionalized HPB-based model
19 compares very well with poly(p-phenylene) bearing oligo-
(oxyethylene) side chains (PPP(EO)) with respect to conduc-
tivity. This is rather surprising, since PPP(EO) consists of a stiff
main chain layer embedded in an amorphous matrix of the EO
side chain segments exhibiting
a nanophase separated
structure.13 However, only the EO side chains were found to
contribute to the conductivity in PPP(EO), and therefore,
conductivity of 19 is in a similar range.
The temperature dependent DC conductivity of the various
model compounds blended with LiTFSI (Figure 5) does not
follow simple Arrhenius law, since a log(σ) ∼ Tꢀ1 behavior
cannot be observed. However, WilliamꢀLandelꢀFerry (WLF)
type behavior is found, which is related to the free volume theory.
Experimental conductivity data obey the WLF equation:
Figure 7. Temperature dependent ionic conductivity of the model
compound 6 blended with different lithium salts with different sized
counterions (molar ratio Li:O = 1:25).
C1ðT ꢀ Tref Þ
σðTÞ ¼ σðT0Þ exp
ð1Þ
C2 þ ðT ꢀ Tref Þ
for this differential behavior yet is currently unclear but may be
related to a different nature of the glass transition.
In eq 1 σ(T) is the ionic conductivity at a given temperature T,
Tref is the reference temperature, and C1 and C2 are two
parameters containing the temperature dependence of the ob-
servable ionic conductivity, which can be derived from the
experimental data simply by plotting (T ꢀ Tg)/log(σ(T)/σ(Tg))
versus T ꢀ Tg. For all model compound blends, the observed glass
transition temperatures Tg serve as reference temperature Tref.
The data are displayed in Table 1. Especially the C1 values are
similar for all considered compounds and are fairly close to the
universal value of C1 = 17.4.36 C2 carries information on the
“character” of the compound (universal C2 = 51.6 K),36 including
the presence of different side groups and/or cores. The full lines
through the data points shown in Figure 5 represent the respective
total fits. In addition, based on this data, a plot of log(σ) versus
(T ꢀ Tg) isleading to a mastercurve, where theexperimentalionic
conductivity data of the blends with LiTFSI fall together within
limits of error (Figure 6), except for the blended oligo-
(oxyethylene) functionalized CTP 10.
Notably, variation of the lithium salt had no significant
influence on the ionic conductivity of the CTP model com-
pounds, which implies, that the CTP models are suitable lithium
ion solvents for various lithium salts, irrespective of the salt
dissociation constants and counterion sizes. Figure 7 shows the
ionic conductivity of model compound 6 blended with LiTFSI,
lithium perchlorate (LiClO4), lithium bis(pentafluoroethane-
sulfone)imide (LiBETI), and lithium bis(oxalate) borate
LiBOB), each in molar ratio Li:O = 1:25.
The ionic conductivity differs in the low temperature regime
about 1 order of magnitude, which can be associated with
variations of the glass transition temperature of the appropriate
blends. The observable differences decrease with increasing
temperature simply because the influence of Tg on conductivity
decreases with increasing temperature. Thus, the blend with
LiClO4 exhibits the highest Tg but the lowest ionic conductivity.
For the HPB-based model compounds, blends of compound 16
with LiClO4 or LiBOB gave no fully homogeneous mixtures in
contrast to the ones with LiTFSI, illustrating the well-known
The fitting curve of compound 10 is shifted along the y-axis
about 1 order of magnitude, in contrast to the HPB-based
counterpart 19, which matches the master curve. The reason
2127
dx.doi.org/10.1021/cm103559d |Chem. Mater. 2011, 23, 2120–2129

Chemistry of Materials
ARTICLE
Since a higher ion mobility and improved salt dissociation is
expected in oligo(oxyethylene) functionalized substances as
compared to cyclic carbonate functionalized models, mixtures
of models 6 and 10 were studied (Figure 9).
For all blends, the conductivity values did not exceed the
values obtained for the pure oligo(oxyethylene) model com-
pound 10. This is in contrast to results reported by West et al. for
hybrid polysiloxanes bearing EO and cyclic carbonate side
chains.40 In the present case, no depression of the glass transition
temperatures as a result of blending could be observed. Identical
results were obtained for the mixtures of the respective HPB
model compounds 16 and 19.
