Thermal decomposition of LiPF6-based electrolytes for lithium-ion batteries

Article abstract of DOI:10.1149/1.2083267

The thermal decomposition of lithium-ion battery electrolytes 1.0 M LiPF₆ in one or more carbonate solvents has been investigated. Electrolytes containing diethyl carbonate (DEC), ethylene carbonate (EC), a 1:1 mixture of EC/dimethyl carbonate (DMC), and a 1:1:1 mixture EC/DMC/DEC have been investigated by multinuclear nuclear magnetic spectroscopy, gas chromatography with mass selective detection, and size exclusion chromatography. Thermal decomposition affords products including: carbon dioxide (CO ₂), ethylene (CH₂CH₂), dialkylethers (R ₂O), alkyl fluorides (RF), phosphorus oxyfiuoride (OPF3), fluorophosphates [OPF₂OR, OPF(OR)₂], fluorophosporic acids [OPF₂OH, OPF(OH)₂], and oligoethylene oxides. The mechanism of decomposition is similar in all LiPF₆/carbonate electrolytes. Trace protic impurities lead to generation of OPF₂OR, which autocatalytically decomposes LiPF₆ and carbonates. The presence of DEC leads to the generation of ethylene, while the presnce of EC leads to the generation of capped oligothylene oxides [OPF₂(OCH ₂CH₂)_nF].

Full text of DOI:10.1149/1.2083267

Journal of The Electrochemical Society, 152 ͑12͒ A2327-A2334 ͑2005͒
A2327
0
013-4651/2005/152͑12͒/A2327/8/$7.00 © The Electrochemical Society, Inc.
Thermal Decomposition of LiPF -Based Electrolytes
6
for Lithium-Ion Batteries
,
z
Christopher L. Campion,* Wentao Li,* and Brett L. Lucht**
Department of Chemistry, University of Rhode Island, Kingston, Rhode Island 02881, USA
The thermal decomposition of lithium-ion battery electrolytes 1.0 M LiPF in one or more carbonate solvents has been investi-
6
gated. Electrolytes containing diethyl carbonate ͑DEC͒, ethylene carbonate ͑EC͒, a 1:1 mixture of EC/dimethyl carbonate ͑DMC͒,
and a 1:1:1 mixture EC/DMC/DEC have been investigated by multinuclear nuclear magnetic spectroscopy, gas chromatography
with mass selective detection, and size exclusion chromatography. Thermal decomposition affords products including: carbon
dioxide ͑CO ͒, ethylene ͑CH CH ͒, dialkylethers ͑R O͒, alkyl fluorides ͑RF͒, phosphorus oxyfluoride ͑OPF ͒, fluorophosphates
2
2
2
2
3
͓
OPF OR, OPF͑OR͒ ͔, fluorophosporic acids ͓OPF OH, OPF͑OH͒ ͔, and oligoethylene oxides. The mechanism of decomposi-
2 2 2 2
tion is similar in all LiPF /carbonate electrolytes. Trace protic impurities lead to generation of OPF OR, which autocatalytically
6
2
decomposes LiPF and carbonates. The presence of DEC leads to the generation of ethylene, while the presnce of EC leads to the
6
generation of capped oligothylene oxides ͓OPF ͑OCH CH ͒ F͔.
2
2
2 n
©
2005 The Electrochemical Society. ͓DOI: 10.1149/1.2083267͔ All rights reserved.
Manuscript submitted May 20, 2005; revised manuscript received July 20, 2005. Available electronically October 24, 2005.
Lithium-ion batteries are one of the most important power
sources for portable electronic devices. Improvements in lithium-
nism of the reaction between the electrolyte or electrolyte decompo-
sition products and the cathode is poorly understood. We have
recently reported that the presence of cathode particles inhibits the
1
ion technology have made them interesting alternatives for nickel
metal hydride and nickel cadmium batteries for high-energy, long-
1
6
thermal decomposition of bulk electrolyte. It is likely that this
inhibition is related to the formation of surface films. The reaction of
cathode particles, in both the charged and uncharged states, with
electrolyte is currently under investigation and will be reported in
2
,3
lifetime electric vehicle ͑EV͒ or satellite applications. However,
the use of lithium-ion batteries for these applications is limited by
operational lifetime and high-temperature stability. The most widely
1
7
used electrolyte for lithium-ion batteries is lithium hexafluorophos-
due course.
While the reactions of electrolyte on the surface of the anode
5
,6
4
phate ͑LiPF ͒ in a mixture of organic carbonate solvents. However,
6
7
,8
and cathode during accelerated aging experiments have been in-
vestigated, the thermal decomposition of the liquid electrolyte has
lithium-ion batteries containing LiPF -based electrolytes have poor
6
stability at elevated temperatures ͑Ͼ60°C͒. The loss of performance
in lithium-ion batteries at elevated temperatures has been attributed
to many factors, including the continued reaction of the electrolyte
with the anode or compounds present in the anode solid electrolyte
9
-11
received less attention.
Most investigations of the thermal de-
composition of electrolytes have focused on rapid decompositions at
very high temperature ͑140–350°C͒ using differential scanning
1
8-21
5
,6
calorimetry ͑DSC͒
or accelerated rate calorimetry
interface ͑SEI͒, which results in growth of the SEI, decomposition
1
8,19,22,23
͑
ARC͒.
These investigations are important for understand-
of the electrolyte on the surface of cathode particles to generate
7
,8
9-11
ing the reactions occurring during overcharge or thermal runaway.
However, developing batteries with extended lifetimes ͑Ͼ10 yr͒
and stability at moderately elevated temperatures ͑50–100°C͒ re-
quires an understanding of the decomposition reactions and products
in this temperature regime. We recently reported a preliminary ac-
resistive films, and bulk electrolyte decomposition.
The formation of the SEI on anode particles in lithium-ion bat-
teries is one of the most important factors in long-term cycling
1
2
stability. SEI formation is attributed to the electrochemical reduc-
tion of the electrolyte solvent and salt to generate a complex mixture
of surface species including LiF, Li CO , ͑CH OCO Li͒ , ROLi,
count of the mechanism of the thermal decomposition of LiPF in
6
2
3
2
2
2
16
carbonate solvents at moderate temperatures ͑85–100°C͒.
and ROCO Li. However, the long-term thermal stability of the SEI
2
In this manuscript, a detailed description of the thermal decom-
position products and reaction mechanisms of electrolytes for
lithium-ion batteries is presented. The electrolytes investigated in-
is uncertain. Several studies have suggested that the SEI is unstable
at elevated temperatures, especially in the presence of LiPF . It has
6
been proposed that the surface film reacts with electrolyte at el-
clude 1.0 M LiPF in diethyl carbonate ͑DEC͒, ethylene carbonate
6
evated temperature to form thicker, more resistive films with high
6
͑EC͒, a 1:1 mixture of EC/dimethyl carbonate ͑DMC͒, and a 1:1:1
mixture EC/DMC/DEC ͑ternary electrolyte͒. Storage of the electro-
lytes at elevated temperatures ͑85–100°C͒ for 1 week or longer re-
sults in significant decomposition. The decomposition products were
characterized by a combination of gas chromatography with mass
concentrations of LiPF decomposition products. Electrochemical
6
impedance spectroscopy suggests that at elevated temperatures PF5,
generated via the thermal dissociation of LiPF , reacts with the SEI,
6
destroying the protective film and freeing up the surface of the
5
graphite for further reduction of the solvent. The reactions of the
1
13
19
31
selective detection ͑GC-MS͒, H, C, F, and P nuclear magnetic
resonance ͑NMR͒ spectroscopy, and size exclusion chromatography
anode with electrolyte are reported to be the primary source of ca-
13
pacity loss in lithium-ion batteries.
