Preparation and characterization of trifluoroethyl aliphatic carboxylates as co-solvents for the carbonate-based electrolyte of lithium-ion batteries

Article abstract of DOI:10.1016/j.jfluchem.2014.02.006

In this work, a series of trifluoroethyl aliphatic carboxylates with different carbon-chain lengths in acyl group are prepared and investigated as the co-solvents for the carbonate-based electrolyte of lithium-ion batteries. The trifluoroethyl aliphatic carbonates are synthesized by a modified one-step approach, using aliphatic carboxylic acid and trifluoroethanol as the raw materials (molar ratio, 1.2:1), hydrogen ion exchange resin as the catalyst and silica gel drier as the de-hydration. The structure and electrochemical properties of the final products have been characterized by FTIR, ¹H NMR, GC-MS, viscosity, conductivity meter and electrochemical measurements. The structure characterizations show that the final products have high purity. Electrochemical tests present that the co-solvents are able to improve the electrochemical performances of graphite electrode at low temperature. In particular, we find that an addition of trifluoroethyl n-hexanoate (TFENH) into 1 M LiPF₆/EC + EMC electrolyte can significantly decrease the Li de-intercalation potential of graphite by 540 mV and achieve a high capacity retention of 92% at 218 K. The electrochemical impedance spectroscopy (EIS) measurements indicate that the observed performance improvement at low temperature is associated with the decreased surface film resistance (R _SEI) by the addition of co-solvents.

Full text of DOI:10.1016/j.jfluchem.2014.02.006

Journal of Fluorine Chemistry 161 (2014) 110–119
Contents lists available at ScienceDirect
Journal of Fluorine Chemistry
journal homepage: www.elsevier.com/locate/fluor
Preparation and characterization of trifluoroethyl aliphatic
carboxylates as co-solvents for the carbonate-based electrolyte of
lithium-ion batteries
*
Wei Lu, Kai Xie, Zhong xue Chen, Yi Pan, Chun man Zheng
College of Aerospace Science and Engineering, National University of Defense Technology, Changsha, Hunan 410073, PR China
A R T I C L E I N F O
A B S T R A C T
Article history:
In this work, a series of trifluoroethyl aliphatic carboxylates with different carbon-chain lengths in acyl
group are prepared and investigated as the co-solvents for the carbonate-based electrolyte of lithium-
ion batteries. The trifluoroethyl aliphatic carbonates are synthesized by a modified one-step approach,
using aliphatic carboxylic acid and trifluoroethanol as the raw materials (molar ratio, 1.2:1), hydrogen
ion exchange resin as the catalyst and silica gel drier as the de-hydration. The structure and
electrochemical properties of the final products have been characterized by FTIR, 1H NMR, GC–MS,
viscosity, conductivity meter and electrochemical measurements. The structure characterizations show
that the final products have high purity. Electrochemical tests present that the co-solvents are able to
improve the electrochemical performances of graphite electrode at low temperature. In particular, we
find that an addition of trifluoroethyl n-hexanoate (TFENH) into 1 M LiPF6/EC + EMC electrolyte can
significantly decrease the Li de-intercalation potential of graphite by 540 mV and achieve a high capacity
retention of 92% at 218 K. The electrochemical impedance spectroscopy (EIS) measurements indicate
that the observed performance improvement at low temperature is associated with the decreased
surface film resistance (RSEI) by the addition of co-solvents.
Received 5 November 2013
Received in revised form 8 February 2014
Accepted 10 February 2014
Available online 19 February 2014
Keywords:
Trifluoroethyl aliphatic carboxylates
Co-solvent
Low temperature
Electrolyte
Lithium-ion battery
ß 2014 Elsevier B.V. All rights reserved.
1. Introduction
Laboratory (JPL) has developed three generations of successful
electrolytes over the last two decades [1,4,5,7,8]. The two previous
At present, the electrolytes composed of lithium hexafluor-
ophosphate (LiPF6) salts and carbonate-based solvents are
commonly used in commercialized lithium-ion batteries. Howev-
er, the performances of graphite electrode in these electrolytes at
low temperature are extremely poor, which can be ascribed to the
following two reasons: (1) the decreasing ionic conductivity of the
electrolytes caused by the increased viscosity at low temperature
[1]; (2) the largely decreased lithium-ion diffusion rate across the
solid electrolyte interface (SEI) film on graphite electrode at low
temperature [2–3].
In order to improve the low temperature performance of current
electrolyte, all-carbonate electrolytes with lower concentrations of
EC have been developed by many groups [1,4–7]. Carbonate-based
electrolyte consisting of 1.0 M LiPF6 in EC + DEC + DMC (1:1:1) has
been shown to have excellent performance at 253 K [4]. Under
sponsorship by the Mars Exploration Program, Jet Propulsion
generations of low temperature electrolytes are based on all-
carbonatemixedsolvents, includedEC:DEC:DMC:EMC(1:1:1:3)and
EC:EMC (1:4) – Gen 1 and 2.
Adding co-solvents with low freezing point, low viscosity and
good electrochemical stability is another effective method to
enhance the low temperature performance of the electrolyte. The
third generation of JPL electrolyte for low temperature is based on
1.0 M LiPF6 in EC + EMC (1:4) mixtures with ester co-solvents (such
as methyl propionate and ethyl butyrate). JPL have reported
several type cells which show outstanding low temperature
performances by using their electrolyte [6–8]. They have done
more research on the polarize voltage and the residences at low
temperature, but not explained the SEI formation mechanism
when the ester co-solvents added. During the 1990s, several
researchers have clearly discussed the surface chemistry on
graphite anodes in several carbonate-based electrolyte systems
which could form SEI film [9–11]. But why aliphatic carboxylates
could enhance the low temperature performance of the carbonate-
based electrolyte? And what role in forming SEI film does the
aliphatic carboxylate co-solvent play? These problems have not
reached agreement.
*
Corresponding author. Tel.: +86 073184573149; fax: +86 073184576578.
E-mail addresses: zhengchunman@hotmail.com, taigold3rd@hotmail.com
(C.m. Zheng).
http://dx.doi.org/10.1016/j.jfluchem.2014.02.006
0022-1139/ß 2014 Elsevier B.V. All rights reserved.