Nevertheless, the absolute conductivity values of the cyclic
carbonate functionalized model compounds exhibit relatively
low values even at 40 °C, despite the comparatively low glass
transition temperatures. While for the HPB models, the elevated
glass transition temperatures appear to be the predominant
factor for the diminished ionic conductivity, for the low-Tg
CTP models, the results suggest that the interaction of the
lithium ions with the solvating 2-oxo-1,3-dioxolane moieties is
actually too strong. This conclusion is also supported by the
conductivity data of the previously mentioned open chain
polymer PDOA (Figure 5). In order to understand the discre-
pancy between low glass transition temperatures and low con-
ductivity values entirely, it is essential to get a closer insight into
the complex interplay between the mobility of the side groups
and their interactions with the lithium ions. In consideration of
what is known about the solvation shell of the lithium ions and
the transport mechanism, the lithium ion exchange between
solvating moieties can be described as being linked to local
rotations of the terminal groups. In the present case, still, at
ambient temperatures, those motion modes are supposed to be
active,41 consistent with the assumption that the coordination
itself comprises the limiting factor. Indeed, extensive solid state
NMR relaxation studies are currently under way to verify this
presumption and in order to obtain more details on the ion
dynamics. In addition, transference number studies of the blends
will provide important information on the dominant contribu-
tion to the observable conductivity.
Figure 8. Ionic conductivity at 30 °C and related glass transition
temperatures of the model compounds 3 and 6 blended with LiTFSI
as a function of molar ratio Li:O.
Figure 9. Temperature dependent ionic conductivity of the model
compounds 6 and 10 in different mixing ratios blended with LiTFSI
(molar ratio Li:O = 1:25).
plasticizing effect of LiTFSI.37 On the basis of these results,
blends with LiTFSI were chosen for further measurements on
the HPB models.
The dependence of the ionic conductivity of the blends
containing compounds 3 and 6 on the LiTFSI concentration is
shown in Figure 8. Furthermore, the development of the glass
transition temperatures is mapped.
The ionic conductivity exhibits a maximum for both blends,
while the glass transition temperatures increase with increasing
salt concentration. Since the ionic conductivity is proportional to the
number of effective charge carriers ni, their mobility qi, and the
electric charge ui,38 at low salt contents, the increase of charge
carriers is apparently responsible for an increase in conductivity. The
potentially displaced counteracting effect of an increase in the glass
transition temperature due to cation complexation and ionꢀdipole
interactions causes the observed maxima in ionic conductivity. Due
to the lower glass transition and most likely enhanced system
mobility of pure compound 6 over 3, the optimum molar Li:O
ratio for maximum ionic conductivity is 1:25 for 6 compared to 1:50
in the case of 3. Similar studies have been recently described for low-
Tg (ca. ꢀ80 °C) cyclosiloxane blends.39
’ CONCLUSIONS
In the present work, well-defined model compounds based on
CTP and HPB cores with tethered Liþ-solvents were successfully
synthesized. Adequate CTPs were obtained upon reaction of
N3P3Cl6 with appropriate alkenoles, followed by epoxidation
with MCPBA and catalytic CO2 insertion. In the epoxidation
step, equimolar amounts of antioxidant BHT are essential to
prevent degradation of the CTP cycle.
The synthesis of the HPB-based model compounds starts by
reacting hexakis-(4-hydroxyphenyl)-benzene with an enoyl
chloride. The buildup of the 2-oxo-1,3-dioxolane side group
follows the description for the CTP model substances; however,
for the catalyzed CO2 insertion, a high CO2 partial pressure of
about 80 bar was significant to obtain full conversion.
The well-defined model compounds show the onset of
decomposition above 250 °C and cover a broad range of glass
transition temperatures from ꢀ79 °C for the oligo(oxyethylene)
functionalized CTP up to þ3.5 °C for the HPB-based model
compound featured by an alkyl chain of eight carbons between a
core and Liþ-solvating group.
