You are on page 1of 10

Journal of The Electrochemical

Society

OPEN ACCESS

Copper Dissolution in Overdischarged Lithium-ion Cells: X-ray


Photoelectron Spectroscopy and X-ray Absorption Fine Structure
Analysis
To cite this article: Christopher E. Hendricks et al 2020 J. Electrochem. Soc. 167 090501

View the article online for updates and enhancements.

This content was downloaded from IP address 193.52.23.190 on 23/02/2022 at 09:47


Journal of The Electrochemical Society, 2020 167 090501

Copper Dissolution in Overdischarged Lithium-ion Cells: X-ray


Photoelectron Spectroscopy and X-ray Absorption Fine Structure
Analysis
Christopher E. Hendricks,1,2,z Azzam N. Mansour,1,* Daphne A. Fuentevilla,1 Gordon
H. Waller,1 Jonathan K. Ko,1 and Michael G. Pecht2
1
Naval Surface Warfare Center Carderock Division, United States of America
2
Center for Advanced Life Cycle Engineering, College Park, Maryland, United States of America

In some applications, such as military or back-up energy applications, lithium-ion batteries can undergo storage for multiple years
without use. If the batteries are not properly maintained, the pack voltage can decrease over time due to cell self-discharge, battery
management system power requirements, and parasitic loads. However, lithium-ion batteries have a recommended discharge
voltage limit corresponding to a nominal 0% state of charge, and if discharged below this limit, they will experience an
overdischarge condition which can lead to dissolution of the copper current collector and introduce potential safety and
performance issues. This paper investigates the nature of copper dissolution in overdischarged lithium-ion batteries including the
relative concentration and chemical state of the copper found in overdischarged batteries through characterization by X-ray
photoelectron spectroscopy and X-ray absorption fine structure spectroscopy.
© 2020 The Author(s). Published on behalf of The Electrochemical Society by IOP Publishing Limited. This is an open access
article distributed under the terms of the Creative Commons Attribution 4.0 License (CC BY, http://creativecommons.org/licenses/
by/4.0/), which permits unrestricted reuse of the work in any medium, provided the original work is properly cited. [DOI: 10.1149/
1945-7111/ab697a]

Manuscript submitted November 21, 2019; revised manuscript received January 3, 2020. Published February 4, 2020. This paper is
part of the JES Focus Issue on Battery Safety, Reliability and Mitigation.

Lithium-ion batteries are increasingly being employed in appli- recovered. The failure mechanism of overdischarge into reversal is
cations ranging from consumer electronics to military and aerospace fairly well understood, as the negative electrode’s copper current
applications. Compared to other rechargeable battery technologies, collector is oxidized to Cu2+ and eventually plates on the positive
lithium-ion batteries have a high cell potential, energy density, and electrode where it is reduced to Cu. Eventually an internal short
cycle life that make them attractive as an energy source. To prevent circuit is formed; however, the state of charge (SOC) is low enough
excessive degradation and potential safety hazards, lithium-ion that it typically does not result in a thermal runaway event.5–9 More
batteries need to be operated within stable voltage, current, interesting is whether a cell that is overdischarged and is recoverable
temperature, and pressure limits, which is often accomplished (i.e. can be recharged and cycled) poses a safety risk. At high SOC,
through a battery management system. Well-designed battery cells are more likely to undergo thermal runaway if an internal short
management systems can prevent batteries from overcharging to circuit develops.17
voltages above the stable voltage limit, however, a battery manage- Maleki and Howard10 repeatedly overdischarged small prismatic
ment system itself cannot prevent a battery from overdischarging cells to voltages between 2.0 and 0 V to induce copper dissolution.
below the stable voltage limit without an external source of power. If Performance was degraded for cells cycled to 1.0 V or below, and
a battery is not maintained and regularly charged, the battery’s cells behaved most erratically when overdischarged to 0 V. The
voltage may slowly decrease due to parasitic loads and internal presence of copper was confirmed on both sides of the separator for
losses, eventually causing an overdischarge condition. Alternatively, one of the 0 V samples using energy dispersive spectroscopy (EDS),
imbalances in cells connected in series can lead to inadvertent but was not found in the other 2 cells overdischarged to 0 V. The
overdischarge of one or more cells during use, potentially resulting authors do not provide an additional explanation of this discrepancy
in the negative electrode reaching a greater potential than the other than the fact that the behavior of the cell was unpredictable
positive electrode. This phenomenon is known as voltage reversal when overdischarged to 0.0 V.
and is considered a more aggressive form of overdischarge. Crompton and Landi11 investigated copper dissolution thresholds
Overdischarge has been studied through electrochemical mea- in 3-electrode pouch cells and explored pre-lithiation strategies for
surements of negative electrode half-cells, three-electrode cells, and zero-volt storage of lithium-ion batteries. Direct measurement of the
full cells. In experiments carried out in half-cells, bare copper foils dissolved copper was not conducted in the study, but the authors
and graphite coated copper foils1–4 were placed in electrolyte found that a negative electrode potential of 3.1 V vs Li/Li+ resulted
solutions, and researchers found that copper dissolution occurs at a in a small oxidative current, with higher voltages leading to
voltage of approximately 3.5 V vs Li/Li+, and that a graphite coating increased currents. While this value is lower than the 3.4–3.6 V vs
limited the amount of copper dissolved when compared to the bare Li/Li+ reported elsewhere,1,2 the authors attributed this to ambient
copper foil. A half-cell voltage of 3.5 V corresponds to roughly a conditions, electrolyte composition, and non-faradaic processes.
full-cell voltage of 0.5 V for cells containing a layered-type Fuentevilla et al.12 and Hendricks et al.13 presented X-ray
transition metal oxide cathode (i.e. having the generic composition photoelectron spectroscopy and X-ray absorption fine structure
LiMO2 where M = Co, Mn, Ni, Al). Copper in the electrolyte was spectroscopy results from large-format (>30 Ah) nickel cobalt
measured through inductively coupled plasma (ICP) analysis. aluminum (NCA) oxide cylindrical cells experiencing overdischarge
In three-electrode and full cell studies, cells can either be to voltages as low as 0 V. Copper was detected on both electrodes
overdischarged into reversal,5–9 which requires a power supply to from cells that were stored at less than 0.5 V, whereas no copper was
force the cell voltage into a negative regime, or overdischarged to observed on electrodes from cells stored at 0.5 V and above.
voltages as low as 0 V.10–16 Depending on the severity of the Li at al.14 overdischarged LiCoO2 cells to voltages between 0 and
overdischarge condition, the cell can be rendered useless or may be 3 V and studied copper deposition on the positive electrode using
inductively coupled plasma (ICP) analysis. Furthermore, the thick-
ness of the cells and the gas composition were studied.
*Electrochemical Society Member. Overdischarge to 0 V led to the generation of gases, most notably
z
E-mail: Christopher.hendricks@navy.mil CO2, and the performance was significantly affected. ICP analysis
Journal of The Electrochemical Society, 2020 167 090501