W. Lu et al. / Journal of Fluorine Chemistry 161 (2014) 110–119
111
Avarietyofcompoundsareinvestigatedastheco-solventsforthe
2. Results and discussion
carbonate electrolytes. Among them, fluorinated aliphatic carbox-
ylateisone ofthemostpromisingcandidates. Ithasbeenprovedthat
partially fluorinated esters can effectively decrease the viscosity of
the electrolyte, and are favorable for the formation of stable SEI film
on graphite electrode at low temperature [6,12,13]. Smith et al.
described favorable results with trifluoroethyl butyrate, ethyl
trifluoroacetate, and trifluoroethyl acetate and methyl pentafluor-
opropionate. The results showed that the electrolyte consisting of
1.0 M LiPF6/EC + EMC + trifluoroethyl butyrate (20:60:20 vol%) had
a better performance in nearly all tests than the other fluorinated
ester containing mixtures [6]. Nakajima et al. investigated methyl
difluoroacetate, methyl hexafluoroisobutyrate and the other three
fluorinated esters. They found a low molecular weight fluoroester
CHF2COOCH3-mixed 1 M LiClO4/EC + DEC demonstrated the much
larger charge capacities than the base-line electrolyte [12].
Trifluoroethyl aliphatic carboxylates were synthesized by a
modified one-step approach using aliphatic carboxylic acid and
trifluoroethanol as the raw materials (molar ratio, 1.2:1). After
˚
drying over anhydrous MgSO4 and 4 A molecular sieves, the
moisture contents of products were controlled at a level of less
than 20 ppm, which could meet the requirements of electrolyte for
lithium-ion batteries. The final yield could reach about 85%. The
four products, trifluoroethyl acetate (CH3COOCH2CF3), trifluor-
oethyl n-butyrate (CH3(CH2)2COOCH2CF3), trifluoroethyl n-hex-
anoate (CH3(CH2)4COOCH2CF3) and trifluoroethyl normal-
octanoate acid (CH3(CH2)6COOCH2CF3) were denoted as TFEA,
TFENB, TFENH and TFENO, respectively.
2.1. Chemical structure characterization
Moreover, most fluorinated esters have much higher flash
points than that of hydrocarbon-based carbonates, which benefit
the safety of lithium-ion battery [14–18]. Sato et al. did some
research on the thermal stabilities of electrolytes containing a
series of fluorinated carboxylic acid esters using differential
scanning calorimeter. Among these fluorinated carboxylic esters,
methyl difluoroacetate exhibited the highest onset temperature
and the smallest amount of exothermic heat at the coexistence of
lithium. In 2011, they investigated that the exothermically
decomposed temperature of the methyl difluoroacetate – based
LiPF6 solution – was higher than 723 K. So, this electrolyte was
considered to be a good candidate for safer Li-ion batteries [16,17].
Yamaki et al. studied the thermal stability of fluorinated ester
electrolytes with and without lithium metal and the positive
electrode material at the charged state. According to DSC
measurement, LiPF6/methyl difluoroacetate also showed the best
stabilization with both lithium metal and Li0.5CoO2 [18].
Trifluoroethyl aliphatic carboxylates, a member of the fluori-
nated aliphatic carboxylates family, are usually synthesized by
liquid phase esterification. Conventional esterification always
involves strong acid catalysts such as H2SO4, HCl, HF, H3PO4 and p-
toluene sulfonic acid, etc. These acids are corrosive, and the
neutralization of the excess acid after the reaction may pollute the
environment. Furthermore, the side reaction caused by strong acid
catalysts may generate acidic impurities which are harmful for
lithium-ion batteries. In contrast, using solid-state catalysts in
liquid phase esterification is one way of avoiding the problems
above, and the catalysts could be easily separated from products by
simple filtration. Hence, considerable attention has been paid to
the heteropolyacid catalysts for the esterification reactions, such as
H4SiW12O40ÁnH2O, etc. [19,20] However, high cost prohibits their
practical use. As alternative catalyst, hydrogen ion exchange resin
has low cost and high efficiency for esterification reactions.
In this paper, a series of trifluoroethyl aliphatic carboxylates with
different carbon-chain lengths in acyl group were prepared by
modified one-step synthesis, using aliphatic carboxylic acid and
trifluoroethanol as the raw materials (molar ratio, 1.2:1). In order to
enhance the yield of ester, hydrogen ion exchange resin was used as
the catalyst, and silica gel was selected as the dehydrating agent to
remove water during esterification process. Both hydrogen ion
exchange resin and silica gel, in this reaction, could be dried for
reuse, showing that the method presented in this paper could be
economic and green. The structure characterizations including FTIR,
1HNMRandGC–MS, revealedthatthefinalproductshadhighpurity.
By adding the trifluoroethyl aliphatic carboxylates into 1 M LiPF6/
EC + EMC (1:4) electrolyte as co-solvents, the electrochemical
performances of graphite electrode at low temperature were greatly
improved. Further EIS measurements indicated that the improve-
ment was related to the decreased surface film resistance (RSEI) with
the addition of co-solvents.
The FTIR spectra of TFEA, TFENB, TFENH and TFENO samples are
shown in Fig. 1; at the same time the attribution of the main
absorption peaks are listed in Table 1.
(1) In the wave number region of 3000–2815 cmÀ1, all the peaks
are attributed to the C–H bond stretching. In TFEA molecule,
due to the strong electron-withdrawing effect of the C55O
group, there is no marked peak in this region attributed to
stretching vibration of –CH3 at
a-site of C55O group. With the
increase of the carbon-chain length in acyl group, the electron-
withdrawing effect of the C55O group is gradually weakened.
For this reason, in TFENH and TFENO, peaks attributed to C–H
stretching vibration of long acyl group gradually become
significant, and at the same time move to lower wave number.
(2) The strong and sharp absorption peak at 1790–1740 cmÀ1
could be assigned to the C55O group stretching. This peak is
shifted to a lower wave number as the alkyl group in the
structure increases, which can be explained by the stronger
electron repulsive effect in longer alkyl group.
(3) There are two sharp strong absorption peaks at 1200–
1050 cmÀ1, which can be indexed to the stretching vibration
of C–O–C in ester group. Peaks at 1200–1130 cmÀ1 with higher
intensity and strong sensitivity onthe lengthof the carbon-chain
in acyl group are attributed to the anti-symmetric stretching
vibration. The wave number of these peaks gradually decreased
[F(ig._1)TD$FIG]
Fig. 1. FTIR spectra of the products.

