Journal of The Electrochemical Society, 164 (14) A3727-A3741 (2017) A3727

Growth of Ambient Induced Surface Impurity Species on Layered

Positive Electrode Materials and Impact on Electrochemical
Performance
Nicholas V. Faenza, a,∗,z Lejandro Bruce,a Zachary W. Lebens-Higgins,b Irene Plitz,a
Nathalie Pereira,a Louis F. J. Piper,b and Glenn G. Amatuccia
a EnergyStorage Research Group, Department of Materials Science and Engineering, Rutgers University, North
Brunswick, New Jersey 08902, USA
b Department of Physics, Applied Physics and Astronomy, Binghamton University, Binghamton, New York 13902, USA

Surface impurity species, most notably Li2 CO3 , that develop on layered oxide positive electrode materials with atmospheric aging
have been reported to be highly detrimental to the subsequent electrochemical performance. LiNi0.8 Co0.15 Al0.05 O2 (NCA) was
used as a model layered oxide compound to evaluate the growth and subsequent electrochemical impact of H2 O, LiHCO3 , LiOH
and Li2 CO3 . Methodical high temperature annealing enabled the systematic removal of each impurity specie, thus permitting the
determination of each specie’s individual effect on the host material’s electrochemical performance. Extensive cycling of exposed
and annealed materials emphasized the cycle life degradation and capacity loss induced by each impurity, while rate capability
measurements correlated the electrode impedance to the impurity species present. Based on these characterization results, this work
attempts to clarify decades of ambiguity over the growth mechanisms and the electrochemical impact of the specific surface impurity
species formed during powder storage in various environments.
© The Author(s) 2017. Published by ECS. This is an open access article distributed under the terms of the Creative Commons
Attribution Non-Commercial No Derivatives 4.0 License (CC BY-NC-ND, http://creativecommons.org/licenses/by-nc-nd/4.0/),
which permits non-commercial reuse, distribution, and reproduction in any medium, provided the original work is not changed in any
way and is properly cited. For permission for commercial reuse, please email: [email protected]. [DOI: 10.1149/2.0921714jes]
All rights reserved.

Manuscript submitted October 12, 2017; revised manuscript received November 22, 2017. Published December 7, 2017.

The initial surface chemistry of Li-ion electrode materials plays the surface impurity species.1,13 Adsorbed H2 O and CO2 react with
a critical role in the development of in-situ derived interphases with the highly active oxygen species to produce hydroxyl, bicarbonate,
the electrolyte and on the subsequent electrochemical performance. and carbonate species. Then lithium from the host structure and the
The surface chemistry can have a beneficial or destructive impact newly formed surface products, react to create LiOH, LiHCO3 and
on performance relative to the bulk material. The principal challenge Li2 CO3 .1 The instability of Ni3+ and the propensity for Ni3+ to be
is controlling the variable surface chemistries which are induced by reduced to Ni2+ encourages the development of impurity species on
impurities during processing or material instability in an ambient at- the surfaces of Ni-rich layered oxides, by easily forming active oxy-
mosphere. Impurity species on the surface of the electrochemically gen sites.1,12,18 In isostructural materials with higher amounts of Co,
active materials have been shown to decrease both the electronic and less LiOH and Li2 CO3 are observed because the increased stabil-
ionic conductivities as well as reduce the amount of electrochemi- ity of Co3+ minimizes the creation of active oxygen species on the
cally viable lithium.1,2 These consequences have a profound negative surface.18 Similarly, when layered oxide materials contain some Mn
impact on the materials’ cycle life and energy density, resulting in atoms, lower concentrations of surface impurities are measured be-
significantly reduced cell performance.3–5 The higher impedance has cause of the preference for Mn3+ to be oxidized to Mn4+ , which
been proposed to create localized heating, which is known to increase simultaneously reduces another transition metal to the +2 valence
the risk of catastrophically destroying the cell or igniting a liquid state and prevents additional transition metal reduction that would
electrolyte.6–9 As a result, understanding the impact and controlling create the active oxygen sites.14,23
the surface impurities on positive electrode materials is of utmost im- A systematic inverse relationship between a material’s redox po-
portance to improve the functionality and safety of secondary lithium- tential and air stability is known.24 For example, the trend of increasing
ion batteries. redox potential during the delithiation process of LiNi0.8 Co0.15 Al0.05 O2
The primary surface impurity species that develop when layered (NCA) < LiCoO2 < LiMn2 O4 leads to the instability of nickelates,
positive electrode materials are exposed to the ambient atmosphere such as NCA, in air relative to the increased stability of LiCoO2 and
include adsorbed H2 O, LiHCO3 , LiOH, and Li2 CO3 .4,10–14 Since the the excellent stability of LiMn2 O4 . While some Li2 CO3 has been
impurity species are limited to the surface of the host particles, sur- observed to form on LiCoO2 with long exposures, it is mostly a by-
face sensitive characterization techniques such as Fourier transform product of excess Li utilized in the initial fabrication of the electrode.37
infrared (FTIR) spectroscopy,2–4,10,15,16 and X-ray photoelectron spec- Similarly, little Li2 CO3 is observed on air exposed LiMn2 O4 because
troscopy (XPS),1,3,12,13,16,17 are frequently used to study these materi- of the material’s high redox potential.24
als. These techniques in conjunction with thermogravimetric analysis The detrimental impact that the chemical and structural changes
(TGA),1,11 X-ray diffraction (XRD)1,4,11–14,18 and Karl-Fisher titra- on the layered oxide’s surface have on the material’s electrochemical
tion (KF)10 have enabled the identification and quantification of these performance is well documented. Lower discharge capacities and re-
impurity species. duced rate capability for NCA electrodes were correlated to carbonate
Transition metal reduction in the surface and subsurface regions species for cells with both solid and liquid electrolytes by Visbal et al.3
of these positive electrode materials, which cause the structural tran- and Zhuang et al.,4 respectively. The increased interfacial resistance
sition from the layered phase to spinel or rocksalt-type phases, has caused by the development of the surface impurity layer impedes the
been observed upon exposure in air.6,19–22 It has been proposed that the lithium-ion transport across the electrode-electrolyte barrier resulting
transition metal reduction from a valence state of +3 to +2 produces in increased deviation from the thermodynamic potential when cy-
active oxygen species which concurrently catalyzes the formation of cling, especially at higher rates.1,4,13,25 Additionally, the newly formed
divalent transition metals migrate from the 3b to the 3a sites in the R-
3m structure. This causes the transformations from layered to spinel
∗ Electrochemical Society Member. and rocksalt-type phases on the material’s surface, which obstructs
z
E-mail: [email protected] both the electronic conductivity and lithium-ion diffusivity.4,19,21,22

Downloaded on 2017-12-15 to IP 134.148.10.12 address. Redistribution subject to ECS terms of use (see ecsdl.org/site/terms_use) unless CC License in place (see abstract).
 A3728 Journal of The Electrochemical Society, 164 (14) A3727-A3741 (2017)

