Reactions in the rechargeable lithium-O2 battery with alkyl carbonate electrolytes

Article abstract of DOI:10.1021/ja2021747

The nonaqueous rechargeable lithium-O₂ battery containing an alkyl carbonate electrolyte discharges by formation of C₃H ₆(OCO₂Li)₂, Li₂CO₃, HCO₂Li, CH₃CO₂Li, CO₂, and H ₂O at the cathode, due to electrolyte decomposition. Charging involves oxidation of C₃H₆(OCO₂Li)₂, Li₂CO₃, HCO₂Li, CH₃CO₂Li accompanied by CO₂ and H₂O evolution. Mechanisms are proposed for the reactions on discharge and charge. The different pathways for discharge and charge are consistent with the widely observed voltage gap in Li-O₂ cells. Oxidation of C₃H₆(OCO ₂Li)₂ involves terminal carbonate groups leaving behind the OC₃H₆O moiety that reacts to form a thick gel on the Li anode. Li₂CO₃, HCO₂Li, CH₃CO ₂Li, and C₃H₆(OCO₂Li)₂ accumulate in the cathode on cycling correlating with capacity fading and cell failure. The latter is compounded by continuous consumption of the electrolyte on each discharge.

Full text of DOI:10.1021/ja2021747

ARTICLE
pubs.acs.org/JACS
Reactions in the Rechargeable LithiumꢀO Battery with Alkyl
2
Carbonate Electrolytes
Stefan A. Freunberger, Yuhui Chen, Zhangquan Peng, John M. Griffin, Laurence J. Hardwick,
†
†
†
†
†
‡
§
,†
Fanny Bard ꢀe , Petr Nov ꢀa k, and Peter G. Bruce*
†
School of Chemistry, University of St. Andrews, North Haugh, St. Andrews, Fife KY16 9ST, U.K.
Toyota Motor Europe, Technical Centre, Hoge Wei 33 B, B-1930 Zaventem, Belgium
‡
§
Electrochemistry Laboratory, Paul Scherrer Institut, CH-5232 Villigen PSI, Switzerland
S Supporting Information
b
ABSTRACT: The nonaqueous rechargeable lithiumꢀO battery
2
containing an alkyl carbonate electrolyte discharges by formation of
C H (OCO Li) , Li CO , HCO Li, CH CO Li, CO , and H O at
3
6
2
2
2
3
2
3
2
2
2
the cathode, due to electrolyte decomposition. Charging involves
oxidation of C H (OCO Li) , Li CO , HCO Li, CH CO Li accom-
3
6
2
2
2
3
2
3
2
panied by CO and H O evolution. Mechanisms are proposed for the
2
2
reactions on discharge and charge. The different pathways for
discharge and charge are consistent with the widely observed voltage
gap in LiꢀO cells. Oxidation of C H (OCO Li) involves terminal
2
3
6
2
2
carbonate groups leaving behind the OC H O moiety that reacts to form a thick gel on the Li anode. Li CO , HCO Li, CH CO Li,
3
6
2
3
2
3
2
and C H (OCO Li) accumulate in the cathode on cycling correlating with capacity fading and cell failure. The latter is
3
6
2
2
compounded by continuous consumption of the electrolyte on each discharge.
1
. INTRODUCTION
widespread use of organic carbonate electrolytes in LiꢀO
cells
2
and their performance, it is important to understand the pro-
cesses that occur on discharge and subsequent charge. What
exactly are the discharge products? How do they form? How do
they decompose on charging, and how does the cell cycle yet also
fade and die? We show that on discharge a mixture of lithium
propyl dicarbonate, C H (OCO Li) , Li CO , HCO Li, CH -
The rechargeable lithiumꢀO battery is receiving a great deal
2
of interest at the present time because, theoretically, it can store
5
ꢀ10 times more energy than current lithium-ion batteries, the
energy density of which can, at best, be doubled. The recharge-
able lithiumꢀO battery could transform energy storage, vital in
2
3
6
2
2
2
3
2
3
order to address global warming; however, it remains some way
from a practical device at present.
CO Li, CO , and H O is formed. Charging involves decomposi-
2
2
2
tion of C H (OCO Li) , HCO Li, CH CO Li, and Li CO
3
6
2
2
2
3
2
2
3
The nonaqueous lithiumꢀO battery was first reported by
2
along with CO and H O evolution. HCO Li, CH CO Li,
2
2
2
3
2
Abraham and Jiang, with recent contributions by a number of
groups, including those listed in the cited references.
1
ꢀ18
Li CO , and C H (OCO Li) accumulate in the cathode on
2 3 3 6 2 2
A
þ
cycling, correlating with capacity fading and cell failure alongside
electrolyte consumption.
typical lithiumꢀO battery consists of a lithium anode, a Li
2
conducting organic electrolyte, and a porous cathode composed
1
ꢀ4,8ꢀ10,17ꢀ27
of carbon black, a catalyst, and binder.
It is believed
2. EXPERIMENTAL SECTION
that the LiꢀO battery operates by reduction of O (from the
2
2
atmosphere) to form Li O within the pores of the cathode, with
2
2
1
ꢀ4,8ꢀ12,15ꢀ27
2.1. Electrochemical Measurements. Propylene carbonate
the process being reversed on charge.
(PC) was distilled over a packed bed column and dried for several days
Organic carbonate-based electrolytes (e.g., LiPF in propylene
6
over freshly activated molecular sieves (type 4 Å). All solvents had a final
water content of e4 ppm (determined using a Mettler-Toledo Karl
Fischer titration apparatus). Electrochemical grade LiPF (Stella) was
6
used for preparing the electrolytes. The electrochemical cells used to
investigate cycling were based on a Swagelok design and composed of a
carbonate) have been the most widely used in LiꢀO cells to
2
1
ꢀ4,8,10,17ꢀ22,24ꢀ26
date.
