You are on page 1of 13

Journal of The Electrochemical

Society

OPEN ACCESS

Increased Disorder at Graphite Particle Edges Revealed by Multi-length


Scale Characterization of Anodes from Fast-Charged Lithium-Ion Cells
To cite this article: Saran Pidaparthy et al 2021 J. Electrochem. Soc. 168 100509

View the article online for updates and enhancements.

This content was downloaded from IP address 186.211.111.10 on 07/12/2021 at 13:16


Journal of The Electrochemical Society, 2021 168 100509

Increased Disorder at Graphite Particle Edges Revealed by Multi-


length Scale Characterization of Anodes from Fast-Charged
Lithium-Ion Cells
Saran Pidaparthy,1,2 Marco-Tulio F. Rodrigues,1,* Jian-Min Zuo,2,3 and Daniel
P. Abraham1,*,z
1
Chemical Sciences and Engineering Division, Argonne National Laboratory, Lemont, Illinois, 60439, United States of
America
2
Department of Materials Science and Engineering, University of Illinois at Urbana-Champaign, Urbana, Illinois, 61801,
United States of America
3
Materials Research Laboratory, University of Illinois at Urbana-Champaign, Urbana, Illinois, 61801, United States of
America

Fast charging of lithium-ion cells increases voltage polarization of the electrodes and creates conditions that are favorable for Li-
deposition at the graphite anode. Repeated fast charging induces changes in the capacity-voltage profiles and increases the
probability of lithium-plating on the electrode. This higher probability results from structural, morphological and chemical
modifications that are revealed by multi-length scale characterization of graphite anodes extracted from discharged lithium-ion
cells, previously charged at rates up to 6 C. The distinct differences between anodes with lithium-plating and as-prepared electrodes
are clearly seen in analytical electron microscopy data. Scanning electrode microscopy (SEM) images show that the fast-charged
anode is significantly thicker, apparently because of the electrolyte reduction/hydrolysis products that accumulate in electrode
pores. High resolution electron microscopy (HREM) images reveal wavy graphite fringes near the particle edges. Analysis of
scanning electron nanodiffraction (SEND) data reveal higher d-spacings and greater lattice rotations, indicating disorder in the
graphite near the particle edges that extend about 20 nm into the bulk. The extent of this disorder is greater near larger internal
pores, highlighting nanoscale heterogeneities within particles. As graphite lithiation occurs primarily through edge planes, this
permanent disorder would hinder Li+ intercalation kinetics and favor Li0 plating during repeated cycling.
© 2021 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/ac2a7f]

Manuscript submitted August 25, 2021; revised manuscript received September 17, 2021. Published October 8, 2021.
Supplementary material for this article is available online

Advances in lithium-ion battery (LIB) technology have fueled the and growth on the graphite electrode.10–12 The situation is further
rise of the electric vehicle (EV) industry.1 Most high-energy LIBs complicated by non-uniformities in local reactivity resulting from
contain a positive electrode (cathode) comprised of a layered oxide factors that include uneven mechanical compression of the separator
with Ni, Co and Mn, a negative electrode (anode) comprised of that affects its porosity/tortuosity, and irregular distribution of active
graphite, and an electrolyte with LiPF6 salt in a mixture of organic particles that affects distribution of local currents within the
solvents. These cell components are being continually optimized so electrodes.13–16 These inhomogeneities cause large lithium concen-
that LIB technology can deliver high energy densities safely and tration gradients in the electrodes, both across the surface and along
reliably during a battery’s 10–15-year lifetime. State-of-the-art LIBs its thickness, promoting non-uniform aging of the cell materials. The
can be fully charged in about 1 h with minimal degradation in cell high charge currents can also enhance structural stresses in the
performance. However, full charging over shorter durations causes electrode matrix causing the coating to peel off the current collector
irreversible damage to the battery via losses in the inventory of and within the active material particles causing their fracture and
mobile Li+ ions (i.e., capacity fade) and resistance increases (i.e., disconnection from the electron conduction network.17
power fade) in the cells.2,3 The U.S. Department of Energy (DOE) is In our recent series of articles we have used operando X-ray
funding research to identify the underlying mechanisms that limit techniques to quantify phase evolution in the graphite and lithium
fast-charge performance of LIB cells and to develop technological concentration gradients that develop in the electrode during fast
solutions that would enable full battery charge in less than 10 min.4–7 charging.18,19 In addition, we have described techniques to identify
Solutions to these limitations will allay range anxiety concerns and lithium deposition on the graphite electrode and detailed electro-
accelerate customer acceptance of EVs. chemical charging protocols that can mitigate this plating.20–22 We
During cell charging, lithium ions are extracted from the observed that the likelihood of Li-plating on the graphite anode
transition metal oxide cathode and intercalated between atomic increases during repeated fast-charge cycling.10 We note here that
planes of the graphite particles. Under fast charge conditions, there is commercial graphite typically contains a mixture of the hexagonal
salt enrichment in the oxide cathode and salt depletion in the (2H) and rhombohedral (3 R) phases: the stacking sequence is
graphite anode, leading to non-uniform distribution of Li+ ions ABAB in the 2H structure and ABCABC in the 3 R structure.23
across the electrode thickness.8,9 These conditions lower Li+ ion The 2H phase is more thermodynamically stable than the 3 R phase
conductivity in the electrolyte within the electrode pores and at room temperature and pressure. The Li intercalation mechanism is
increase impedance, causing polarization in both electrodes. reported to be similar for both phases: during Li+ insertion, the
Cathode polarization causes the cell to reach its upper cut-off adjacent graphene sheets glide relative to each other to lower the
voltage (UCV) before Li+ ions are fully extracted from the oxide, overall energy of the system and attain AAAA layer stacking.24 Any
while anode polarization causes the cell to dip below the Li-plating material feature that hinders the movement of graphene sheets will
potential creating conditions that are favorable for lithium nucleation prevent Li+ intercalation. During fast charging, Li+ intercalation
into the graphite competes with Li0 nucleation on the particles.
Greater layer-to-layer stacking disorder in the graphite in the form of
added strain, local misfits, and misorientation angles in the stack,
*Electrochemical Society Member.
z
E-mail: abraham@anl.gov
would make Li+ intercalation more difficult and expedite Li0
Journal of The Electrochemical Society, 2021 168 100509