1
1
͑
SEC͒. Two-dimensional NMR spectroscopy, including H– H cor-
While the formation of the anode SEI is essential, the formation
of surface films on cathode materials results in a significant increase
1
13
relation spectroscopy ͑COSY͒ and H– C heteronuclear correlation
spectroscopy ͑HETCOR͒ contributed to the characterization of de-
composition products. The effect of trace protic impurities, ethanol
or water, on the thermal decomposition of electrolytes is also dis-
cussed. Protic impurities in commercial electrolytes may arise from
7,8
in the impedance of cathodes. The presence of surface films on
cathode particles is attributed to both thermal reactions and electro-
chemical reactions of the cathode with electrolyte. However, film
formation appears to be more sensitive to temperature than state of
1
4,15
the highly hygroscopic LiPF salt or residual alcohol from the syn-
6
charge ͑SOC͒.
The structure of cathode surface films has been
thesis of carbonate solvents and have a profound effect on the ther-
mal stability.
analyzed by various techniques and is reported to contain decompo-
sition products of both LiPF and carbonates. However, the mecha-
6
Experimental
Battery-grade carbonate solvents obtained from EM Industries
*
Electrochemical Society Student Member.
*
* Electrochemical Society Active Member.
were used without further purification. Battery-grade LiPF obtained
6
z
E-mail: blucht@chm.uri.edu
from Hashimoto Chemical Corporation, Tomiyama High Purity
Downloaded on 2014-04-11 to IP 140.120.135.222 address. Redistribution subject to ECS terms of use (see ecsdl.org/site/terms_use) unless CC License in place (see abstract).

A2328
Journal of The Electrochemical Society, 152 ͑12͒ A2327-A2334 ͑2005͒
Chemicals, or Advance Research Chemicals, Inc., was used without
further purification. Samples for GC-MS or NMR spectroscopy
were prepared in a nitrogen-filled glove box followed by flame seal-
ing under reduced pressure in ampoules or NMR tubes. The sealed
samples were heated in a silicon oil bath. Samples for GC-MS
analysis were vacuum transferred prior to analysis to remove all
nonvolatile components. Samples for SEC analysis were obtained
by decanting the remaining liquid away from the precipitate and
dissolving the solids in tetrahydrofuran ͑THF͒.
Table I. 1H and 3C NMR spectroscopic data for ternary elecro-
lyte decomposition products.
1
13C ppm
Species
H ppm ͑JHH͒
CO2
EC
—
4.51͑s͒
124.1
64.6
1
56.1
13.1
63.2
54.9
53.7
DEC
1.24͑t͒, 7.1 Hz
.14͑q͒, 7.1 Hz
4
EC is a solid at room temperature and required alternative han-
1
dling. Glass ampoules were prepared in a N -filled glove box with
2
DMC
EMC
3.72͑s͒
1.0 M LiPF6 in EC and flame sealed. Ampoules were placed in
155.6
13.6
53.4
63.0
155.5
13.9
67.4
56.8
13.9
65.3
59.1
69.7
67.6
15.0
79.7
14.8
65.2
14.7
66.3
14.7
66.1
53.8
122.1
nesting aluminum cans and heated in an oven at 240°C. After heat-
ing the ampoules were frozen in liquid nitrogen, opened in a glove
bag, transferred to a Shlenck flask, analyzed for gaseous products,
1.24͑t͒, 7.1 Hz
3.71͑s͒
4.13͑q͒, 7.1 Hz
and transferred to a N -filled glove box for solvent extractions. The
2
residual solid was sequentially washed with organic solvents ͑di-
Et2O
1.15͑t͒, 7.1 Hz
3.43͑q͒, 6.9 Hz
3.30͑s͒
ethyl ether and THF͒ or hydrolyzed and extracted with CDCl or
3
EtOMe
C D . Samples were analyzed with a combination of NMR, GC-MS,
6
6
1
3
.16͑t͒, 7.1 Hz
.51͑q͒, 6.9 Hz
and SEC.
Lecture bottles of PF5 and OPF3 were obtained from Fluoro-
chem. Addition of PF and OPF to carbonates was conducted by
Me O
3.28͑s͒
3.56͑m͒
2
5
3
–
͑CH CH O͒ –
2
2
n
bubbling the gas through liquid carbonates. The apparatus was con-
nected in sequence to mineral oil and aqueous 1.0 M NaOH bub-
blers. The concentration of PF or OPF is related to addition time
3
.62͑s͒
1.25͑t͒
.51͑q͒
1.39͑t͒, 7.3 Hz
.32͑q͒
1.37͑t͒, 7.3 Hz
.22͑q͒
EtF
5
3
4
and flow rate. Upon addition of the desired quantity of gas, the
headspace was purged with nitrogen and the carbonate solution of
PF5 or OPF3 was transferred to a Teflon-capped storage flask.
Samples for NMR or GC-MS analysis were prepared in the glove
box under a protective atmosphere of N . PF and OPF are highly
OPF OEt
2
4
OPF͑OEt͒2
4
2
5
3
OPF͑OEt͒͑OMe͒
1.37͑t͒, 7.3 Hz
4.22͑q͒
3.93͑s͒
reactive gasses, which fume on contact with atmospheric moisture
and should be handled with care in a fume hood.
OPF ͑OMe͒
2
GC-MS analyses were conducted on an Agilent Technologies
CH CH
5.31͑s͒
2
2
6
890 GC with a 5973 mass selective detector and a HP-5MS col-
umn. Helium was used as the carrier gas with a flow rate of
1
5
.0 mL/min. Liquid samples were first ramped from 30 to 50°C at
°C/min followed by ramping at 10°C/min to 250°C. SEC analysis
forded via a combination of 19F and 31P NMR spectroscopy and
GC-MS. Continued investigation of the thermal decomposition of
LiPF6 in DEC along with the effect of common impurities and pro-
posed intermediates has been conducted.
was conducted on an Agilent 1100 series high-pressure liquid chro-
matograph with a PLgel 5 ␮m–10 Å column, G1362A differential
refractometer detector, THF as the eluting solvent, and polystyrene
standards.
3
1
Analysis of thermally decomposed 1.0 M LiPF6 in DEC by H
and 1 C NMR spectroscopy has provided additional support for the
3
Multinuclear NMR analyses were conducted on a JEOL
1
4
00-MHz NMR spectrometer. NMR samples were either flame
structure of decomposition products. Identification of the H and
1
13
sealed or sealed with Teflon needle valves. H NMR resonances
C resonances characteristic of Et O and EtF are straightforward.