However, recently it has been reported
that such electrolytes decompose in LiꢀO cells on discharge,
2
2
8ꢀ31
rather than form Li O .
Despite such electrolyte decom-
2
2
position, prototype LiꢀO batteries with organic carbonate
2
Li metal anode, an electrolyte (1 M LiPF in PC) impregnated into a
6
32
electrolytes have been described. Organic carbonate cells have
glass fiber separator (Whatman), and a porous cathode. The porous
cathode consisted of carbon black (Super P, TIMCAL), R-MnO
ꢀ1
been shown to exhibit specific energies >1000 W h kg (based
on total mass of the electrode) and, in some cases, can sustain
more than 100 cycles.
may even be sufficient for some applications. Given the
2
2
4,28
High capacity for a few 10 s of cycles
Received: March 9, 2011
Published: April 27, 2011
r 2011 American Chemical Society
8040
dx.doi.org/10.1021/ja2021747
|J. Am. Chem. Soc. 2011, 133, 8040–8047

Journal of the American Chemical Society
ARTICLE
8
nanowires, and polyvinylidenefluoride (PVdF-HFP, Kynarflex 2801)
(mass ratios of 24:42:34) cast on glass from a slurry made with acetone.
The cathodes used to explore charging of model compounds contained,
in addition, the relevant finely ground compound. The cell was gastight
except for the Al grid window that exposed the porous cathode to the O2
atmosphere. The cell was operated in 1 atm of O . Electrochemical
2
measurements were performed at room temperature using a Maccor
battery cycler.
2
.2. Synthesis. Model discharge compounds based on Li alkyl
carbonate salts were synthesized by routes described in the Supporting
Information.
2
.3. Mass Spectrometry (MS). The cells again consisted of a
lithium anode, electrolyte (1 M LiPF in PC) impregnated into a glass
6
fiber separator, and a porous cathode. MS analysis was carried out to
examine the gases evolved on discharge and charge, by purging the cell
with a carrier gas (N or Ar) after charge or discharge, so that the gases
2
evolved were transferred into the mass spectrometer. The process of
2
Figure 1. FTIR spectrum of a composite electrode (Super P/R-MnO /
charging Li
DEMS) involving stepping the current while continuously flowing Ar
to carry the gases evolved to the mass spectrometer. The experimental
2 3
CO was followed by in situ differential electrochemical MS
Kynar) discharged in 1 M LiPF /PC to 2 V in O . The reference spectra
6 2
ꢀ1
(
2 2 2 3
for Li O (small impurity peaks at 1080, 1450, and 1620 cm ), Li CO ,
and the electrode before discharge are also shown.
33
setup is described in detail elsewhere.
.4. Physical Characterization. All procedures were performed
2
severe decomposition of the organic carbonate electrolyte on
in an Ar-filled glovebox. Examination of electrodes involved first
disassembling the cell; if not otherwise stated, rinsing the cathode twice
2
8ꢀ31
discharge, in accord with recent studies.
that the presence of some electrolyte degradation is to be
expected, based on studies of O reduction in alkyl carbonates
It is worth noting
1
13
with CH
3
CN; and removing the solvent under vacuum. H- and
C
solid-state NMR experiments were performed on the cathode using a
Bruker Avance III spectrometer operating at a magnetic field strength of
2
3
4,35
carried out some years ago.
The data in Figure 1 and in
previous investigations were obtained on an electrode with a
catalyst. To confirm that the electrolyte decomposition is not
simply due to the catalyst, FTIR data were collected on glassy
carbon and a porous electrode without a catalyst. The FTIR
spectra at the end of discharge are shown in Figure S1 and
confirm the dominance of electrolyte decomposition in the
absence of catalyst.
1
2
4.1 T. Experiments were performed using airtight rotors in a Bruker
.5 mm probe, with a magic-angle spinning (MAS) rate of 30 kHz. For
liquid-phase NMR analysis the previously rinsed and dried electrodes
were extracted with D O, and then the extracted solution was examined
on a Bruker Avance II 400 spectrometer. Spectra were referenced to the
3
CH peak of propylene glycol (found in the discharged electrodes)
measured separately in CDCl . Powder X-ray diffraction (PXRD) was
carried out using a STOE STADI/P diffractometer operating in trans-
mission mode with a primary beam monochromator and position
2
3
Previous studies did not identify the specific discharge prod-
ucts that accompany Li CO ; however, identifying such prod-
2
3
1
sensitive detector. Fe KR radiation (λ = 1.936 Å) was employed. The
ucts is important in order to understand how the LiꢀO cell
2
samples were contained in an airtight X-ray holder. FTIR measurements
were carried out on a Nicolet 6700 spectrometer (Thermo Fisher
Scientific) either in transmission with a CsI pellet or on an ATR unit
operates in the absence of Li O formation. Although FTIR
2
2
spectroscopy has demonstrated decomposition and has identi-
fied the presence of Li CO , it is more difficult to distinguish
2
3
in a N -filled glovebox.
2
between the other possible Liꢀorganic products by FTIR alone.
2
.5. In Situ Surface Enhanced Raman Spectroscopy (SERS).