These structural changes further increase the positive electrode’s elec- impedance and improved cycling stability, once surface Li2 CO3 is
tronic and charge transfer impedances.20 The presence of Li2 CO3 on removed. The ultimate goal of the present investigation is to identify
the surface of LiCoO2 was shown to be highly detrimental to the elec- if Li2 CO3 is the main surface species leading to poor performance or
trochemical performance and that such performance was renewed with whether it is one of the other species that are present and integral to
the high temperature removal of the Li2 CO3 .26,27 The development of the formation of Li2 CO3 .
impurity species also influences the reversible capacity by trapping In this investigation NCA electrodes were exposed to ambient
lithium from the host structure in the newly formed surface layer. atmosphere, the resulting surface species were identified, and these
These trapped lithium ions can be at least partially restored by charg- species were systematically removed one by one by way of careful
ing the electrode to sufficient potentials to extract the lithium from the thermal annealing. The annealed materials were then electrochemi-
impurity species.2,15 However, charging to the high potentials neces- cally characterized and benchmarked with unexposed NCA to iden-
sary to retrieve the lithium atoms from the impurity species may gen- tify each species’ role in the electrochemical degradation of NCA.
erate adverse consequences such as electrolyte oxidation, or reactions By exploring how the impurity species develop on positive electrode
between the deintercalated impurity species and the electrolyte.28–31 materials and by correlating individual species to changes in the elec-
Furthermore, it has been proposed that growth of Li2 CO3 can in- trochemical performance, this work attempts to clarify decades of
duce some positive electrode particle isolation by establishing either ambiguity over the growth mechanisms and impact of different sur-
an electronic or ionic barrier around particles or by preventing elec- face impurity species. The results presented here outline the optimal
trolyte from accessing a sufficient number of pores.4 While positive atmospheric conditions to store and process layered positive electrode
electrode particle isolation hasn’t yet been conclusively attributed to materials for lithium-ion batteries.
other impurity species, the low electronic and lithium conductivities
of LiOH or Li2 O are likely sufficient to duplicate the isolation mech-
Experimental
anisms proposed for Li2 CO3 .
The presence of impurity species at the electrode-electrolyte in- Materials preparation.—Samples were fabricated by exposing
terface can also initiate adverse reactions between the electrolyte and LiNi0.8 Co0.15 Al0.05 O2 (NAT1050, TODA America) for two weeks
either the impurity species or the reduced transition metals on the in either an Ar-filled glove box, a dryroom (dew point of −35 to
surface of the positive electrode.16,32–34 Aurbach et al. observed that −40◦ C), a sealed bag of dry CO2 (<10 ppm H2 O, Airgas), or by
the initial layer of Li2 CO3 on their pristine LiNiO2 samples is lost dur- exposing to ambient air in a hood with constant air flow. All an-
ing storage in electrolyte solutions containing Li salts, and replaced neals were performed using a box furnace (48000 Furnace, Barn-
by an assortment of surface species including organic and inorganic stead Thermolyne) in the dryroom. Electrode tapes were made us-
carbonates, alkoxides, and salt decomposition products.16 Their FTIR ing the Bellcore method and were cast in the dryroom to mini-
and XPS measurements show that when submersed in an LiPF6 so- mize exposure to moisture.37 Casting slurries were a mixture of
lution the Li2 CO3 layer is replaced by LiF, which is a product of the poly(vinylidene fluoride-co-hexafluoropropylene) (PVDF-HFP, Ky-
HF present in the electrolyte, and that this new surface layer signifi- nar 2801, Elf Atochem), carbon black (Super P (SP), MMM), propy-
cantly raises the electrode impedance.16 Similar results were obtained lene carbonate (Aldrich), and acetone (Aldrich). Homogenous slurries
for LiMn2 O4 16 , and NCA15,35 positive electrodes. Our previous work were cast, allowed to air dry, and then the propylene carbonate plas-
demonstrates that layered oxide materials with higher contents of ticizer was extracted by soaking the electrode in 99.8% anhydrous
surface impurities have an amplified propensity for transition metal diethyl ether (Aldrich). All electrodes had a composition of 80.0 wt%
dissolution, likely caused by the increase of reduced transition met- active material, 7.0 wt% SP, and 13.1 wt% PVDF-HFP. Free stand-
als that are present at the material’s surface.33 Impurity species other ing electrodes were subsequently dried at 120◦ C under vacuum for a
than Li2 CO3 can also have adverse effects on a cell’s performance. minimum of 10 hours, and stored in an Ar atmosphere glove box to
For example, it is well known that adsorbed and absorbed H2 O will avoid atmospheric exposure.
react with fluorinated electrolyte salts to produce HF, which then de-
grades the positive electrode.32,34 Furthermore, the reactions between Electrochemical characterization.—Coin cells (2032, Hohsen)
HF and the positive electrode are autocatalytic, leading to significant were assembled in an Ar-filled glove box using Whatman GF/D
destruction of the electrode’s surface.32,36 glass fiber separators saturated with a 1M LiPF6 ethylene carbonate:
While the formation of Li2 CO3 and other impurity species on dimethyl carbonate (EC:DMC) (1:1 volume ratio) electrolyte (BASF)
the surface of layered oxide materials is frequently documented, the (<20 ppm of H2 O), and a Li metal (FMC) negative electrode. Poten-
exact growth mechanisms are not well understood. As previously tiostatic and galvanostatic electrochemical testing was controlled by
mentioned, many publications support the conclusion that adsorbed a VMP3 (Biologic) or Maccor battery cycling system at 24◦ C. Cells
H2 O and CO2 react with the positive electrode because reduction of were galvanostatically charged at 20 mA/g (per g of active material)
the transition metal atoms creates active oxygen sites. However, there to Vmax (4.2 or 4.5 V), held under potentiostatic conditions until the
is uncertainty in the literature over whether Li2 CO3 can form by a current dropped below 10 mA/g, and then discharged at 15 mA/g to
direct reaction between CO2 and the layered oxide material or if inter- 2.75 V. Starting on the 6th cycle, and every 19 cycles afterwards, the
mediary phases, such as LiHCO3 or LiOH, are necessary. Matsumoto cells were discharged at 150 mA/g to 2.75 V.
et al. suggest that NCA readily reacts with CO2 to form Li2 CO3 ,
but acknowledge that moisture promotes the carbonate formation at Physical characterization.—Thermogravimetric analysis (TGA)
low temperatures.11 Both Liu et al. and Zhang et al. proposed reac- experiments were performed in a TA instruments TGA 2050, with Pt
tion mechanisms that require intermediary impurity phases before the pans and a flowing dry air atmosphere. Tests consisted of constant
development of Li2 CO3 .1,14 The topic is further complicated by the heating rate and isothermal segments. All heating was at 5◦ C/min and
temperature dependence of these reactions. Li2 CO3 has been shown isothermal segments were at 150◦ C, 350◦ C, 500◦ C, and 750◦ C. The
to form on NCA when exposed to dried 100% CO2 at 400◦ C, but the first three isotherms each had durations of two hours and the 750◦ C
authors did not demonstrate that this reaction could take place at lower isothermal segment was held for four hours. Additional samples were
temperatures.12 also cooled at 5◦ C/min to 30◦ C. As described later in the text, some
In order to elucidate the mechanisms for the formation of impurity experiments use slightly adjusted profiles, and a dry air with <0.1
species on the surface of layered oxide materials and to clarify the ppm of CO2 . KF titration (831 Coulometer and 860 Thermoprep,
understanding of the atmospheric environments necessary for growth, Metrohm) was used to measure the moisture content in all samples and
a comprehensive analysis of the changes to the surface of NCA as a in relevant standard materials. Samples were sealed in KF coulometry
function of atmosphere has been investigated. As noted above, a num- vials while inside the Ar-filled glove box to eliminate any exposure to
ber of studies, including those from our own group, have documented the ambient or dryroom atmospheres. KF tests were run at 250◦ C with
evidence of distinct performance improvements, mostly in reduced flowing N2 gas to ensure that all moisture was desorbed. Successive

Downloaded on 2017-12-15 to IP 134.148.10.12 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, 164 (14) A3727-A3741 (2017) A3729

KF experiments were performed on each sample until the measured Figure S3 (g-h) shows a decline in the magnitude of the 1576–1714
moisture content was below 10 ppm. cm−1 peak after annealing at 150◦ C, supporting that the peak is from
Attenuated Total Reflectance (ATR) Fourier Transform InfraRed water. The surfaces of the baseline materials may have also contained
(FTIR) experiments were run in the MidIR range using a Smart Or- LiHCO3 , but because of the similarities to the Li2 CO3 spectra, the
bit accessory with a TypeIIa diamond crystal (Nicolet 6700 FTIR presence of LiHCO3 could not be conclusively determined.
Spectrometer, Thermo). All experiments used a background scan per- In an Ar-filled glove box, each of the four baseline materials were
formed the same day as the sample test. Tests consisted of 250 scans mixed with approximately 26 wt% PE and ground together by a mortar
with a resolution of 4 cm−1 , unless otherwise noted. Exposed materials and pestle. Extra precautions were taken to ensure that no additional
were mixed with 26 ± 1 wt% polyethylene (PE) powder (Sigma) and moisture was added to the sample during creation of the PE mixtures.
ground using a mortar and pestle. The mixtures were then sealed on the PE was used as an internal standard to normalize the peak heights be-
ATR stage with an adhesive backed lead tape while inside the Ar-filled tween samples and enable relative quantitative comparisons between
glove box. However, the sampling area was free of lead tape adhesive materials. The FTIR spectra of these mixtures are shown in Figure
in order to prevent any contamination. To support surface impurity 1a–1d. Peaks corresponding to PE, Li2 CO3 , and LiOH · xH2 O are de-
identification, Li2 CO3 (Johnson Mattey), LiOH (Sigma), LiOH · H2 O noted with black, red, and blue symbols, respectively. To enable direct
(Sigma) and Li2 O (Alfa Aesar) standards were also examined. After visual comparison, the y-axes of Figure 1a–1d are normalized to the
TGA analysis, exposed materials were held in a closed chamber with area of each scan’s 2912 cm−1 PE peak (not shown). The bending
flowing dry air to prevent the adsorption of moisture. Samples were mode LiHCO3 /Li2 CO3 (870 cm−1 ) peak was used for quantitative
then quickly transferred to the ATR stage and sealed with lead tape. analysis because it is clearly distinct from any PE peaks. It is evident
Ambient exposure was less than 30 seconds during transfer. from Figure 1a–1d that little LiHCO3 /Li2 CO3 is present on either
A Bruker D8 Advance diffratometer (Cu Kα, λ = 1.5406 Å) was the Pristine or Dryroom samples, only slightly more on CO2 NCA,
used for all X-ray diffraction (XRD) characterization. XRD samples but that the Ambient material has an appreciable amount. Relative to
of exposed materials were assembled in an Ar-filled glove box, by Pristine NCA, the carbonate peak areas were around 23% and 617%
sealing the powder onto a glass slide with Kapton film and grease to larger for CO2 and Ambient materials, respectively. While LiOH and
prevent ambient exposure during testing. XRD scans were conducted H2 O peaks were observed by FTIR in the baseline materials, they
over a two-theta range of 15–70o at a scan rate of 0.6o /min. TOPAS were obscured by the PE and too small for a conclusive quantitative
software (version 5, Bruker AXS) was used for all Rietveld refinement. analysis.
XPS measurements were performed using a Phi VersaProbe 5000 Figure 1e presents the results of TGA experiments for each baseline
system on powders mixed with carbon black. A flood gun was used material. During these experiments, samples were heated at 5◦ C/min
to charge compensate insulating samples, and the carbon black C 1s to 750◦ C with 2 h isothermal segments at 150◦ C, 350◦ C and 500◦ C,
peak was aligned to 284.5 eV to account for possible charging effects. and a 4 h isothermal hold at 750◦ C. Throughout the tests, dry air was
Exposed materials were deposited on conductive tape and transferred flowed through the sample chamber to carry away any evolved gases.
into the XPS chamber in under 30 seconds. Similar surface species Dry air was utilized not only because the lack of moisture prevents the
were observed when select samples were transferred using a vacuum adsorption of additional H2 O or the formation of LiHCO3 or LiOH
suitcase set-up between the glove box and vacuum chamber. impurities, but also because the atmosphere doesn’t induce transition
metal reduction and oxygen gas evolution at higher temperatures. The
temperatures of the isothermal segments were selected to isolate the
mass loss from each impurity specie and the duration of the isothermal
Results
holds provided significant time to remove the majority of the particular
Effect of atmosphere exposure.—NCA powder was exposed to specie. The 150◦ C, 350◦ C, 500◦ C, and 750◦ C isothermal segments
four different atmospheric environments; an Ar-filled glove box (Pris- were designed to remove the adsorbed and absorbed H2 O, LiHCO3 ,
tine), ambient air (Ambient), a dryroom (Dryroom), and dry CO2 gas LiOH, and Li2 CO3 species, respectively.
(CO2 ). Each sample was exposed continuously for two weeks, and Adsorbed water should be highly unstable at 150◦ C and is expected
intermittent recordings of the temperature and humidity are presented to evaporate from the sample relatively quickly. Using temperature
in Figure S1 (a) and (b), respectively. While all of the environments programmed desorption-mass spectroscopy to measure the evolved
were at room temperature, the relatively humidity (RH) varied sub- H2 O, O2 and CO2 while heating LiNi0.8 Co0.2 O2 , Liu et al. showed
stantially. The Pristine and CO2 environments should only have trace that lithium bicarbonate and adsorbed carbonate species on positive
amounts of moisture since they were in sealed enclosures filled with electrode materials will decompose at 270◦ C and 350◦ C.1 Other stud-
very dry gasses. Although the dryroom was measured to have a dew ies have demonstrated that LiOH is thermally degraded at 410◦ C,
point of −35 to −40◦ C, the moisture content is slightly higher than the and Li2 CO3 is evolved at 710◦ C.37,38 The thermal decompositions of
Pristine dew point of glove box is < −80◦ C) or CO2 environments. In LiHCO3 , LiOH and Li2 CO3 are expected to follow Equations 1–3,
contrast, the average dew point of the ambient laboratory was +8.0◦ C respectively.
(45.8% RH).
XRD scans of the four baseline samples and of the initial commer- 2LiHCO3 → Li2 O + H2 O + 2CO2 [1]
cial NCA material are shown in Figure S2 (a-e). There was no evidence
for the development of impurity phases or of structural degradation to
the initial material within the detection limitations for powder XRD. 2LiOH → Li2 O + H2 O [2]
Rietveld refinement of the diffraction profiles (Table SI) revealed lit-
tle evidence of significant bulk lattice parameter or transition metal
occupancy changes as a function of exposure. Li2 CO3 → Li2 O + CO2 [3]
FTIR spectra on the baseline materials in Figure S3 (a-d) re-
vealed the presence of LiOH · xH2 O (x denotes an unknown con- To maximize the mass loss while still isolating a particular impurity
stant), H2 O and Li2 CO3 . Ambient, Dryroom, and CO2 NCA samples specie, the isothermal temperatures were chosen to be slightly higher
showed broad peaks centered at ∼3700–3745 cm−1 which are from than the previously measured degradation temperatures. The duration
the hydroxyl stretching mode of a LiOH · xH2 O-type material.38–41 of each isotherm was determined by multiple tests and represents the
The CO3 group bending, asymmetric and symmetric C-O stretching amount of time necessary for the mass loss to plateau at the isotherm
modes associated with Li2 CO3 were clearly present at 870 cm−1 , 1419 temperature. To ensure that the mass loss at high temperatures is
cm−1 , and 1471 cm−1 , respectively.4,15,42 Ambient and CO2 NCA also due to the loss of Li2 CO3 and not from oxygen loss associated with
showed a broad peak between 1576 cm−1 and 1714 cm−1 , which co- Ni reduction, additional TGA tests with a cooling segment were per-
incides with the expected location for the H2 O bending mode.38–40 formed on Pristine and Ambient NCA, and are presented in Figure S4.