2
13
were referenced to EC at 4.510 ppm, C NMR resonances were
While the resonances characteristic of OPF ͑OEt͒ and OPF͑OEt͒
2
2
referenced to EC at 64.64 ppm, 1 F NMR resonances were refer-
9
1
13
observed by H and C NMR spectroscopy are difficult to assign
with certainty, the coupling patterns and chemical shifts are consis-
tent with the proposed assignments. The products are assigned in
31
enced to LiPF at 65.00 ppm, and P NMR spectra were referenced
to LiPF at −145.0 ppm. C distortionless enhancement by polar-
ization transfer ͑DEPT͒ spectroscopy was used to determine the
number of H nuclei bound to each C nucleus ͑C, CH, CH , or
CH ͒. COSY was used to correlate which H resonances were
coupled. HETCOR spectra were used to correlate C nuclei with
directly attached protons.
6
13
6
1
13
Table I. In addition to the species mentioned above, H and
C
1
13
NMR resonances at 5.31 ppm ͑s͒ and 112.1 ppm ͑s͒, respectively,
are characteristic of ethylene. This is particularly interesting because
the presence of ethylene is typically attributed to electrochemical
2
1
3
1
3
2
5
24
decomposition of EC as reported by Aurbach and co-workers. Fur-
ther analysis, as discussed below, confirms that generation of ethyl-
ene via thermal decomposition of ternary electrolyte can be attrib-
uted to DEC as opposed to EC.
Results and Discussion
Continued storage of 1.0 M LiPF in DEC at 85°C in excess of
6
Thermal decomposition of 1.0 M LiPF in DEC.— Due to the
6
19
4
weeks results in broadening of the F NMR resonances charac-
complexity of the thermal decomposition product distribution of
mixed carbonate electrolytes, the investigation began with the ther-
mal decomposition of 1.0 M LiPF6 in single carbonates. 1.0 M
−
6
teristic of OPF ͑OEt͒, OPF͑OEt͒ , and PF . Variable-temperature
2
2
1
9
F NMR spectroscopy indicates that the linewidth of the reso-
nances are dependent upon sample temperature. Low-temperature
LiPF in DEC affords straightforward identification of many decom-
6
1
9
͑
−15°C͒ F NMR spectroscopy unveils additional resonances at-
26
position products formed during thermal storage and provides a base
for understanding the more complex reaction products of mixed car-
tributed to OPF2OH and OPF͑OH͒2. The chemical shifts and cou-
11
bonate electrolytes. As previously reported, the thermal decompo-
pling constants are confirmed by addition of OPF OH ͑Strem
2
sition of LiPF6 in DEC upon storage at elevated temperature
Chemical Co.͒ to 1.0 M LiPF in DEC. The presence of OPF OH
6
2
͑
85–100°C for 1–4 weeks͒ generates a mixture of products, includ-
and ethylene ͑CH =CH ͒ are consistent with an elimination reaction
2 2
ing CO , POF , PF , fluoroethane, diethyl ether, OPF ͑OEt͒, and
OPF͑OEt͒ . Identification of the decomposition products was af-
͑Eq. 1͒. The generation of difluorophosphoric acid and the conse-
quent reactions of OPF OH with LiPF ͑Eq. 2͒ provides a partial
2
3
5
2
2
2
6
Downloaded on 2014-04-11 to IP 140.120.135.222 address. Redistribution subject to ECS terms of use (see ecsdl.org/site/terms_use) unless CC License in place (see abstract).

Journal of The Electrochemical Society, 152 ͑12͒ A2327-A2334 ͑2005͒
A2329
Detailed analysis of the rate of formation of thermal degradation
products of LiPF6 in DEC provides a basic understanding of the
mechanism of decomposition. A preliminary report of the mecha-
nism of thermal decomposition of 1.0 M LiPF in DEC has been
6
1
6
reported recently. Initial observations of the electrolyte decompo-
sition reveal that the concentration of decomposition products, EtF
and OPF OEt, increase exponentially as a function of time. The
2
appearance of products in an increasing exponential function is at-
tributed to an autocatalytic reaction.
The source of autocatalysis and the effect of common impurities
in LiPF -based electrolytes were investigated via the addition of
6
protic impurities including water and ethanol. The thermal decom-
position products upon addition of low concentrations of protic im-
purities ͑EtOH or H O, 500–2500 ppm͒ are similar to those ob-
2
served in 1.0 M LiPF6 in DEC. However, the addition of low
concentrations of protic impurity results in a significant increase in
the rate of electrolyte decomposition along with a decrease in the
initiation time ͑Fig. 1͒. This suggests that the reaction of LiPF with
6
protic impurities generates the autocatalytic species. The reproduc-
ibility of decomposition rates with added ethanol impurities was
superior to that of water. Therefore, the rate investigations were
conducted with added ethanol.
Figure 1. Increase in the OPF OC H 19F NMR signal as a function of time
2
2
5
for different initial concentrations of ethanol in 1.0 M LiPF in DEC stored
6
at 85°C: ͑b͒ no added ethanol; ͑᭺͒ 500 ppm added ethanol; ͑᭢͒ 1500 ppm
Proposed mechanism of DEC decomposition.— The proposed
added ethanol. The lines are the fits to the kinetic model using k
autocatalytic mechanism for the decomposition of LiPF in DEC is
6
−
1
2
=
0.011 h , r = 0.95.
shown in Scheme 1. Trace impurities of alcohol or water react with
explanation for the enhanced rate of decomposition for DEC vs
DMC. The temperature-dependent linewidths are consistent with a
rapid exchange of fluoride between OPF ͑OEt͒, OPF ͑OH͒, and
2
2
−
6
PF . Lower temperatures slow the exchange rate and sharpen the
resonances. The exchange of fluoride between fluorophosphates pro-
vides insight into the mechanism of electrolyte decomposition, as
discussed below
OPF OCH CH → OPF OH + CH =CH
͓1͔
2
2
3
2
2
2
OPF OH + LiPF → OPF OLi + HF + PF
͓2͔
2
6
2
5
In order to explore the role of PF5 and OPF in the thermal
3
decomposition of DEC, DEC solutions of PF and OPF were pre-
5
3
pared. Incorporation of an internal standard, hexafluorobenzene
HFB͒, provides quantitative determination of the concentration of
the dissolved gasses in DEC. Upon addition of PF , high concentra-
͑
5
−
2
tions of PF ͑1.2 ϫ 10 M͒ along with low concentrations of OPF
1.4 ϫ 10 M͒ are observed by F NMR spectroscopy. Collection
of F NMR spectra at low temperature ͑−35°C͒ reveals two reso-
nances characteristic of PF in a 10:1 ratio. An increase in the probe
5
3
−
3
19
͑
1
9
Scheme 1.
5
temperature results in a coalescence of the PF resonances. Storage
5
at room temperature for 1 week results in partial conversion of PF5
LiPF /PF to generate phosphorus oxyflouride ͑OPF ͒. The POF
3
reacts with DEC to generate OPF2OC2H5 along with CO2 and
6
5
3
to OPF3 ͑5:1͒ without appearance of OPF ͑OEt͒ or OPF͑OEt͒ .