This is illustrated by examining the FTIR spectra of several
An electrochemical cell with a sapphire window was employed. The
working electrode was exposed to an excitation wavelength of 632.8 nm
using a Renishaw Raman system with collector optics based on a Leica
inverted microscope.
possible discharge products. On discharge, O is first reduced to
2
ꢀ
O2 , which then attacks the PC and could, in principle, give rise
to a variety of products. A number of these have been synthesized
directly (as described in Supporting Information) and their
spectra compared with those of the discharged cathode (see
Figure S2). The spectra of some of these compounds are
indistinguishable, e.g., f and g in Figure S2. Many of the spectra
are too similar when present in a mixture with Li CO and Kynar
3
. RESULTS AND DISCUSSION
3
.1. The Discharge Process. A typical LiꢀO cell, consisting
2
2
3
of a lithium metal anode, a 1 M solution of LiPF in propy-
lene carbonate as the electrolyte and a porous cathode (Super
P/R-MnO /Kynar) was constructed as described in the Experi-
mental Section. After galvanostatic discharge in O , the cell was
disassembled, and FTIR data were collected from the cathode
without exposure to air, Figure 1. The FTIR data extend down to
(as they are in the discharged electrode) to determine unam-
biguously which compounds form on discharge, Figure S2.
In order to make progress in identifying the specific
6
2
1
3
1
discharge products, solid-state C and H NMR (ssNMR)
data were collected on the discharged cathode, a and b of
Figure 2, respectively; details of the NMR experiments are presen-
ted in the Experimental Section. The peak assignments are given
in the figure. The spectra indicate the presence of several
moieties, including carbonate, propyl, and HꢀCdO groups.
While these results do give more insight into the nature of the
products than FTIR, the ssNMR spectra alone cannot identify
complete compounds. To address this problem the cathodes
2
ꢀ
1
2
50 cm , in order to include the spectral region within which
the three bands of Li O occur. Previous FTIR studies did not
2
2
28,31
extend down to such low wavenumbers.
in Figure 1 shows that the discharge products are not dominated
by Li O , that there is a significant proportion of Li CO , and
The FTIR spectrum
2
2
2
3
that peaks associated with CdO, CꢀO, CꢀC, and CꢀH groups
are also present. In other words, the FTIR data demonstrate
were washed with D O, and the resulting solution was subjected
2
8
041
dx.doi.org/10.1021/ja2021747 |J. Am. Chem. Soc. 2011, 133, 8040–8047

Journal of the American Chemical Society
ARTICLE
3
6,37
work.
Mechanisms have been proposed previously for the
34,38
reaction between reduced O species and PC.
Those in the
2
þ
absence of Li cannot, of course, reproduce the products
observed here. Some reports have suggested that in the presence
þ
of Li reduced oxygen species lead to the formation of com-
34,38
pounds containing alkoxy and carbonate groups.
In sum-
mary, none of the mechanisms proposed previously can explain
the range of products observed here. To address this, we propose
a possible reaction path in Scheme 1. For clarity we shall describe
the scheme in three parts.
(
i) The reactions on discharge commence with O reduction
2
ꢀ
ꢀ
5,39
to O2 . Early work regarding the reactivity of O toward
2
3
esters suggested attack on the C of the carbonyl.
More
ꢀ
recently, a DFT study on the nucleophilic attack of O2 on
PC reported a prohibitively high activation energy for
attack at this position, whereas attack at the CH group via
2
a S 2 mechanism is much more favorable, in accord with
N
40,41
earlier experimental studies.
Such a S 2 addition of
N
ꢀ
O2 to the methylene group, reactions 2 and 3 in Scheme 1,
results in ring-opening and the formation of the peroxor-
adical species 2. This then loses O , reaction 4, to form the
2
bifunctional Li alkoxycarbonate radical 3, which is easily
reduced to form, in the presence of CO , the final product,
2
42
Li propylene dicarbonate 4. Reaction 4 proceeds via a
43,44
dimer of 2.
Note that the CO is generated by the
2
13
Figure 2. (a) C MAS NMR spectrum of discharged composite
reactions described in (ii), and its presence was demon-
strated by MS analysis on discharge, Figure S4.
1
electrode (Super P/R-MnO /Kynar). (b) H MAS NMR spectrum of
2
1
the same electrode. (c) H solution NMR of the same electrode
(
ii) Intermediate 2 is known, in the presence of O , to easily
2
2 6
extracted with D O. Electrodes were discharged in 1 M LiPF /PC to
4
4ꢀ46
undergo oxidative decomposition reactions.
There
2
V under O2.
are many possible reaction pathways; in modeling such
oxidative decomposition reactions (e.g., combustion re-
actions) 100 s of pathways are often considered, which
makes it virtually impossible to formulate a unique reaction
1
to H NMR spectroscopy. The spectrum is shown in Figure 2c
and is consistent with the presence of propylene glycol, formic
acid, acetic acid, and residual PC. Propylene glycol is the product
expected from the reaction between Li propyl dicarbonate and
water, according to the process 2ROCO Li þ D O f 2ROD þ
44
path. Incomplete decomposition leads to the formation
þ
of formic and acetic acids. In the environment of the Li
2
2
electrolyte the equivalents are Li formate and Li acetate,
which were observed as decomposition products in our
studies.
Li CO þ CO . Similarly, formic and acetic acids arise from the
2
3
2
reaction between Li formate, Li acetate, and D O. Integration of
2
the areas under the peaks yields the following mole ratio of
Li propyl dicarbonate/Li acetate/Li formate, 1:0.22:0.21 (corres-
ponding to a mass ratio of 1:0.084:0.062). The presence of Li
propyl dicarbonate, Li acetate, and Li formate in the dis-
charged product is consistent with the ssNMR results and with
the FTIR spectrum of the discharged electrode, Figure S3
(
iii) Finally, formation of Li CO can occur by reaction be-
2
3
ꢀ
47,48
tween O2 and CO , see reactions 6 and 7 in Scheme 1.