plating. Hence, an in-depth understanding of changes in the graphite


particles subjected to fast charging is crucial for the development of
materials that enable intercalation and mitigate deterioration of cell
performance during the electrochemical cycling.
In this work, we examine the morphological, chemical, and
structural changes to the graphite materials and electrodes, using a
suite of electrochemical and physicochemical techniques. The
characterization was performed across multiple length-scales on
as-prepared electrodes and on numerous electrodes harvested from
cycled cells; only representative data from the various techniques
are shown to highlight key changes that result from the cycling. The
data in this manuscript are from an as-prepared graphite (Gr-AP)
electrode that was never exposed to the electrolyte, graphite
electrode with no plating (Gr-NP) harvested from a cell after a
few low-rate cycles and graphite electrode with lithium plating (Gr-
LP) harvested from a fast-charged cell. For purposes of clarity and
contrast, we show only the Gr-AP and Gr-LP data in the main text;
data from the Gr-NP sample are included in the Supporting
Information (SI). Note that all Tables and Figures in the SI are
marked with the designator S, as in Fig. S1 (available online at
stacks.iop.org/JES/168/100509/mmedia).
We start by describing electrochemical tests, conducted on full
cells with LiNi0.5Co0.2Mn0.3O2 (NCM523) cathodes and graphite
(Gr) anodes, which include charging rates up to 6 C (here, 1 C is the
current needed to fully discharge the cell in 1 h). Following this, we
compare electrochemical behavior of the Gr-AP electrode with that
of the Gr-LP electrode. Then, after presenting high-resolution X-ray
diffraction (XRD) data from the materials, we show microscopy and
spectroscopy results obtained on electrode cross-sections fabricated
via ion milling and on sections prepared with a focused ion beam
(FIB). At the bulk-scale, we use scanning electron microscopy
(SEM) and energy dispersive X-ray spectroscopy (EDS) to probe the Figure 1. In this escalating C-rate experiment, the NCM523//Gr cell is
morphological and chemical changes in cross-sections of the charged with increasing currents to a fixed capacity of 87 mAh g−1. Each
graphite anode. At the nano- to atomic-scale, we rely on both charge is followed by 1 h rest and C/5 discharge to 3 V and a hold. The cell
conventional transmission electron microscopy (TEM) and scanning voltage changes during the cycles are shown in (a), while the graphite-
electrode potential changes are shown in (b). Lithium plating on the graphite
TEM (STEM) to reveal local structural disorder and chemical
particles becomes likely when the electrode potential dips below 0 V vs
variation within individual graphite particle sections through a Li/Li+; this Li-plating condition (LPC) is indicated by the dotted line in (b).
combination of high-resolution imaging, EDS, and scanning electron In the above experiment, Li-plating on the graphite electrode becomes likely
nanodiffraction (SEND). Our investigations reveal that anodes from when the charge rate exceeds ∼2 C.
the fast-charged cells have lower capacity, present a sloping voltage
profile, display an increase in thickness and show greater lattice °C. The Gr-LP electrode was harvested from a cell that experienced
disorder in localized areas near the graphite particle edges. multiple sequences of an “escalating C-rate” protocol (Fig. 1). To
accelerate cell aging, three of the sequences were at 30 °C, one at 40
Experimental Section °C and one at 50 °C. Including the initial cycles and periodic
Electrode materials, cell assembly and electrochemical reference performance tests (C/25 rate), the cell experienced a total
testing.—In the NCM523//Gr cells, the positive electrode has a of 50 cycles; of these, 25 charge segments were at rates of 2 C and
71 μm thick coating, with 90 wt% Li(Ni0.5Co0.2Mn0.3)O2 (Toda), above.
5 wt% C45 carbon (Timcal) and 5 wt% PVDF binder (KF-9300, After cycling, cells were discharged to (and held for 10 h at) 3 V
Kureha), on a 20 μm thick Al current collector.10 The negative before disassembly in an argon-atmosphere glove box. Various
electrode has a 70 μm thick coating, with ∼92 wt% SLC1506T diagnostic tests were conducted on the harvested cell materials to
graphite (Superior Graphite), 2 wt% C45 carbon and 6 wt% PVDF determine the effect of cycling on the graphite material. For the
binder, on a 10 μm thick Cu current collector. The electrolyte (aka electrochemical tests, 2032-type coin cells were assembled with 1.6
Gen 2, purchased from Tomiyama) is a solution of 1.2 M lithium cm2 area harvested-electrode samples, lithium metal counter elec-
hexafluorophosphate (LiPF6) salt dissolved in a 3:7 w/w solvent trode, fresh Celgard 2320 separator, and fresh Gen2 electrolyte.
mixture of ethylene carbonate (EC) and ethyl methyl carbonate
(EMC). Celgard 2320 is the separator between the electrodes. Before Cross-section polishing.—Cross-sections of the graphite elec-
cell assembly, the electrodes and separator are dried in vacuum trode samples were prepared via argon ion milling using the Gatan
ovens set at 120 °C and 70 °C, respectively. The 3-electrode Model 685.O Precision Etching Coating System II (PECS II). The
measurements are conducted in custom-made stainless-steel cells, electrode was gently handled with metal tweezers and cut with sharp
which are assembled and tested inside an Ar-atmosphere glove box. metal scissors to leave at least one flat-side of ∼5 mm; all metal
The reference electrode (RE) probe is placed between two separa- tools were cleaned using an ethanol and acetone rinse prior to
tors, sandwiched between 20.3 cm2 disks of the oxide-positive and handling the electrode. The cut electrode was mounted onto the
graphite-negative electrodes (Fig. 1): the RE consists of a copper PECS II sample blade (Gatan, Product No. 693.08500) using fast-
wire (25 μm diameter) with a thin layer of Li metal plated on the drying silver paint (Ted Pella, Product No. 16040–30) and adjusted
exposed tip.9 with a micrometer to ensure ∼200 μm overhang of the flat-side of
Electrochemical cycling is typically conducted at 30 °C with a the electrode from the top of the sample blade. Once the electrode
Maccor Model 2400 cycler. The Gr-NP electrode is from a cell that was secure on the blade, the electrode/blade polishing sample was
experienced two C/10 and five C/2 cycles between 3 and 4.2 V at 30 placed onto the cross-section specimen mount (Gatan, Product No.
Journal of The Electrochemical Society, 2021 168 100509