2
2
19
Low-temperature F NMR spectra reveal a shift in intensity of the
two PF resonances. The shift in the concentration of the two PF
C H F. The ethylfluorophosphate interacts with PF to transiently
2
5
5
5
5
16,26,27
form a complex, similar to those previously reported.
Ex-
resonances coincides with the increase in concentration of OPF3.
change of fluoride for alkoxide generates POF3 and the unstable
PF OC H , which undergoes an Arbuzov rearrangement to generate
The observation of two forms of PF , a strong Lewis acid, at low
5
temperature suggests an equilibrium between two different PF -base
4
2
5
5
29
a second equivalent of POF and C H F.
complexes. The two resonances are attributed to DEC and OPF3
3
2
5
The kinetic data was modeled assuming a second-order reaction
as an approximation for the early reactions in the catalytic cycle ͑Eq.
complexes of PF which rapidly exchange at room temperature ͑Eq.
5
1
6,27,28
3
͒. Similar complexes of PF have been previously reported
5
4
, Scheme 1͒. The high concentration of DEC, relative to the other
PF –OPF + OC͑OEt͒ ꢀ PF –OC͑OEt͒ + OPF
͓3͔
5
3
2
5
2
3
reactants, allows the use of pseudo-first-order kinetics. The initial
OPF2OEt concentration is assumed to be proportional to the initial
ethanol concentrations due to rapid reaction of EtOH with LiPF6
͑Eq. 5͒, as discussed below. The rate equation for the proposed
decomposition reaction ͑Eq. 4͒ is provided in Eq. 6. The data was
simultaneously fit for different concentrations of EtOH ͑OPF2OEt͒,
−
2
Generation of a DEC solution of OPF ͑3.3 ϫ 10 M͒ results in
3
−
3
the formation of trace quantities of OPF ͑OEt͒ ͑2.2 ϫ 10 M͒
2
shortly after sample preparation. Storage at room temperature ͑RT͒
for 1 week results in conversion of 10% of OPF to OPF ͑OEt͒.
3
2
While PF and OPF react with carbonates at comparable rates, the
5
3
formation of OPF OEt, the observable product of electrolyte de-
assuming the concentration of LiPF to be 1.0 M. The pseduo-first-
order rate constant was determined to be 0.011 h , where
2
6
−
1
composition, proceeds by way of OPF formation.
3
Downloaded on 2014-04-11 to IP 140.120.135.222 address. Redistribution subject to ECS terms of use (see ecsdl.org/site/terms_use) unless CC License in place (see abstract).

A2330
Journal of The Electrochemical Society, 152 ͑12͒ A2327-A2334 ͑2005͒
a 19F resonance at 87.68 ppm ͑tt, J 47.6 and 31.2͒. Storage of this
sample at 150°C for 75 min results in the disappearance of the
k = kЈ ͓DEC͔ ͑Fig. 1͒. Divergence from the fit at higher percent
HF
conversion reflects the continued reactions of OPF OEt forming
2
OPF OH and OPF͑OEt͒ . The correlation of the data with the
resonance associated with FCH CH OH and the appearance of new
resonances with similar chemical shifts and coupling constants to
2
2
2
2
model supports the proposed autocatalytic mechanism ͑Scheme 1͒
the decomposition products of pure 1.0 M LiPF in EC. Resonances
6
kЈ
in the fluorophospate region ͓53.3 ͑d, 1009 Hz͒, 54.7 ͑d, 973 Hz͒,
LiPF + 2 ϫ DEC + OPF OEt → 2 ϫ OPF OEt + 2 ϫ CO
6
2
2
2
5
6.3 ͑d, 967 Hz͒, 58.5 ͑d, 946 Hz͔͒ and the alkyl fluoride region
͑triplets of triplets at −85.10, −85.11, −87.15, −87.61, −87.63,
87.87͒ with similar coupling constants ͑JHF 47 and 29 Hz͒ are
observed. The resonances have coupling constants and splitting pat-
+
3 ϫ EtF + LiF
͓4͔
−
fast
LiPF + 2 ϫ EtOH ——→ OPF OEt + LiF + 2 ϫ HF + EtF
6
2
1
terns consistent with alkyl fluorides coupled to two pairs of
H
͓
5͔
nuclei characteristic of fluorine at the terminus of an ethylene link-
age ͑–CH CH F͒. The chemical shifts and coupling patterns of
the F NMR resonances suggest oligoethylene oxides capped
with alkyl fluoride and fluorophosphate end groups
2
2
d͓OPF OEt͔
19
2
=
− k͓LiPF ͔͓OPF OEt͔
͓6͔
6
2
dt
͓
OPF ͑OCH CH ͒ F͔. Similar species have been recently
2 2 2 n
To confirm the effect of protic impurities on the thermal decom-
3
3
31
reported. Attempts to obtain P NMR spectra of the extract
proved difficult due to the low concentration of the decomposition
products and low sensitivity of P compared to F NMR spectros-
copy.
position of LiPF -based electrolytes, the direct reaction of ethanol
6
with LiPF was investigated. Dry absolute ethanol was added to a
6
3
1
19
reaction flask charged with LiPF . Storage at RT for 1 month results
6
in a partial conversion of LiPF6 to fluorophosphates including
The decomposition products were also analyzed by SEC. Extrac-
tion of the residual solid with THF followed by SEC suggests low-
molecular-weight oligomers ͑Mw = ca. 660͒, providing additional
support of the assignments of the NMR spectroscopic data.
OPF OEt and OPF͑OEt͒ , while storage at 85°C results in nearly
2
2
quantitative conversion to OP͑OEt͒ and OPF͑OEt͒ after only 1 h.
3
2
1
9
F NMR spectroscopy also supports the presence of HF at
17 ppm and LiF at −49 ppm. The observation of fluorophosphates,
−
OPF ͑OEt͒ and OPF͑OEt͒ , in the direct reaction of LiPF with
2
2
6
Proposed mechanism of EC decomposition.— The thermal de-
ethanol suggests that the rate enhancements of the thermal decom-
composition mechanism of 1.0 M LiPF in EC is similar to that of
6
position of 1.0 M LiPF in DEC upon addition of ethanol impurities
LiPF6 in DEC. Initiation occurs upon generation of the reactive
Lewis acids PF and OPF via thermal dissociation of LiPF and
6
may be related to fluorophosphate production.
5
3
6
reaction with protic impurities. Upon generation, OPF reacts with
3
Thermal decomposition of 1.0 M LiPF in EC.— While 1.0 M
the EC to produce CO and OPF OCH CH F. Subsequent reaction
6
2
2
2
2
LiPF in EC is typically not used as an electrolyte in lithium-ion
of EC with OPF OCH CH F inserts an ethylene oxide unit into the
6
2 2 2
batteries, EC is used as a cosolvent in almost all lithium-ion batter-
P–O bond with liberation of CO . Continued reaction produces the
capped oligoethylene oxides depicted in Scheme 2. Evidence for the
2
4
,30,31
ies due to its importance in stable anode SEI formation.
There-
fore, understanding the thermal decomposition of EC in the presence
of LiPF6 is essential to the understanding of electrolyte stability.