2
In addition to the mechanism shown in Scheme 1, H O
2
formed in the oxidative decomposition reactions and was
observed in MS analysis (Figure S4) and can react with
the Li propylene dicarbonate to form Li CO , CO , and
2
3
2
(
although, as stated above, it would have been difficult to identify
1
the corresponding alcohol.
The proposed reaction mechanisms are consistent with the
discharge products observed by NMR, FTIR, and MS: C H -
all the discharge products from FTIR alone). Note that H NMR
of the electrolyte removed from the cell after discharge showed
only PC, confirming that there are no discharge products in the
liquid phase. The gases evolved on discharge were analyzed by
MS (see Experimental Section) and shown to consist of H2,
H O, and CO , Figure S4. H can arise by reaction of evolved
3
6
(
OCO Li) , Li CO , HCO Li, CH CO Li, CO and H O.
2 2 2 3 2 3 2 2 2
Prior to the introduction of the nonaqueous LiꢀO cell,
2
several authors reported that organic carbonates and esters are
2
2
2
35,36,39
susceptible to attack by reduced O species,
yet it was
H O with the Li metal anode. Therefore, combining all of the
2
2
widely believed that operation of the LiꢀO cell in such elec-
results, we have identified that the discharged product in the
cathode contains Li propyl dicarbonate, Li CO , Li formate, and
Li acetate and some residual PC trapped within the solid
products as well as H O and CO .
Previous studies of reduction in Li ꢀPC electrolytes have
focused mainly on reactions aimed at understanding formation
of the solidꢀelectrolyte interface (SEI) layer on Li metal and
graphite electrodes. As a result, such studies have taken place at
much lower voltages than used here and in the absence of O2;
therefore, the mechanisms are not applicable to the present
2
trolytes was dominated by Li O formation on discharge. This
2 2
2
3
was in part due to the cell’s ability to sustain chargeꢀdischarge
cycling and because of ex situ Raman spectra on a cathode after
2
2
þ
1
discharge, which showed evidence of Li
O
. In contrast,
2
2
recent reports based on FTIR and MS concluded that Li O
2 2
2
8,31
is not formed.
However, as shown in the FTIR spectrum
of the discharged cathode in Figure 1 (extended down to
ꢀ
1
250 cm ), FTIR can clearly demonstrate extensive electro-
lyte decomposition, but it cannot rule out the formation of a
8
042
dx.doi.org/10.1021/ja2021747 |J. Am. Chem. Soc. 2011, 133, 8040–8047

Journal of the American Chemical Society
ARTICLE
Scheme 1. Proposed Reaction Scheme on Discharge To Explain Formation of the Compounds: Li Propyl Dicarbonate, Li
a
Formate, Li Acetate, Li CO , CO , and H O
2
3
2
2
a
ꢀ
þ
ꢀ
2
The O2 generated in the first step could combine rapidly with Li , forming LiO , which may act as a nucleophile instead of, or as well as, O . The
2
subsequent reactions in the scheme would be the same. However, no LiO
in Figure 3.
2
was detected on discharge by surface-enhanced Raman spectroscopy (SERS)
6 2
Figure 3. SERS on an Au electrode in 1 M LiPF /PC under O at open
circuit potential and after discharge to 2 V. The dashed line marks the
position where Li O , if present, would occur.
2
2
minor amount of Li O because of extensive overlap in the
2
2
region of the three Li O peaks with peaks from Li CO and
2
2
2
3
the other Liꢀorganic compounds. The previous MS studies
were carried out on charging a discharged electrode in He; it
was concluded that Li O was not formed on the basis of the
2
2
absence of O in the gases evolved on charging. These are
2
excellent studies; however, we now know (see discussion in
the next section) that charging the discharge products can
involve reaction with O so that, if Li O had formed on
2
2
2
discharge, the O evolved on subsequent charging may have
2
been consumed by oxidation of the Liꢀorganic compounds.
In view of the conflicting reports concerning the possible
presence of Li O , we carried out in situ surface-enhanced
Figure 4. (a) FTIR spectra of a pristine electrode (Super P/R-MnO
2
/
Kynar) and after the first discharge, then charge in 1 M LiPF in PC
6
2
2
under O . (b) Variation of voltage as cell is discharged then charged.
2
Raman spectroscopy (SERS). Raman is particularly sensitive
to Li O , and the in situ nature of the experiment ensures no
contamination from air. SERS data collected at 2 V on a
roughened Au electrode at the end of discharge in 1 M LiPF6
2
2
apparent. We conclude that no significant quantity of Li O
can be detected on discharge.
2
2
in PC saturated with O are reported in Figure 3. There is no
3.2. The Charging Process. Although the recently reported
2
evidence of any Li O , but a band associated with Li CO is
MS data on charging an organic carbonate LiꢀO cell after the
2
2
2
3
2
8
043
dx.doi.org/10.1021/ja2021747 |J. Am. Chem. Soc. 2011, 133, 8040–8047

Journal of the American Chemical Society
ARTICLE
Figure 5. Composite electrodes (Super P/R-MnO
2 6
/Kynar) that contain the discharge products individually were subjected to charging in 1 M LiPF in
PC under O . (a) FTIR spectra of the as-prepared electrodes and the charged electrodes for each of the compounds, together with the spectrum of a
2
pristine electrode. (b) The corresponding charging curves at 70 mA/g (based on mass of carbon). Since the theoretical capacities of the different
ꢀ
compounds vary, to aid comparison the capacities are all normalized to unity (theoretical capacities: Li propyl dicarbonate 1000 mA h/g, 2 e /mol;
ꢀ
ꢀ
ꢀ
Li
under O
evolution measured by DEMS on oxidation of a composite electrode containing Li
2
CO
3
1500 mA h/g, 2e /mol; CH
3
CO
2
Li 750 mA h/g, 1e /mol; HCO
2
Li 750 mA h/g, 1e /mol). (cꢀe) MS gas analysis at the end of charging
2
of CH CO Li (c), C (OCO
3
2
3
H
6
2
Li)
2
(d), and HCO
2
Li (e). Note that unmarked peaks arise from fragments of CO
2
, H
2
O, O
2
, and Ar. (f) Gas
2
3
CO in response to a stepwise increased current under Ar.