showed that approximately 200 μg of copper per g of positive including: Cu(OH)2, Cu2O, CuO, CuF2, LiNiO2, NiO, LiCoO2, and
electrode material was detected at voltages of 3.0, 1.0, and 0.4 V. In LiNi0.8Co0.15Al0.05O2. The choice of compounds was based on
contrast, cells overdischarged to 0 V contained a copper content of potential copper compounds that could be formed in a lithium-ion
approximately 900 μg g−1 of positive electrode material. battery under overdischarge conditions as well as the primary
Kumai et al.15 repeatedly overdischarged LiCoO2 cells to –10% lithium-ion battery active positive electrode material of interest for
SOC with no voltage limits implemented, with the cell reaching 0 V this study. Reference samples were prepared by manually grinding
by the 50th cycle. While copper dissolution was not measured, gas powders with a mortar and pestle. An ultrasonic sifter was then used
evolution was attributed to accelerated electrolyte decomposition as to collect particles less than 20 μm in diameter. The amount
a result of copper plating that inhibits lithium intercalation. of powder used for each sample was calculated to coincide with a
Love and Gaskins16 overdischarged pouch cells with a LiCoO2 K-edge step of approximately 1. The powder was then combined
positive electrode and graphite negative electrode. Their report with boron nitride and mixed with a magnetic stir rod until a uniform
primarily focuses on swelling due to gas generation and reduction mixture was attained. Approximately 80 mg of each mixture was
in charge/discharge cycles. When a cell was repeatedly over- then placed into a pellet mold and compressed using a force of 1.5
discharged to 1.2 V, the thickness increased by over 3.5 mm after metric tons. The mold was disassembled to remove the 12-mm-long
200 cycles. Furthermore, the cells reached 20% capacity loss within by 5-mm-wide rectangular sample pellet. The pellet was then
57 cycles, which is at least 4 times fewer cycles than the same cells mounted on Scotch-brand tape to avoid contamination.
operated in their normal cycling range. Copper analysis was not Samples of lithium-ion battery electrodes were transferred to the
performed, and may not have occurred based on the voltage ranges beamline in sealed plastic sample holders and Scotch-brand tape was
tested. Instead, solid electrolyte interphase (SEI) layer decomposi- used to mount samples for analysis. The tape was used to peel one
tion, reformation, and gas generation may have consumed lithium, side of the electrode off of the current collector. In the case of the
resulting in a reduced cell capacity. overdischarged negative electrodes, the graphite was collected as
Overdischarge of lithium-ion batteries has the potential to lead to large flakes that delaminated from the electrode.
a number of side reactions within a lithium-ion battery including
dissolution of the copper current collector, degradation of the SEI XAFS analysis.—X-ray absorption fine structure (XAFS) ana-
layer on the negative electrode, and gas generation. Depending on lysis was performed at the Advanced Photon Source (APS) at
the severity of the overdischarge condition, the battery could Argonne National Laboratory on beamline 5-BM-D. Experiments
experience more rapid capacity fade or develop internal short were run under a general user proposal submitted to the APS.
circuits. Typically, the internal short circuits do not cause a Beamline 5-BM-D uses bending magnets to direct X-rays at the
catastrophic failure, due to the low SOC of the cell at the time the samples located within the beam’s path. A Cambera passivated
short initiates; however, it is not clear from the literature whether implanted planar silicon (PIPS) fluorescence detector was used to
cells experiencing a single overdischarge can be recharged and measure copper on the negative electrode of the battery. Ion chamber
reused. This paper seeks to utilize both surface- sensitive (X-ray detectors were used to measure the intensities of the incident X-rays,
photoelectron spectroscopy, XPS) and bulk (X-ray absorption fine the transmitted X-rays through the sample, and the transmitted
structure, XAFS) analytical techniques to learn about the chemical X-rays through the reference foil. Nickel and cobalt in the positive
state, amount, and location of copper in cells that were over- electrode were measured in transmission mode, and copper in the
discharged for more than 9 months to identify the risk for short negative electrode could also be measured in transmission mode
circuiting. simultaneously with the fluorescence signal.
The monochromator was tuned based on the reference foil
Experimental selected. The reference foil was chosen as copper when investigating
the negative electrode (without current collector) for evidence of
Cells for this study were selected from battery packs containing
copper dissolution. The reference foil was chosen as nickel and
series-connected 52 Ah cells. The cells are cylindrical (54 mm
cobalt when investigating the positive electrode for potential
diameter × 208 mm length) and the positive and negative electrodes
changes in the oxidation state of the positive electrode
contain nickel cobalt aluminum (NCA) oxide and graphite, respec-
(LiNi0.8Co0.15Al0.05O2) as a result of overdischarge. This required
tive, as the active material. A population of 15 cells was over-
2 sets of scans for each sample, once at the nickel K-edge, and again
discharged to 0.5 V, 0.25 V, or 0 V for over 9 months at room
at the cobalt K-edge.
temperature of 25 ± 3 °C. The cells were then split into groups to
The X-ray cross-section was approximately 1 mm × 7 mm, and
either undergo disassembly without recharge, recharged once at a
samples were cut in excess of this cross section. It was possible to
C/5 C-rate (10.4 A) and then discharged at the same rate to 0% SOC
locate the center of the sample by performing vertical and horizontal
prior to disassembly, or set aside to cycle between 2.5 V and 4.1 V at
scans while monitoring the transmitted and fluoresced signal and
a C/2 C-rate (26 A). A sample of cells above 2.5 V were not
selecting the location that provided either the maximum (fluores-
overdischarged and simply stored at open circuit to be used as
cence) or minimum (transmission) signal. Scans for the copper
control samples. All cells were disassembled in a discharged state
K-edge were performed between 8.779 keV and 10.079 keV with
inside of an argon-filled glovebox with oxygen content less than
5 eV step in the pre-edge region, 0.5 eV step in the XANES region,
1 ppm. As the electrode sheets were unwound, the negative and
and 0.05 Å–1 step in the EXAFS region. Scans were performed with
positive electrodes were separated, and wound individually around
the same pre-edge, XANES, and EXAFS step settings for the nickel
separate mandrels. Samples were removed corresponding to dif-
K-edge between 8.133 keV and 9.433 keV, and scans for the cobalt
ferent locations within the electrode winding, including areas close
K-edge were performed between 7.509 keV and 8.809 keV.
to the cell casing (exterior), after unwinding approximately 2 m of
electrode material (middle) and after unwinding approximately 4 m
XPS sample preparation.—The same reference powders studied
of electrode material (interior). Roughly 50 mm × 50 mm square
under XAFS were examined by XPS. Electrode samples were
samples were cut from each section of the electrode for XAFS, and
attached to the sample holder with double-sided tape inside of the
two 11-mm-diameter samples were punched out of the electrode for
argon-filled glovebox and transferred to the XPS intro chamber in an
XPS. All samples were stored in the glovebox and either transferred
inert sample transfer vessel. The sample transfer vessel was
to the XPS in an airtight transfer vessel or spent up to 5 days in
connected directly to the XPS and placed under vacuum before
sealed plastic sample holders outside of the glovebox during XAFS.
opening the vessel. At no point was the sample exposed to air.