Downloaded on 2017-12-15 to IP 134.148.10.12 address. Redistribution subject to ECS terms of use (see ecsdl.org/site/terms_use) unless CC License in place (see abstract).
 A3730 Journal of The Electrochemical Society, 164 (14) A3727-A3741 (2017)

Figure 1. FTIR measurements of the (a) Pristine, (b) Ambient, (c) Dryroom and (d) CO2 baseline materials mixed with 26 wt% PE. The spectra focus on the
LiHCO3 /Li2 CO3 (red star), PE (black dot), and hydroxyl (blue caret) peaks in the ranges of 750 to 1650 cm−1 , and 3550–3850 cm−1 . Both absorbance ranges for
the FTIR spectra are normalized to the size of each spectrum’s 2912 cm−1 PE peak. Mass loss curves from (e) TGA and the moisture content from cumulative
(f) KF experiments on Pristine (black), Ambient (red), Dryroom (blue) and CO2 (green) NCA baseline materials are also presented. TGA experiments had 2 h
isothermal holds at 150◦ C, 350◦ C, and 500◦ C to systematically remove H2 O, LiHCO3 , and LiOH, respectively. The fourth and final isotherm, at 750◦ C, lasted 4
h and measured the mass loss from Li2 CO3 decomposition.

During cooling there is negligible mass gain indicating that the high the material during storage. Quantitative results from FTIR, TGA,
temperature mass loss is not from a reversible Ni redox transition. KF and electrochemistry experiments on all baseline and annealed
The loss curves in Figure 1e show that Pristine and Dryroom NCA materials are presented in Table I.
are virtually indistinguishable by TGA. This indicates that the two Electrodes fabricated from each of the baseline positive electrode
materials have very similar amounts of each impurity, and that expo- materials were dried overnight at 120◦ C under vacuum prior to storage
sure to the dryroom atmosphere did not result in significant formation in an Ar-filled glove box. Coin cells containing these positive elec-
of surface species. On the contrary, Ambient NCA lost over 7 times trodes were charged at 20 mA/g to Vmax (4.2 V or 4.5 V), held under
as much mass as the Pristine sample. While the Ambient material lost potentiostatic conditions until the current dropped below 10 mA/g and
more mass than Pristine at every isotherm, the largest differences in then discharged to 2.75 V at −15 mA/g, all at 24◦ C. To investigate
mass loss corresponded to H2 O, LiHCO3 and Li2 CO3 . CO2 NCA also the capacity retention as a function of rate for each material, cells
lost more mass than the Pristine or Dryroom samples, but the majority were discharged at −150 mA/g starting on the 6th cycle, and every
of the additional mass lost was evolved during the 150◦ C isotherm, 19 cycles afterwards. The discharge capacity results for each mate-
indicating that it was H2 O. rial are presented in Figure 2. Duplicate cells were made to ensure
KF coulometry results presented in Figure 1f perfectly supports reproducibility.
the water desorption measurements from TGA. Pristine (253.3 ppm) Pristine NCA demonstrated good initial specific capacities to both
and Dryroom (461.3 ppm) samples have small moisture contents, 4.2 (Figure 2a) and 4.5 V (Figure 2b), and maintained robust capac-
while CO2 NCA (1125.7 ppm) has marginally more, and the Ambient ity retention over many cycles. As expected, there is small capacity
material (7125.4 ppm) has over 28 times as much moisture as Pristine. fade when charging to 4.2 V, and a slightly faster degradation rate
The incrementally higher moisture content on CO2 NCA relative to when charging to 4.5 V. The cycling stability of Dryroom NCA was
the Pristine material is believed to be from contamination by the nearly identical to that of the Pristine material, which is expected be-
plastic container and/or the interior surface of the bag used to enclose cause there was negligible differences in the surface impurity species

Downloaded on 2017-12-15 to IP 134.148.10.12 address. Redistribution subject to ECS terms of use (see ecsdl.org/site/terms_use) unless CC License in place (see abstract).
Downloaded on 2017-12-15 to IP 134.148.10.12 address. Redistribution subject to ECS terms of use (see ecsdl.org/site/terms_use) unless CC License in place (see abstract).

Table I. Quantitative results from KF, TGA, FTIR and electrochemistry experiments on all baseline and annealed materials. KF results are the cumulative ppm of H2 O, while the FTIR results are
listed as an approximation of the peak area relative to the area of that scan’s main PE peak. Some materials had insufficiently sized LiOH · xH2 O peaks to determine the relative peak area, and are
labeled as “N/A”. The % capacity loss results were calculated by using the maximum discharge capacity and the discharge capacity of the 50th cycle. The % capacity loss for Ambient NCA was
“N/A” when cycled to 4.2 V because the discharge capacity was negligible for all cycles, and the % capacity loss was 0.00 when charged to 4.5 V since the 50th cycle also had the largest discharge
capacity of the first 50 cycles. The discharge rate capacity retention data was obtained by dividing the discharge capacity of the 25th cycle (150 mA/g) by the discharge capacity of the 24th cycle (15

Journal of The Electrochemical Society, 164 (14) A3727-A3741 (2017)

mA/g). Since Ambient NCA had no discharge capacity when cycled to 4.2 V, the capacity retention is listed as “N/A”. The average OCV values for all electrodes are listed in the rightmost column.

Technique: KF TGA TGA TGA FTIR TGA FTIR Electrochemistry

Material H2 O H2 O Loss LiHCO3 LiOH LiOH · Li2 CO3 LiHCO3 / % % 4.2V 4.5V Average
(ppm) (wt%) Loss Loss xH2 O Loss Li2 CO3 Capacity Capacity Cells, Cells, OCV vs
(wt%) (wt%) (Relative (wt%) (Relative loss after loss after Discharge Discharge Li/Li+ (V)
to PE) to PE) 50 cycles 50 cycles Rate Rate
to 4.2V to 4.5V Capacity Capacity
(%) (%) Retention Retention
(%) (%)
Pristine NCA 253.3 0.06 0.02 0.00 0.098 0.31 0.044 3.26 8.31 89.42 89.98 3.391
150C - Pristine 131.2 0.01 0.00 0.00 0.181 0.30 0.200 4.12 8.37 89.15 89.43 3.407
350C - Pristine 97.4 0.02 0.05 −0.01 0.221 0.58 0.207 5.61 9.75 87.81 88.17 3.410
500C - Pristine 86.4 −0.01 0.04 0.07 N/A 0.79 0.154 3.87 9.50 86.91 88.40 3.398
750C - Pristine 178.4 0.02 0.04 −0.01 N/A 0.77 0.065 2.56 10.91 87.90 88.83 3.380
Ambient NCA 7125.4 0.47 1.41 0.24 0.032 0.72 0.271 N/A 0.00 N/A N/A 3.441
150C - Ambient 407.3 0.16 0.78 0.22 0.088 0.89 0.272 25.69 28.29 79.73 77.78 3.489
350C - Ambient 103.6 0.03 0.07 0.13 0.014 0.98 0.192 9.26 15.46 87.47 87.17 3.375
500C - Ambient 88.7 0.00 0.05 0.07 N/A 0.97 0.252 3.33 9.61 86.87 87.91 3.368
750C - Ambient 175.4 0.01 0.05 0.06 N/A 0.44 0.095 10.13 10.16 89.06 89.06 3.373
Dryroom NCA 461.3 0.05 0.02 0.01 0.089 0.32 0.039 2.91 8.25 89.42 89.73 3.405
150C - Dryroom 96.4 0.02 0.00 −0.01 0.044 0.31 0.068
350C - Dryroom 66.2 0.01 0.04 0.04 N/A 0.72 0.140
500C - Dryroom 107.7 0.03 0.05 0.06 N/A 0.86 0.135
750C - Dryroom 117.5 0.00 0.04 −0.01 N/A 0.69 0.079
CO2 NCA 1125.7 0.17 0.06 0.00 N/A 0.37 0.054 0.39 3.42 88.21 89.86 3.434
150C - CO2 509.9 0.09 0.07 0.00 N/A 0.44 0.114
350C - CO2 45.0 −0.01 0.04 0.00 N/A 0.53 0.099
500C - CO2 103.4 0.01 0.05 0.04 N/A 0.76 0.109
750C - CO2 112.9 0.01 0.03 −0.01 N/A 0.62 0.092

A3731
 A3732 Journal of The Electrochemical Society, 164 (14) A3727-A3741 (2017)

Figure 2. Comparison of the discharge capacities as a function of cycle number for all four baseline materials. Electrodes were charged to an upper cutoff potential
of either (a) 4.2 V or (b) 4.5 V. Cells were charged at 20 mA/g to Vmax (4.2 V or 4.5 V), held under potentiostatic conditions until the current dropped below 10
mA/g and then discharged to 2.75 V at −15 mA/g. Starting on the 6th cycle, and every 19 cycles afterwards, the cells were discharged at −150 mA/g instead of
−15 mA/g. All of the cycling was done in a temperature controlled chamber set at 24◦ C.