112
W. Lu et al. / Journal of Fluorine Chemistry 161 (2014) 110–119
Table 1
Attribution of absorption peaks in FTIR spectra.
Attribution
TFEA
–
TFENB
TFENH
TFENO
More information
Stretching vibration of C–H in acyl group
2976
2882
2962
2937
2876
1760
1169
1110
1411
976
2960
2932
2860
1760
1169
1110
1411
976
Symmetrical stretching vibration of C–H in b-site
Symmetrical stretching vibration of C–H in –CH2–
Symmetrical stretching vibration of C–H in terminal –CH3
Stretching vibration of C5O
1782
1767
1175
1104
1411
976
Stretching vibration of C–O–C
1185
Anti-symmetrical stretching vibration of C–O–C
Symmetrical stretching vibration of C–O–C
Deformation vibrations in O–CH2–CF3
1090
Deformation vibration of C–H
C–F stretching vibration
1411
976
External deformation vibrations of C–H
1290 1225
1290
1290
1290
with the increasing carbon-chain length. The peaks at 1130–
1050 cmÀ1 with lower intensity belong to the C–O–C symmetric
stretching vibration. With the extension of the carbon-chain, the
peaks move to higher wave number gradually and ultimately
the 1H NMR spectra of the four fluoro-esters respectively, which
can be ascribed to the protons in –O–CH2– group. As TFEA, TFENB,
TFENH and TFENO have the same trifluoro-alcohol group –O–CH2–
CF3 which stays in similar chemical environment, the correspond-
ing chemical shift of the protons in O–CH2– structure is identical.
Due to the strong electronegativity of the oxygen atoms and
electron withdrawing effect of the –CF3 group, the shielding effects
overlap with the peak at 1200–1130 cmÀ1
.
(4) The peaks at 1411, 1290 and 976 cmÀ1 are attributed to the
deformation vibrations of –CH2–, the stretching vibrations of
C–F in –CF3, and the external deformation vibrations of C–H in
trifluoro-alcohol group, respectively. Stretching vibrations of
C–F in –CF3, often form multi-bands and peak group with high
intensity from 1350 to 1120 cmÀ1 [21]. These peaks in the FTIR
spectra of TFEA, TFENB, TFENH and TFENO all hold at the
identical wave number, for the same alcohol group –O–CH2–
CF3 and similar chemical environment in molecules of the four
esters.
of protons can be reduced, and then higher chemical shift (d values:
4–5 ppm) will be gained in –O–CH2– group. Further, this peak split
into a quartet, which is affected by the three 19F atoms in
consecutive –CF3 group.
According to Fig. 2, peaks at 2.50–2.00 ppm can be ascribed to
the protons in
sharp single peak emerged at 2.15 ppm in 1H NMR spectrum of
TFEA, because there is only one methyl at the -site of the C55O
group. By peak integration lines, atomic ratio of protons in
trifluoro-alcohol group and -CH3 could be determined; therefore,
we could conclude that the molecule structure formula of TFEA
a-site C–H of the C55O group. As shown in Fig. 2a, a
a
1H NMR spectra of the four products are shown in Fig. 2. There
are quartets with the chemical shift between 4.50 and 4.43 ppm in
a
[(Fig._2)TD$FIG]
Fig. 2. 1H NMR spectra.

W. Lu et al. / Journal of Fluorine Che[(F.giT)3_F$DIG] mistry 161 (2014) 110–119
113
sample is CH3COOCH2CF3. But in TFENB, TFENH, TFENO molecule
structures, there is a methylene (–CH2–) at the -site of the C55O
group. As another –CH2– stay at the -site of C55O group, so the
peak related to protons in -CH2 will split into a triplet. As a result,
a
b
a
sharp triplets at 2.39–2.42 ppm in 1H NMR spectra of TFENB,
TFENH and TFENO can be seen in Fig. 2b,c and d.
As shown in Fig. 2b,c and d, sharp multiplets at 1.75–1.60 ppm
can be attributed to protons in –CH2– at the
group. In TFENB molecule structure, as there are one and one
terminal-CH3 -site of the C55O group) close together, the peak
related to protons in -CH2– will split into a sextet; in TFENH and
TFENO structure, the -CH2– is next to the -CH2– and -CH2–,
respectively. For this reason, the peak related to protons in -CH2–
will split into a quintet. With the carbon-chain extension in acyl
groups, electron repulsive effect of -CH2– is increasing, which
leads to the enhancement of the shielding effect in -CH2– group.
-CH2– values could move to lower
b-site of the C55O
(
g
b
b
a
g
b
g
b
So the multiplets related to
chemical shifts.
b
As shown in Fig. 2b, c and d, sharp split triplets below 1.00 ppm
are ascribed to protons in terminal methyl (–CH3) of n-butyryl, n-
caproyl and n-capryl group for the consecutive methylene. AS the
weak shielding effect, the chemical shift value is close to the
internal standard (TMS). With carbon-chain extension in the acyl
group, terminal –CH3 could get less electron withdrawing effects
from C55O group. Thus, shielding effects of terminal –CH3 become
stronger, and the triplets move to lower chemical shift.
Fig. 3. GC–MS spectra of TFEA samples.
GC–MS spectra have been used to further analyze the purities of
products and the types of impurities, as shown in Figs. 3–5. It can
be seen from Fig. 3, TFEA sample contains two components. By
contrast to standard mass spectra, the tiny peak whose retention
time arrives at 1.5 min is attributed to the CF3CH2OH impurity. But
there is no standard mass spectrum that could match the
prominent peak related to the major constituent whose retention
time arrives at 1.7 min. Analysis of mass spectrometry is shown in
Table 3. The molecular weight of the main component is about 142.
Connecting with 1H NMR and MS spectrum analysis it can be
speculated that the peak at 1.7 min is attributed to CH3COOCH2CF3.