first discharge demonstrated CO evolution, suggesting decom-
HCO Li on charging are described below and presented in
Scheme 2.
2
2
31
position of carbonates, it did not provide direct evidence of
carbonate decomposition and in particular did not report how
the different discharge products decompose on charging. Here
we examine how each of the discharge products behaves on
charge using a combination of DEMS, MS, FTIR, and PXRD.
Before considering each product separately, FTIR spectra
were collected on a pristine composite electrode, an electrode
after discharge, and then subsequent charge, Figure 4. The
spectra from the pristine electrode and the electrode after one
cycle are identical, indicating that the products formed on dis-
charge are decomposed on charging. Composite electrodes
containing each of the discharge products separately were prepared
and charged, Figure 5. The FTIR spectra in Figure 5a confirm the
decomposition of each compound on charging. The charging
voltages for Li CO , HCO Li, and CH CO Li are in a similar
3.2.1. C H (OCO Li) . The charging mechanism for C H -
3
6
2
2
3 6
(OCO Li) , Scheme 2, involves oxidation of the two terminal
2
2
carbonate groups to yield an ORO diradical, 5, which can react
3
3
in two ways. It can either undergo oxidative decomposition
reactions in the presence of O to yield CO and H O, reaction
2
2
2
10, or abstract a proton (e.g., from H O) to form propylene
2
45
glycol, reaction 11. The proposed reaction scheme is consistent
with the MS analysis on charging, Figure 5d, which shows
evolution of CO and H O; it is also in accord with the amount
2
2
of CO evolved, Table 1. If oxidation involved decomposition of
2
5 only via reaction 11, then 2 mol of CO per mol of C H -
2
3 6
(OCO Li) is predicted; if reaction 10 also occurs, then addi-
2
2
tional CO is predicted. Experimentally, Table 1, 2.3 mol of CO
2
2
is observed, indicating both paths occur. The formation of
propylene glycol on charging C H (OCO Li) was demon-
2
3
2
3
2
range, close to the charging voltage of the cell in Figure 4; the
charging voltage of C H (OCO Li) is somewhat lower,
3
6
2
2
strated as follows. An electrode containing C H (OCO Li)
3
6
2
2
3
6
2
2
Figure 5b. As a result, when C H (OCO Li) is present in the
was charged in 1 M LiPF in tri-ethylene glycol dimethyl ether. A
3
6
2
2
6
mixture of discharge products, it is subjected to a higher over-
potential on charging, Figure 4b. Possible mechanisms for
decomposition of C H (OCO Li) , Li CO , CH CO Li, and
thick gel formed on the Li metal anode, which was scraped off at
1
the end of charging, washed with CH CN, and analyzed by H
3
NMR in D O, see Figure S5. Propylene glycol was detected along
3
6
2
2
2
3
3
2
2
8
044
dx.doi.org/10.1021/ja2021747 |J. Am. Chem. Soc. 2011, 133, 8040–8047

Journal of the American Chemical Society
ARTICLE
Scheme 2. Proposed Mechanisms of Charging C H -
3
6
(
OCO Li) , Li CO , HCO Li, and CH CO Li
2 2 2 3 2 3 2
Table 1. Moles of CO Evolved per Mole of Compound on
2
a
Oxidation (charging)
mol CO
2
/mol compound
experimental
Figure 6. (a) Discharge and charge capacity vs cycle number for a
composite electrode (Super P/R-MnO
2
/Kynar) cycled between 2 and
.2 V in 1 M LiPF in PC under O . (b) FTIR spectra of composite
compound
(OCO Li)
theoretical
4
6
2
electrodes. Pristine and after the indicated number of cycles between 2
C
3
H
6
2
2
g2
g1
1
2.3 ( 0.12
1.7 ( 0.05
0.9 ( 0.05
2.0 ( 0.14
and 4.2 V at the end of charge. Spectra of Li acetate, formate, and
Li
HCO
CH
2
CO
3
1
carbonate are shown for comparison. (c) H solution NMR spectrum of
2
Li
D O extract from the composite electrode after 30 cycles at the end of
2
3
CO
2
Li
2
charge. Integration of the areas under the peaks yields a mole ratio of Li
propyl dicarbonate/Li acetate/Li formate of 1:3:1.1 (corresponding to a
mass ratio of 1:1.09:0.32).
a
Theoretical values are based on the reactions in Scheme 2; experi-
mental values are obtained from the MS analysis of gas evolution on
charge, Figure 5.
could form by direct contact between Li and PC electrolyte. We
have observed large amounts of a gel-like substance on the Li
with some remaining ether. The process can be understood as
follows. Oxidation of C H (OCO Li) produces propylene
glycol, Scheme 2. The gel then forms from the direct reduction
of the propylene glycol on the Li surface, resulting in products
such as Li propoxide, which can account for part of the H2
metal anode on cycling many LiꢀO cells with PC electrolytes,
3
6
2
2
2
36,49
which far exceeds the usual SEI layer formation.