685.08540) and loaded into the PECS II instrument. After the sample with a dwell time of 10 s per point. Line profile data are further
was loaded, the instrument was set to “Etch by Time” with a “Single binned for every 5 μm step.
Modulation” etching condition, a Gun Tilt angle of 0° for both ion
guns, and a rotation rate of 0.5 RPM. The ion milling was performed Focused-ion beam milling.—Samples for nano- to atomic-scale
in two stages: first, a high-voltage milling at 8 kV (with current electron imaging and diffraction were prepared with a ThermoFisher
readings in the range 20–40 μA) that was run for 4 h to create a Scientific Scios 2 FIB-SEM. The electrodes were cut with sharp
rough cross-section; second, a low-voltage milling at 4 kV (with metal scissors and mounted onto a flat sample holder via carbon tape
current readings of 7–15 μA) that ran for 30 min to apply a low-kV and loaded into the instrument. The cross-section samples were
fine polishing. selected from areas near the electrode surface and prepared via
standard focused-ion beam lift-out procedure provided by
Synchrotron powder diffraction.—Powder diffraction data were ThermoFisher Scientific onboard user guidance (included with the
collected at beamline 11-BM of the Advanced Photon Source, instrument software). The samples were mounted onto one of four
Argonne National Laboratory. Coatings from the as-prepared posts of a PELCO FIB lift-out TEM copper grid and thinned to a
electrodes and cycled electrodes, scraped of the copper current final thickness around 100–200 nm.
collectors, were loaded into sealed capillaries for examination. An
X-ray wavelength of 0.41283 Å was used to collect the diffraction Transmission electron microscopy.—Transmission electron mi-
data, over an angular range from –6.0° to 28.0° with a step size of croscopy (TEM) was performed on a Hitachi 9500 TEM equipped
0.001° and a time per step of 0.1 s. with a LaB6 emitter and operated at 300 kV. Image acquisition was
done using the K2 direct electron detector operated in counting
Scanning electron microscopy and energy-dispersive X-ray mode.
spectroscopy.—Both imaging and elemental examinations were
performed using a JEOL 7000 F analytical SEM. The electrode Scanning transmission electron microscopy, energy dispersive
samples were prepared in the following ways: (1) for top-down X-ray spectroscopy, and scanning electron nanodiffraction.—
imaging, a small portion of the electrode is mounted onto a flat Nano- and atomic-scale imaging, chemical analysis, diffraction
sample holder via carbon tape; (2) for cross-sectional imaging, the data were acquired using the ThermoFisher Scientific Titan
samples prepared by the PECS II ion milling are loaded onto a Themis Z STEM equipped with a Schottky field emission gun
standard cross-section specimen mount designed for the SEM. For operated at 300 kV accelerating voltage. The electrode sample was
the SEM imaging, the instrument was set to 7 kV and operated under loaded into a standard double-tilt holder and, during operation, tilted
secondary electron imaging mode with a medium probe current of close to the 〈 100〉 zone axis or the 〈 210〉 zone axis (with respect to
∼0.3 nA. For EDS acquisition, the sample was aligned to an optimal the Graphite-2H phase), as both orientations allowed us to track the
working distance of 10 mm and the probe current was adjusted to a graphite interplanar spacing in both imaging and diffraction modes.
setting of ∼15 nA. The mapping data were acquired using the Imaging was performed at 15–30 pA screen current with a
ThermoFisher NSS 2.0 software using its “Spectral Imaging” mode. convergence angle of 18.0 mrad and acquired using a high-angle
The datasets were summed for 150 frames, each acquired for 10 s annular dark field (HAADF) detector. The EDS acquisition was
per frame with a 50 μs dwell time. The mapping data are reported as performed using the above mentioned imaging conditions via the
counts per element and the quantification was performed by the Velox software (ThermoFisher Scientific) with an acquisition time
software automatically using a built-in database and algorithm based of ∼1 h using a continuous scan with a frame time of ∼1 s. SEND
on the standard phi-rho-z method. The line profile data for the cross- data were collected in microprobe STEM (μP STEM) mode with a
section samples are acquired using the “X-ray Linescans” mode with convergence angle of 0.46 mrad. The SEND probe size is 1.7–-
a single scan with 50 points from the bottom to the top of the anode 1.8 nm at full-width half maximum. The diffraction datasets were
acquired at a camera length of 145 mm over a real-space scan area of

Figure 2. (a) Capacity-potential profiles of the graphite electrode from a NCM532//Gr cell with a reference electrode. The lithiation profiles are at the indicated
rates; the 0.2 C rate delithiation profile shown is after lithiation at the 2 C rate followed by a 1 h rest. (b) Profiles for as-prepared (black) and harvested electrodes
(red); the latter are from a cell subjected to multiple “escalating C-rate” cycling. The as-prepared electrode data are at C/100 rate, whereas the harvested electrode
plots are at both C/100 (solid) and C/10 (dashed) rates. In both (a) and (b), the specific capacities are obtained by dividing the cell capacity (mAh) by the graphite
weight (gGr−1) in the anode.
Journal of The Electrochemical Society, 2021 168 100509

Figure 2b compares plots from a lithium-plated graphite elec-


trode (Gr-LP) harvested from the cell in Fig. S1, with plots from an
as-prepared graphite (Gr-AP) electrode. The lithiation profile of the
Gr-AP electrode (at C/100 rate) displays distinctive plateaus around
210, 117 and 83 mV vs Li/Li+. These plateaus are also very distinct
in the C/10 rate lithiation data (not shown), though they occur at
slightly lower potentials because of polarization arising from
electrode impedance. In contrast, the plateaus are not distinct in
the lithiation profile of the Gr-LP electrode, even at a C/100 rate.
The capacity of this aged-harvested electrode (326 mAh g−1) is also
smaller than that of the as-prepared electrode (357 mAh g−1). The C/
10 lithiation data of this Gr-LP electrode resembles the 2 C graphite-
lithiation data in Fig. 2a; the sloped profile, the barely visible staging
plateaus and the lower capacity all suggest changes in the graphite
particles that alter lithium insertion into the material. Interestingly,
the staging plateaus are visible both in the C/100 and C/10
delithiation profiles of the harvested material, which is consistent
with the distinct differences between the lithium insertion and
extraction processes mentioned earlier.
The theoretical capacity of graphite is 372 mAh g−1 but the Gr-
AP electrode yields only 357 mAh g−1 even at very low rates. The
underlying reason for this smaller capacity is the inherent turbos-
tratic stacking disorder in the commercial graphite, which limits Li+
ion intercalation into the material.31,32 Fast-charging appears to
create additional disorder, further decreasing capacity. Moreover, the
increased structure-disorder increases the number of electronically
and geometrically non-equivalent sites for Li-intercalation causing
the sloping lithiation profile of the Gr-LP even at extremely low
rates.33 In contrast, most of the sites in Gr-AP are thermodynami-
Figure 3. Synchrotron powder diffraction data from the following graphite cally equivalent for Li+ intercalation, thereby producing the
electrodes: (a) as-prepared (Gr-AP) and (b) cycled with lithium plating (Gr- observed staging behavior at low lithiation rates.
LP). The Bragg reflections are indexed according to JCPDS cards 01–071-
–3739 for Graphite-2H, 01–075–2078 for Graphite-3R, and 01–071–3743 for
Ensemble phase analysis via powder X-ray diffraction.—We
LiF. The phases are marked using the following symbols: # for Graphite-2H,
* for Graphite-3R, and ♢ for LiF. evaluated the bulk-scale structure and ensemble lattice spacing of the
graphite anode coatings using synchrotron X-ray diffraction.
Diffraction patterns obtained on the Gr-AP and Gr-LP electrodes
150 nm × 150 nm using 75 × 75 total steps with drift correction are shown in Figs 3a and 3b, respectively. Figure 3a shows that the
performed after each line acquisition. as-prepared electrode contains two distinct graphite phases: the 2H
phase (marked as #) and the 3 R phase (marked as *). The 3 R phase
Results and Discussion is not observed in the Gr-LP electrode pattern (Fig. 3b); we postulate
Electrochemical cycling.—Capacity-potential profiles of the that this phase transforms into the more stable 2H form during
graphite electrode from a NCM523//Gr cell are shown in Fig. 2a. repeated Li+ insertion into and extraction from the graphite, which
The graphite is lithiated and delithiated during cell charge and requires gliding of the graphene planes as mentioned in the
discharge, respectively. At low lithiation rates (<C/5), the inter- Introduction.
calation progresses in distinctive stages seen as plateaus in the plots; We calculated the graphite interplanar d-spacing from the strong
the Li+ ions fill spaces between the carbon sheets forming distinct diffraction peak at ∼1.87 Å–1, which corresponds to the (002)
LixC6 phases (x = 0.22, 0.33, 0.5, 1). At higher lithiation rates reflection for the Graphite-2H phase or the (003) reflection for the
(⩾1 C), the plateaus are not distinct as intercalation occurs under far- Graphite-3R phase. For both Gr-AP and Gr-LP electrode samples we
from-equilibrium conditions. Multiple mixed and strained phases are obtained 〈 d interplanar〉 ≅ 3.356 Å, which indicates that the latter
probably formed so that the profile appears continuous, rather than sample is fully delithiated and that the amount of residual Li+
staged; note that LixC6 phases of various compositions have been ions, within the graphite of the fully discharged cell, is negligible.
reported in the research literature.25–29 During the C/5 delithiation, However, diffraction pattern of the Gr-LP sample (Fig. 3b) shows
staging plateaus are always observed, regardless of the behavior new peaks at ∼2.70 Å–1 and ∼4.40 Å–1, which arise from lithium
during the preceding lithiation cycle (Fig. 2a). We observed this fluoride (LiF) (marked as ♢). The presence of LiF was confirmed by
difference between the lithiation and delithiation behavior during selected area electron diffraction (SAED) data on a FIB cross-
operando XRD studies; therein, we attributed the blurring of phase section prepared from the top surface of the Gr-LP sample (Fig. S2).
boundaries during lithiation to lattice strain and the formation of LiF is known to be a component of the graphite SEI and forms
distinct LixC6 phases during delithiation to abrupt separation from through decomposition reactions of the LiPF6 salt.34,35 There was no
oversaturated solid solutions.18 evidence of other known SEI phases in the Gr-LP pattern, which
As indicated earlier, the cells in this study were subjected to indicates that these compounds are either fully amorphous or lack
multiple applications of the escalating C-rate protocol. Capacity- the long-range order needed to coherently scatter the X-ray beam.
voltage profiles from a representative cell, before and after the
protocol, are shown in Fig. S1. The electrochemical cycling causes a SEM imaging and compositional analysis.—Figure 4 reveals the
significant loss in cell capacity. The Li+ ions are lost to the solid dramatic change to the graphite anode surface after being subjected
electrolyte interphase (SEI) formed during electrolyte reduction on to cycling under the fast-charge, lithium-plating condition. Figure 4a
the graphite electrode and to the dead lithium that accumulates in the shows a photograph of the Gr-AP sample (cut to a 14.2 mm
electrode because of repeated excursions below the Li-plating diameter); the EDS data in Figs. 4b and S3 indicate that the coating
potential (Fig. 1b).22,30 contains only C and F, reflecting the graphite, C45 and PVDF
Journal of The Electrochemical Society, 2021 168 100509