XAFS sample preparation.—Reference powders were collected XPS analysis.—XPS analysis was performed at Naval Surface
for a number of relevant copper, nickel, and cobalt compounds Warfare Center Carderock Division using a Physical Electronics
Journal of The Electrochemical Society, 2020 167 090501

Figure 1. Cu K-edge XANES of some copper reference compound, namely,


Cu, Cu2O, CuO, Cu(OH)2, CuF2. Figure 2. Cu K-edge XANES of different negative electrodes collected
from cells overdischarged to 0 or 0.25 V as well as after being recharged and
discharged to 0% state of charge.
VersaProbe II scanning X-ray photoelectron spectroscopy microp-
robe. The VersaProbe features a monochromatic Al-Kα X-ray source
Cu(OH)2 reference compounds as seen in Fig. 3a. The pre-edge
that can precisely perform analysis on an identified sample region.
feature illustrated with the dashed vertical line labeled A corre-
Samples are transferred into the XPS ultrahigh vacuum chamber
sponds with the pre-edge shoulder observed in Cu2O, whereas the
through an intro chamber that is pumped down to 4 × 10–4 Pa. The
dashed vertical lines B and C match those observed for Cu(OH)2.
XPS data collection commenced when the ultrahigh vacuum analysis
Figure 4 illustrates this further by looking at the Fourier transforms
chamber reached ∼2 × 10–8 Pa. The sample height is adjusted to
of the extended X-ray absorption fine structure (EXAFS) oscillations
maximize the signal intensity and survey and multiplex scans were
that are associated with interactions and collisions between ejected
performed on all of the samples. Multiplex scans of the negative
electrons and the surrounding atoms. There is close agreement with
electrodes focused on Li1s, C1s, O1s, F1s, P2p, and Cu2p3/2
the Cu2O and Cu(OH)2 spectra; however, the reduced amplitude of
photoelectron lines, and the positive electrode scans additionally
the peaks above 2 Å illustrates the amorphous nature of the Cu(OH)2
included Co2p and Ni2p orbitals. Scans were taken over a 200 μm ×
found in the negative electrode.
1000 μm area with an electron take-off angle of 45°. The 200-μm X-
To explore the differences between the different storage condi-
ray beam was electronically rastered to minimize X-ray damage of
tions, data from both fluorescence and transmission mode measure-
the samples. The analyzer pass energy was set to 117.4 eV for
ments were evaluated. Table I contains the copper K-edge step
survey spectra and 23.5 eV for multiplex spectra, which enabled
magnitude for the different storage conditions as measured in
resolutions of 1.76 eV and 0.35 eV, respectively. Argon ion
transmission mode and the calculated copper loading in each of
sputtering was conducted using an ion gun set to a power of 3 kV,
the samples. Values range from 0.37 mg cm−2 for a 0 V cell to
which has been calibrated to mill at a rate of approximately
0.070 mg cm−2 for a 0.25 V cell. The K-edge step for 0.5 V and
8 nm min−1 of SiO2.
2.5 V cells is sufficiently low that copper can be assumed to be
absent from these samples. The concentrations were calculated using
Results and Discussion
the absorption cross section at 30 eV above and below the Cu edge
XAFS.—XAFS data were analyzed using the Demeter software energy edge of 8979 as calculated by the software code
suite for calibration, pre-edge background subtraction, normalization Hephaestus.18 The magnitude of the K-edge step relates to the
and extraction of the EXAFS spectra (Athena package), and model relative concentration of copper in each of the samples, and clearly
fitting (Artemis package). All of the spectra were calibrated and shows a trend toward higher copper species concentrations in more
aligned using the measured reference foil (Cu for the negative severely overdischarged cells (0 V and 0.25 V) compared to healthy
electrode and Ni/Co for the positive electrode). The X-ray absorp- (2.5 V) or mildly overdischarged (0.5 V) cells. To determine the
tion near edge structure (XANES) for various reference copper contribution of different copper species to the overall copper signal,
compounds is shown in Fig. 1, and those for cells held at 0 V and the EXAFS spectra for each of the cells is fitted to a linear
0.25 V are shown in Fig. 2. Measurements of negative electrodes combination of contributions from the first coordination sphere of
from the 0.5 V and 2.5 V conditions demonstrated negligible copper copper in Cu2O and Cu(OH)2. This assumes that all samples contain
K-edges, indicating that copper dissolution only occurs under 0.5 V the same two copper compounds in varying concentrations. This
for this cell design. The spectra in Fig. 1 clearly display the model is supported by the observation made on the basis of the
sensitivity of the XANES region to composition, changes in XANES spectra. The first shell scattering paths for Cu(OH)2 and
oxidation state, and changes in local coordination geometry. The Cu2O were calculated using Feff within the Artemis software
spectra in Fig. 2 have differences in the intensity of the pre-edge package. Crystal structure data for Cu(OH)2 and Cu2O were obtained
shoulder at 8981 eV as well as at and above the K-edge. The changes from literature.19,20 The Cu2O contribution was limited to the first
in intensity as a function of energy are unique for each compound as shell which consists of 2 oxygen atoms at a distance of 1.83 Å. On the
seen in Fig. 1, and samples containing multiple compounds can be other hand, the Cu(OH)2 contribution was limited to four oxygen
represented as a linear combination of the two, with the intensity atoms at 1.96 Å and one oxygen atom at 2.36 Å. The many-body
depending on the mole fraction of each compound. amplitude reduction factor was constrained to the value derived from
Comparisons between the overdischarged negative electrodes and fitting the Cu reference data. After fitting the entire data set to the
the reference compounds are made in Fig. 3 by looking at the raw model, the model agreed quite well with the experimental data as
XANES data as well as the first derivative of the XANES data. Data seen in Fig. 5. The model parameters are summarized in Table II
from all 0 V and 0.25 V cells most closely matched the Cu2O and and Table III. Table II provides the refined parameters for the
Journal of The Electrochemical Society, 2020 167 090501