3.2.2. Li CO . Since Li CO is readily detected by powder
2
3
2
3
X-ray diffraction, such data were collected before and after
charging, Figure S6, confirming the disappearance of the Li CO
3
evolution, Figure 5d. On washing the gel with D O, this converts
2
2
back to propylene glycol. Tri-ethylene glycol dimethyl ether was
chosen for the electrolyte to avoid possible confusion between
the products of C H (OCO Li) oxidation and products that
in accord with the FTIR data, Figure 5a. To investigate the
oxidation of Li CO DEMS data were collected during charging
2
3
under Ar, Figure 5f, in order to probe possible O evolution;
3
6
2
2
2
8
045
dx.doi.org/10.1021/ja2021747 |J. Am. Chem. Soc. 2011, 133, 8040–8047

Journal of the American Chemical Society
ARTICLE
CO was evolved, but there is almost no O . Charging was also
ratio of 1:0.084:0.062). This is why the acetate and formate can
be clearly seen in the FTIR in Figure 6 but not in Figure 1. It
2
2
carried out under O , and the MS results showed that the only
2
additional gases were CO , H O, and H . Previous studies of
appears that the capacity fading of the LiꢀO cell with organic
2
2
2
2
carbonate oxidation in aqueous solutions have described a
carbonate electrolyte is associated with the accumulation of these
products in the cathode. PC also becomes increasingly trapped
50
mechanism involving peroxydicarbonate as an intermediate.
A
ꢀ1
similar mechanism is proposed in Scheme 2 and is consistent with
on cycling, as evident from the growing band at 1780 cm
,
the evolution of CO but not O , as observed. After formation of
Figure 6b, and the NMR data, Figure 6c.
2
2
ꢀ
the peroxydicarbonate by 1e oxidation and dimerization, reactions
5
0
1
2 þ 13, further oxidation and decarboxylation, reactions 14 þ 15,
ꢀ
4. CONCLUSIONS
can yield the CO
radical 6. The equilibrium, reaction 16,
4
3
ꢀ
In conclusion, we have shown that the LiꢀO cell, containing
between the radical 6 and CO þ O has been described
2
2
2
4
7
ꢀ
alkyl carbonate electrolyte, cycles by the decomposition of the
previously. The O2 will in turn react with, e.g., PC to evolve
34
electrolyte on discharge to form Li CO , C H (OCO Li) ,
additional CO rather than being oxidized to dioxygen. To-
2
3
3
6
2
2
2
CH CO Li, HCO Li, CO , and H O and oxidation of Li CO ,
gether, these data demonstrate that it is indeed possible to
3
2
2
2
2
2
3
C H (OCO Li) , CH CO Li, HCO Li on charging, with no
charge, and hence decompose, Li CO within the voltage range
3
6
2
2
3
2
2
2
3
evidence of reversible oxygen reduction to Li O . These reac-
studied here. Analysis of the gases evolved on charging under O2
show that 1.7 mol of CO per mol Li CO is obtained, in accord
2 2
tions are not reversible, since the pathways for reduction (discharge)
are different from oxidation (charge), correlating with the widely
observed voltage gap (charging voltage greater than discharge
even at low rates) in LiꢀO cells. Oxidation of C H (OCO Li)
2
2
3
with the mechanism in Scheme 2.
3
.2.3. HCO Li and CH CO Li. After removal of one electron
2 3 2
and loss of CO , then H and CH3 radicals can form,
2
3
6
2
2
2
3
3
occurs at the terminal carbonate groups, leaving behind the
respectively. In the presence of O they may react to produce
2
OC H O moiety that reacts to form a thick gel-like deposit on
HO2 and CH OO , which then further decompose to H O and
3 6
3
3
3
2
45,51
the Li anode. Li CO , C H (OCO Li) , CH CO Li, and HCO
CO , in accord with previous studies.
This is supported by
2
3
3
6
2
2
3
2
2-
2
Li, accumulate in the cathode on cycling, resulting in capacity
fading and eventually cell failure. Such a failure mechanism is
compounded by consumption of the electrolyte on each cycle.
Recog-
mass spectra of the gas phase after charging HCO Li and
2
CH CO Li in O , which show H and H O evolution along
3
2
2
2
2
with CO , Figure 5cþe. The close match between the amount
2
of CO predicted to be evolved based on Scheme 2 and the
2
nition that organic carbonates are not suitable as electrolytes
observed CO evolution, Table 1, supports the proposed reac-
2
for LiꢀO cells has led to interest in other solvents. Ethers
tion paths. Again, H can arise by reaction of evolved H O with
2
2
2
are attractive because they can function with Li anodes, are of
low cost, and are safe; however, we shall show in a future paper
that, although more stable than carbonates, they too exhibit
the Li metal anode.
.3. Cycling and Capacity Fading. Not only is it possible to
3
recharge the LiꢀO cell, even in the absence of reversible oxygen
2
decomposition of the electrolyte. Successful Li O formation/
reduction to Li O , it is also possible to cycle the cell. Cycling
2 2
2
2
5
2
decomposition has been demonstrated in acetonitrile, but this
is not a suitable solvent for practical lithium cells, especially with
Li metal anodes.
involves, on discharge, repeated decomposition of the electrolyte
and, on charge, oxidation of the decomposition products; how-
ever, why does the capacity fade on cycling? One obvious mecha-
nism is electrolyte starvation: when the entire electrolyte is
consumed, then discharge, and hence cycling, cannot take place.