By peak area ratio, it can be determined that the purity of TFEA
sample is about 98%. However, due to the close boiling point, it is
difficult to separate CF3CH2OH from TFEA completely by rectifica-
As shown in Fig. 2c and d, a series of broad peaks at 1.40–1.20
attributed to the protons in several –CH2– groups between b-CH2–
and terminal –CH3. For the similar chemical environments, these
multiplets couldn’t be separated easily from broad peaks. But
we are able to get atomic ratio of protons in different structure by
the integration line. It can be concluded that the molecule
structure formulas of major component of TFENB, TFENH, TFENO
samples are CH3(CH2)2COOCH2CF3, CH3(CH2)4COOCH2CF3 and
CH3(CH2)6COOCH2CF3, respectively.
According to Table 2, with the acyl carbon-chain extension from
TFEA to TFENO, the melting points, boiling points, and the
refractive indexes are increased gradually. TFEA, TFENB and
TFENH all have very low freezing point (<197 K). Meanwhile, at
both room temperature and low temperatures, they have good
mutual solubility in ethyl methyl carbonate (EMC), which is one of
the main solvent components in lithium-ion battery electrolyte.
But TFENO only shows partial solubility in EMC at room
temperature. When temperature was below 263 K, phase separa-
tion would be occurred between TFENO and EMC. For this reason,
only TFEA, TFENB and TFENH may be suitable as low temperature
co-solvents of lithium-ion battery electrolyte.
tion. There are reactive hydrogen atoms in the
a-CH3 (pKa = 12.5)
[22], which could react with LixC on graphite negative electrode
and affect the cycle performance of lithium-ion battery. Therefore,
it is essential for TFEA sample to remove the CF3CH2OH impurity
before adding it to the lithium-ion battery electrolyte as a co-
solvent. We have selected CaH2 to remove impurities with
hydroxyl group. After rectification, CF3CH2OH impurity could be
removed from TFEA sample and meet the requirements of co-
solvent, which can be confirmed by gas chromatogram.
From the physical properties of aliphatic carboxylic acid
(reactants) and aliphatic carboxylate (products) listed in Table 2,
we can see that RCOOCH2CF3 could be easily separated from
RCOOH by distillation as they have quite different boiling points.
From the gas chromatogram in Fig. 4, the purity TFENB sample
is 99.1%, but also contains trace impurities. Compared the standard
mass spectrum, main peak at 4.0 min is attributed to the TFENB,
Table 2
Some physical properties of the reactants and RCOOCH2CF3 products.
Reactant
B.p./K
M.p./K
nD (293 K)
CF3CH2OH
346.6
390.9
436.5
478.0
512.3
228.0
289.6
265.1
269.6
289.0
1.2907
1.3716
1.3991
1.4170
1.4268
CH3COOH
CH3(CH2)2COOH
CH3(CH2)4COOH
CH3(CH2)6COOH
RCOOCH2CF3
MWt
B.p./K
M.p./K
nD (293 K)
Solubility in (EMC)
TFEA
142.08
170.13
198.18
226.24
350.0
385.0
423.0
463.5
<193.0
<193.0
196.8
1.3038
1.3436
1.3798
1.3808
Mutual solubility
Mutual solubility
Mutual solubility
Partial solubility
TFENB
TFENH
TFENO
232.5

[
W. Lu et al. / Journal of Fluorine Che
1
mistry 161 (2014) 110–119
Scheme 1. Possible reaction between aliphatic ketone and lithium or LixC6
alcohols, at room temperature. However, there is equilibrium
between ketoform and enolic form in aliphatic ketone [22], and the
hydroxyl group from enolic form may react with LixC6, as shown in
Scheme 1. Therefore, the synthetic product was contacted with
lithium metal at reflux temperature for 5 h to eliminate the 4-
heptanone impurity. After redistillating, middle distillate was
selected as a co-solvent of lithium-ion battery electrolyte.
Fig. 4. GC–MS spectra of TFENB samples.
[(Fig._5)TD$FIG]
As shown in Fig. 5, the purity of TFENH sample is about 98.3%,
containing one impurity. By the standard mass spectrum we could
make a judgment that the impurity peak at 3.8 min may be
attributed to TFENB. This impurity may stem from the small amount
of n-butyric acid in the analytical reagent n-caproic acid. Main peak
at 5.9 min could not be ascribed to any standard mass spectrum,
whose analysis is shown in Table 4. The molecular weight of the
corresponding component is about 198. Connecting MS and 1H NMR
analysis, it can be confirmed that the peak at 5.9 min is attributed to
TFENH. Low content of impurities TFENB could not significantly
affect the action of TFENH as a co-solvent of electrolyte.
2.2. Properties of the modified electrolyte with co-solvents
As a co-solvent, the purified product was added into original
electrolyte 1 mol LÀ1 LiPF6/EC + EMC (1:4, vol). The obtained
modified
electrolytes
0.75 mol LÀ1
LiPF6/EC + EMC + TFEA
(15:60:25, vol), 0.75 mol LÀ1 LiPF6/EC + EMC + TFENB (15:60:25,
vol) and 0.75 mol LÀ1 LiPF6/EC + EMC + TFENH (15:60:25, vol) were
presented in simplified form as electrolyte A, B and C. As a contrast,
the base-line electrolyte 1 mol LÀ1 LiPF6/EC + EMC (1:4, vol) was
named electrolyte D. In order to eliminate the influence of lower EC-
content and lower LiPF6 concentration after adding co-solvent, two
electrolytes composed by 1 mol LÀ1 LiPF6/EC + EMC (15:85, vol) and
0.75 mol LÀ1 LiPF6/EC + EMC (15:85, vol) was prepared and assigned
as electrolyte E and F, which added the same volume of EMC instead
of RCOOCH2CF3 co-solvent. However, electrolyte E kept an equal salt
concentration withbase-line electrolyte. On the contrary, electrolyte
F held a similar salt concentration in accord with three modified
electrolytes (electrolyte A, B and C).
Fig. 5. GC–MS spectra of TFENH samples.
Table 3
Mass spectrum analysis of TFEA sample (t = 1.7 min).
M/Z
Species
+
142
122
83
CH3COOCH2CF3
+
CH3COOCH-CF2
ꢀCH2CF3
+
+
Table 4
69
ꢀCF3
63
CH3COOꢀ+
Mass spectrum analysis of TFENH sample
(t = 5.9 min).