between the samples. When charged with a cutoff potential of 4.2 V, at higher temperatures for all samples. Starting with the baseline ma-
Ambient NCA had negligible discharge capacity and was functionally terials, which are represented as the samples annealed at 24◦ C, to the
useless. When the upper potential was increased to 4.5 V, the positive 750◦ C annealed samples the a and c lattice parameters each increased
electrode became capable of reproducibly delivering some capacity, by ∼0.05%, causing the cell volume to grow by ∼0.15%. Even though
but the discharge capacity was still only about a fourth of the Pris- the change in the lattice parameters is near the margin of error, the
tine material. The initial discharge capacities of the CO2 NCA cells clear trends support the results’ validity. Coincident with the devel-
were significantly lower than the Pristine samples, and even though opment of the lattice parameters, Rietveld analysis calculated a rough
the capacity of the 4.2 V cells rose steadily over the first 45 cycles, trend of increasing amounts of Ni in the Li 3a site. This, along with the
the material still well underperformed the Pristine NCA cells. The increased cell volume may be associated to the formation of divalent
dependence of the discharge capacity on the upper cutoff potential Ni or Co which subsequently migrated to the 3a layer.
was much larger for CO2 NCA, than it was for Pristine or Dryroom While Li2 CO3 was imperceptible by XRD, FTIR easily detected
NCA. Li2 CO3 peaks on all of the annealed materials. Figure 4 compares the
Figure S5 displays the first cycle voltage profile of a cell with a PE normalized magnitudes of the 870 cm−1 peak as a function of base-
4.5 V cutoff for each baseline material. The degree of polarization line material and annealing conditions. Since this peak corresponds
varies significantly between materials, with Ambient and CO2 NCA to the CO3 group bending mode and is relatively broad, it can arise
electrodes showing substantial polarization when current is applied from the presence of either LiHCO3 or Li2 CO3 . After being annealed
while Dryroom and Pristine NCA showed little to none. The initial at 150◦ C and 350◦ C, the carbonate content on Pristine and Dryroom
polarization is only temporary for Dryroom NCA, which quickly re- NCA increased steadily, but slowly began to decline on the 500◦ C an-
gained a voltage profile similar to the Pristine material. However, the nealed materials, and rapidly decreased after being annealed at 750◦ C.
over-potentials for Ambient and CO2 NCA only partially relaxed dur- The carbonate peak on Ambient NCA, which is initially very high,
ing charging, and both materials reach the cutoff potential well before showed little change after annealing at 150◦ C, but otherwise steadily
being sufficiently delithiated. The discharge capacity retention of the decreased with increasing annealing temperature. Similar to Pristine
25th cycle (150 mA/g) vs. the 24th cycle (15 mA/g) was calculated for and Dryroom NCA, the CO2 material had an increase in the size of
each material when electrodes were charged up to either 4.2 or 4.5 the Li2 CO3 peak after annealing at 150◦ C, but then the Li2 CO3 con-
V. The resulting data is listed in Table I. The rate dependent capacity tent remained relatively consistent until it declined slightly after the
retention for 4.2 V cells was best for Pristine and Dryroom NCA, only 750◦ C anneal. For all materials, regardless of exposing atmosphere,
slightly diminished for CO2 NCA, and immeasurable for Ambient annealing at 750◦ C caused a reduction of the Li2 CO3 content relative
NCA since there was no discharge capacity at either rate. to the 500◦ C annealed samples. While Ambient and CO2 NCA do
show incrementally more Li2 CO3 after being annealed at 750◦ C than
Isolation and analysis of impurity species.—To further investigate the 750◦ C Pristine or Dryroom samples, these materials started with
the growth mechanisms for the surface impurity species and isolate higher amounts of Li2 CO3 and this excess Li2 CO3 could likely be
the impact that each specie has on electrochemical performance, all fully removed with a longer anneal at 750◦ C.
materials were separately annealed in the dryroom at 150◦ C, 350◦ C, Table I shows the average OCV (open circuit voltage) for elec-
or 500◦ C for 2 h, or at 750◦ C for 4 h. Similar to the TGA experiments, trodes comprised of each baseline material and every annealed sam-
the temperature and duration of the anneals were designed to remove ple of Pristine and Ambient NCA. Of the baseline materials, Pristine
species one at a time. Rietveld analysis was performed on the XRD NCA had the lowest OCV (3.391 V), followed by Dryroom (3.405
scans of each annealed sample, and the a and c lattice parameters as V), CO2 (3.434 V) and Ambient NCA (3.441 V). The OCV values
well as the calculated cell volume are presented in Figure 3. The Ri- are intimately related to the LiHCO3 and Li2 CO3 results listed in
etveld calculated lattice parameters and transition metal occupancies Table I and shown in Figure 4. Careful analysis of the OCVs for
are listed in Table SI. While none of the XRD scans showed any signs annealed Pristine and Ambient electrodes shows the same trend as
of impurity phases, the a (Figure 3a) and c (Figure 3b) lattice param- the normalized carbonate peak area results in Figure 4. Samples with
eters and the cell volume (Figure 3c) increased steadily by annealing higher (bi)carbonate contents have increased OCVs, which may be

Downloaded on 2017-12-15 to IP 134.148.10.12 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, 164 (14) A3727-A3741 (2017) A3733

Figure 4. FTIR data showing the normalized 870 cm−1 carbonate peak area
as a function of the baseline material and annealing conditions. Carbonate peak
intensities are normalized to the area of the 2912 cm−1 PE peak in each scan,
and show the changes in the LiHCO3 /Li2 CO3 content as a result of annealing.

and Li2 CO3 , TGA experiments identical to those previously described

were run on all annealed samples. Mass loss results for each of the
main impurity species is plotted as a function of annealing temperature
for each baseline material in Figure 5a–5d and listed in Table I. From
Figure 5a, which shows the impurity mass loss for all variations of
Pristine NCA, it is clear that there is very little moisture, LiHCO3 , or
LiOH on the baseline material, and that there are minimal changes in
the quantity of these species for the annealed samples. 0.31 wt% of
Li2 CO3 was observed on Pristine NCA, and the quantity of Li2 CO3
grew with higher temperature anneals until it reached a maximum
of 0.79 wt% on the 500◦ C annealed material. The 750◦ C annealed
sample had a slightly lower Li2 CO3 content than the 500◦ C material,
but had a smaller decline in the amount of Li2 CO3 than was suggested
by the FTIR results.
In contrast to the Pristine material, Ambient NCA (Figure 5b)
had large quantities of every impurity specie. 0.47 wt% H2 O was
measured on the Ambient baseline sample, however after annealing
at 150◦ C and 350◦ C the amount of water dropped to 0.16 and 0.03
wt%, respectively. Similar trends were observed for the LiHCO3 and
LiOH contents on Ambient NCA, which initially had 1.41 and 0.24
wt% of LiHCO3 and LiOH. The LiHCO3 content was negligible
after annealing Ambient NCA at 350◦ C and a 500◦ C anneal proved
sufficient to eliminate the vast majority of the LiOH. Supporting the
FTIR results, the amount of Li2 CO3 on baseline Ambient NCA was
0.72 wt%, much higher than what was observed on Pristine NCA
Figure 3. Plots of the (a) a and (b) c lattice parameters, and (c) cell volume as (0.31 wt%). The measured Li2 CO3 content on Ambient materials
determined by Rietveld analysis of XRD scans of each annealed sample. The increased with higher temperature annealing, reaching a maximum
calculated parameters for the baseline materials are shown as having an anneal after the 500◦ C anneal, comparable to the development seen on the
temperature of 24◦ C. TGA experiments of Pristine NCA. After the 750◦ C anneal, much less
Li2 CO3 was observed on Ambient NCA (0.44 wt%) than what was
on either the baseline Ambient (0.72 wt%) or 500◦ C Ambient (0.97
associated with the partial delithiation of the NCA. OCV results for wt%) materials, confirming that annealing at such high temperatures
the four baseline materials may help explain the discrepancies in the causes a reduction in the Li2 CO3 content. FTIR results from post-
c lattice parameters determined by XRD, since materials with higher TGA Pristine and Ambient baseline materials are displayed in Figure
OCVs have less Li in the interlayer, which is known to cause ex- S6. The absence of the 870 cm−1 peak and the peaks in the range of
pansion of the c-axis for mostly lithiated layered oxides.44 There is a 1419–1480 cm−1 or 1576–1714 cm−1 , signify the complete removal
clear correlation between the OCV and c lattice parameter for each of the Li2 CO3 , LiHCO3 and H2 O species. Similar FTIR experiments
baseline material. Changes to the c lattice parameter as a function of were run on all post-TGA materials, none of which had signs of any
annealing can be explained by a combination of changes in the OCV remaining moisture, hydroxide or carbonate species.
and increased degrees of transition metal mixing. Dryroom NCA (Figure 5c) was nearly identical to the Pristine
To provide a more rigorous quantitative analysis of the impurity material. Very little H2 O, LiHCO3 or LiOH was observed and with a
species present, and to distinguish between the evolution of LiHCO3 Li2 CO3 content of 0.32 wt% Dryroom NCA had essentially the same

Downloaded on 2017-12-15 to IP 134.148.10.12 address. Redistribution subject to ECS terms of use (see ecsdl.org/site/terms_use) unless CC License in place (see abstract).
 A3734 Journal of The Electrochemical Society, 164 (14) A3727-A3741 (2017)