Figure 3. Cu K-edge XANES (left) and its derivative (right) of 0 V cell compared to reference compounds, a.) Cu2O and Cu(OH)2, and b.) CuF2, Cu, and CuO.
The pre-edge feature for the 0 V cell aligns more closely with that of Cu2O (a) than that of Cu (b) as seen by the dashed line labeled A. Examination of the 1st
derivative of XANES data demonstrates that the post-edge features for the 0 V cell align with those of Cu(OH)2 (a) and not those of Cu, CuO or CuF2 (b) as seen
by the dashed lines B and C.

coordination distance and the disorder for the various contributions energy scales for the negative electrode and positive electrode samples
from Cu2O and Cu(OH)2 as seen for copper in the negative electrode were calibrated based on the hydrocarbon C1s peak occurring at
samples. Table III demonstrates how the concentration of Cu2O and 284.8 eV. Once calibrated, the spectra were deconvoluted using a
Cu(OH)2 vary with respect to electrochemical test conditions. The Gauss-Lorentz function with an iterative Shirley background correc-
concentration of Cu2O is higher for the 0 V cells compared to the tion and fit to symmetric Gauss-Lorentz curves. For the positive
0.25 V cells. Furthermore, recharging the cell results in higher electrode C1s fit, an asymmetric curve was fit for the carbon black
concentrations of Cu2O as well. In addition, the XANES data signal. Each of the multiplex scans for the different electrode samples
for dissolved Cu were also analyzed in terms of a linear combination was fit and a summary of the negative electrodes Cu2p3/2 fit is
of XANES data for Cu2O and Cu(OH)2, and the results were summarized in Table IV.
qualitatively similar to the XAFS curve fit analysis results. XPS measurements of the negative electrode samples harvested
The positive electrodes were scanned at the copper, nickel, and from 2.5 V cells did not reveal any copper, consistent with
cobalt K-edges. The copper fluorescence signal could not be expectations. Only one 0.5 V sample exhibited a well-defined copper
resolved due to interference from the other metals contained in the peak, with about 0.1 atomic percent of copper detected. Samples from
positive electrode. Both the nickel and cobalt K-edges did not show 0.25 V and 0 V cells contained copper of 1–3 atomic percent. The
any differences between different storage conditions and the control samples are summarized in Table V. It is apparent that 0.5 V is close
samples. The absence of a change in oxidation state in the positive to the copper dissolution point and even at this voltage, some limited
electrode indicates that overdischarge does not lead to degradation of dissolution is possible; however, as no copper is found on the positive
the positive electrode active material as seen in Fig. 6. electrode side, extensive copper dissolution is likely not occurring.
The 3 samples (exterior, middle, and interior) from each cell did not
XPS.—The XPS data were analyzed using the Multipak software demonstrate significant differences in the atomic percentage of
from Physical Electronics. Copper reference compounds were measured elements and the multiplex spectra were consistent as well.
measured and the Cu2p3/2 region is shown in Fig. 7. CuF2 was Table V provides values for the scanned elements.
consistent with literature and exhibited a high binding energy The negative electrode multiplex data are shown in Fig. 8, with
(935.8 eV) compared to all of the measured samples.21 The binding the C1s spectra used for energy scale calibration on the top and the
Journal of The Electrochemical Society, 2020 167 090501

Figure 4. Fourier transform of EXAFS data for overdischarged cell and Figure 5. Comparison of Fourier transforms of experimentally measured
reference compounds Cu2O and Cu(OH)2. The Fourier transforms were k3-weighted EXAFS data (solid curve) and fitted model (dash curve) for the
performed from k3-weighted EXAFS spectra over the k-range of 3-12.0 Å−1 0 V negative electrode.
and a Hanning window of 1.0 Å−1.