1
.0 M LiPF in pure EC treated under N reacts very slowly when
6
2
heated. At both 85 and 100°C, 1.0 M LiPF in EC produces trace
6
quantities of fluorophosphates before rapidly decomposing to an in-
soluble solid. The autocatalytic nature of the reaction and the sensi-
tivity of reaction rates to protic impurities makes analysis difficult.
Upon initiation, the decomposition reaction occurs rapidly. In order
to obtain reproducible results in a timely fashion, high temperatures
͑
240°C͒ and short reaction times ͑20–60 min͒ were utilized.
Samples of 1.0 M LiPF in EC were stored in sealed ampoules at
6
2
40°C for 20, 40, and 60 min. The products were similar for all
samples, although the concentration of decomposition products sys-
tematically increases with increased storage time. Samples stored in
excess of 60 min exploded due to excess pressure from generation
of CO . The residual solid was extracted with anhydrous Et O,
2
2
1
9
evaporated to dryness, and redissolved in anhydrous d benzene.
F
6
NMR spectra of the decomposition extract reveal resonances char-
acteristic of decomposition products in two different regions
of the spectrum. In the fluorophosphate region of the
spectrum ͑49–62 ppm͒ five doublets are observed. We assign the
resonances to OPF OH, OPF OR, OPF͑OH͒ , OPF͑OR͒ , and
Scheme 2.
generation of oligoethylene oxides capped with alkyl fluoride and
fluorophosphate end groups ͓OPF ͑OCH CH ͒ F͔ is provided by a
combination of H, F, P NMR spectroscopy and SEC. While
previous reports suggest the generation of oligocarbonates such as
dimethyl-2,5-dioxahexane carboxylate ͑DMOHC͒ in thermally
treated electrolytes,
consistent with oligoethylene oxides ͑Table I͒. An understanding of
the source of the fluorophosphoric acids, OPF OH and OPF͑OH͒ ,
2
2
2
2
2
2
2 n
OPF͑OH͒͑OR͒. Several ill-defined resonances, consistent with trip-
1
19
31
1
9
lets of triplets, are observed in the alkyl fluoride region of the
F
NMR spectrum ͑−70 to − 130 ppm͒. Attempts to determine the
relative concentrations of fluorophosphates and alkyl fluorides by
9,10
1
the chemical shifts of the H resonances are
1
9
F integration were unsuccessful due to low signal-to-noise ratio of
the decomposition products.
2
2
While the alkyl fluoride resonances could be attributed to
is incomplete. These species could be generated via reaction of trace
protic impurities in the electrolyte or solvents used during the ex-
tractions 2.
͑
FCH CH ͒ O or FCH CH OH, neither of these volatile liquids are
2
2 2
2
2
1
8,32
observed by GC-MS.
Independent addition of low concentra-
tions of FCH CH OH ͑Aldrich Chemical Co.͒ to EC reveals that
2
2
FCH CH OH is not an observed decomposition product. Addition
Thermal decomposition of LiPF in 1:1 EC/DMC.— Difficulties
2
2
6
of FCH CH OH to 1.0 M LiPF in EC results in the appearance of
with the analysis of the decomposition products of 1.0 M LiPF in
2
2
6
6
Downloaded on 2014-04-11 to IP 140.120.135.222 address. Redistribution subject to ECS terms of use (see ecsdl.org/site/terms_use) unless CC License in place (see abstract).

Journal of The Electrochemical Society, 152 ͑12͒ A2327-A2334 ͑2005͒
A2331
Figure 2. Alkylfluoride region of the 19F NMR spectrum of 1.0 M LiPF in
6
Figure 3. Gas chromatogram of ternary electrolyte, 1.0 M LiPF6 in
EC:DEC:DMC ͑1:1:1͒ stored at 85°C for 5 days.
1
:1 EC/DMC after storage at 85°C for 3 weeks.
decomposition products mentioned above, storage of ternary elec-
trolyte at 85°C in excess of seven days results in the appearance of
small peaks in the gas chromatograms characteristic of fluo-
EC lead to the exploration of a binary electrolyte, 1.0 M LiPF , in
6
1:1 EC/DMC. DMC was selected as the cosolvent due to fewer side
reactions compared to DEC. Storage of 1.0 M LiPF6 in 1:1 EC/
DMC at 85°C for 3 weeks lead to a significant graying of the solu-
romethane, dimethylfluorophosphate ͓OPF͑OCH ͒ ͔, diethylfluoro-
3 2
1
13
phosphate ͓OPF͑OCH CH ͒ ͔, and ethylmethylfluorophosphate
tion along with formation of precipitate. H and C NMR spectra of
the solution reveal resonances associated with EC and DMC along
with Me O, –͑CH CH O͒–, and CO . However, no ethylene was
2 3 2
͓
OPF͑OEt͒͑OMe͔͒. These species were previously characterized
upon thermal decomposition of LiPF in dialkyl carbonates and are
6
2
2
2
2
1
9
of particular concern because they are structurally and physiologi-
observed. F NMR spectra reveal two sharp doublets ͑51.1 ppm,
3
5
cally related to Sarin.
1
009 Hz; 51.3 ppm, 964 Hz͒ along with several broad resonances in
We previously reported the difficulties of interpretation of NMR
the fluorophosphate region. Additional resonances characteristic of
spectra of ternary electrolyte after storage for 1 week at 85°C due to
HF ͑−16.2 ppm͒, MeF ͑−130.23 ppm, quart. 46 Hz͒, and
11
the large number of different and overlapping resonances ͑Fig. 4͒.
–
CH CH F ͑−87.2 ppm, tt, 45, 28 Hz͒ are observed ͑Fig. 2͒. Poorly
2 2
Investigation of this system by NMR spectroscopy has continued
and a better understanding of the species present in thermally de-
composed ternary electrolyte has been obtained by a combination of
multinuclear NMR spectroscopy and correlation techniques.
resolved resonances at −85 ppm are consistent with additional
CH CH F species. However, none of the –CH CH F resonances
–
2
2
2
2
3
1
have a chemical shift consistent with HOCH CH F. P NMR spec-
2
2
tra reveal severely broadened resonances characteristic of rapidly
exchanging fluorophosphates. Low-temperature 31P NMR spectra
could not be acquired due to solidification of the samples below
1
13
H, C, and correlation NMR spectroscopy.— The first reaction
observed during thermal decomposition of ternary electrolyte is the
ester exchange reaction of DEC and DMC to generate ethylmethyl-
carbonate ͑EMC͒. Storage of ternary electrolyte at room temperature
for several months does not result in the generation of EMC; how-
ever, EMC is generated within a day of storage at 85°C. In lithium-
ion cells EMC production has been observed after cell
1
9
room temperature. The presence of two sharp
F doublets
in the fluorophosphate region and –CH CH F resonances is
2
2
consistent with an alkyl fluoride-terminated fluorophosphate
͓
OPF ͑OCH CH ͒ F͔ and OPF OMe. The severely broadened
2 2 2 n 2
resonances are attributed to fluorophosphoric acids ͓OPF OH and
2
3
6,37
formation.