Figure 4. Comparison of the anode surface morphology and composition. The as-prepared graphite (Gr-AP) anode surface details are elucidated via: (a)
photograph at 1x magnification, (b) EDS elemental composition, in weight percent, and (c) SEM image at 2000 × magnification. Similarly, the Gr-LP anode
surface details are elucidated via: (d) photograph at 1x magnification, (e) EDS elemental composition in weight percent, and (f) SEM image at 2000 ×
magnification. Scale bars in (a) and (d): 5 mm.

materials used for electrode preparation. Figure 4c is an SEM image than that of the Gr-AP electrode. Irreversible increases in graphite
of the Gr-AP sample surface at 2k × magnification to illustrate that electrode thickness have been noted previously in electrochemical
the electrode is comprised of densely packed, potato-shaped graphite dilatometry studies; these typically range from 2%–5% after a few
particles up to 10 μm in size. cycles.36–38 EDS scans reveal carbon, oxygen, fluorine and phos-
SEM and EDS data from the Gr-NP electrode are displayed in phorous signals, distributed across the electrode. These elements are
Fig. S4. The graphite particles appear unchanged by the low-rate from the electrolyte decomposition products that accumulate in the
cycles; the elemental analyses reveal the presence of C, O, F and P, electrode.
which arise from electrolyte decomposition products and residues on By contrast, the Gr-LP electrode has an average coating thickness
the electrode. By contrast, Fig. 4d shows a photograph of the Gr-LP of 91.3 μm, which does not include the ∼15–20 μm thick layer
sample (cut to 14.2 mm diameter) with an uneven white-coloration resulting from lithium plating on the anode surface (see Fig. 5d).
across the surface; the average elemental composition (in wt%) of This coating thickness is 19.2% and 16.9% greater than that of the
the electrode surface (Fig. 4e) is 21% O, 49% F, 27% P and 3% C. Gr-AP and Gr-NP electrodes, respectively. We attribute this
Figure 4f shows the surface morphology of the Gr-LP sample at 2k electrode thickness increase to SEI buildup in the electrode pores
× magnification; in this image, there is complete coverage of the and rearrangement of graphite particles during the electrochemical
graphite particles by products formed by reactions of the plated cycling. The process is exacerbated by Li plating and subsequent
lithium with the electrolyte (also see Fig. S5). electrolyte reduction, which increases the amount of SEI. This SEI
To further determine extent of changes, we examined the buildup is also seen on individual graphite particles, on the exterior
electrodes in cross-section to study morphological and elemental surfaces and in the internal pores (Fig. 5e). Moreover, EDS line-
characteristics throughout their depth (see Figs. 5, S6, S7 and S8). scans show that the elemental composition is non-uniform across the
Details of the coating thickness calculations are provided in Section electrode thickness (Figs. 5f and S8). The topmost portion, which
4 of the SI and Fig. S9. Figure 5a shows that the Gr-AP sample has has no graphite, contains mainly oxygen, fluorine and phosphorus
an average coating thickness of 76.6 μm. Figure 5b shows a FIB indicating that the products of lithium plating are mainly inorganic—
cross-section of a typical graphite particle, revealing the presence of for example, LiF and lithium (oxy)fluoro-phosphate compounds.
internal pores. The elemental composition across the electrode Below this portion, in the electrode coating, carbon is the dominant
thickness was measured via EDS line-scans. For the Gr-AP sample, element but we also observe an abundance of fluorine and
the carbon signal, which arises predominantly from the graphite, phosphorus, possibly (oxy)fluoro-phosphate compounds resulting
measures an average of 97 wt% across the depth of the coating layer from hydrolysis reactions of the electrolyte salt.39,40
above the copper current collector, while the fluorine signal, which
arises from PVDF binder, measures an average of 3 wt% (Figs. 5c STEM-EDS compositional analysis.—Using STEM-EDS, we
S6). Cross-sections of the Gr-NP electrode (Fig. S7) reveal an further show that the (oxy)fluoro-phosphate compounds only coat
average coating thickness of 78.1 μm, which is about 2% greater the surface of the graphite particle and do not penetrate the bulk.
Journal of The Electrochemical Society, 2021 168 100509

Figure 5. SEM imaging and elemental mapping of the electrode cross-sections. The as-prepared graphite (Gr-AP) sample details are elucidated via (a) SEM
image of the electrode cross-section, (b) SEM image of an individual particle section, and (c) EDS elemental data acquired across the thickness from bottom of
the current collector to the top of the coating surface—see white arrow in (a). Similarly, the lithium-plated graphite electrode (Gr-LP) sample details are
elucidated via (d) SEM image of the electrode cross-section, (e) SEM image of an individual particle section, and (f) EDS elemental data acquired across the
thickness depth from bottom of the current collector to the top surface—see white arrow in (d). Scale bars: (a), (d) are 25 μm; (b), (e) are 2 μm.