M/Z
Species
+
198
169
155
141
127
99
CH3(CH2)4COOCH2CF3
ꢀCH2(CH2)2COOCH2CF3
+
but the negligible peak at 5.7 min may be ascribed to the 4-
heptanone impurity, which exists in starting material n-butyric
acid. Unlike TFEA samples, CF3CH2OH can be completely separated
from TFENB by distillation and not be found in the gas
chromatogram for the larger boiling points difference (about
ꢀCH2CH2COOCH2CF3
+
ꢀCH2COOCH2CF3
+
ꢀCOOCH2CF3
+
ꢀOCH2CF3
+
ꢀCH2CF3
+
83
ꢀ+
57
CH3CH2CH2CH2
38 K). The pKa value of
a-H in the aliphatic ketone carbonyl (in
ꢀ+
43
CH3CH2CH2
DMSO) is about 27.1–28.3 [23], which are far less than that of

[
W. Lu et al. / Journal of Fluorine Chemistry 161 (2014) 110–119
115
Temperature dependence of the ionic conductivity of carbon-
ate-based electrolytes and modified electrolytes with RCOOCH2CF3
co-solvent is shown in Fig. 7, including the fitting curve by VTF
equation [24–26]. Adapting to the lowering temperature, the
conductivities of the six electrolytes were all decreased. The
electrolyte D could perform a higher ionic conductivity above
223 K. Comparing electrolyte E with electrolyte F, a little higher
salt concentration would be helpful to hold a larger ionic
concentration at room temperature. But at low temperature,
higher salt concentration means larger viscosity, which is not
helpful to enhance the ionic conductivity. In contrast, the
decreasing ionic conductivity of the electrolyte with RCOOCH2CF3
co-solvent was slowed down significantly, comparing with that of
the carbonate-based electrolyte (electrolyte A, B and C). In keeping
with the temperature dependence of the viscosity, electrolyte A
containing TFEA has a better performance, due to the lower
viscosity at lower temperature. It could be conjectured that
RCOOCH2CF3 with shorter acyl group in molecules would
participate in Li+ solvation process and reduce the ionic radius
of the solvated Li+, which could provide weaker internal friction in
the fluid during the ionic migration.
Fig. 6. Temperature dependence of the viscosity.
According to the fitting results by VTF equation in Table 5, T0 of
the three modified electrolytes were markedly lower than that of
the carbonate-based electrolyte. Furthermore, the values of T0
went down with the decreasing of the number of carbon atoms in
the co-solvents. At the same time, after trifluoroethyl aliphatic
carboxylate co-solvents added, n values of modified electrolytes
were close to zero, indicating that the mechanism of Li+ migration
in electrolyte A at low temperature was similar to that at room
temperature. Ion-migration in modified electrolytes could be
described by ionic atmosphere model in the test temperature
range, which is related to the structure of Li+ solvation.
Temperature dependence of ionic conductivity and viscosity of
the modified electrolytes containing co-solvents suggests that
trifluoroethyl aliphatic carboxylate with shorter carbon-chain in
acyl group (such as TFEA) should be fit for the co-solvents to reduce
the viscosity and enhance the ionic conductivity of the electrolyte
at low temperature.
Temperature dependence of the viscosity of base-line electro-
lyte and electrolyte with one co-solvent is shown in Fig. 6. It could
be seen from the figure that the viscosity of the six electrolytes
were all increased obviously below 278 K. However, electrolytes
containing co-solvent (included RCOOCH2CF3 or EMC) could
clearly slow the increasing trend of viscosity at lowing-tempera-
ture. It might be ascribed to the lower content of EC which has a
high viscosity. Viscosity of electrolyte E was a little higher than that
of electrolyte F, it is concluded that higher concentration of LiPF6
could slightly increase the viscosity. From electrolyte A, B, C and F,
it could be found that trifluoroethyl aliphatic carboxylates are
more suitable as the co-solvent to reduce the viscosity of
electrolyte than aliphatic carbonate. Electrolyte A (with TFEA)
exhibited the lowest viscosity increasing during the cooling
process. From 298 K to 238 K, the viscosity of the electrolyte only
increased four times. The viscosity is macro-reflection of internal
migrating friction of solvent molecules in the electrolyte. Due to its
smaller molecule-scale than that of TFENB or TFENH, and weaker
solvent-solvent molecular interaction than that of linear carbon-
ate, TFEA could reduce the viscosities of the electrolyte systems
more obviously.
2.3. Electrochemical characterization
Fig. 8 shows the charge/discharge curves of graphite electrode
inthecarbonate-based electrolyteand threemodifiedelectrolytesat
0.2 C rate under different temperatures. As can be seen in Fig. 8a, the
initial coulombic efficiency of the graphite in three carbonate-based
electrolytes (electrolyte D, E and F) was slightly higher than that in
the three modified electrolytes (electrolyte A, B and C) at 298 K,
which suggested trifluoroethyl aliphatic carboxylate could cause a
larger irreversible capacity loss than linear carbonate co-solvent.
After five cycles, according to Fig. 8b, both of them had an identical
coulomb efficiency above 99%, which indicated the electrolyte
containing trifluoroethyl aliphatic carboxylate co-solvents could
form effective passivation of graphite electrode but not affect the
cycling performance of graphite at room temperature.
[(Fig._7)TD$FIG]
When tested at low temperature of 238 K, the charge capacity of
graphite in base-line electrolyte (electrolyte D) dropped to about
Table 5
Non-linear fitting result by VTF function.
Electrolyte
T0/K
n
A
Ea/J molÀ1
A
B
C
D
E
F
122.8 Æ 5.5
130.2 Æ 6.3
132.2 Æ 9.2
156.9 Æ 13.0
152.5 Æ 13.8
151.0 Æ 12.3
À0.03 Æ 0.00
À0.02 Æ 0.00
À0.23 Æ 0.09
À0.50 Æ 0.01
À0.18 Æ 0.01
À0.12 Æ 0.01
À0.86 Æ 0.09
À0.87 Æ 0.29
1.22 Æ 0.24
1.90 Æ 0.19
À1.53 Æ 0.06
À1.40 Æ 0.07
3972 Æ 406
4207 Æ 616
4473 Æ 534
2970 Æ 153
2496 Æ 153
1739 Æ 115
Fig. 7. Temperature dependence of the ionic conductivity.