However fading and cell failure can occur even when there is
ample electrolyte present. To investigate this question, a series of
FTIR spectra were collected as a function of cycle number and at
the end of charge; the data are presented in Figure 6b, along with
the specific capacity as a function of cycle number. The FTIR
spectrum at the end of charge after two cycles is identical to the
as-prepared electrode; however, after six cycles, prominent bands
’
ASSOCIATED CONTENT
S
Supporting Information. Experimental procedures, sup-
b
porting figures, and reaction schemes. This material is available
free of charge via the Internet at http://pubs.acs.org.
’ AUTHOR INFORMATION
ꢀ
1
ꢀ1
Corresponding Author
p.g.bruce@st-andrews.ac.uk
at 1430 and 1530 cm appear, and the band at 1610 cm grows
significantly in relative intensity. Associated with the changes in
the FTIR spectra, a significant decrease in the specific capacity is
observed, Figure 6a. After 10 and 30 cycles the changes in the
FTIR become even more prominent, and the capacity continues
to fade. The FTIR data indicate the accumulation of Li CO and
’
ACKNOWLEDGMENT
P.G.B. is indebted to the EPSRC, EU, and Toyota Motor
Europe for financial support.
2
3
suggest the increasing presence of Li acetate and Li formate. To
1
probe the products that accumulate on cycling further, a H
solution NMR spectrum of the D O extract from an electrode
2
’
REFERENCES
after 30 cycles at the end of charge was obtained, Figure 6c. Peaks
indicate that the charged electrode after 30 cycles contains Li
propyl dicarbonate, Li acetate, and Li formate in the ratios 1:3:1.1
(
(
(
1) Abraham, K. M.; Jiang, Z. J. Electrochem. Soc. 1996, 143, 1.
2) Read, J. J. Electrochem. Soc. 2002, 149, A1190.
3) Read, J.; Mutolo, K.; Ervin, M.; Behl, W.; Wolfenstine, J.;
(
corresponding to a mass ratio of 1:1.09:0.32), along with the
Driedger, A.; Foster, D. J. Electrochem. Soc. 2003, 150, A1351.
(4) Ogasawara, T.; Debart, A.; Holzapfel, M.; Novak, P.; Bruce, P. G.
J. Am. Chem. Soc. 2006, 128, 1390.
(5) Kuboki, T.; Okuyama, T.; Ohsaki, T.; Takami, N. J. Power Sources
2005, 146, 766.
Li CO clearly identified by FTIR. The relative proportions of Li
2
3
acetate and Li formate are much higher than in the cathode after
the first discharge where the ratios of Li propyl dicarbonate/Li
acetate/Li formate were 1:0.22:0.21 (corresponding to a mass
8
046
dx.doi.org/10.1021/ja2021747 |J. Am. Chem. Soc. 2011, 133, 8040–8047

Journal of the American Chemical Society
ARTICLE
(
6) Visco, S. J.; Katz, B. D.; Nimon, Y. S.; De Jonghe, L. C.; U.S.
Patent 7,011,7007, 2007.
7) Imanishi, N.; Hasegawa, S.; Zhang, T.; Hirano, A.; Takeda, Y.;
Yamamoto, O. J. Power Sources 2008, 185, 1392.
8) D ꢀe bart, A.; Paterson, A.; Bao, J.; Bruce, P. Angew. Chem., Int. Ed.
008, 47, 4521.
9) Yang, X.-H.; He, P.; Xia, Y.-Y. Electrochem. Commun. 2009,
1, 1127.
(39) San Filippo, J.; Romano, L. J.; Chern, C.-I.; Valentine, J. S.
J. Org. Chem. 1976, 41, 586.
(40) Bryantsev, V. S.; Blanco, M. J. Phys. Chem. Lett. 2011, 379.
(41) McDonald, R. N.; Chowdhury, A. K. J. Am. Chem. Soc. 1985,
107, 4123.
(42) Behrendt, W.; Gattow, G.; Draeger, M. Z. Anorg. Allg. Chem.
1973, 397, 237.
(43) Matsumoto, S.; Sugimoto, H.; Sawyer, D. T. Chem. Res. Toxicol.
1988, 1, 19.
(44) Curran, H. J.; Gaffuri, P.; Pitz, W. J.; Westbrook, C. K. Combust.
Flame 1998, 114, 149.
(45) Atkinson, R. Int. J. Chem. Kinet. 1997, 29, 99.
(46) Sinha, A.; Thomson, M. J. Combust. Flame 2004, 136, 548.
(47) Roberts, J. L.; Calderwood, T. S.; Sawyer, D. T. J. Am. Chem.
Soc. 1984, 106, 4667.
(
(
2
1
2
(
(
10) Beattie, S. D.; Manolescu, D. M.; Blair, S. L. J. Electrochem. Soc.
009, 156, A44.
(
11) Laoire, C. O.; Mukerjee, S.; Abraham, K. M.; Plichta, E. J.;
Hendrickson, M. A. J. Phys. Chem. C 2009, 113, 20127.
12) Laoire, C. O.; Mukerjee, S.; Abraham, K. M.; Plichta, E. J.;
Hendrickson, M. A. J. Phys. Chem. C 2010, 114, 9178.
(
(13) Wang, Y.; Zhou, H. J. Power Sources 2010, 195, 358.
(
14) Kumar, B.; Kumar, J.; Leese, R.; Fellner, J. P.; Rodrigues, S. J.;
(48) Wadhawan, J. D.; Welford, P. J.; Maisonhaute, E.; Climent, V.;
Lawrence, N. S.; Compton, R. G.; McPeak, H. B.; Hahn, C. E. W. J. Phys.