Table I. Copper K-edge step measured in transmission mode. Cu2p3/2 spectra on the bottom. The C1s spectra contains 3 primary
contributions to the signal including an overlapping region con-
Transmission Cu loading sisting of hydrocarbon contamination and carbon singly bonded to
Storage Condition Cu K-Edge Step (mg cm−2) carbon. The other two peaks are representative of SEI layer
components including lithium alkyl-carbonates (carbon singly
0V 0.0722 0.304 bonded to oxygen) and lithium carbonate or carbon in the carboxylic
0.25 V 0.0165 0.070 group (carbon singly bonded to two oxygen atoms and doubly
0V, recharged 0.0873 0.368 bonded to one oxygen). The Cu2p3/2 spectra highlights contributions
0.25 V, recharged 0.0169 0.071 from both Cu2O and Cu(OH)2 with primary peaks occurring at
0.5 V, recharged 0.0002 0.000 approximately 933 eV and 935 eV, respectively. The satellite peaks
2.5 V 0.00002 0.000 between 939 and 947 eV are only seen in compounds with Cu2+
oxidation state, and can be associated with Cu(OH)2. In the 0.25 V
recharged cell, a low intensity C1s peak around 282.5 eV is observed
and can be attributed to intercalated graphite. The same cell also has
Cu2p3/2 spectra that are shifted to lower binding energies, with the
Table II. Model parameters.
first peak occurring closer to 932.5 eV. This binding energy is more
representative of Cu2O found in literature.22 While the binding
E0
energies of metallic copper and Cu2O are close together, as seen in
S 02 N R (Å) σ2 (Å2) (eV) Fig. 7, metallic copper has a Cu2p3/2 peak with a narrower FWHM
(0.96 for metallic copper vs 1.35 for Cu2O) based on our own
Cu(OH)2 4 1.939 0.0031 measurements. Furthermore, the presence of metallic copper was not
(0.004) (0.0010) detected during the XAFS analysis. Even though the XAFS analysis
0.86 1 2.335 0.036(0.018) 0.5 was performed in open air, metallic copper would have still been
(0.004) (0.4) detected even if the surface was oxidized. The positive electrode C1s
Cu2O 2 1.805 0.0028 and Cu2p3/2 spectra can be found in Fig. 9. The copper signal is
(0.004) (0.0014) weaker on the positive electrode, resulting in a noisy Cu2p3/2 signal.

Table III. Mole fraction of copper compounds in cells held at different storage conditions.

Storage Condition Mole Fraction (Cu(OH)2) Mole Fraction (Cu2O) R-Factor

0V 0.87(0.04) 0.13(0.04) 0.0018


0.25 V 0.93(0.05) 0.07(0.05) 0.014
0 V, recharged 0.78(0.04) 0.22(0.04) 0.0020
0.25 V, recharged 0.84(0.05) 0.16(0.05) 0.032
Journal of The Electrochemical Society, 2020 167 090501

Figure 7. Cu2p XPS spectra for some copper reference compounds, namely,
Cu, Cu2O, CuO, Cu(OH)2 and CuF2.

surface film on the cell cycled 40 times following an overdischarge


event. The presence of copper reemerges following the sputtering
and the location and FWHM of the peak are consistent with Cu2O.
Argon sputtering below 5 keV can reduce CuO to Cu2O23; however,
the presence of CuO was not observed during XAFS experiments.
Therefore, the Cu2O observed is not attributed to reduction of CuO.
Cu(OH)2 has been shown to reduce to Cu2O after exposure to X-rays
as well as with Argon ion sputtering,24 which may explain why
Cu(OH)2 is not detected in the cell cycled 40 times.
Taken together with the XAFS data, it is clear that Cu2O and
Cu(OH)2 are present within the battery. During overdischarge, the
copper current collector is oxidized to both Cu+ and Cu2+. The
copper ions are carried through the separator by the electrolyte,
which is supported by the presence of copper species on both the
Figure 6. Cobalt and nickel K-edges for NCA reference material, and negative and positive electrodes. The exact mechanism by which the
positive electrodes extracted from 2.5 V and 0 V cells. Cu2O and Cu(OH)2 are formed was not the focus of this study,
however, a number of mechanisms are possible. While trace
Similarly to the negative electrode, the Cu2p3/2 signal from the amounts of H2O can lead to the formation of Cu(OH)2, there is
positive electrode can be deconvoluted into peaks representing not sufficient water in the electrolyte to support this as the primary
contributions from Cu2O and Cu(OH)2, albeit shifted to a slightly mechanism. Instead, it is hypothesized that a reaction with the
higher binding energy. The presence of a satellite structure further electrolyte solvents is responsible for the formation of Cu2O and
supports the presence of Cu(OH)2. The resolution of the peaks is Cu(OH)2. Upon recharge, the formation of additional Cu2O can
difficult due to the weak copper signal which can explain some of be the result of reduction of copper species at the surface. This
the uncertainty in the peak location and full width at half maximum explains the XAFS observation that the samples primarily contain
(FWHM). Cu(OH)2 with increased Cu2O upon recharge. The XPS data shows
One of the 0.25 V cells cycled 40 times between 2.5 V and 4.1 V the opposite, with Cu2O dominating the Cu 2p3/2 spectra. While
following overdischarge was disassembled and the negative elec- the bulk of the sample may contain more Cu(OH)2, Cu2O on the
trode was examined using XPS. Initially, very little copper was surface will primarily show up in the XPS measurements. A
detected on the surface of the electrode, and argon ion sputtering was qualitative assessment of Cu(OH)2 and Cu2O solubility was con-
employed to remove surface layers to determine if decomposition ducted by suspending reference species in vials of electrolyte. The
products on the surface of the electrode obscured the copper signal. powders were found to settle out of suspension quickly with no
Figure 10 shows the C1s and Cu2p3/2 spectra before and after 5 min obvious changes in the volume of powder or color of the solution
of Argon ion sputtering. Sputtering was performed to remove the indicating solubility.