2128
dx.doi.org/10.1021/cm103559d |Chem. Mater. 2011, 23, 2120–2129

Chemistry of Materials
ARTICLE
All considered model compounds represent good solvents for
LiTFSI. Notably, for the CTP-based compounds even, no
significant differences in ionic conductivity were observed for
the blends with various lithium salts including LiBETI, LiBOB, or
LiClO4. In addition, the temperature-dependent ionic conduc-
tivities of the blends follow a WilliamꢀLandelꢀFerry (WLF)
type behavior with the corresponding glass transition tempera-
tures as the reference temperature. The highest obtained con-
ductivity value was 6.0 ꢁ 10ꢀ6 at 40 °C for the CTP model
bearing the longest alkyl spacer of eight carbons between the core
and solvating group, while shorter alkyl spacers reveal reduced
ionic conductivities. The HPB models exhibit higher glass
transition temperatures and conductivities at 40 °C being about
2 orders of magnitude lower than their CTP analogues. However,
compared to the previously studied poly(meth)acrylate analo-
gues with much higher glass transitions,12 especially the con-
ductivity of the low-Tg CTP models was unexpectedly low. The
discrepancy between low glass transition temperatures and rather
low ionic conductivities most likely indicates too strong interac-
tions between lithium ions and the solvating 2-oxo-1,3-dioxolane
moieties. This may hinder fast lithium ion mobility, thereby
causing lower ionic conductivities. Following the Pearson con-
cept, softer solvating moieties should be taken into account for
future studies. Furthermore, in order to prove the discussed
statements and to obtain more detailed information on the
system dynamics, extended solid state NMR relaxation studies
are currently in progress as well as transference number
measurements.
(14) Allcock, H. R. Chemistry and Applications of Polyphosphazenes;
Wiley-Interscience: New York, 2003; p 528.
(15) Kaskhedikar, N.; Burjanadze, M.; Karatas, Y.; Wiemh€ofer, H. D.
Solid State Ionics 2006, 177, 3129.
(16) Tada., Y.; Sato, M.; Takeno, N. Macromol. Chem. Phys. 1994,
195, 1923.
(17) Andrianov, A. K.; Marin, A.; Peterson, P. Macromolecules 2005,
38, 7972.
(18) DeCollibus, D. P.; Marin, A.; Andrianov, A. K. Biomacromole-
cules 2010, 11, 2033.
(19) Jimenez-Garcia, L.; Kaltbeitzel, A.; Pisula, W.; Gutmann, J. S.;
Klapper., M.; Muellen, K. Angew. Chem., Int. Ed. 2009, 48, 9951.
(20) Sugimoto, H.; Kuroda, K. Macromolecules 2008, 41, 312.
(21) Sakurai, T.; Yamamoto, K.; Naito, H.; Nakamoto, N. Bull.
Chem. Soc. Jpn. 1976, 49, 3042.
(22) Paddock, R. L.; Hiyama, Y.; McKay, J. M.; Nguyen, S. T.
Tetrahedron Lett. 2004, 45, 2023.
(23) Allcock, H. R.; Ravikiran, R.; O’Connor, S. J. M. Macromolecules
1997, 30, 3184.
(24) Gaul, C.; Njardarson, J. T.; Shan, D.; Dorn, D. C.; Wu, K.;
Tong, W. P.; Huang, X.-Y.; Moore, M. A. S.; Danishefsky, S. J. J. Am.
Chem. Soc. 2004, 126, 11326.
(25) Kobayashi, K.; Shirasaka, T.; Sato, A.; Horn, E.; Furukawa, N.
Angew. Chem., Int. Ed. 1999, 38, 3483.
(26) Lu, Y.; Suzuki, T.; Zhang, W.; Moore, J. S.; Marinas, B. J. Chem.
Mater. 2007, 19, 3194.
(27) Buizide, A.; LeBerre, N.; Sauve, G. Tetrahedron Lett. 2001,
42, 8781.
(28) Tohda, Y.; Sonogashira, K.; Hagihara, N. J. C. S. Chem.
Commun. 1975, 2, 54.
(29) Fantin, G.; Medici, A.; Fogagnolo, M.; Pedrini, P.; Gleria, M.