The production of EMC has been attributed to reac-
OPF͑OH͒ ͔. Spectral analysis suggests that the thermal decomposi-
2
38
tion of DEC/DMC with the strongly reducing anode.
tion of 1.0 M LiPF6 in 1:1 EC/DMC affords similar products to
EMC is observed before detectable concentrations of other de-
composition products. Ester exchange reactions are typically attrib-
uted to either acid or base-catalyzed processes. Therefore, we were
those observed during the thermal decomposition of LiPF in EC or
6
11
DMC alone.
interested in the role of the strong Lewis acid PF in EMC produc-
Thermal decomposition of LiPF in 1:1:1 EC/DEC/DMC.— As
described earlier, one of the preferred electrolytes for commercial
5
6
tion. Spectroscopic analysis of the reaction of PF with 1:1 DEC/
5
DMC provides insight into the mechanism of ester exchange reac-
lithium-ion batteries is 1.0 M LiPF in 1:1:1 EC/DEC/DMC, herein
6
1
tions of ternary electrolyte. H NMR spectroscopy and GC-MS
referred to as the ternary electrolyte. Investigation of the thermal
decomposition of single carbonate electrolytes allows straightfor-
ward analysis of decomposition products and mechanisms. A basic
understanding of the reactions described above allows deconvolu-
tion of the complex thermal decomposition reactions of the ternary
electrolyte.
indicate that the concentration of EMC remains low ͑ϳ1%͒ after
1 month of storage at room temperature in the presence of high
concentrations of PF5͑ϳ0.1 M͒. While this observation is surpris-
ing, it suggests that ester exchange reaction requires either thermal
or electrochemical energy to facilitate the exchange.
1
13
The volatile components produced during thermal storage of this
electrolyte were analyzed by GC-MS and structurally assigned
through matching to the National Institutes of Standards ͑NIST͒
library or by deconvolution of mass-spectral patterns. The products
in order of volatility ͑early to late retention times͒ are characterized
as CO2, CH CH F, ͑CH ͒ O, CH OCH CH , OPF OCH ,
H and C NMR spectra of partially decomposed ternary elec-
trolyte contain a number of overlapping resonances that were de-
convoluted by a combination of comparison with single-carbonate
spectra, addition of known decomposition products such as diethyl
ether, along with more advanced spectroscopic techniques including
COSY, DEPT, and HETCOR. The characterized products include
ethers ͑Et O,Me O,EtOMe͒, EMC, fluorophosphates ͑OPF OMe,
3
2
3 2
3
2
3
2
3
3
4
͑
CH CH ͒ O, and OPF OCH CH ͑Fig. 3͒. In addition to the
3 2 2 2 2 3
2
2
2
Downloaded on 2014-04-11 to IP 140.120.135.222 address. Redistribution subject to ECS terms of use (see ecsdl.org/site/terms_use) unless CC License in place (see abstract).

A2332
Journal of The Electrochemical Society, 152 ͑12͒ A2327-A2334 ͑2005͒
Figure 5. Proton decoupled 13C– H HETCOR of ternary electrolyte stored
1
at 85°C for 7 days.
13C DEPT 135° spectra confirmed the assignment of many of the
carbon resonances ͑Fig. 7͒. Resonances associated with CH and
CH carbons are positive, resonances associated with CH carbons
3
2
are inverted, and resonances associated with tertiary carbons are not
Figure 4. 1H spectrum of ternary electrolyte, 1.0 M LiPF in EC:DEC:DMC
6
͑
1:1:1͒ before ͑top͒ and after heating for 7 days at 85°C ͑bottom͒.
OPF OEt, etc.͒, oligoethylene oxides ͑–OCH CH –͒, ethylene, and
2
2
2
CO . The chemical shifts and coupling constants of the decomposi-
2
tion products are summarized in Table I. The structures of many of
the decomposition products correlate with the products character-
ized by GC-MS.
1
13
Figure 5 contains a portion of the H– C HETCOR spectrum of
ternary electrolyte heated for 7 days at 85°C. The two-dimensional
spectrum allows assignment of many of the species described above.
Cross peaks are observed in the HETCOR spectrum, indicating
1
13
H nuclei are coupled to C nuclei. For example, the strong
1
13
cross peak is observed for EC ͑ H, 4.51 ppm, s; C, 64.63͒.
The presence of oligoethylene oxides is supported by1
a
correlation between several small resonances in the 3C
spectrum ͑67–69 ppm͒ with several small multiplets in
1
the H͑3.5–3.6 ppm͒. Other strong correlations are observed for ter-
minal methyl groups ͓Me O, EtOMe, and OPF O͑CH CH ͒ OCH ,
2
2
2
2 n
3
1
13
respectively,
H, C:
͑3.28 ppm, 59.1 ppm͒,
͑
3.30 ppm, 56.8 ppm͒, ͑3.32 ppm, 57.4 ppm͔͒.
The assignment of decomposition products was further facili-
1
1
tated by H– H COSY data. The region of the COSY spectrum
highlighting the ethyl fragment connectivity is provided in Fig. 6.
Cross peaks are observed indicating the CH –CH connections. In
3
2
particular, the resonances for EtF ͑4.51 ppm, q;1.25 ppm, t͒, would
have been difficult to assign without the COSY spectrum due to the
overlapping resonances of EC and DEC.
1
1
Figure 6. Portion of H– H COSY of ternary electrolyte stored at 85°C for
7 days.
Downloaded on 2014-04-11 to IP 140.120.135.222 address. Redistribution subject to ECS terms of use (see ecsdl.org/site/terms_use) unless CC License in place (see abstract).

Journal of The Electrochemical Society, 152 ͑12͒ A2327-A2334 ͑2005͒
A2333
Figure 7. 13C DEPT spectrum of ternary electrolyte stored at 85°C for
days.
7
observed. All of the resonances below 60 ppm are positive and thus
CH resonances, while all of the resonances above 60 ppm are in-
3
verted and thus CH resonances.
2
The observation of ethylene is related to that observed for LiPF6
in DEC, as described above, and electrochemical decomposition of
25
EC-containing electrolytes. Investigation of the thermal decompo-
sition of binary electrolytes suggest DEC not EC to be the source of
ethylene. Solutions of EC and DMC do not produce ethylene after
3
weeks of heating at 85°C; however, samples containing DEC and
EC or DEC and DMC contain resonances characteristic of ethylene.
The concentration of ethylene in all four electrolytes containing
DEC ͑DEC:DMC:EC, DEC:EC, DEC:DMC, and DEC alone͒ are
similar.
Figure 8. 13P NMR spectrum of the ternary electrolyte after 5 days of heat-
19F and 31
P NMR spectroscopy.— 19F and 31P NMR spectros-
1
9
ing at 85°C ͑top͒ and F NMR spectrum of the ternary electrolyte after
days of heating at 85°C ͑bottom͒.
copy has been critical for understanding the thermal decomposition
5
1
9
of LiPF -based electrolytes. Strong scalar coupling between F and
6
3
1
P allows determination of P–F connectivity. After storing ternary
electrolyte at 85°C for 5 days, three species are observed in the
addition, neither 2-fluoroethanol or bis͑2-fluoroethyl͒ether are ob-
served in GC-MS of thermally decomposed ternary electrolyte.
fluorophospate region of 19F and P NMR spectra ͑Fig. 8͒. The
31
19
F
NMR resonances are doublets consistent with a single phosphorus
coupled to each fluorine. 31P NMR spectra contain three overlapping
triplets, suggesting that each phosphorus is coupled to two flourines.