[
W. Lu et al. / Journal of Fluorine Chemistry 161 (2014) 110–119
Fig. 8. Specific capacity – voltage curve at different temperature.
80% of that at 298 K. Electrolytes E and F with lower EC-content
could provide higher Li+ de-intercalation capacity at 238 K. By
comparing electrolyte E with electrolyte F, it could be observed
that higher salt concentration could present a little lower initiate
polarizing voltage but a slightly smaller final capacity. In contrast,
the capacity retention in electrolyte A, B and C could reach 85%, 92%
and 95%, respectively. Furthermore, the typical charge potential
was lowered from 0.95 V to 0.90 V, 0.80 V to 0.70 V by the addition
of co-solvents. This contrast was magnified at 218 K, as we could
see that the capacity retention of graphite in electrolyte F was only
50%, far lower than that in electrolyte B and C. Especially the
capacity retention maintain at 92% in electrolyte C, which was the
highest value reported so far as we know. At the same time, the
charge potential was decreased from 1.30 V in electrolyte F to
0.89 V in electrolyte C. These data demonstrate that the addition of
trifluoroethyl aliphatic carboxylate co-solvents in carbonate-based
electrolyte can effectively improve the electrochemical perfor-
mance of graphite at low temperature.
In order to gain further insight into the electrochemical
mechanisms operating for the improvement, EIS measurements
were conducted by discharging graphite to 0.1 V at five different
temperatures (298, 278, 258, 238 and 218 K). As shown in Fig. 9,
the Nyquist plots can be fitted by the equivalent circuit model
comprising four parts: (i) lithium-ion transport in an electrolyte
and electron transport in electrode (R0), (ii) lithium-ion transport
in an SEI (RSEI), (iii) lithium-ion (charge) transfer at a graphite/
electrolyte interface (Rct), (iv) lithium-ion diffusion inside graphite
electrode. In this case, equivalent circuit which has been shown in
Fig 9f as used for fitting the Nyquist plots [27–29].
At 298 K and 278 K, there were straight lines with an angle of
about 608 from the Zrm axis at low frequency region, which refers to
the Li+ solid-phase diffusion. But at 258 K, 238 K and 218 K,
Warburg impedance didn’t appear notably in the Nyquist plots.
The reason might be that the ability of lithium-ion diffusion inside
graphite electrode was significantly decreased at low temperature.
Lithium-ion diffusion inside graphite electrode should be pre-
sented below a much lower frequency. At every testing tempera-
ture, impedance in all-carbonate electrolytes (electrolyte E and F)
was close to that in base-line binary electrolyte (electrolyte D).
Electrolytes contained trifluoroethyl aliphatic carboxylates pro-
vided obviously lower impedance. The fitting results of the six ones
have been listed in Table 6.
According to Table 6, when the cells were evaluated at one
temperature (both room temperature and low temperature), it was
observed modified electrolytes with RCOOCH2 CF3co-solvents
performed a higher R0 compared with the all-carbonate solution.
Adapting to the increasing number of carbon atoms in the co-
solvent structure, R0 was continuously raised. These results could
be confirmed by the temperature dependences of the ionic
conductivities of electrolytes in Fig. 7. It could be concluded that
the co-solvent with long carbon-chain in acyl group was harmful to
increase the ionic conductivities.
But the values of RSEI were provided another order: when the
impedance spectra of the cells containing modified electrolytes

[
W. Lu et al. / Journal of Fluorine Chemistry 161 (2014) 110–119
117
Fig. 9. Nyquist plots of graphite electrode in every electrolyte at different temperature.
were compared to the three carbonate-based electrolytes at room
temperature, the electrodes in contact with electrolyte A, B and C
generally displayed low film polarization characteristics as
determined from the low RSEI values observed. Especially,
electrode in electrolyte C containing TFENH with longer carbon-
chain performed the lowest RSEI values among those in electrolyte
B and A. When impedance measurements were performed on the
cells at low temperatures, more dramatically lower film resis-
tances were observed for the modified electrolyte containing cells.
TFENH with longer carbon-chain could also helped to form a SEI
Table 6
Fitting result of AC impedance spectroscopy.
Temperature/K
R0/V
RSEI/V
298
278
258
238
218
298
278
258
87.3
238
218
Electrolyte A
Electrolyte B
Electrolyte C
Electrolyte D
Electrolyte E
Electrolyte F
2.6
3.1
3.7
2.5
2.6
2.9
3.7
4.0
6.1
3.2
3.3
3.5
5.5
5.7
8.2
4.4
4.4
4.5
7.2
8.0
13.8
18.8
21.5
11.7
11.5
10.8
18.7
13.2
8.6
42.0
31.7
24.7
68.1
70.2
69.3
324.4
206.3
128.1
728.8
699.4
705.2
3114.0
1854.0
538.3
67.8
36.9
11.7
6.4
32.9
30.4
31.3
267.9
304.6
288.7
6690
6.3
6809.2
6470.1
6.4

[
W. Lu et al. / Journal of Fluorine Chemistry 161 (2014) 110–119
From Table 6, it was generally observed that the R0 and RSEI
were continuously increased upon going to lower temperatures in
each electrolyte. But adapting to the decreasing of temperature,
RSEI was growing exponentially, which increased much faster than
R0. Comparing with the low temperature performances of the
graphite electrode, a conclusion could be observed that the nature
of the SEI layer on the graphite electrode plays a more significant
role in determining the low temperature performance than the
bulk resistivity of the electrolyte.
The composition and microstructure of SEI film in the modified
electrolytes containing trifluoroethyl aliphatic carboxylate co-
solvents has not been studied clearly. Further research on the
formation mechanism is needed in the future.
3. Conclusions
(1) Hydrogen ion exchange resin as a catalyst and silica gel drier as
a de-hydration, a series of trifluoroethyl aliphatic carboxylates
were prepared by one-step approach with different carbon-
chain length in acyl group.
(2) By FTIR, 1H NMR and GC–MS analysis, the products have the
expected chemical structures and high yields by the synthetic
route. Species and structures of the impurities in the products
were confirmed and then feasible approaches for purification
were founded.