essentially annealed a second time, allowing for additional Li2 CO3

to develop or reform on the surface at temperatures below 710◦ C,
and causing the Li2 CO3 mass loss to be inflated from what was orig-
inally on the sample. To confirm that Li2 CO3 was developing on
these materials as a result of the TGA experimental setup, additional
TGA tests where the individual isotherms were systematically re-
moved were performed on 750◦ C Pristine NCA. Figure S7 compares
the measured Li2 CO3 mass loss when using the standard TGA pro-
tocol, and when removing each isotherm one by one. In the absence
of the 500◦ C isotherm the Li2 CO3 content on 750◦ C Pristine NCA
decreased by 23%, but when the 350◦ C or 150◦ C isotherms were also
removed the quantity of Li2 CO3 was little changed. These results sug-
gest that Li2 CO3 readily develops on the material’s surface at higher
temperatures, but the growth rate is insignificant at lower tempera-
tures (≤350◦ C). Figure S7 also shows that when using a specialty
dry air gas that is devoid of CO2 (<0.1 ppm) the Li2 CO3 mass loss
from 750◦ C Pristine NCA is considerably reduced. The Li2 CO3 mass
loss was comparable between samples that used the dry air with no
CO2 (patterned bar) and ones that used standard dry air but had no
500◦ C isotherm (solid bars), confirming that Li2 CO3 growth is preva-
lent only at higher temperatures when in the presence of a dry atmo-
sphere with CO2 . Matsumoto et al. measured the growth of Li2 CO3 on
LiNi0.81 Co0.16 Al0.03 O2 during heating to 800◦ C in air by in-situ XRD.
Their results show that Li2 CO3 began to develop around 525◦ C, and
reached a maximum concentration at 675◦ C before decomposing.11 In
their study of Li2 CO3 growth on Li2 O, Mosqueda et al. proposed that
the formation of Li2 CO3 was limited at lower temperatures because
of the inability for Li atoms to diffuse through the thin Li2 CO3 shell.
When their samples were heated to 530◦ C and higher, they observed
rapid Li2 CO3 development due to the temperature promoted increase
in diffusion through the Li2 CO3 layer.43
Figure 5. Quantification of H2 O, LiHCO3 , LiOH, and Li2 CO3 contents as The moisture content of every annealed and baseline sample was
determined by TGA. The surface impurity mass loss results of the baseline measured by KF coulometry and the results are shown in Figure 6.
and annealed samples for Pristine, Ambient, Dryroom, and CO2 NCA materials For each sample the moisture determined by the first KF test (black
are displayed in panels (a-d), respectively. The magnitude of the H2 O (black bars) and the cumulative result from successive tests (red bars) are
square), LiHCO3 (red circle), LiOH (green upward triangle), and Li2 CO3 provided. For all four baseline materials, the amount of moisture de-
(blue inverted triangle) contents were determined by the mass loss from the creased significantly by annealing at 150◦ C, and decreased further by
20–150◦ C, 150–350◦ C, 350–500◦ C, and 500–750◦ C regions, respectively.
annealing at 350◦ C. Annealing at 500◦ C and 750◦ C was not effective
in continuing to reduce the moisture content, likely because longer
cool down times enabled more moisture to re-adsorb onto the surface.
amount of Li2 CO3 as the Pristine baseline (0.31 wt%). Analogous When comparing samples from different exposures it is critical
to the annealed Pristine NCA samples, Li2 CO3 continued to develop to take note of the scale, which for clarity was made different for
after the 150◦ C, 350◦ C, and 500◦ C anneals, but decreased after being each baseline sample. Not surprisingly, the Pristine material had the
annealed at 750◦ C. The quantity of the other impurity species was lowest moisture content and Ambient NCA had the most. The cumu-
little changed as a result of annealing, similar to what was observed on lative ppm of H2 O for Ambient NCA was over 28 times the Pristine
the Pristine NCA samples. While difficult to see, CO2 NCA, shown in measurement, and well above either the Dryroom or CO2 baseline
Figure 5d, had about 3 times as much moisture and lithium bicarbonate samples. The slight (∼200 ppm) increase in H2 O that Dryroom NCA
than either the Pristine or Dryroom baseline materials. Additionally, had compared to the Pristine material is a result of the 2 week ex-
the Li2 CO3 content of 0.37 wt% found on CO2 NCA is slightly (19%) posure to the dryroom atmosphere, which contains a minute amount
higher than the analogous Pristine sample, supporting the relative of moisture. KF measurements show that CO2 NCA had a moisture
increase that was measured by FTIR. From the TGA results, CO2 content about 4 times that of Pristine NCA, which supports the TGA
NCA had lower concentrations of Li2 CO3 on the 500◦ C and 750◦ C experiments that observed around 3 times higher wt% loss from water.
samples than either of the corresponding Pristine or Dryroom NCA For all samples the moisture measured from the first test was a good
materials did. indicator of the cumulative quantity, and the two bars always moved
The insight provided by TGA helps explain the FTIR results shown in lock-step with one another.
in Figure 4. Ambient NCA was the only material not to have a large XPS measurements on baseline Pristine, Ambient exposed, and
increase in the carbonate peak after annealing, which is attributed to a annealed Ambient NCA samples are presented in Figure 7. The O 1s
saturation of LiHCO3 and Li2 CO3 by previous exposure. The reduc- region (Figure 7a and Figure S8 (a)) shows peaks centered at 531–
tion of the carbonate peak on the 350◦ C annealed Ambient material 532 eV, and at 529 eV. The 531–532 eV peak location and breadth
can be explained by the loss of LiHCO3 , which was more than enough for Ambient and 150◦ C Ambient NCA corresponds very well to a
to offset any increase in Li2 CO3 . However, annealing at 350◦ C caused Li2 CO3 reference material as shown in Figure S8 (b). For our refer-
an increase in the carbonate peak for Pristine and Dryroom NCA ence Li2 CO3 , the main O 1s lattice peak around 531.5 eV is slightly
because there was an insignificant amount of LiHCO3 to lose. lower in energy than previously reported for in-situ measurements of
Discrepancies in the observed Li2 CO3 content between FTIR and 531.9 eV, which may indicate the presence of a thin bicarbonate sur-
TGA can be partially attributed to how each technique was performed. face layer on the Li2 CO3 Reference 45. The peak at 529 eV is from the
For FTIR, the samples were only exposed to an Ar atmosphere, and lattice oxygen in NCA and is used to normalize the O 1s spectra in-
were kept at room temperature during testing. However, in the TGA tensities, with the exception of Ambient NCA whose surface impurity
experiments all samples were heated to temperatures as high as 750◦ C layer was thick enough to obscure the lattice oxygen signal. Results
and were continually exposed to dry air. As a result, the samples were of the peak intensity ratios between surface (from impurity species)

Downloaded on 2017-12-15 to IP 134.148.10.12 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, 164 (14) A3727-A3741 (2017) A3735

Figure 6. KF coulometry results for all anneals of (a) Pristine, (b) Ambient, (c) Dryroom, and (d) CO2 NCA. Results from the measured moisture content of
the first test (black) and the cumulative result from successive experiments (red) are provided. Note that the scale of the measured H2 O axis varies significantly
between samples. All KF samples were prepared and sealed in an Ar-filled glove box to prevent atmospheric exposure.

and lattice oxygen are presented in Figure 7b, and exhibit a trend that thick to clearly see the lattice signal, there are no significant differences
closely mirrors the FTIR results presented in Figure 4. After annealing in the Ni or Co signals between samples, suggesting no change in the
at 150◦ C the percent composition of surface oxygen decreases slightly, redox state of the transition metals. The samples spectra clearly differ
mainly because the impurity layer is now not completely obscuring the from the NiO reference, which seems to suggest that the average
lattice oxygen signal. Annealing at 350◦ C, which removes LiHCO3 valence state of the Ni is closer to +3. This is in contrast to other
and partially forms Li2 CO3 and CO2 gas, causes a large decline in the studies which have observed an increase in Ni2+ on the surface after
normalized surface oxygen peak, while annealing at 500◦ C has little ambient exposure. Importantly, it should be noted that the Ni 2p line
impact on the surface to lattice oxygen ratio. As expected the 750◦ C shapes for materials such as Ni3+ OOH and Ni2+ (OH)2 , which could
anneal, which removes Li2 CO3 , induced another substantial drop in form upon proton exchange or insertion, are shifted to higher energies
the surface oxygen signal. With each anneal the degree of change to compared to NiO, and are similar to Ni3+ in layered oxides.46 While
the surface to lattice oxygen ratio closely resembles the magnitude of the Ni 2p lineshape for samples annealed at temperatures ≥350◦ C
impurity species lost as determined by TGA. It was also observed that compare well with the Pristine NCA, the similarity of NiOOH and
as H2 O, LiHCO3 , LiOH and Li2 CO3 are subsequently removed from Ni(OH)2 peak positions, makes conclusive determination of the nickel
the Ambient exposed materials, the O1s 531–532 eV peak is con- redox state exceedingly difficult.47 The steady decrease of the Li 1s
sequentially shifted to higher energies. The peak position at around Li2 CO3 peak with higher temperature annealing confirms the loss of
531.9 eV is consistent with primarily Li2 CO3 remaining on the 500◦ C the carbonate signal observed in the O 1s spectra.
and 750◦ C annealed samples, supporting the TGA results presented To complement the electrochemistry results of the baseline mate-
earlier. rials (Figure 2), electrodes were fabricated from all of the annealed
Figures 7c and 7d display the XPS spectra in the Ni 2p, Co 3p, and analogs of the Pristine and Ambient NCA materials. The discharge
Li 1s regions. Other than Ambient NCA, whose surface layer was too capacities versus cycle number for coin cells using these materials

Downloaded on 2017-12-15 to IP 134.148.10.12 address. Redistribution subject to ECS terms of use (see ecsdl.org/site/terms_use) unless CC License in place (see abstract).
 A3736 Journal of The Electrochemical Society, 164 (14) A3727-A3741 (2017)

Figure 7. XPS spectra of the (a) O 1s region for select samples and (b) the percent composition of surface and lattice oxygen for Pristine NCA and all Ambient
materials. The amount of surface and lattice oxygen were determined by the relative intensities of the higher binding energy peaks (530–533 eV) and NCA (529
eV) peaks. Peak intensities of the spectra in the Figure 7a are normalized to the size of the NCA peak. The peak locations from a Li2 CO3 reference and NCA are
denoted by vertical bars. Figure 7c–7d show the Ni 2p, Co 3p, and Li 1s regions for the same samples shown in (a). To compare the results to a Ni2+ lineshape, the
Ni 2p region of a NiO reference material is displayed in (c).

are plotted in Figure 8a–8d. Each material was cycled using the capacities similar between the two materials, but so is the capacity
same procedure as the previously shown cells. Figure 8a shows that retention performance. These results are mimicked when comparing
150◦ C annealed (H2 O removed) Ambient NCA performs drastically the 500◦ C (Figure 8c) and 750◦ C (Figure 8d) annealed materials.
better than the Ambient baseline material, but still underperforms While the discharge capacities of the 350◦ C Pristine material were a
150◦ C annealed and baseline Pristine NCA samples. Annealing at bit lower than what was measured for the 150◦ C Pristine cells, they
150◦ C had minimal impact on the cycling capabilities of Pristine had a nearly identical performance as the 500◦ C Pristine samples.
NCA relative to the baseline material, and only a slight reduction in The 750◦ C Pristine cells had higher capacities than any of the other
discharge capacity (<2%) was observed. As a result of the 150◦ C annealed Pristine materials, since the 750◦ C anneal removed most
anneal, 150◦ C Ambient NCA has discharge capacities as high as of the Li2 CO3 that developed at lower temperatures. As a result of
152 and 168 mAh/g when charged to 4.2 and 4.5 V. However, there the Li2 CO3 decomposition, the 750◦ C Pristine electrodes had similar
is a rapid capacity fade that quickly degrades the electrochemical capacities as the baseline Pristine NCA.
performance, and causes the discharge capacity to drop below 100 The change in the discharge capacity of the Pristine NCA samples
mAh/g after only 69 and 86 cycles. While the discharge capacity of as a result of annealing follows the inverse trend of the Li2 CO3 content
150◦ C Ambient NCA is higher when the cutoff potential is 4.5 V identified from the FTIR and TGA measurements (Figure 4 and Figure
versus 4.2 V, the capacity increase is relatively minor compared to 5). As the Li2 CO3 content increases the discharge capacity is reduced,
the increase in capacity between 4.5 and 4.2 V for 150◦ C Pristine and after annealing at 750◦ C, when the Li2 CO3 is mostly decomposed,
NCA. the discharge capacity increases. As shown by the percent capacity
As shown in Figure 8b, annealing at 350◦ C (LiHCO3 removed) had loss after 50 cycles for all Pristine and Ambient materials in Figure 8e
a profound impact on the electrochemical performance of Ambient (and in Table I), there are significant increases in the capacity retention
NCA. When cycled to 4.2 V, the cycling stability of the 350◦ C Pristine of Ambient NCA as H2 O, LiHCO3 , and LiOH are removed. The %
and 350◦ C Ambient are indistinguishable, and even when charged to capacity loss for Pristine electrodes are relatively invariant because of
4.5 V the discharge capacity of 350◦ C Ambient is only marginally the minimal surface species present on the material. Comparison of
worse than the Pristine counterpart. Not only are the initial discharge the 500◦ C annealed Pristine and Ambient materials (H2 O, LiHCO3 ,

Downloaded on 2017-12-15 to IP 134.148.10.12 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, 164 (14) A3727-A3741 (2017) A3737

Figure 8. Discharge capacity vs. cycle number for cells made with positive electrodes of (a) 150◦ C, (b) 350◦ C, (c) 500◦ C, and (d) 750◦ C annealed Pristine and
Ambient NCA materials. All cells were cycled at 24◦ C by charging to Vmax (4.2 or 4.5 V) at 20 mA/g, holding at constant voltage until the current fell below 10
mA/g, and then discharging at 15 mA/g to 2.75 V. Starting on the 6th cycle and every 19 cycles afterwards, the cells were discharged at 150 mA/g instead of 15
mA/g. (e) The 50th cycle discharge capacity as a function on annealing temperature for all Pristine and Ambient materials is provided. A breakdown of the surface
impurity species on Ambient exposed materials as a function of anneal, and the total impurity species on Pristine materials is shown in part f.