The resonances are consistent with three different difluorophosphate
or difluorophosphoric acids. From the similarity of the chemical
shifts and coupling constants to previously reported or commercially
Table II. 19F and 31P NMR spectral data of ternary electrolyte
decomposition products.
31P␦ ͑mult, JFP͒
19F␦ ͑mult, JFP͒
Compound
11
available compounds, the resonances are assigned to OPF OEt,
2
LiPF6
−145.0 ͑sept, 709͒
−35.7 ͑d, 1068͒
−20.5 ͑t, 1008͒
−20.1 ͑t, 1008͒
−19.6 ͑t, 970͒
−9.63 ͑d, 962͒
−8.44 ͑d, 961͒
−10.8 ͑d, 963͒
−10.0 ͑d, 933͒
—
65.0 ͑d, 709͒
49.4 ͑q, 1068͒
53.1 ͑d, 1008͒
51.2 ͑d, 1008͒
54.4 ͑d, 970͒
56.0 ͑d, 962͒
51.4 ͑d, 961͒
53.7 ͑d, 963͒
61.5 ͑d, 933͒
var. −17 ͑s͒
var. −49 ͑s͒
−73.43
OPF OMe, and OPF OH ͑Table II͒. The assignments of OPF OEt
2
2
2
3
1
OPF3
and OPF OMe were confirmed via acquisition of P NMR spectra
2
OP͑F͒ OEt
1
31
1
2
without H decoupling at −25°C. Additional P– H fine coupling
t, 8 Hz; q, 14 Hz; respectively͒ are detected. In addition, reso-
nances characteristic of EtF, MeF, –CH CH F, LiF, and HF
OP͑F͒ OMe
2
͑
OP͑F͒ ͑OH͒
2
2
2
OPF͑OEt͒2
OPF͑OMe͒2
OPF͑OMe͒͑OEt͒
OPF͑OH͒2
HF
19
are observed by F NMR spectroscopy ͑Table II͒. Continued
storage of ternary electrolyte at 85°C leads to further
decomposition of the electrolyte. New resonances characteristic
of OPF͑OEt͒ , OPF͑OMe͒ , OPF͑OMe͒͑OEt͒, OPF͑OMe͒͑OH͒,
2
2
LiF
EtF
—
—
and OPF͑OEt͒͑OH͒ are observed.
The resonances assigned to the alkyl fluoride termini of
a
͑
t-q, 47.7, 26.5͒
OPF ͑OCH CH͒ F ͑−87.4 ppm, tt, J = 48, 29 Hz͒ are charac-
2
2
n
HF
MeF
—
−130.34
teristic of an ethylene fluoride fragment ͑–CH CH F͒. This frag-
a
2
2
͑
q, 46.3͒
ment is present in 2-fluoroethanol and bis͑2-fluoroethyl͒ether, previ-
–
͑OCH CH ͒ F
—
−87.45
t-t, 48.7, 29.5͒
2
2 n
a
ously reported decomposition products of lithium-ion battery
͑
1
8,31
electrolytes.
However, independent addition of 2-fluoroethanol
a
to ternary electrolyte confirms the absence of 2-fluoroethanol. In
Reported coupling constants are for H–F coupling ͑JH–F͒.
Downloaded on 2014-04-11 to IP 140.120.135.222 address. Redistribution subject to ECS terms of use (see ecsdl.org/site/terms_use) unless CC License in place (see abstract).

A2334
Journal of The Electrochemical Society, 152 ͑12͒ A2327-A2334 ͑2005͒
mized via minimization of protic impurities ͑H O and EtOH͒ and
2
low concentrations of DEC vs other carbonates. Investigation of the
thermal decomposition of LiPF -based electrolytes in the presence
6
of electrode materials and in full cells is in progress and will be
presented in due course.
Acknowledgment
We thank the Lithion/Yardney Technical Products, the United
States Air Force ͑contract no. F33615-03-D-2323͒ and NSF/CIA
͑
award no. DMR-0442024͒ for financial support of this research.
The University of Rhode Island assisted in meeting the publication costs
of this article.
References
1
2
. Lithium Ion Batteries: Fundamentals and Performance, M. Wakihara and O.
Yamamoto Editor, Wiley-VCH, New York 1998.
. G. Sarre, P. Blanchard, and M. Broussely, J. Power Sources, 127, 65 ͑2004͒.
Figure 9. Increase in ͓OPF OC H ͔ as detected by 19F signal as a function
of time for ternary electrolyte with added alcohol ͑filled circles, 1000 ppm͒
and without added alcohol ͑open circles͒.
2
2
5
3. D. P. Abraham, J. Liu, C. H. Chen, Y. E. Hyung, M. Stoll, N. Elsen, S. MacLaren,
R. Twesten, R. Haasch, E. Sammann, I. Petrov, K. Amine, and G. Henriksen, J.
Power Sources, 119–121, 511 ͑2003͒.
4
5
6
. K. Xu, Chem. Rev. (Washington, D.C.), 104, 4303 ͑2004͒.
. H. H. Lee, C. C. Wan, and Y. Y. Wang, J. Electrochem. Soc., 151, A542 ͑2004͒.
. M. Herstedt, D. P. Abraham, J. B. Kerr, and K. Edstrom, Electrochim. Acta, 49,
5097 ͑2004͒.
SEC.— The precipitate generated via the thermal decomposition
of ternary electrolyte was analyzed by SEC to establish the presence
of polymers. Ternary electrolyte stored at 85°C for 6 days produced
significant quantities of solid. The THF soluble fraction of solids
contains low-molecular-weight oligomers on the order of 1000 mass
units. The THF insoluble material is uncharacterized.
7
8
. D. P. Abraham, R. D. Twesten, M. Balasubramanian, J. Kropf, D. Fischer, J.
McBreen, I. Petrov, and K. Amine, J. Electrochem. Soc., 150, A1450 ͑2003͒.
. D. Aurbach, K. Gamolsky, B. Markovsky, G. Salitra, Y. Gofer, U. Heider, R.
Oesten, and M. Schmidt, J. Electrochem. Soc., 147, 1322 ͑2000͒.
9. S. E. Sloop, J. K. Pugh, S. Wang, J. B. Kerr, and K. Kinoshita, Electrochem.
Solid-State Lett., 4, A42 ͑2001͒.
1
0. S. E. Sloop, J. B. Kerr, and K. Kinoshita, J. Power Sources, 119–121, 330 ͑2003͒.
1
1. B. Ravdel, K. M. Abraham, R. Gitzendanner, J. DiCarlo, B. L. Lucht, and C.
Mechanism of ternary electrolyte decomposition.— Monitoring
Campion, J. Power Sources, 119–121, 805 ͑2003͒.
12. D. Aurbach, B. Markovsky, A. Shechter, Y. Ein-Eli, and H. Cohen, J. Electrochem.
Soc., 143, 3809 ͑1996͒.