(3) Temperature dependence of ionic conductivity and viscosity of
the modified electrolytes containing co-solvents suggests that
trifluoroethyl aliphatic carboxylate with shorter carbon-chain
in acyl group (such as TFEA) should be fit for the co-solvents to
reduce the viscosity and enhance the ionic conductivity of the
electrolyte at low temperature.
(4) According to the Li+ de-intercalation curves of the graphite
electrode and analysis of AC impedance spectroscopy, modified
electrolyte containing TFENH could improve the low tempera-
ture performances of graphite electrode obviously. It could be
confirmed that the nature of the SEI layer on the graphite
electrode plays a more significant role in determining the low
temperature performance than the bulk resistivity of the
electrolyte.
Fig. 10. Temperature dependences of SEI film conductivities (1/RSEI) at 0.1 V.
film with low polarization on the graphite electrode, for the lowest
RSEI values were obtained during the tests at 258 K, 238 K and
218 K.
Fig. 10 shows the temperature dependences (Arrhenius plots) of
the SEI film conductivities (1/RSEI) at 0.1 V in all-carbonate and
three modified electrolytes. The activation energies were evaluat-
ed from the slopes of the Arrhenius plots with the Arrhenius
equation [30], ln (1/RSEI) = A–Ea/RT. The symbols A, Ea, R, and T
denote the frequency factor, activation energy, gas constant, and
absolute temperature, respectively. In three carbonate electrolytes,
the activation energy was close to 35 kJ molÀ1. However, electro-
lyte A, B and C gave a smaller activation energy of 33, 32,
27 kJ molÀ1. Smaller activation energy suggests that it is much
easier for the diffusion of lithium-ions in the SEI film. It could be
conjectured that RCOOCH2CF3 might change the mechanism of
forming SEI film in the carbonate-based electrolyte. Although
TFENH with long carbon-chain in acyl group isn’t fit for reducing
the viscosity of the electrolyte, it is helpful to form a better SEI film
with higher Li+ conductive ability.
4. Experiments
[(Fig._11)TD$FIG]
4.1. Reagents
Trifluoroethanol, acetic acid, n-butyric acid, n-hexanoic acid, n-
octanoic acid, sodium ion exchange resin, NaHCO3, anhydrous
˚
MgSO4, 4 A molecular sieves, CaH2, silica gel, hydrochloric acid,
ethanol, all the reagents were of analytical grade, and directly used
forthesynthesisreactionwithoutanytreatment. Originalelectrolyte
1 mol LÀ1LiPF6/EC + EMC (1:4) was provided by Novolyte Company.
4.2. Co-solvent synthesis
Preparation of hydrogen ion exchange resin catalyst: sodium
ion exchange resin about 5 g was mixed with 50 mL 1 mol LÀ1 HCl.
After stirring for 6 h, ion exchange resin was washed by deionized
water till the pH value is about 7, and then dried at 350 K for 5 h.
The schematic diagram of esterification reaction device is
shown in Fig. 11. 0.1 mol CF3CH2OH was added into the dropping
funnel, and 50 g silica gel was added into the Soxhlet’s extractor.
Three-necked flask (100 mL volume) with 2 g hydrogen ion
exchange resin and 0.12 mol fatty acid was heated to the reflux
temperature with magnetic stirring. After adding CF3CH2OH drop
by drop, the mixture was heated at reflux for about 6 h, and then
Soxhlet’s extractor was replaced by the distillation apparatus. The
Fig. 11. Schematic diagram of esterification reactor.

W. Lu et al. / Journal of Fluorine Chemistry 161 (2014) 110–119
119
intermediate fraction was collected as the primary product, and
added NaHCO3 powder until no gas emission. Then collected
filtrate was added an appropriate amount of anhydrous MgSO4,
and dried for 24 h. Purified RCOOCH2CF3 were obtained by
collecting the intermediate fraction during distillation. Finally,
ac voltage of 10 mV. Measurements were conducted for each cell
after the formation process (after five cycles), the polarization
potential of the graphite electrode was kept on 0.1 V. The analysis
of resultant Nyquist plots was conducted by Zview software.
˚
activated 4 A molecular sieves were added into the distilled
Acknowledgements
RCOOCH2CF3.
The moisture content of the product was measured by the Karl-
Fisher coulometric moisture analyzer (831 KF Coulometer,
Metrohm). Boiling point was observed by distillation. Moiling
point was tested in low temperature environment which was
provided by cryostat circulation tank (YU HUA, DFY-80, minimum
temperature 193 K). Refractive index was measured by using an
Abbe refractometer (2 W Abbe refractometer) at 293 K.
We are grateful to the support of aid program for Science and
Technology Innovative Research Team in Higher Educational
Institutions of Hunan Province, the National Science Foundation
of China (No. 21303262) and the Hunan Provincial Natural Science
Foundation of China (No. 13JJ4004).
We are appreciating to the help of Sinopoly Battery Ltd.
Appendix A. Supplementary data
4.3. Structure characterization
Supplementary data associated with this article can be found, in
the online version, at http://dx.doi.org/10.1016/j.jfluchem.2014.02.
006http://dx.doi.org/10.1016/j.jfluchem.2014.02.006.
FTIR spectrum was obtained by Fourier Transform Infrared
Spectrometer (Bruke V70, KBr tablet, wave number 4000–
600 cmÀ1, resolution 4 cmÀ1
)
1H NMR characterization was dependent upon the Fourier
transform NMR spectrometer (Bruker AVANCE 600), using CDCl3 as
a solvent and tetramethylsilane (TMS) as the internal standard.
MS–GC spectrum was characterized in gas chromatography–
mass spectrometry (Agilent Technologies 7890A GC System and
5975C inert MSD), using He as eluant gas. The temperature in
vaporizer was set 30 K higher than the boiling point of the sample.
References
[1] M.C. Smart, B.V. Ratnakumar, L.D. Whitcanack, K.B. Chin, S. Surampudi, H. Croft, D.
Tice, R. Staniewicz, J. Power Sources 119–121 (2003) 349–358.
[2] M. Contestabile, M. Morselli, R. Paraventi, R.J. Neat, J. Power Sources 119–121
(2003) 943–947.
[3] B. Wu, Y. Ren, D. Mu, C. Zhang, X. Liu, F. Wu, Int. J. Electrochem. Sci. 8 (2013) 8502–
8512.