Figure 6a shows a survey of the Gr-LP particle section (prepared by signature, which indicates a lack of (oxy)fluoro-phosphate com-
FIB) with EDS elemental analysis performed in the marked region, pounds in the graphite interior. Figure 6c(ii) shows the spectra
which is magnified in Fig. 6b. corresponding to the whole area in Fig. 6b. The elemental maps
We reveal compositional heterogeneity by comparing the local (Fig. 6d) further confirm the spatial heterogeneity between carbon
EDS spectra from graphite edges (marked as Area #1 and Area #3) and the (oxy)fluoro-phosphates; the C K signal appears only in the
to the graphite center (marked as Area #2) shown in Fig. 6c(i). The graphite bulk, while the O, F, and P signals appear on the graphite
spectra from Area #1 and Area #3 show signals for O, F, and P in band edges. Moreover, the O, F, and P signals appear in the same
addition to the signal from C. However, Area #2 only shows C
Journal of The Electrochemical Society, 2021 168 100509

Figure 6. STEM-EDS of a graphite particle cross-section from the Gr-LP sample. HAADF-STEM images of (a) graphite particle section and (b) a close-up of
the area where the EDS maps were acquired. (c) EDS spectra represented by the following regions: (i) from selected areas 1 (top-edge of graphite edge), 2 (center
of graphite band), and 3 (bottom-edge of graphite band) as marked in (b); (ii) whole-area EDS spectrum. (d) EDS maps of (i) C, (ii) O, (iii) F, and (iv) P signals.
Scale bars: (a) 2 μm, (b) 100 nm, (d) (i)–(iv) 50 nm.

locations across the EDS maps. Additional EDS maps for different The localized Fast-Fourier Transform (FFT) of the Region 1 and
regions of the Gr-LP electrodes are provided in Figs. S10 and S11. Region 2 images are shown in Figs. 7d and 7e, respectively;
The above observations indicate that the SEI only accumulates on information to explain features observed in the HRTEM images is
the graphite particle edges and does not penetrate the bulk material. contained in Section 6 and Fig. S12 of the SI. Briefly, the FFT of a
Electrolyte solvent co-intercalation would cause excessive strain real-space intensity image (with distances x in nm) yields a
and/or damage to the host graphite. We do not see such damage in representative image in the corresponding spatial frequency domain
the bulk graphite and, therefore, do not anticipate extensive (with distances 1/ x in nm–1).42,43 By applying FFT to the image in
electrolyte penetration. We also observe that smaller pores within Fig. 7b, the periodic stripe pattern of the graphite lattice (with
the graphite yield weaker electrolyte residue signatures compared to measured spacing ∼ 0.33 nm ) is mapped to points in Fig. 7d
the larger pores. The electrolyte is known to penetrate the intra- corresponding to the frequency of the lattice spacing (i.e.,
particle pores during electrochemical cycling.41 Larger pores pro- ∼ 1 / 0.33 nm−1 from the center). Moreover, because the intensity
vide easier access to the electrolyte; hence, more SEI would be variation of the fringes lies along the direction normal to the fringes
expected to accumulate in the larger pores than in the smaller pores. in Fig. 7b (specified as direction 1 on the shown axis), the
corresponding spots in the frequency domain of Fig. 7d lie along
Revealing localized disorder via high-resolution TEM.—High- this same direction. The higher frequency spots in Fig. 7d that lie
resolution TEM (HRTEM) data reveal the nature of the turbostratic along the same line (direction 1) correspond to additional harmonics
disorder that manifests in graphite particles at the lattice-level (i.e., n × ∼ 1 / 0.33 nm−1 where n is an integer) of the real-space
(Fig. 7). Figure 7a, spanning a 50 nm × 50 nm field-of-view, shows lattice; observation of such harmonics implies that the lattice is
that the Gr-AP sample has straight and uniform graphite lattice relatively straight and uniform (i.e., ordered) within Region 1. In the
fringes corresponding to the constituent graphene planes. From this specified direction 2 (normal to direction 1), spots are occasionally
image, we can qualitatively assess the degree of uniformity by observed within the lattice fringes (Fig. 7b); these spots likely arise
studying Region 1 (Fig. 7b) and Region 2 (Fig. 7c). The lattice- from atomic columns within the graphite and the distance between
fringe spacing, measured directly from the images of both regions, is them correspond to the spacing between such columns. Similar
∼0.33 nm, which is consistent with the interplanar d-spacing of observations can be made for the Region 2 image (Fig. 7c) and its
graphite determined by XRD. corresponding FFT (Fig. 7e), thereby illustrating spatial uniformity
Fast-Fourier Transform (FFT) analysis is useful for quantifying of the Gr-AP lattice.
geometric and periodic patterns in electron microscopy images.42,43
Journal of The Electrochemical Society, 2021 168 100509

Figure 7. Direct electron imaging of the graphite lattice via HRTEM. The as-prepared graphite (Gr-AP) electrode lattice details are elucidated via (a) TEM
imaging with highlighted Regions 1 and 2; real-space magnification of (b) Region 1 and (c) Region 2; localized Fast-Fourier Transforms of (d) Region 1 and (e)
Region 2. Correspondingly, the Gr-LP lattice details are elucidated via (f) TEM imaging with highlighted Regions 1 and 2; real-space magnification of (g)
Region 1 and (h) Region 2; and localized Fast-Fourier Transforms of (i) Region 1 and (j) Region 2. Scale bars: (a), (f) are 10 nm; (b), (c), (g), (h) are 2 nm; (d),
(e), (i), (j) are 5 nm–1.

By contrast, HRTEM images of the Gr-LP sample (Fig. 7f), show the interplanar d-spacing from Fig. 7g measures up to ∼0.37 nm,
significant nonuniformity. This nonuniformity is illustrated by indicating that the lattice in this region does not return to its original
example from the magnified Region 1 (Fig. 7g) and Region 2 delithiated state. This nanoscopic increase in interplanar spacing is
(Fig. 7h); localized FFTs of Regions 1 and 2 are shown in Figs. 7i not captured in the powder diffraction data (Fig. 3) because the latter
and 7j, respectively. Region 1, which is taken near the surface of an technique provides an average from all portions of the sample. On
internal pore within a graphite particle, is different from both the other hand, Region 2, which is in the bulk of the Gr-LP sample,
Regions 1 and 2 of the Gr-AP sample. From Fig. 7g, we find that shows straight, uniform lattice fringes (Fig. 7h) similar to those
the Gr-LP lattice of Region 1 appears wavy; based on Fig. 7f, we observed in the Gr-AP sample (Figs. 7b, 7c). The corresponding FFT
estimate that this waviness extends about 20 nm into the graphite of Region 2 (Fig. 7j) has spots along direction 1 and a high-
bulk. The FFT of Region 1 (Fig. 7i) confirms lattice disorder in the frequency harmonic, which suggests greater ordering within the
following ways: (i) arcs are found instead of spots along direction 1; graphite bulk. Taken together, the HRTEM images and FFT results
(ii) the high frequency harmonic spot (marked by the white arrow) is indicate there is greater disorder in the graphite lattice near the
relatively faint compared to those in the Gr-AP sample. Furthermore,
Journal of The Electrochemical Society, 2021 168 100509