1
1
the rate of formation of decomposition products, OPF OEt and EtF,
2
in ternary electrolyte provided similar results to that observed for
3. P. Arora, R. E. White, and M. Doyle, J. Electrochem. Soc., 145, 3647 ͑1998͒.
4. A. M. Andersson, D. P. Abraham, R. Haasch, S. MacLaren, J. Liu, and K. Amine,
J. Electrochem. Soc., 149, A1358 ͑2002͒.
5. Y. Wang, X. Guo, S. Greenbaum, J. Liu, and K. Amine, Electrochem. Solid-State
Lett., 4, A68 ͑2001͒.
6. C. Campion, W. Li, W. E. Euler, B. L. Lucht, B. Ravdel, J. DiCarlo, R. Gitzen-
danner, and K. M. Abraham, Electrochem. Solid-State Lett., 7, A194 ͑2004͒.
7. B. L. Lucht, C. L. Campion, W. Li, B. Ravdel, J. F. Dicarlo, R. Gitzendanner, and
K. M. Abraham, in Proceedings of the 41st Power Sources Conference, June 14–17
2004, p. 442 ͑2004͒.
1
.0 M LiPF in DEC. Product formation follows an increasing ex-
6
ponential function with increased storage time ͑Fig. 9͒. Addition of
ethanol to the ternary electrolyte increases the rate of electrolyte
decomposition and decreases the initiation period, consistent with
generation of the autocatalytic species. The rate constant
1
1
1
−
1
͑
k = 0.016 h ͒ is similar to that observed for the thermal decom-
position of LiPF in DEC, supporting similarity of the decomposi-
6
tion mechanisms. Unfortunately, due to the many different reactions
and decomposition products, accurate integrations of individual de-
composition products is difficult. Thus, a full mechanistic analysis
of the thermal decomposition of ternary electrolyte was not con-
ducted. However, the detailed decomposition product analysis
coupled with the cursory rate data strongly suggest that the mecha-
nism of ternary electrolyte decomposition is a combination of the
1
1
8. J. S. Gnanaraj, E. Zinigrad, L. Asraf, H. E. Gottlieb, M. Sprecher, M. Schmidt, W.
Geissler, and D. Aurbach, J. Electrochem. Soc., 150, A1533 ͑2003͒.
9. D. D. MacNeil and J. R. Dahn, J. Electrochem. Soc., 150, A21 ͑2003͒.
20. T. Kawamura, A. Kimura, M. Egashira, S. Okada, and J.-I. Yamaki, J. Power
Sources, 104, 260 ͑2002͒.
2
2
1. G. G. Botte, R. E. White, and Z. Zhang, J. Power Sources, 97–98, 570 ͑2001͒.
2. H. Maleki, G. Deng, A. Anani, and J. Howard, J. Electrochem. Soc., 146, 3224
͑
1999͒.
reactions of LiPF in pure DEC, DMC, and EC.
23. D. H. Doughty, E. P. Roth, and C. C. Crafts, ITE Letters on Batteries, New Tech-
nologies & Medicine, 4, 158 ͑2003͒.
6
Conclusions
24. Spectrometric Identification of Organic Compounds, 6th ed., R. M. Silverstein and
F. X. Webster, John Wiley & Sons, New York ͑1998͒.
25. D. Aurbach, Y. Gofer, M. Ben-Zion, and P. Aped, J. Electrochem. Soc., 139, 451
͑1992͒.
The thermal decomposition of lithium battery electrolytes com-
posed of LiPF in organic carbonates produces carbon dioxide, eth-
6
2
2
2
6. A. V. Plakhotnyk, L. Ernst, and R. Schmutzler, J. Fluorine Chem., 126, 27 ͑2005͒.
7. G. Ceccarelli and F. Andruzzi, Makromol. Chem., 180, 1371 ͑1979͒.
8. W. Li, C. Campion, B. L. Lucht, B. Ravdel, J. DiCarlo, and K. M. Abraham, J.
Electrochem. Soc., 152, A1361 ͑2005͒.
ylene, alkyl fluorides, dialkyl ethers, fluorophosphates, fluorophos-
phoric acids, HF, LiF, and oligoethylene oxides. The product
distribution is supported by a combination of nuclear magnetic reso-
1
13
19
31
2
3
9. G. M. Kosolapoff, Organophosphorus Compounds, pp. 213–214, John Wiley &
Sons, New York ͑1950͒.
0. N. Katayama, T. Kawamura, Y. Baba, and J. Yamaki, J. Power Sources, 109, 321
nance ͑NMR͒ spectroscopy ͑ H, C, F, P, DEPT, COSY, and
HETCOR͒, GC-MS, and SEC. A mechanistic investigation of the
thermal decomposition of electrolyte suggests that the reaction is
autocatalytic. The presence of protic impurities results in generation
͑2002͒.
31. M. C. Smart, B. V. Ratakumar, J. F. Whitacre, L. D. Whitcanack, K. B. Chin, M. D.
Rodriguez, D. Zhao, S. G. Greenbaum, and S. Surampudi, J. Electrochem. Soc.,
of alkylfluorophosphates ͑OPF OR͒, the autocatalytic species, and
2
1
52, A1096 ͑2005͒.
accelerates the electrolyte degradation. Upon generation of
OPF OR, a rapid fluoride alkoxy exchange between OPF OR and
3
2. A. Hammami, N. Raymond, and M. Armand, Nature (London), 424, 635 ͑2003͒.
2
2
33. S. Laruelle, S. Pilard, P. Guenot, S. Grueon, and J.-M. Tarascon, J. Electrochem.
Soc., 151, A1202 ͑2004͒.
PF occurs, followed by an Arbuzov rearrangement which exponen-
tially increases the concentration of OPF and other decomposition
5
3
4. B. L. Lucht, C. L. Campion, B. Ravdel, J. F. DiCarlo, and K. M. Abraham, in
Electrolytes for Lithium-Ion Batteries, PV 2003-4, The Electrochemical Society
Proceedings Series, Pennington, NJ, In press.
3
products ͑Scheme 1͒. The decomposition of electrolytes containing
DEC is further accelerated by the competing elimination reaction of
OPF OEt to generate OPF OH and ethylene. The presence of
35. B. C. Saunders and G. J. Stacey, J. Chem. Soc., 1948, 695.
36. M. Terasaki, H. Yoshida, H. Tukamoto, M. Mizuani, and M. Yamachi, Denki Ka-
gaku oyobi Kogyo Butsuri Kagaku, 61, 1417 ͑1993͒.
2
2
OPF OH results in further production of PF and HF ͑Eq. 2͒.
2
5
3
3
7. A. Ohta, H. Koshina, H. Okuno, and H. Murai, J. Power Sources, 54, 6 ͑1995͒.
8. E. S. Takeuchi, H. Gan, M. Palazzo, R. A. Leising, and S. M. Davis, J. Electro-
chem. Soc., 144, 1944 ͑1997͒.
This investigation suggests that the thermal stability of lithium-
ion batteries composed of LiPF in carbonate solvents can be maxi-
6
Downloaded on 2014-04-11 to IP 140.120.135.222 address. Redistribution subject to ECS terms of use (see ecsdl.org/site/terms_use) unless CC License in place (see abstract).