Chem. B 2001, 105, 10659.
Abraham, K. M. J. Electrochem. Soc. 2010, 157, A50.
15) Hummelshoj, J. S.; Blomqvist, J.; Datta, S.; Vegge, T.; Rossmeisl,
(
J.; Thygesen, K. S.; Luntz, A. C.; Jacobsen, K. W.; Norskov, J. K. J. Chem.
Phys. 2010, 132, 071101.
(49) Verma, P.; Maire, P.; Nov ꢀa k, P. Electrochim. Acta 2010,
55, 6332.
(
16) Bryantsev, V. S.; Blanco, M.; Faglioni, F. J. Phys. Chem. A 2010,
14, 8165.
17) Lu, Y.-C.; Xu, Z.; Gasteiger, H. A.; Chen, S.; Hamad-Schifferli,
K.; Shao-Horn, Y. J. Am. Chem. Soc. 2010, 132, 12170.
18) Xu, W.; Xiao, J.; Wang, D.; Zhang, J.; Zhang, J.-G. Electrochem.
Solid-State Lett. 2010, 13, A48.
19) Mohamed, S. N.; Johari, N. A.; Ali, A. M. M.; Harun, M. K.;
Yahya, M. Z. A. J. Power Sources 2008, 183, 351.
20) Lu, Y.-C.; Gasteiger, H. A.; Crumlin, E.; Robert McGuire, J.;
Shao-Horn, Y. J. Electrochem. Soc. 2010, 157, A1016.
21) Lu, Y.-C.; Gasteiger, H. A.; Parent, M. C.; Chiloyan, V.;
Shao-Horn, Y. Electrochem. Solid-State Lett. 2010, 13, A69.
22) Peled, E.; Golodnitsky, D.; Mazor, H.; Goor, M.; Avshalomov,
S. J. Power Sources 10.1016/j.jpowsour.2010.09.104.
23) Aleshin, G. Y.; Semenenko, D. A.; Belova, A. I.; Zakharchenko,
T. K.; Itkis, D. M.; Goodilin, E. A.; Tretyakov, Y. D. Solid State Ionics
(50) Zhang, J.; Oloman, C. W. J. Appl. Electrochem. 2005, 35, 945.
(51) Chin, D. H.; Chiericato, G.; Nanni, E. J.; Sawyer, D. T. J. Am.
Chem. Soc. 1982, 104, 1296.
(52) Peng, Z.; Freunberger, S. A.; Hardwick, L. J.; Chen, Y.;
Giordani, V.; Bard ꢀe , F.; Nov ꢀa k, P.; Graham, D.; Tarascon, J.-M.; Bruce,
P. G. Angew. Chem., Int. Ed. 2011; accepted.
1
(
(
(
(
(
(
(
2
011, 184, 62.
(24) Cheng, H.; Scott, K. J. Power Sources 2010, 195, 1370.
(25) Zhang, S. S.; Foster, D.; Read, J. J. Power Sources 2010,
195, 1235.
(26) Zhang, D.; Fu, Z.; Wei, Z.; Huang, T.; Yu, A. J. Electrochem. Soc.
2
010, 157, A362.
(
27) Andrei, P.; Zheng, J. P.; Hendrickson, M.; Plichta, E. J.
J. Electrochem. Soc. 2010, 157, A1287.
28) Mizuno, F.; Nakanishi, S.; Kotani, Y.; Yokoishi, S.; Iba, H.
Electrochemistry 2010, 78, 403.
29) Freunberger, S. A.; Hardwick, L. J.; Peng, Z.; Giordani, V.;
Chen, Y.; Maire, P.; Nov ꢀa k, P.; Tarascon, J.-M.; Bruce, P. G. In IMLB
010 ; 15th International Meeting on Lithium Batteries, Montreal,
Canada,June 27 to July 2, 2010.
30) Freunberger, S. A.; Peng, Z.; Hardwick, L. J.; Chen, Y.; Barde,
F.; Bruce, P. ECS Meet. Abstr. 2010, 1002, 340.
31) Xu, W.; Viswanathan, V. V.; Wang, D.; Towne, S. A.; Xiao, J.;
Nie, Z.; Hu, D.; Zhang, J.-G. J. Power Sources 2011, 196, 3894.
32) Dobley, A.; Morein, C.; Abraham, K. M. ECS Meet. Abstr. 2006,
02, 823.
33) Ufheil, J.; Wursig, A.; Schneider, O. D.; Novak, P. Electrochem.
Commun. 2005, 7, 1380.
34) Aurbach, D.; Daroux, M.; Faguy, P.; Yeager, E. J. Electroanal.
Chem. 1991, 297, 225.
35) Gibian, M. J.; Sawyer, D. T.; Ungermann, T.; Tangpoonpholvivat,
R.; Morrison, M. M. J. Am. Chem. Soc. 1979, 101, 640.
36) Aurbach, D.; Daroux, M. L.; Faguy, P. W.; Yeager, E.
J. Electrochem. Soc. 1987, 134, 1611.
37) Fong, R.; Reid, M. C.; McMillan, R. S.; Dahn, J. R. J. Electro-
chem. Soc. 1987, 134, 516.
38) Aurbach, D.; Gofer, Y.; Langzam, J. J. Electrochem. Soc. 1989,
36, 3198.
(
(
2
(
(
(
5
(
(
(
(
(
(
1
8
047
dx.doi.org/10.1021/ja2021747 |J. Am. Chem. Soc. 2011, 133, 8040–8047