surface of its internal pores and that this disorder can extend up to disorder. The average rotation gradient in this region is ̃0.30 °/ nm,
∼20 nm into the particle bulk. while the particle edges have values > 4 °/ nm (Fig. 9c(i)). Moreover,
the frequency of lattice rotation (i.e., the “wavy” nature of the
Quantification of nanoscopic turbostratic disorder via SEND.— lattice) occurs up to ∼20 nm into the graphite bulk from the pore
In Fig. 8 and 9, we quantify the nanoscale heterogeneity of the edge. A similar behavior is observed in Fig. 9c(ii), which shows
graphite by measuring the interplanar d-spacing and local lattice elevated d-spacings near the pore edge and extending into the
rotation using SEND.44 In this technique, a ∼2 nm probe (at full- particle bulk. Figure 9c(iii) shows that the mean d-spacing of Region
width half-maximum) is rastered across a 150 nm × 150 nm real- 2 is 0.35 nm, which is larger than that calculated from the X-ray
space area generating an electron diffraction pattern for each probe data. The d-spacing distribution also shows a tail that extends from
position. SEND provides position-resolved structural information ∼0.36 nm to 0.40 nm, which indicates that the graphite edges in
regarding turbostratic disorder in the materials. Details of the SEND Region 2 have sustained greater damage than those in Region 1. This
data acquisition are briefly described in the experimental section; result also points to the spatial heterogeneity in lattice disorder
data analysis information is provided in Section 7 of the SI, which within each graphite particle—the edges with greater exposure to the
includes Fig. S13. In the SEND experiments, we quantify both the electrolyte will experience more Li+ ion (de)intercalation reactions
graphite interplanar spacing and the lattice rotation by tracking and apparently greater damage, especially during high-rate charging.
changes to the Ginterplanar vector (with respect to the Graphite-2H Damage localized at the graphite particle edges is not entirely
phase) for each real-space sample coordinate of the SEND scan. The unexpected.45 Prior studies have pointed out that Li+ ion intercala-
quantification is done by determining (1) changes to the length of the tion into graphite occurs primarily through the edge planes.46,47 At
vector ∣ Ginterplanar ∣ = 1 d interplanar to measure position-to-position high charge rates, significant build-up of Li+ ions is expected at the
nanoscopic interplanar d-spacing, which is then represented as a edges planes; the concentration gradients between the surface and
d interplanar map to highlight local d-spacing changes; (2) changes to bulk sites could induce stresses that are significant enough to deform
the in-plane orientation θ of each vector Ginterplanar , which is then the graphite lattice near the particle edges.
Mitigating damage at the graphite particle edges is crucial to
represented as a map of ∣∇θ ∣ to identify locations where rotational prolonging the life of LIB cells subjected to fast charge. One
changes in the lattice occur. solution would be to raise the lower cutoff voltage (LCV) for cell
Figure 8a(i) contains HAADF-STEM image from a Gr-AP cycling (e.g. from 2.8 to 3.3 V) to prevent complete delithiation of
particle cross-section, with marked Region 1 (magnified in the graphite. The residual dilute LixC6 phases have higher d-spacings
Fig. 8a(ii)) and Region 2 (magnified in Fig. 8a(iii)) where SEND than the pristine graphite,29 which would facilitate Li+ ion inter-
data were acquired. First, we discuss the nanoscale structure of calation and (partially) mitigate the stresses induced during the
Region 1 in Fig. 8b. The data indicate that the lattice rotation shows process. Alternative solutions could include the use of graphite with
negligible variation (∼ 0.09° nm−1) across the sample (Fig. 8b(i)) and higher d-spacings in the anode, as has been suggested for both Li+
that the interplanar d-spacing map across Region 1 is relatively and Na+ ion cells.48,49 However, such “expanded” graphite materials
uniform (Fig. 8b(ii)). Figure 8b(iii) is a plot of the overall d-spacing typically have lower electronic conductivity, which would limit the
from the map as a histogram; the data show a distribution of d- overall rate of intercalation. In addition, these materials have higher
spacings in the sample with a mean of 0.334 nm and a standard surface areas and a proclivity to trap more lithium in the SEI formed
deviation of 0.006 nm. Figure 8c illustrates that Region 2 is during electrolyte reduction reactions. Other solutions proposed to
comparable to Region 1 in terms of both lattice rotation facilitate rapid cycling include the use of functionalized-graphite that
(Fig. 8c(i)) and interplanar spacings (Figs. 8c(ii) and (iii)). have higher d-spacings at the particle edges,50 graphite with particle
However, we find a slightly elevated value for rotational variation edges aligned perpendicular to the current collector,51 graphite with
in the particle bulk of Fig. 8c(i). This observation points to an carbon coatings that have larger interlamellar spacing,52 and anodes
important characteristic of graphite, which is that the graphene containing a mix of graphite and hard carbons.7 These efforts, along
planes tend to group as bands of 10–20 nm thickness (Fig. S14). with research on cathode and electrolyte materials, electrode
Within each graphite band, the magnitude of the rotational variation architectures and cycling protocols, are important steps towards
is negligible; between band-to-band, the in-plane lattice angle can making fast-charging of LIB cells, a commercial reality.
exhibit sudden jumps. Overall, though, the results for the Gr-AP
sample are consistent with the HRTEM observations, which also Conclusions
indicated a uniform lattice. Furthermore, SEND data from the Gr-NP
electrode (Fig. S15) are very similar to those from the Gr-AP Fast charging of LIB cells can degrade their energy and power
electrode, which indicate that the low-rate cycles have negligible performance characteristics. Here we show representative data from
impact on the graphite particle structure. as-prepared graphite electrodes (Gr-AP) and from electrodes ex-
By contrast, a non-uniform lattice and an increase in lattice tracted from (discharged) cells that were previously charged at rates
disorder at the particle edges are observed in particles from the Gr- up to 6 C (Gr-LP). Electrochemical measurements show that the
LP electrode. Figure 9a(i) contains a HAADF-STEM image with harvested electrodes have lower capacities than the Gr-AP electrode
areas marked Region 1 (magnified in Fig. 9a(ii)) and Region 2 and display a sloping lithiation profile, rather than a staged profile,
(magnified in Fig. 9a(iii)) where SEND data were collected. In even at low rates. Analysis of X-ray diffraction data from the Gr-AP
Fig. 9b(i), we show position-to-position changes of the in-plane and Gr-LP electrodes yield similar interplanar d-spacings, indicating
rotation angle of Ginterplanar in Region 1, which is near a relatively that the graphite bulk is not altered by the electrochemical cycling.
thin pore region. The graphite bulk in Region 1 has a uniform lattice Similar conclusions about the graphite particle bulk can be drawn
from the analysis of HREM and HAADF-STEM images and SEND
rotation (∼ 0.14 ° nm−1), while the graphite edge has greater disorder
data. However, these nanoscale data clearly indicate damage at the
(with rotational changes > 2° nm−1). Figure 9b(ii) shows the corre- graphite edges, which extend up to ∼20 nm into the particle bulk for
sponding interplanar d-spacing reconstruction map; the d-spacing is the Gr-LP sample. This damage is seen as arcs (rather than spots) in
elevated (> 0.37 nm ) only within ∼2 nm of the particle edge while the FFT data of HREM images and as increased rotations and higher
the bulk has d-spacings similar to those in the Gr-AP sample. The d- d-spacings in the SEND data; both datasets suggest increased
spacing distribution (Fig. 9b(iii)) shows a mean around 0.337 nm turbostratic disorder and buckling of the graphene planes at particle
and a standard deviation of 0.008 nm, only slightly larger than those edges. In addition to changes in graphite-edge structure, the Gr-LP
for the Gr-AP sample. electrode shows a significant thickness increase from the build-up of
On the other hand, Region 2 (acquired near the edge of a larger SEI and particle rearrangements that occur during the cycling. This
graphite pore) of the Gr-LP sample (Fig. 9c) shows greater lattice SEI contains a mix of compounds, including LiF (seen in the XRD
Journal of The Electrochemical Society, 2021 168 100509