Downloaded on 2017-12-15 to IP 134.148.10.12 address. Redistribution subject to ECS terms of use (see ecsdl.org/site/terms_use) unless CC License in place (see abstract).
 A3738 Journal of The Electrochemical Society, 164 (14) A3727-A3741 (2017)

and LiOH removed) shows that there is a negligible difference in the that the CO2 exposed sample had nearly 3 times the amount of H2 O
capacity retention after 50 cycles. Figure 8f, which depicts a break- as Pristine NCA, despite using dry CO2 gas and storing the sealed
down of the surface impurity species on each Ambient exposed sample container in the dryroom. The CO2 NCA material also had slightly
(bars) and the cumulative impurity content on the annealed Pristine higher Li2 CO3 and LiHCO3 contents than Pristine NCA, and it is
materials (line), is a useful visual for understanding the relationships believed that the slight increases of these impurities are all related.
between each impurity specie and the impact on the material’s cycle Moisture from the plastic container or bag used to seal the CO2 gas
performance. could have caused the formation of some LiHCO3 and LiOH. Once
The voltage profiles of the 1st and 50th cycles from cells charged to bicarbonate or hydroxide species are created, they could interact with
4.5 V for all baseline and annealed versions of Pristine and Ambient either the CO2 gas (Eq. 4 and Eq. 5) or other impurity species (Eq. 6
NCA are presented in Figure S9. The 50th cycle voltage profiles in and Eq. 7) to produce new LiHCO3 , Li2 CO3 and H2 O.1,46,14,48,49
Figure S9 (a) show little difference between Pristine, Dryroom, and
LiOH + CO2 → LiHCO3 [4]
CO2 NCA materials. This suggests that the impurity species on CO2 ,
which caused the initial over-potential, are lost during cycling and
didn’t cause significant damage to the positive electrode. However,
even on the 50th cycle the Ambient material still shows significant 2LiOH + CO2 → Li2 CO3 + H2 O [5]
polarization, which results from its excess of impurity species. 1st
cycle voltage profiles for the annealed Pristine NCA samples, shown
in Figure S9 (b), display a strong correlation between the initial over- LiOH + LiHCO3 → Li2 CO3 + H2 O [6]
potential and the carbonate content presented in Figure 4. By the 50th
cycle, there are minimal differences in polarization between samples,
confirming the loss of the carbonate species with cycling. The Ambi-
2LiHCO3 → Li2 CO3 + H2 O + CO2 [7]
ent materials annealed at ≥350◦ C (Figure S9 (c)) have substantially
reduced polarization on their 50th cycle relative to the 1st cycle. Yet, Since many of the reactions between surface impurities (Eqs. 5–7)
due to the exceedingly large amount of impurity species still on 150◦ C produce water as a by-product, these mechanisms of surface impurity
annealed Ambient NCA significant polarization even on the 50th cycle development would likely be autocatalytic and would continue until
was observed. the source of Li or CO2 is impeded. Even though the LiHCO3 and
When charged to either 4.2 or 4.5 V, the rate dependent capacity Li2 CO3 contents did increase from exposure to 100% CO2 the degree
retention (Table I) for Pristine NCA decreases after being annealed of formation was minor relative to the Ambient samples, which were
at 350◦ C, and subsequently increases after being annealed at 750◦ C. exposed to an atmosphere that only had a few hundred ppm of CO2 .
The decline in high rate performance after the 350◦ C anneal may be These results, in conjunction with the Dryroom NCA results, strongly
attributed to the formation of Li2 CO3 by the degradation of the trace suggest that the formation of surface species at room temperature
LiHCO3 initially present whereas the increase in rate capability after requires both moisture and CO2 , and that Li2 CO3 is mainly created
annealing at 750◦ C is likely associated to the reduced Li2 CO3 content. through intermediary species and not, as suggested by some in the past,
In contrast, the high discharge rate performance of Ambient NCA in- by direct reaction between CO2 and the layered oxide material.11,34,50
creased significantly after annealing at 150◦ C, 350◦ C, and 750◦ C. The Thus dry atmospheres, especially ones with low CO2 concentrations,
large increases in rate capability after the 150◦ C and 350◦ C anneals is have a negligible impact on the growth of impurity species and are
likely because Ambient NCA originally had so much excess moisture safe for handling layered oxide electrode materials.
and bicarbonate species which formed an exceedingly thick cathode- As shown in Figure 6, annealing at 150◦ C for 2 h significantly
electrolyte interface upon reaction with the electrolyte. Even though reduces the moisture content of the baseline materials, and by 350◦ C
Li2 CO3 and Li2 O likely formed from LiHCO3 decomposition, the cu- there is only a negligible amount remaining. TGA results demonstrate
mulative interface impedance was still reduced. The improvement in that the vast majority of LiHCO3 and LiOH can be eliminated by
rate capability after annealing at 750◦ C is associated with the loss of annealing at 350◦ C and 500◦ C, respectively. While annealing in dry
Li2 CO3 , similar to the Pristine materials. Baseline, 150◦ C and 350◦ C air had the unintended consequence of promoting Li2 CO3 growth,
annealed Ambient NCA electrodes had steady impedance growth with especially at high temperatures, annealing above 710◦ C is known to
extended cycling, as measured by the high discharge rate capacity re- cause the degradation of Li2 CO3 and all of our 750◦ C annealed ma-
tention, while the rate capabilities of all other samples only decreased terials had a reduction in the Li2 CO3 content. Eliminating all of the
slightly. The large quantities of surface impurities on baseline, 150◦ C Li2 CO3 could have been achieved by using either a higher temper-
and 350◦ C annealed Ambient materials likely created an excess of HF ature or longer anneal. By choosing the right annealing temperature
which continually degraded the positive electrode’s surface. and duration, one can selectively remove impurity species. However,
it was observed that when annealing at higher temperatures the ma-
terial’s lattice parameters increased, and the resulting impact on the
Discussion
electrode’s stability and transport properties needs to be taken into
Comparison of the four baseline materials offers valuable insight account.
into the formation mechanisms of surface impurity species on layered Defining the mechanism of Li2 CO3 formation on the surface of
oxide materials and their impact on the host materials’ cycle stabil- layered compounds has been widely investigated and the main con-
ity. Exposure to the dryroom atmosphere (CO2 and O2 sans H2 O) had clusions are summarized in the introduction of this paper. While it was
negligible impact on the quantity of the impurity species or the electro- not our primary goal to define the specific details of this mechanism
chemical performance relative to the Pristine material. While storage within this paper, we have accumulated a body of evidence that can be
in a pure CO2 atmosphere induced the development of some additional useful to further advance the collective understanding. As noted above,
moisture, Li2 CO3 and hindered the material’s capacity retention, the the fact that exposure to dry CO2 does not enhance the formation of
magnitude of degradation was far less than what was observed in the Li2 CO3 by a great degree clearly shows that at room temperature di-
Ambient sample. The ambient atmosphere caused a nearly 30-fold rect reaction of CO2 with NCA to extract Li and form Li2 CO3 is not
increase in the moisture content, as well as ample LiHCO3 , LiOH and the active mechanism as proposed by others. Undoubtedly, moisture
Li2 CO3 development. As a result of the extreme increase in the sur- is required and is the catalyst for Li2 CO3 growth.
face impurity species concentrations, Ambient NCA was incapable of It is evident that the OCV increases systematically with Li2 CO3
cycling with a cutoff potential of 4.2 V, and performed terribly when formation at the surface of the NCA. This in itself can be utilized
charged to 4.5 V. as a valuable quality check for layered oxide materials without any
While it is intuitive that storage of NCA in the dryroom only caused advanced characterization to examine the degree of contamination of
a marginal increase in the moisture content, it was surprising to find the material’s surface with ambient induced impurity species. With