Eur. Polym. J. 1993, 29, 1571.
(30) Baba, A.; Nozaki, T.; Matsuda, H. Bull. Chem. Soc. Jpn. 1987,
60, 1552.
(31) Ochiai, B.; Hatano, Y.; Endo, T. J. Polym. Chem. Part A 2009,
47, 3170.
’ AUTHOR INFORMATION
Corresponding Author
*E-mail: meyer@mpip-mainz.mpg.de. Tel.: þ49 (0)6131 379 400.
Fax: þ49 (0)6131 379 480.
(32) Movassaghi, M.; Jacobsen, E. N. J. Am. Chem. Soc. 2002,
124, 2456.
(33) Bonsignore, S.; Dalcanale, E.; Martinengo, T. Synth. Commun.
1988, 18, 2241.
(34) Regla, I.; Luviano-Jardꢀon, A.; Demare, P.; Hong, E.; Torres-
Gavilꢀan, A.; Lꢀopez-Munguía, A.; Castillo, E. Tetrahedron: Asymm. 2008,
19, 2439.
’ ACKNOWLEDGMENT
The authors gratefully acknowledge assistance in the synthetic
work by A. Manhart and in retrieving impedance spectroscopy
data by C. Sieber. P. R€ader is acknowledged for assistance in
obtaining thermal analytical data of the model compounds as is M.
Wagner for supporting the 31P NMR spectroscopy. This research
was supported by the International Max Planck Research School.
(35) Doughty, S. W.; Fitzsimmons, B. W.; Reynolds, C. A. J. Chem.
Soc., Dalton Trans. 1997, 367.
(36) Williams, M. L.; Landel, R. F.; Ferry, J. D. J. Am. Chem. Soc.
1955, 77, 3701.
(37) Armand, M.; Gorecki, W.; Andreani, R.; Scrosati, B. Second
International Symposium on Polymer Electrolytes; Elsevier Applied
Science: New York, 1989; p 91.
’ REFERENCES
(38) Ratner, M. A. Polymer Electrolyte Reviews; MacCallum, J. R.,
Vincent, C. A., Eds.; Elsevier: London/New York, 1987; Chapter 7.
(39) Zhang, Z.; Lyons, L. J.; Jin, J. J.; Amine, K.; West, R. Chem.
Mater. 2005, 17, 5646.
(40) Zhang, Z.; Lyons, L. J.; West, R.; Amine, K.; West, R. Silicon
Chem. 2005, 3, 259.
(1) Winter, M.; Brodd, R. J. Chem. Rev. 2004, 104, 4245.
(2) Scrosati, B. JEC Battery Newsletters 1993, 6, 44.
(3) Schalkwijk, van, W., Scrosati, B., Eds. Advances in Lithium-Ion
Batteries; Kluwer Academic/Plenum: New York, 2002.
(4) Zhang, S. S. J. Power Sources 2007, 164, 351.
(5) Xu, K. Chem. Rev. 2004, 104, 4303.
(41) McCrum, N. G.; Read, B. E.; Williams, G. In Anelastic and
Dielectric Effects in Polymeric Solids; Wiley-Interscience: New York, 1967;
Chapter 8, p 238.
(6) Tsutsumi, H.; et al. Solid State Ionics 2003, 160, 131.
(7) Vogdanis, L.; Martens, B.; Uchtmann, H.; Hensel, F.; Heitz, W.
Macromolekul. Chem. 1990, 191, 465.
(8) Fenton, D. E.; Parker, J. M.; Wright, P. V. Polymer 1973, 14, 589.
(9) Meyer, W. H. Adv. Mater. 1998, 10, 439.
(10) Lewandowski, A.; Swiderska-Mocek, A. J. Power Sources 2009,
194, 601.
(11) Stephan, A. M.; Nahm, K. S. Polymer 2006, 47, 5952.
(12) Britz, J.; Meyer, W. H.; Wegner, G. Macromolecules 2007,
40, 7558.
(13) Lauter, U.; Meyer, W. H.; Wegner, G. Macromolecules 1997,
33, 2092.
2129
dx.doi.org/10.1021/cm103559d |Chem. Mater. 2011, 23, 2120–2129