Figure 8. Quantitative lattice rotation and interplanar d-spacing analysis of the as-prepared graphite (Gr-AP) sample. (a) HAADF-STEM images (i) at low-
magnification, showing Regions 1 and 2 where SEND acquisition was performed; (ii), (iii) magnified images of Regions 1 and 2, respectively. Diffractive image
reconstructions and analysis of (b) Region 1 and (c) Region 2 containing the following: (i) in-plane rotation gradient map, (ii) interplanar spacing map, (iii)
interplanar spacing distribution. Scale bars: (a) (i) is 2 μm; (a) (ii), (a) (iii), (b) (i), (b) (ii), (c) (i), (c) (ii) are 50 nm.

patterns) and (oxy)fluoro-phosphates suggested by the EDS data. internal pores of the graphite, which likely have easier access to the
Larger quantities of this SEI are present adjacent to the larger electrolyte than the smaller pores. Furthermore, the SEI is localized
Journal of The Electrochemical Society, 2021 168 100509

Figure 9. Quantitative lattice rotation and interplanar d-spacing analysis of the fast-charged graphite (Gr-LP) electrode. (a) HAADF-STEM images (i) at low-
magnification showing Regions 1 and 2 where SEND data were acquired; (ii), (iii) magnified images of Regions 1 and Region 2. Diffractive image
reconstructions and analysis of (b) Region 1 and (c) Region 2 containing the following: (i) in-plane rotation gradient map, (ii) interplanar spacing map, (iii)
interplanar spacing distribution. Scale bars: (a) (i) is 2 μm; (a) (ii), (a) (iii), (b) (i), (b) (ii), (c) (i), (c) (ii) are 50 nm.

at the particle edges and does not penetrate the graphite bulk, suggesting that only Li+ ions (no electrolyte components) intercalate
into the particles (effective desolvation of ions).
Journal of The Electrochemical Society, 2021 168 100509