[4] M.C. Smart, B.V. Ratnakumar, S. Surampudi, J. Electrochem. Soc. 146 (1999) 486–
492.
4.4. Electrochemical characterization
[5] E.J. Plichta, M. Hendrickson, R. Thompson, G. Au, W.K. Behl, M.C. Smart, B.V.
Ratnakumar, S. Surampudi, J. Power Sources 94 (2001) 160–162.
[6] K.A. Smith, M.C. Smart, G.K.S. Prakash, B.V. Ratnakumar, ECS Transactions 11
(2008) 91–98.
[7] M.C. Smart, B.V. Ratnakumar, V.S. Ryan-Mowrey, S. Surampudi, G.K.S. Prakash, J.
Hu, I. Cheung, J. Power Sources 119–121 (2003) 359–367.
[8] M.C. Smart, J.F. Whitacre, B.V. Ratnakumar, K. Amine, J. Power Sources 168 (2007)
501–508.
[9] D. Aurbach, A. Zaban, Y. Ein-Eli, I. Weissman, O. Chusid, B. Markovsky, M. Levi, E.
Levi, A. Schechter, E. Granot, J. Power Sources 68 (1997) 91–98.
[10] D. Aurbach, B. Markovsky, I. Weissman, E. Levi, Y. Ein-Eli, Electrochim. Acta 45
(1999) 67–86.
[11] D. Aurbach, K. Gamolsky, B. Markovsky, Y. Gofer, M. Schmidt, U. Heider, Electro-
chim. Acta 47 (2002) 1423–1439.
As a co-solvent, the purified product was added into original
electrolyte 1 mol LÀ1 LiPF6/EC + EMC (volume ratio 1:4) in a
volume ratio of 3:1, to obtain a new electrolyte whose composition
could be approximately represented by 0.75 mol LÀ1 LiPF6/
EC + EMC + RCOOCH2CF3 (volume ratio 15:60:25). The low-tem-
perature ionic conductivity of the electrolyte was measured by
conductivity meter (RIDAO, DDS-307), using conductivity elec-
trode (conductivity cell constant was about 1, which was
calibrated by 0.1 mol LÀ1 KCl solution before the experiment).
The viscosity of the electrolyte was tested by Ubbelohde
viscometer in a constant temperature water bath (298 K) and
cryostat circulation tank (<273 K).
[12] T. Nakajima, K. Dan, M. Koh, J. Fluorine Chem. 87 (1998) 221–227.
[13] R. Chandrasekaran, M. Koh, Y. Ozhawa, H. Aoyoma, T. Nakajima, Indian Acad. Sci.
121 (2009) 339–346.
The graphite electrode was prepared by combining 80 wt.%
graphite negative electrode material (AGP-8, BTR New Energy
materials INC.), 10 wt.% acetylene black, and 10 wt.% polyvinyli-
dene fluoride (Kynar 741, Arkema Inc., USA) in N-methyl-2-
pyrrolidone. Then the slurry was uniformly coated on a copper foil
(highly pure) and fully dried in vacuum oven. A separator (Celgard
2320) was placed between the cathode and the anode. Lithium foil
(China Energy Lithium Co., Ltd., China) was used as counter-
electrode. The CR2016 type (20 mm diameter, 1.6 mm thick)
graphite/lithium cells were assembled in a high purified argon-
filled dry box. Charge and discharge performances were tested by
LAND testing system (CT2001-A). At 298 K, both charge and
discharge performances of graphite electrodes were measured at
0.2 C rate. The Li+ de-intercalation test at low temperature (238 or
218 K) was carried out in the following procedure: firstly, the
graphite-Li cells were discharged to 0.02 V at 298 K; secondly, the
batteries were held at low temperature for 2 h, and then charged to
cut-off voltage 1.5 V at 0.2 C rate.
[14] Y. Matsuda, T. Nakajima, Y. Ohzawa, M. Koh, A. Yamauchi, M. Kagawa, H. Aoyama,
J. Fluorine Chem. 132 (2011) 1174–1181.
[15] M. Ihar, B.T. Hang, K. Sato, M. Egashira, S. Okada, J. Yamaki, J. Electrochem. Soc. 150
(2003) A1476–A1483.
[16] K. Sato, I. Yamazaki, S. Okada, J. Yamaki, Solid State Ionics 148 (2002) 463–
466.
[17] K. Sato, L. Zhao, S. Okada, J. Yamaki, J. Power Sources 196 (2011) 5617–5622.
[18] J. Yamaki, I. Yamazaki, M. Egashira, S. Okada, J. Power Sources 102 (2001) 288–
293.
[19] K.M. Parida, S. Mallick, J. Mol. Catal. A: Chem. 275 (2007) 77–83.
[20] F.T. Sejidov, Y. Mansoori, N. Goodarzi, J. Mol. Catal. A: Chem. 240 (2005) 186–190.
[21] Y.K. Ke, H.R. Dong, second ed., Handbook of Analytical Chemistry, Vol. 3,
Chemical Industry Press, Beijing, 2004.
[22] H.B. Gao, Organic Chemistry, fourth ed., Higher Education Press, Beijing, 2005.
[23] F.G. Bordwell, Acc. Chem. Res. 21 (1988) 456–463.
[24] J. Vila, C. Franjo, J.M. Pico, L.M. Varela, O. Cabeza, Portugaliae Electrochim. Acta 25
(2007) 163–172.
[25] A. Horky, N.P. Kherani, G. Xu, J. Electrochem. Soc. 150 (2003) A1219–A1224.
[26] W. Lu, K. Xie, Y. Pan, Z. Chen, C. Zheng, J. Fluorine Chem. 156 (2013) 136–
143.
[27] A. Funabiki, M. Inaba, Z. Ogumi, J. Power Sources 68 (1997) 227–231.
[28] A. Funabiki, M. Inaba, Z. Ogumi, S. Yuasa, J. Otsuji, A. Tasaka, J. Electrochem. Soc.
145 (1998) 172–178.
[29] M. Wagner, Electrochim. Acta 42 (1997) 1623–1625.
[30] Y. Yamada, Y. Iriyama, T. Abe, Z. Ogumi, Langmuir 25 (2009) 12766–12770.
AC impedance spectroscopy was carried out by Versa STAT MC
(Princeton) over a frequency range of 106–0.1 Hz with an applied