Table IV. Binding energy and full-width at half maximum (FWHM) values for negative and positive electrode Cu2p3/2 fits.

Storage Condition Cu2p3/2 Negative Electrode Binding Energy (FWHM) Cu2p3/2 Positive Electrode Binding Energy (FWHM)

0V 933.0 eV (1.66) – Cu2O 935.0 eV (2.24) – Cu(OH)2 933.3 eV (1.97) – Cu2O 935.2 eV (3.06) – Cu(OH)2
0.25 V 933.0 eV (1.87) – Cu2O 935.1 eV (1.62) – Cu(OH)2 933.6 eV (1.72) – Cu2O 935.3 eV (1.94) – Cu(OH)2
0 V, recharged 933.0 eV (1.49) – Cu2O 934.8 eV (2.91) – Cu(OH)2 932.0 eV (1.62) – Cu2O 934.5 eV (3.71) – Cu(OH)2
0.25 V, recharged 932.7 eV (1.88) – Cu2O 935.6 eV (0.62) – Cu(OH)2 933.6 eV (2.07) – Cu2O 935.2 eV (2.97)– Cu(OH)2
Journal of The Electrochemical Society, 2020 167 090501

Table V. Atomic percentage for different storage conditions and locations within each cell.

Atomic Percentage, % (Exterior, Middle, Interior)

Storage condition Location Li 1 s C 1s O 1s F 1s P 2p Cu 2p3/2

0V Exterior 7.7 35.3 35.4 13.6 4.1 3.8


Middle 8.8 46.0 25.3 14.5 3.1 2.3
Interior 10.1 40.1 27.4 15.7 3.5 3.2
0.25 V Exterior 12.4 40.3 26.1 16.4 3.9 1.0
Middle 17.4 39.1 17.9 21.9 3.0 0.6
Interior 11.7 41.3 26.1 15.7 4.1 1.3
0 V, recharged Exterior 20.3 34.9 21.9 19.3 3.3 0.3
Middle 25.1 28.8 16.7 26.9 2.3 0.3
Interior 12.6 40.1 26.2 16.1 3.7 1.4
0.25 V, recharged Exterior 16.1 34.5 22.0 22.9 3.0 1.4
Middle 15.2 38.7 22.2 20.3 2.7 1.0
Interior 18.0 36.5 20.3 22.4 2.0 0.8
0.5 V Exterior 16.7 40.8 17.2 21.8 3.6 0
Middle 17.4 38.2 17.5 23.4 3.5 0
Interior 14.3 40.7 22.8 17.9 4.2 0.1
2.5 V Exterior 17.4 40.1 23.8 16.1 2.7 0
Middle 18.2 38.2 23.8 17.1 2.8 0
Interior 18.4 40.5 24.6 13.9 2.6 0

Figure 8. C1s (top) and Cu2p3/2 (bottom) XPS multiplex spectra of negative Figure 9. C1s (top) and Cu2p3/2 (bottom) XPS multiplex spectra of positive
electrodes extracted from cells overdischarged to 0 or 0.25 V as well as after electrodes extracted from cells overdischarged to 0 or 0.25 V as well as after
being recharged and discharged to 0% state of charge. The spectra were being recharged and discharged to 0% state of charge. The spectra were
calibrated to hydrocarbon at 284.8 eV. calibrated to hydrocarbon at 284.8 eV.
Journal of The Electrochemical Society, 2020 167 090501

Figure 11. Following overdischarge to 0, 0.25, or 0.5 V, cells were cycled


and their capacity fade behaviors were compared to that of a control/healthy
cell. The average of capacity from 3 cells at each overdischarge condition is
plotted with error bars representing 1 standard deviation.