The implication of our study is that even a small number of high- 10. I. A. Shkrob, M.-T. F. Rodrigues, D. W. Dees, and D. P. Abraham, J. Electrochem.
rate cycles appears to be sufficient to induce significant and Soc., 166, A3305 (2019).
11. M.-T. F. Rodrigues, I. A. Shkrob, A. M. Colclasure, and D. P. Abraham,
permanent disorder in graphite domains that are closest to the J. Electrochem. Soc., 167, 130508 (2020).
electrolyte, be it at particle surfaces or the inner pores. Such disorder 12. T. Waldmann, B.-I. Hogg, and M. Wohlfahrt-Mehrens, J. Power Sources, 384, 107
could affect the lithiation kinetics, preventing the graphite particles (2018).
from accepting charge and favoring the occurrence of Li plating. 13. L. Somerville, J. Bareño, S. Trask, P. Jennings, A. McGordon, C. Lyness, and
I. Bloom, J. Power Sources, 335, 189 (2016).
Moreover, we show that these structural changes are highly non- 14. J. Cannarella and C. B. Arnold, J. Electrochem. Soc., 162, A1365 (2015).
uniform across the various pores within a same particle, which could 15. J. S. Okasinski, I. A. Shkrob, A. Chuang, M.-T. F. Rodrigues, A. Raj, D. W. Dees,
either be a cause or consequence of reaction heterogeneities and D. P. Abraham, Phys. Chem. Chem. Phys., 22, 21977 (2020).
commonly observed at high rates. Awareness and mitigation of the 16. S. Müller, J. Eller, M. Ebner, C. Burns, J. Dahn, and V. Wood, J. Electrochem. Soc.,
165, A339 (2018).
structural evolution at the graphite/electrolyte interfaces is crucial to 17. A. Mukhopadhyay and B. W. Sheldon, Prog. Mater Sci., 63, 58 (2014).
the development of LIBs that can endure repeated fast-charging and 18. A. Raj, I. A. Shkrob, J. S. Okasinski, M.-T. Fonseca Rodrigues, A. C. Chuang,
yet reliably deliver high-performance over the 10 + year lifespan of X. Huang, and D. P. Abraham, J. Power Sources, 484, 229247 (2021).
electric vehicles. 19. K. P. C. Yao, J. S. Okasinski, K. Kalaga, I. A. Shkrob, and D. P. Abraham, Energy
Environ. Sci., 12, 656 (2019).
20. M. T. F. Rodrigues, S.-B. Son, A. M. Colclasure, I. A. Shkrob, S. E. Trask, I.
Acknowledgments D. Bloom, and D. P. Abraham, ACS Appl. Energy Mater., 4, 1063 (2021).
21. M.-T. Fonseca Rodrigues, V. A. Maroni, D. J. Gosztola, K. P. C. Yao, K. Kalaga, I.
SP acknowledges support from the U.S. Department of Energy A. Shkrob, and D. P. Abraham, ACS Appl. Energy Mater., 2, 873 (2019).
(DOE) Graduate Student Research (SCGSR) program. The SCGSR 22. P. P. Paul et al., , Adv. Energy Mater., 11, 2100372 (2021).
program is administered by the Oak Ridge Institute for Science and 23. H. Shi, J. Barker, M. Y. Saïdi, R. Koksbang, and L. Morris, J. Power Sources, 68,
Education for DOE under contract number DE‐SC0014664. This 291 (1997).
24. A. Van der Ven, Z. Deng, S. Banerjee, and S. P. Ong, Chem. Rev., 120, 6977
work was carried out in part at the Materials Research Laboratory (2020).
Central Research Facilities, University of Illinois. DA and MTFR 25. H. Fujimoto, H. Kiuchi, S. Takagi, K. Shimoda, K. Okazaki, Z. Ogumi, and T. Abe,
acknowledge support from DOE’s Vehicle Technologies Office J. Electrochem. Soc., 168, 40509 (2021).
(VTO). The electrodes used in this article were fabricated at 26. H. He, C. Huang, C.-W. Luo, J.-J. Liu, and Z.-S. Chao, Electrochim. Acta, 92, 148
(2013).
Argonne’s Cell Analysis, Modeling and Prototyping (CAMP) 27. S. Schweidler, L. de Biasi, A. Schiele, P. Hartmann, T. Brezesinski, and J. Janek,
Facility, which is fully supported by the VTO. We are grateful to J. Phys. Chem. C, 122, 8829 (2018).
our many colleagues, especially Stephen Trask and Andrew Jansen 28. J. Wilhelm, S. Seidlmayer, S. Erhard, M. Hofmann, R. Gilles, and A. Jossen,
at Argonne. Use of the Advanced Photon Source (APS) at Argonne J. Electrochem. Soc., 165, A1846 (2018).
29. E. Hazrati, G. A. de Wijs, and G. Brocks, Phys. Rev. B, 90, 155448 (2014).
National Laboratory was supported by the U.S. Department of 30. S.-B. Son, D. Robertson, Z. Yang, Y. Tsai, S. Lopykinski, and I. Bloom,
Energy, Office of Science, Office of Basic Energy Sciences, under J. Electrochem. Soc., 167, 140506 (2020).
Contract No. DE-AC02–06CH11357. We are grateful to Saul 31. R. E. Franklin, Acta Crystallogr., 4, 253 (1951).
Lapidus at the APS for help with the high-resolution powder 32. J. R. Dahn, A. K. Sleigh, H. Shi, J. N. Reimers, Q. Zhong, and B. M. Way,
Electrochim. Acta, 38, 1179 (1993).
diffraction experiments. The submitted manuscript has been created 33. J. R. Dahn, R. Fong, and M. J. Spoon, Phys. Rev. B, 42, 6424 (1990).
by UChicago Argonne, LLC, Operator of Argonne National 34. P. Verma, P. Maire, and P. Novák, Electrochim. Acta, 55, 6332 (2010).
Laboratory (“Argonne”). Argonne, a U.S. Department of Energy 35. A. M. Andersson and K. Edström, J. Electrochem. Soc., 148, A1100 (2001).
Office of Science laboratory, is operated under Contract No. DE- 36. D. Sauerteig, S. Ivanov, H. Reinshagen, and A. Bund, J. Power Sources, 342, 939
(2017).
AC02–06CH11357. The U.S. Government retains for itself, and 37. B. Rieger, S. Schlueter, S. V. Erhard, J. Schmalz, G. Reinhart, and A. Jossen,
others acting on its behalf, a paid-up nonexclusive, irrevocable J. Energy Storage, 6, 213 (2016).
worldwide license in said article to reproduce, prepare derivative 38. A. Y. R. Prado, M.-T. F. Rodrigues, S. E. Trask, L. Shaw, and D. P. Abraham,
works, distribute copies to the public, and perform publicly and J. Electrochem. Soc., 167, 160551 (2020).
39. R. Wagner, M. Korth, B. Streipert, J. Kasnatscheew, D. R. Gallus, S. Brox,
display publicly, by or on behalf of the Government. M. Amereller, I. Cekic-Laskovic, and M. Winter, ACS Appl. Mater. Interfaces, 8,
30871 (2016).
ORCID 40. M. Stich, M. Göttlinger, M. Kurniawan, U. Schmidt, and A. Bund, J. Phys. Chem.
C, 122, 8836 (2018).
Saran Pidaparthy https://orcid.org/0000-0002-0783-3094 41. D. J. Miller, C. Proff, J. G. Wen, D. P. Abraham, and J. Bareño, Adv. Energy
Marco-Tulio F. Rodrigues https://orcid.org/0000-0003-0833-6556 Mater., 3, 1098 (2013).
Jian-Min Zuo https://orcid.org/0000-0002-5151-3370 42. A. Marion, , ed. A. Marion (Springer, US, Boston, MA) p. 255 (1991).
43. R. C. Gonzalez and R. E. Richard E. Woods, Digital image processing 2nd ed.
Daniel P. Abraham https://orcid.org/0000-0003-0402-9620 (Prentice Hall, Upper Saddle River, N.J) (2002).
44. J. M. Zuo and J. C. H. Spence, Advanced Transmission Electron Microscopy
References (Springer, New York, NY) (2017).
45. V. A. Sethuraman, L. J. Hardwick, V. Srinivasan, and R. Kostecki, J. Power
1. A. Masias, J. Marcicki, and W. A. Paxton, ACS Energy Lett., 6, 621 (2021). Sources, 195, 3655 (2010).
2. A. Raj, M.-T. F. Rodrigues, and D. P. Abraham, J. Electrochem. Soc., 167, 120517 46. R. J. Rice and R. L. McCreery, Anal. Chem., 61, 1637 (1989).
(2020). 47. K. Persson, V. A. Sethuraman, L. J. Hardwick, Y. Hinuma, Y. S. Meng, A. van der
3. A. Tomaszewska et al., eTransportation, 1, 100011 (2019). Ven, V. Srinivasan, R. Kostecki, and G. Ceder, J. Phys. Chem. Lett., 1, 1176 (2010).
4. S. Ahmed et al., J. Power Sources, 367, 250 (2017). 48. T.-H. Kim, E. K. Jeon, Y. Ko, B. Y. Jang, B.-S. Kim, and H.-K. Song, J. Mater.
5. A. M. Colclasure et al., Electrochim. Acta, 337, 135854 (2020). Chem. A, 2, 7600 (2014).
6. X.-G. Yang, T. Liu, Y. Gao, S. Ge, Y. Leng, D. Wang, and C.-Y. Wang, Joule, 3, 49. Y. Wen, K. He, Y. Zhu, F. Han, Y. Xu, I. Matsuda, Y. Ishii, J. Cumings, and
3002 (2019). C. Wang, Nat. Commun., 5, 4033 (2014).
7. K.-H. Chen, V. Goel, M. J. Namkoong, M. Wied, S. Müller, V. Wood, J. Sakamoto, 50. J.-S. Park, M.-H. Lee, I.-Y. Jeon, H.-S. Park, J.-B. Baek, and H.-K. Song, ACS
K. Thornton, and N. P. Dasgupta, Adv. Energy Mater., 11, 2003336 (2021). Nano, 6, 10770 (2012).
8. D. W. Dees, M.-T. F. Rodrigues, K. Kalaga, S. E. Trask, I. A. Shkrob, D. 51. J. Billaud, F. Bouville, T. Magrini, C. Villevieille, and A. R. Studart, Nat. Energy,
P. Abraham, and A. N. Jansen, J. Electrochem. Soc., 167, 120528 (2020). 1, 16097 (2016).
9. M.-T. F. Rodrigues, K. Kalaga, S. E. Trask, D. W. Dees, I. A. Shkrob, and D. 52. W. Cai, C. Yan, Y.-X. Yao, L. Xu, R. Xu, L.-L. Jiang, J.-Q. Huang, and Q. Zhang,
P. Abraham, J. Electrochem. Soc., 166, A996 (2019). Small Struct., 1, 2000010 (2020).

You might also like