Downloaded on 2017-12-15 to IP 134.148.10.12 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, 164 (14) A3727-A3741 (2017) A3739

respect to the Li2 CO3 formation mechanism, there are two likely and unique to each type of transition metal and this will have a large effect
relevant scenarios that can lead to the increase of the OCV and the on the driving force for the proposed redox mechanism.
production of LiHCO3 /LiOH, which is the necessary precursor for Subsequent Reaction:
Li2 CO3 . One scenario is the deintercalation of Li+ and subsequent
oxidation of the NCA, the other is the formation of surface protonated
NCA with no change in oxidation state of the transition metals. It is LiHCO3 + LiNi0.8 Co0.15 Al0.05 O2 → (1-x) LiHCO3 +xLi2 CO3
known that NiOOH and CoOOH both have higher OCV vs. Li/Li+
than the lithiated analogues of LiNiO2 and LiCoO2 .44,51 In the first + Li1-x Hx Ni0.8 Co0.15 Al0.05 O2 [14]
scenario, based on redox of the transition metal atom (Eq. 8), H2 O
is reduced at the surface of NCA producing LiOH, H and a partially
oxidized NCA. This is quickly followed by either the loss of H2 The primary motivation for investigating the surface impurity
gas (Eq. 9) or subsequent oxidation of the H to reduce NCA back species is to understand the impact that each specie has on the elec-
to its original state and partially intercalate a proton (Eq. 10). The trochemical performance of the electrode. The cycling results for the
transition metal redox scenario is attractive because it is known that baseline (Figure 2) and annealed materials (Figure 8) highlight the
the propensity to form Li2 CO3 inversely scales with the initial redox impact that surface impurities have on the stability and capacities
potential of the intercalation compound. of layered oxide materials. Analysis of the improvement in capacity
Scenario 1 (redox): retention when Ambient NCA was annealed at 150◦ C demonstrates
the importance of ensuring that the electrodes are completely dry.
H2 O + LiNi0.8 Co0.15 Al0.05 O2 → (1-x) H2 O + xLiOH + xH Even trace amounts of water can react with the fluorinated electrolyte,
forming HF and leading to the decomposition of the positive elec-
+ Li1-x Ni0.8 Co0.15 Al0.05 O2 [8] trode surface structure.16,34,53 The increase in capacity retention that
occurred when Ambient NCA was annealed at 350◦ C is from fur-
ther reduction in the material’s moisture content and, more impor-
(1-x) H2 O + xLiOH + xH + Li1-x Ni0.8 Co0.15 Al0.05 O2 tantly, the removal of LiHCO3 , which besides increasing impedance
→ (1-x) H2 O + xLiOH + (x/2) H2 + Li1-x Ni0.8 Co0.15 Al0.05 O2 is also expected to react with fluorinated salts, producing HF. Af-
ter annealing at 500◦ C, all of the surface species other than Li2 CO3
[9] were removed and the Ambient material performs identically to the
500◦ C annealed Pristine samples. However, annealing below 750◦ C
was shown to have caused the formation of Li2 CO3 , and this Li2 CO3
(1-x) H2 O + xLiOH + xH + Li1-x Ni0.8 Co0.15 Al0.05 O2 layer is responsible for the reduced (∼5%) discharge capacity of
→ (1-x) H2 O + xLiOH + Li1-x Hx Ni0.8 Co0.15 Al0.05 O2 the 500◦ C annealed materials compared to the baseline Pristine NCA
cells. When annealed at 750◦ C, the Li2 CO3 layer was mostly removed
[10] and the discharge capacities for both Pristine and Ambient cells im-
proved slightly. An important takeaway is that with respect to cycling
The second explanation relies on ion exchange. In this case stability and capacity, Li2 CO3 has a minimal impact. Previous obser-
(Eq. 11), OH− is strongly adsorbed to the surface of NCA via co- vations of improved performance of layered electrodes when Li2 CO3
ordination with either Li or a transition metal charge density. This was removed are more likely attributed to the simultaneous removal
in turn enables a local ion exchange between a proton and Li atom, of LiHCO3 , LiOH, and possibly moisture (which was not removed
concomitant with the formation of LiOH (Eq. 12). An advantage of during the standard electrode processing before cell fabrication) and
this model is that such ion exchange is very likely to occur. Indeed, the not from Li2 CO3 . Chen and Dahn demonstrated that a similar per-
reverse reaction, LiOH + CoOOH or NiOOH has been used to form formance increase of LiCoO2 can be obtained by annealing at 550◦ C
both LiCoO2 and LiNiO2 at low temperatures.52 An alternative model instead of 800◦ C, supporting our results that surface species other than
(Eq. 13), which bypasses the formation of LiOH completely, relies on Li2 CO3 induce the majority electrochemical stability and capacity
the formation of carbonic acid and a direct exchange between H+ and degradation.26
Li+ . Regardless of the ion exchange scenario, the question remains
as to why the lower voltage nickelates more frequently induce this
reaction compared to the higher voltage LiCoO2 or LiMn2 O4 .
Scenario 2 (ion exchange): Conclusions
 
H2 O + LiNi0.8 Co0.15 Al0.05 O2 → (1-x) H2 O + x H+ + OH− Counter to some reports, Li2 CO3 on the surface of layered oxides
+ LiNi0.8 Co0.15 Al0.05 O2 [11] is not formed by direct reaction with ambient CO2 at room tempera-
ture. Moisture is a definitive entity required to trigger the formation
of Li2 CO3 at room temperature. In contrast to almost all previous
 
(1-x) H2 O + x H+ +OH- +LiNi0.8 Co0.15 Al0.05 O2 → (1-x) H2 O reports, including those from our group, Li2 CO3 on the surface of
layered compounds, such as NCA, is not detrimental to the material’s
+ xLiOH + Li1-x Hx Ni0.8 Co0.15 Al0.05 O2 [12] cycle life. The previously observed degradation can now be attributed
Or to the species integral to the formation of Li2 CO3 . Specifically H2 O,
LiHCO3 and likely LiOH each have their own contribution to the sig-
H2 O + CO2 +LiNi0.8 Co0.15 Al0.05 O2 →H2 CO3 +LiNi0.8 Co0.15 Al0.05 O2 nificant degradation of capacity and cycle stability. Removal of these
species while retaining Li2 CO3 resulted in excellent cycling stabil-
→ Li1-x Hx Ni0.8 Co0.15 Al0.05 O2 + (x/2) Li2 CO3 [13]
ity, comparable to the initial pristine NCA. It should be noted that
From either the redox (Eqs. 8–10) or 1st ion exchange scenario the presence of Li2 CO3 may have other impacts, both positive and
(Eq. 12) the subsequent reactions to form Li2 CO3 can be envisioned negative, beyond cycling stability which were not addressed here. Re-
to be straight forward. LiOH will react with ambient CO2 to either gardless of impact, it is necessary for battery manufacturers to gauge
directly form Li2 CO3 (Eq. 5) or create LiHCO3 (Eq. 4). The proton the degree of exposure and formation of Li2 CO3 on as received and
from LiHCO3 can ion exchange with the Li in NCA to form Li2 CO3 stored powders. It is imperative to the material’s electrochemical per-
(Eq. 14) or simply decompose along a pathway explained by either formance that the magnitude of surface impurities are measured and
Eq. 6 or 7. Both degradation scenarios, laid out in Eqs. 8–14, are not controlled. The most quantitative of all the characterization methods
limited to NCA and the basic mechanisms can be applied to other is TGA, while a more sensitive and practical approach to measure
R-3m layered oxide materials. However, the propensity for redox is the degree of surface impurities is simply monitoring the initial OCV

Downloaded on 2017-12-15 to IP 134.148.10.12 address. Redistribution subject to ECS terms of use (see ecsdl.org/site/terms_use) unless CC License in place (see abstract).
 A3740 Journal of The Electrochemical Society, 164 (14) A3727-A3741 (2017)