supported through the XPS analysis showing the presence of copper


on the surface of the electrodes. The capacity fade can also be
attributed to loss of adhesion between the negative electrode and the
copper current collector.6 Although the electrode jelly roll is under
compression, cyclic stresses due to intercalation can lead to localized
separation of the electrode from the current collector. Once parts of
the electrode are removed from the conductive network, the battery
loses some of its charge storage capabilities. This is further
supported through destructive physical analyses in which the cell
was disassembled and the electrode sheets unwound. In over-
discharged cells, the negative electrode either adhered to the
Figure 10. C1s (top) and Cu2p3/2 (bottom) XPS multiplex spectra of the
negative electrode extracted from a cell overdischarged to 0.25 V and
separator or delaminated in large flakes from the copper current
subjected to 40 cycles of charge and discharge between 2.5 V and 4.1 V. The collector. This is in contrast to the cells operated at 0.5 V and above,
spectra were collected before and after 5 min of sputtering. in which the electrode had to be peeled off of the current collector
for XAFS measurements. It is likely that a combination of these
failure mechanisms is responsible for the capacity fade observed by
While the presence of Cu2O and Cu(OH)2 on the surface of the
the overdischarged cells.
electrodes will certainly lead to performance loss, the formation of a
Failure due to internal short circuit was not observed, unlike
hard short circuit is unlikely due to the non-conductive nature of the
some previous studies.6,7,9,10 First and foremost, the cell never
copper species; however, the existence of copper ions in the
entered into voltage reversal, which has been shown to deposit
electrolyte solution does not completely preclude the formation of
copper on the electrodes and form an internal short circuit.9 In these
metallic copper. Metallic copper and other copper compounds such
instances, it is assumed that the copper is metallic in nature.
as CuO and CuF2 are not detected during both XPS and XAFS
Secondly, the cell only experienced a single overdischarge event,
analysis. This is evident when comparing the XAFS reference
limiting the amount of copper that could be dissolved. Finally, long-
spectra in Fig. 1 and the XPS reference spectra in Fig. 7.
term storage of the batteries at the overdischarge condition in this
Performance degradation is observed in the capacity fade plots
study provided opportunities for the copper ions to react with the
obtained by cycling the cells repeatedly within their nominal range
electrolyte rather than form metallic copper on the anode by
(2.5 V to 4.1 V) following the single overdischarge event as seen in
immediately recharging an over-discharged cell. Under the condi-
Fig. 11. Overdischarged cells demonstrate as much as 10% capacity
tions tested, no metallic copper was found; however, this does not
fade within 40 cycles, compared to the control sample which showed
preclude metallic copper formation under different conditions.
no degradation in capacity. In general, the lower the overdischarge
Further investigation into copper dissolution is warranted because
voltage, the greater the capacity fade. Capacity loss can be attributed
it suggests that the formation of metallic copper is not guaranteed
to a number of failure mechanisms, while several possibilities can be
following an overdischarge event even when eventually recharged.
eliminated. All of the cells experienced only a single overdischarge
This opens the possibility of continuing to use batteries that
event prior to cycling within their nominal voltage range, therefore,
experience a single overdischarge event with the knowledge that
repeated degradation of the SEI layer and significant copper
capacity fade will increase and some residual risk may remain.
dissolution are unlikely to contribute to the longer-term loss of
capacity. The initial loss of capacity can be attributed to the
Conclusions
reformation of the SEI layer, which can undergo dissolution when
the battery was overdischarged.10,11 Further capacity fade is likely The occurrence and chemical state of copper dissolution in large
due to other failure mechanisms, as the cells were not over- format lithium-ion batteries subjected to over-discharge voltages of
discharged repeatedly. One potential contribution to the continued 0 V, 0.25 V, and 0.5 V, were studied using surface- (XPS) and bulk-
capacity fade is that the presence of copper species on the electrodes sensitive (XAFS) analytical techniques, and compared to cells that
can inhibit lithium intercalation, resulting in accelerated electrolyte did not experience overdischarge. The presence of copper was
decomposition and a reduced discharge capacity.6,15 This is further evident in all samples held below 0.5 V using both XPS and XAFS
Journal of The Electrochemical Society, 2020 167 090501