of the as fabricated battery, since the OCV was found to rise with fer impedance with surface phase transformations,” Appl. Phys. Lett., 108, 5
ambient exposure and subsequent surface impurity development. (2016).
21. D. P. Abraham, R. D. Twesten, M. Balasubramanian, J. Kropf, D. Fischer,
J. McBreen, I. Petrov, and K. Amine, “Microscopy and Spectroscopy of Lithium
Nickel Oxide-Based Particles Used in High Power Lithium-Ion Cells,” J. Elec-
Acknowledgments trochem. Soc., 150, A1450 (2003).
The authors acknowledge Frederic Cosandey, Pinaki Mukherjee, 22. D. P. Abraham, R. D. Twesten, and M. Balasubramanian, “Surface changes on
LiNi0.8 Co0.2 O2 particles during testing of high-power lithium-ion cells. 4, 620
and Gerbrand Ceder for useful discussions. This work was supported (2002).
in full as part of NECCES, an Energy Frontier Research Center funded 23. K. Shizuka, C. Kiyohara, K. Shima, K. Okahara, K. Okamoto, and Y. Takeda, “Effect
by the U. S. Department of Energy, Office of Science, Office of Basic of CO2 on Layered Li1+z Ni1-x-y Cox My O2 (M = Al, Mn) Cathode Matierals,” Trans.
Energy Sciences under Award No. DE-SC0012583. E C S Soc. Electrochem., 11, 7 (2008).
24. W. Li, W. R. McKinnon, and J. R. Dahn, “Lithium Intercalation from Aqueous So-
lutions,” MRS Proc, 369, 69 (1994).
25. A. Grenier, H. Liu, K. M. Wiaderek, Z. W. Lebens-Higgins, O. J. Borkiewicz,
ORCID L. F. J. Piper, P. J. Chupas, and K. W. Chapman, “Reaction Heterogeneity in
LiNi0.8 Co0.15 Al0.05 O2 Induced by Surface Layer,” Chem. Mater., 29, 7345
Nicholas V. Faenza https://orcid.org/0000-0002-2620-4592 (2017).
26. Z. Chen and J. R. Dahn, “Improving the Capacity Retention of LiCoO2 Cycled to
4.5V by Heat-Treatment,” Electrochem. Solid-State Lett., 7, A11 (2004).
References 27. N. Pereira, C. Matthias, K. Bell, F. Badway, I. Plitz, J. Al-Sharab, F. Cosandey,
P. Shah, N. Isaacs, and G. G. Amatucci, “Stoichiometric, Morphological, and Elec-
1. H. Liu, Y. Yang, and J. Zhang, “Investigation and improvement on the storage property trochemical Impact of the Phase Stability of Lix CoO2 ,” J. Electrochem. Soc., 152,
of LiNi0.8 Co0.2 O2 as a cathode material for lithium-ion batteries,” J. Power Sources, A114 (2005).
162, 644 (2006). 28. L. Hu, Z. Zhang, and K. Amine, “Electrochemical investigation of carbonate-
2. Y. Saito, M. Shikano, and H. Kobayashi, “State of charge (SOC) dependence of based electrolytes for high voltage lithium-ion cells,” J. Power Sources, 236, 175
lithium carbonate on LiNi0.8 Co0.15 Al0.05 O2 electrode for lithium-ion batteries.” J. (2013).
Power Sources, 196, 6889 (2011). 29. a. R. Armstrong, M. Holzapfel, P. Novák, C. S. Johnson, S. Kang, M. M. Thackeray,
3. H. Visbal, S. Fujiki, Y. Aihara, T. Watanabe, Y. Park, and S. Doo, “The influence and P. G. Bruce, “Demonstrating oxygen loss and associated structural reorganization
of the carbonate species on LiNi0.8 Co0.15 Al0.05 O2 surfaces for all-solid-state lithium in the lithium battery cathode Li[Ni0.2 Li0.2 Mn0.6 ]O2 ,” J. Am. Chem. Soc., 128, 8694
ion battery performance,” J. Power Sources, 269, 396 (2014). (2006).
4. G. V. Zhuang, G. Chen, J. Shim, X. Song, P. Ross, and T. J. Richardson, “Li2 CO3 30. H. Wang, E. Rus, T. Sakuraba, J. Kikuchi, Y. Kiya, and H. D. Abruña, “CO2
in LiNi0.8 Co0.15 Al 0.05 O2 cathodes and its effects on capacity and power,” J. Power and O2 Evolution at High Voltage Cathode Materials of Li-Ion Batteries: A
Sources, 134, 293 (2004). Differential Electrochemical Mass Spectrometry Study.," Am. Chem. Soc., 6197
5. N. Mijung, Y. Lee, and J. Cho, “Water Adsorption and Storage Characteristics of (2014).
Optimized LiCoO2 and LiNi1/3 Co1/3 Mn1/3 O2 Composite Cathode Material for Li- 31. K. T. Lee, S. Jeong, and J. Cho, “Roles of surface chemistry on safety and electro-
Ion Cells,” J. Electrochem. Soc., 153, A935 (2006). chemistry in lithium ion batteries,” Acc. Chem. Res., 46, 1161 (2013).
6. S. Hwang, W. Chang, S. M. Kim, D. Su, D. H. Kim, J. Y. Lee, K. Y. Chung, 32. A. Blyr, C. Sigala, G. Amatucci, D. Guyomard, Y. Chabre, and J-M. Tarascon, “Self-
and E. A. Stach, “Investigation of Changes in the Surface Structure of Discharge of LiMn2 O4 /C Li-Ion Cells in Their Discharged State,” Electrochem. Soc.,
Lix Ni0.8 Co0.15 Al0.05 O2 Cathode Materials Induced by the Initial Charge,” Chem. 145, 194 (1998).
Mater., 26, 1084 (2014). 33. N. V. Faenza, Z. W. Lebens-Higgins, P. Mukherjee, S. Sallis, N. Pereira, F. Badway,
7. C. K. Lin, Y. Ren, K. Amine, Y. Qin, and Z. Chen, “In situ high-energy X-ray A. Halajko, G. Ceder, F. Cosandey, L. F. J. Piper, and G. G. Amatucci, “Electrolyte-
diffraction to study overcharge abuse of 18650-size lithium-ion battery,” J. Power Induced Surface Transformation and Transition-Metal Dissolution of Fully Delithi-
Sources, 230, 32 (2013). ated LiNi0.8 Co0.15 Al0.05 O2 ,” Langmuir, 33, 9333 (2017)
8. J. Vetter, P. Novák, M. R. Wagner, C. Veit, K.-C. Möller, J. O. Besenhard, M. Winter, 34. A. M. Andersson, D. P. Abraham, R. Haasch, S. MacLaren, J. Liu, and K. Amine,
M. Wohlfahrt-Mehrens, C. Vogler, and A. Hammouche, “Ageing mechanisms in “Surface Characterization of Electrodes from High Power Lithium-Ion Batteries,” J.
lithium-ion batteries,” J. Power Sources, 147, 269 (2005). Electrochem. Soc., 149, A1358 (2002).
9. D. D. MacNeil, Dahn, and J. R. , “The Reaction of Charged Cathodes with Non- 35. M. MeF́neF́trier, C. Vaysse, L. Croguennec, C. Delmas, C. Jordy, F. Bonhomme, and
aqueous Solvents and Electrolytes: I. Li0.5 CoO2 ,” J. Electrochem. Soc., 148, A1205 P. Biensan, “7 Li and 1 H MAS NMR Observation of Interphase Layers on Lithium
(2001). Nickel Oxide Based Positive Electrodes of Lithium-Ion Batteries,” Electrochem.
10. J. Kim, Y. Hong, S. Ryu, M. G. Kim, and J. Cho, “Washing Effect of a Solid-State Lett., 7, A140 (2004).
LiNi0.83 Co0.15 Al0.02 O2 Cathode in Water,” Electrochem. Solid-State Lett., 9, A19 36. D. Aurbach, A. Zaban, A. Schechter, Y. Ein-Eli, and E. Zinigrad, ∗ B. Markovsky,
(2006). “The Study of Electrolyte Solutions Based on Ethylene and Diethyl Carbonates for
11. K. Matsumoto, R. Kuzuo, K. Takeya, and A. Yamanaka, “Effects of CO2 in air Rechargeable Li Batteries,” J. Electrochem. Soc., 142, 2882 (1995).
on Li deintercalation from LiNi1−x−y Cox Aly O2 ,” J. Power Sources, 81–82, 558 37. J. M. Tarascon, A. S. Gozdz, C. Schmutz, F. Shokoohi, and P. C. Warren, “Perfor-
(1999). mance of Bellcore’s plastic rechargeable Li-ion batteries,” Solid State Ionics, 86–88,
12. K. Shizuka, C. Kiyohara, K. Shima, and Y. Takeda, “Effect of CO2 on layered 49 (1996).
Li1+z Ni1-x-y Cox My O2 (M = Al, Mn) cathode materials for lithium ion batteries,” J. 38. J. W. Ward, “A Spectroscopic Study of the Surface of Zeolite Y. II. Infrared Spectra
Power Sources, 166, 233 (2007). of Structural Hydroxyl Groups and Adsorbed Water on Alkali, Alkaline Earth, and
13. H. S. Liu, Z. R. Zhang, Z. L. Gong, and Y. Yang, “Origin of Deterioration for LiNiO2 Rare Earth Ion-Exchanged Zeolites," J. Phys. Chem., 73, 4211 (1968).
Cathode Material during Storage in Air,” Electrochem. Solid-State Lett., 7, A190 39. S. F. Parker, K. Refson, R. I. Bewley, and G. Dent, “Assignment of the vibrational
(2004). spectra of lithium hydroxide monohydrate, LiOH · H2O.," 941, 1 (2011).
14. X. Zhang, W. J. Jiang, X. P. Zhu, A. Mauger, Qilu, and C. M. Julien, “Aging of 40. L. H. Jones, “The Infrared Spectra and Structure of LiOH, LiOH-H2 O and the Deu-
LiNi1/3 Mn1/3 Co1/3 O2 cathode material upon exposure to H2 O,” J. Power Sources, terium Species.," Remark on Fundamental Frequency of OH−, 217, (2011).
196, 5102 (2011). 41. K. A. Wickersheim, “Infrared Absorption Spectrum of Lithium Hydroxide,” J. Chem.
15. S.-W. Song, G. V. Zhuang, and P. N. Ross, “Surface Film Formation on Phys., 31, 863 (1959).
LiNi0.8 Co0.15 Al0.05 O2 Cathodes Using Attenuated Total Reflection IR Spectroscopy,” 42. D Aurbach, Y. Ein-Ely, and A. Zaban, “The Surface Chemistry of Lithium Electrodes
J. Electrochem. Soc., 151, A1162 (2004). in Alkyl Carbonate Solutions,” J. Electrochem. Soc., 141, L1 (1994).
16. D. Aurbach, K. Gamolsky, B. Markovsky, G. Salitra, Y. Goer, U. Heider, R. Oesten, 43. H. A. Mosqueda, C. Vazquez, P. Bosch, and H. Pfeiffer, “Chemical Sorption
and M. Schmidt, “The Study of Surface Phenomena Related to Electrochemical of Carbon Dioxide (CO2 ) on Lithium Oxide (Li2 O),” Chem. Mater., 18, 2307
Lithium Intercalation into Lix MOy Host Materials (M = Ni, Mn),” J. Electrochem. (2006).
Soc., 147, 1322 (2000). 44. G. G. Amatucci, J. M. Tarascon, and L. C. Klein, “CoO2 , The End Member of the
17. J.-C. Dupin, D. Gonbeau, P. Vinatier, and A. Levasseur, “Systematic XPS studies LiCoO2 Solid Solution,” Electrochem. Soc., 143, 1114 (1996).
of metal oxides, hydroxides and peroxides,” Phys. Chem. Chem. Phys., 2, 1319 45. A. C. Kozen, A. J. Pearse, C.-F. Lin, M. A. Schroeder, M. Noked, S. B. Lee, and
(2000). G. W. Rubloff, “Atomic layer deposition and in situ characterization of ultraclean
18. H. Kobayashi, S. Emura, Y. Arachi, and K. Tatsumi, “Investigation of inorganic lithium oxide and lithium hydroxide,” J. Phys. Chem. C, 118, 27749 (2014).
compounds on the surface of cathode materials using Li and O K-edge XANES,” J. 46. I. A. Shkrob, J. A. Gilber, P. J. Phillips, R. Klie, R. T. Haasch, J. Bareño, and
Power Sources, 174, 774 (2007). D. P. Abraham, “Chemical Weathering of Layered Ni-Rich Oxide Electrode Ma-
19. S. Watanabe, M. Kinoshita, T. Hosokawa, K. Morigaki, and K. Nakura, “Capac- terials," Evidence for Cation Exchange., 164, (2017).
ity fade of LiAly Ni1-x-y Cox O2 cathode for lithium-ion batteries during accelerated 47. A. P. Grosvenor, M. C. Biesinger, R. S. C. Smart, and N. S. McIntyre, “New in-
calendar and cycle life tests (surface analysis of LiAly Ni1-x-y Cox O2 cathode after terpretations of XPS spectra of nickel metal and oxides,” Surf. Sci., 600, 1771
cycle tests in restricted depth of discharge ranges),” J. Power Sources, 258, 210 (2006).
(2014). 48. Y. Sun, X. Song, J. Wang, and J. Yu, “Preparation of Li2 CO3 by gas-liquid reactive
20. S. Sallis, N. Pereira, P. Mukherjee, N. F. Quackenbush, N. Faenza, C. Schlueter, crystallization of LiOH and CO2 ,” Cryst. Res. Technol., 47, 437 (2012).
T.-L. Lee, W. L. Yang, F. Cosandey, G. G. Amatucci, and L. F. J. Piper, “Sur- 49. W. Yi, C. Yan, and P. Ma, “Crystallization kinetics of Li2 CO3 from LiHCO3 solu-
face degradation of Li1-x Ni0.80 Co0.15 Al0.05 O2 cathodes: Correlating charge trans- tions,” J. Cryst. Growth, 312, 2345 (2010).

Downloaded on 2017-12-15 to IP 134.148.10.12 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, 164 (14) A3727-A3741 (2017) A3741

50. H. Liu, Y. Yang, and J. Zhang, “Reaction mechanism and kinetics of lithium 52. G. G. Amatucci, J. M. Tarascon, D. Larcher, and L. C. Klein, “Synthesis of elec-
ion battery cathode material LiNiO2 with CO2 ,” J. Power Sources, 173, 556 trochemically activer LiCoO2 and LiNiO2 at 100◦ C,” Solid State Ionics, 84, 169
(2007). (1996).
51. R. Moshtev, P. Zlatilova, S. Vasilev, I. Bakalova, and A. Kozawa, “Synthesis, XRD 53. A. du Pasquier, A. Blyr, A. Cressent, C. Lenain, G. Amatucci, and J. M. Tarascon,
characterization and electrochemical performance of overlithiated LiNiO2 ,” J. Power “An update on the high temperature ageing mechanism in LiMn2 O4 -based Li-ion
Sources, 81–82, 434 (1999). cells,” J. Power Sources, 81–82, 54 (1999).

Downloaded on 2017-12-15 to IP 134.148.10.12 address. Redistribution subject to ECS terms of use (see ecsdl.org/site/terms_use) unless CC License in place (see abstract).