measurements. This is consistent with values reported in literature Laboratory under Contract No. DE-AC02–06CH11357. Christopher
and indicates that at voltages below 0.5 V, the copper current E Hendricks and Michael G Pecht would like to thank the more than
collector began dissolving from both the exposed foil and along the 150 companies and organizations that support research activities at
current collector/electrode interface and enters the electrolyte solu- the Center for Advanced Life Cycle Engineering (CALCE) at the
tion phase. University of Maryland annually.
In all cases where the cell was overdischarged beyond 0.5 V,
adhesion between the negative electrode and the copper current References
collector was reduced, contributing in part to a capacity fade as 1. M. Zhao, S. Kariuki, H. D. Dewald, F. R. Lemke, R. J. Staniewicz, E. J. Plichta, and
much as 10% in 40 cycles compared to control samples with no R. A. Marsh, “Electrochemical stability of copper in lithium-ion battery electro-
capacity fade during the same timeframe. Loss of adhesion was lytes.” J. Electrochem. Soc., 147, 2874 (2000).
observed while disassembling the cell, in which large flakes of 2. M. Zhao, H. D. Dewald, F. R. Lemkie, and R. J. Staniewicz, “Electrochemical
stability of graphite-coated copper in lithium-ion battery electrolytes.”
graphite adhered to the separator leaving bare copper current J. Electrochem. Soc., 147, 3983 (2000).
collector behind. This only occurred in overdischarged cells. 3. M. Zhao, H. D. Dewald, and R. J. Staniewicz, “Quantitation of the dissolution of
Furthermore, copper species deposition on the electrode surfaces battery-grade copper foils in lithium-ion battery electrolytes by flame atomic
contributes to a reduction of capacity due to blocking of intercalation absorption spectroscopy.” Electrochim. Acta, 49, 683 (2004).
4. M. Zhao, H. D. Dewald, and R. J. Staniewicz, “Quantitation of the dissolution of
sites. Taken together, the primary failure mechanism in cells held at the graphite-coated copper foil in lithium-ion battery electrolytes by flame
low voltages, recovered, and subsequently cycled is a combination absorption spectroscopy.” Electrochim. Acta, 49, 677 (2004).
of degradation of electrode connectivity and reduction in intercala- 5. L. Wu, Y. Liu, Y. Cui, Y. Zhang, and J. Zhang, “In situ temperature evolution and
tion sites. The magnitude of this effect is directly tied to the amount failure mechanisms of LiNi0.33Mn0.33Co0.33O2 cell under over-discharge condi-
tions.” J. Electrochem. Soc., 165, A2162 (2018).
of copper dissolved and redeposited on the electrode surfaces, which 6. C. Fear, D. Juarez-Robles, J. A. Jeevarajan, and P. P. Mukherjee, “Elucidating
is a function of storage voltage. copper dissolution phenomenon in li-ion cells under overdischarge extremes.”
The copper species that were observed are non-conductive in J. Electrochem. Soc., 165, A1639 (2018).
nature. This reduces the risk of an internal short circuit when a 7. H. He, Y. Liu, Q. Liu, Z. Li, F. Xu, C. Dun, Y. Ren, M-x. Wang, and J. Xie,
“Failure investigations of LiFePO4 cells in over-discharge conditions.”
battery is overdischarged a single time such that the potential of the J. Electrochem. Soc., 160, A793 (2013).
negative electrode reaches the copper dissolution limit but is not 8. J. Kasnatscheew, M. Börner, B. Steipert, P. Meister, R. Wagner, I. C. Laskovic, and
pushed further into reversal. For this particular cell, this corre- M. Winter, “Lithium ion battery cells under abusive discharge conditions: electrode
sponded to full cell voltages below 0.5 V. Formation of metallic potential development and interactions between positive and negative electrode.”
J. Power Sources, 362, 278 (2017).
copper is still possible under different abuse scenarios, such as 9. R. Guo, L. Lu, M. Ouyang, and X. Feng, “Mechanism of the entire overdischarge
overdischarge into reversal; however, during repeated cycling over process and overdischarge-induced internal short circuit in lithium-ion batteries.”
40 cycles following a single overdischarge event, no indication of an Sci Rep-UK, 6, 1 (2016).
internal short circuit or metallic copper formation was identified. 10. H. Maleki and J. N. Howard, “Effects of overdischarge on performance and thermal
stability of a Li-ion cell.” J. Power Sources, 160, 1395 (2006).
Accelerated capacity fade was the primary failure mode observed as 11. K. R. Crompton and B. J. Landi, “Opportunities for near zero volt storage of lithium
a result of copper dissolution when cells were overdischarged to ion batteries.” Energy Environ. Sci., 9, 2219 (2016).
voltages as low as 0 V. 12. D. Fuentevilla, C. Hendricks, and A. Mansour, “Quantifying the impact of
Lithium batteries can sit in storage for extended periods of time. overdischarge on large-format lithium-ion cells.” ECS Trans., 69, 1 (2015).
13. C. Hendricks, A. Mansour, D. Fuentevilla, J. Ko, and G. Waller, “Copper
Through a combination of self-discharge and parasitic loads such as dissolution investigations in large-format lithium-ion cells.” 48th Power Sources
the battery management system, the voltage of the battery can fall Conf. (Denver, CO) June 11 (2018).
below the manufacturers’ recommended lower voltage cutoff. If this 14. H.-F. Li, J.-K. Gao, and S.-L. Zhang, “Effect of overdischarge on swelling and
voltage is sufficiently low, copper dissolution can occur. Previously recharge performance of lithium ion cells.” Chinese J. Chem., 26, 1585 (2008).
15. K. Kumai, H. Miyashiro, Y. Kobayashi, K. Takei, and R. Ishikawa, “Gas generation
it was thought that dissolved copper will form metallic copper and mechanism due to electrolyte decomposition in commercial lithium-ion cell.”
pose a risk for internal short circuit, effectively rendering the battery J. Power Sources, 81–82, 715 (1999).
useless. This paper demonstrates that under some usage conditions, 16. C. Love and A. Gaskins, “Performance loss of lithium ion polymer batteries
recovery is possible without the formation of metallic copper, albeit subjected to overcharge and overdischarge abuse.” Memorandum Report, NRL/MR/
6110 - 12-9455, 1 (2012).
with accelerated performance degradation. This may not be accep- 17. D. H. Doughty and E. P. Roth, “A general discussion of li-ion battery safety.”
table for all applications, and ultimately the tradeoffs need to be Electrochem. Soc. Interface, 21, 37 (2012).
weighed against the operational requirements. 18. B. Ravel and M. Newville, “ATHENA, ARTEMIS, HEPHAESTUS: data analysis
for X-ray absorption spectroscopy using IFEFFIT.” J. Synchrotron Rad., 12, 537
(2005).
Acknowledgments 19. H. R. Oswald, A. Reller, H. W. Schmalle, and E. Dubler, “Structure of copper (II)
hydroxide.” Cu(OH)2, Acta Cryst. Sec. C, 46, 2279 (1990).
The authors acknowledge support under the Naval Innovative 20. P. Villars and L. D. Calvert, in Pearson’s Handbook of Crystallographic Data for
Science and Engineering (NISE) program at the Naval Surface Intermetallic Phases, 2 (American Society for Metals, Metals Park, OH) p. 19943
Warfare Center Carderock Division. The XAS experiments were (1985).
21. J. C. Klein, C. P. Li, D. M. Hercules, and J. F. Black, “Decomposition of copper
performed at the DuPont-Northwestern-Dow Collaborative Access compounds in X-ray photoelectron spectrometers.” Appl. Spectr., 38, 729 (1984).
Team (DND-CAT) located at Sector 5 of the Advanced Photon 22. N. S. Mcintyre and M. G. Cook, “X-ray photoelectron studies on some oxides and
Source (APS). DND-CAT is supported by Northwestern University, hydroxides of cobalt, nickel, and copper.” Anal. Chem., 47, 2208 (1975).
E.I. DuPont de Nemours & Co., and the Dow Chemical Company. 23. G. Panzner, B. Egert, and H. P. Schmidt, “The stability of CuO and Cu2O surfaces
during argon sputtering studied by XPS and AES.” Surf. Sci., 151, 400 (1985).
This research used resources of the Advanced Photon Source, a U.S. 24. E. Cano, C. L. Torres, and J. M. Bastidas, “An XPS study of copper corrosion
Department of Energy (DOE) Office of Science User Facility originated by formic acid vapour at 40% and 80% relative humidity.” Mater.
operated for the DOE Office of Science by Argonne National Corros., 52, 667 (2001).

You might also like