XCEL:

eXtreme Fast Charge

Cell Evaluation of

Lithium-ion Batteries

October 2019 Report Period: July–September 2019

2

Table of Contents

Project Overview ......................................................................................................... 3

XCEL R&D: CAMP, Testing, Post-test Characterization, and Modeling

(Argonne National Laboratory) ....................................................................................................... 4

XCEL R&D: Extreme Fast Charging (National Renewable Energy Laboratory) .................................. 8

XCEL R&D: Electrochemical, Atomistic, and Techno-Economic (BatPaC)

(Argonne National Laboratory) ..................................................................................................... 16

XCEL R&D: Performance Characterization and Post-Test Analysis

(Argonne National Laboratory) ..................................................................................................... 25

XCEL R&D: Extreme Fast Charging R&D: Battery Testing Activities

(Idaho National Laboratory) .......................................................................................................... 31

XCEL R&D: SLAC National Accelerator Laboratory ......................................................................... 33

XCEL R&D: SLAC National Accelerator Laboratory ......................................................................... 38

XCEL R&D: Detecting Lithium Plating during Fast Charging, and Graphite Anodes

with Directional Pore Channels Made by Freeze Tape Casting

(Lawrence Berkeley National Laboratory).................................................................................... 40

3

Project Overview

Venkat Srinivasan (Argonne National Laboratory), and Samuel Gillard (U.S. Department of Energy)

The U.S. Department of Energy’s Office of Energy Efficiency and Renewable Energy (DOE-EERE) has

identified fast charge — with a goal of 15-min recharge time — as a critical challenge to pursue in ensuring

mass adoption of electric vehicles (EVs). Present-day, high-energy cells with graphite anodes and transition

metal cathodes in a liquid electrolyte are unable to achieve this metric without negatively affecting battery

performance. There are numerous challenges that limit such extreme fast charging (XFC) at the cell level,

including lithium (Li) plating, rapid temperature rise, and possible particle cracking. Of these, Li plating is

thought to be the primary culprit. This project aims to gain an understanding of the main limitations during fast

charge using a combined approach involving cell builds, tests under various conditions, characterization, and

continuum-scale mathematical modeling. Researchers from five national laboratories are contributing their

expertise to make progress on the eXtreme Fast Charge Cell Evaluation of Lithium-ion Batteries (XCEL)

project.

Cells are built at the Cell Analysis, Modeling, and Prototyping (CAMP) Facility at Argonne National

Laboratory (Argonne) using various carbons and different cell designs, in both half-cell and full-cell

configurations and with reference electrodes. Cells are tested at both Idaho National Laboratory (INL) and

Argonne under various operating conditions (e.g., C-rate, temperature) and under different charging protocols

with the aim of identifying the onset of plating, quantifying the extent of the problem, and determining

parameters and test data for mathematical models. After testing, cells are opened, and various advanced

characterizations are performed at Argonne to determine the extent of plating and to determine whether other

failure models, such as particle cracking, also play a role.

A critical part of the project is the use of continuum-scale mathematical models so XCEL participants can

understand the limitations at high charge rates and, therefore, suggest possible solutions to pursue. Both

macro-scale approaches and microstructure-based simulations are pursued and serve to complement each

other. Macromodeling at the National Renewable Energy Laboratory (NREL) is used to test cell designs,

accompanied by development of microstructure models to provide deeper insights into the electrochemical

phenomena in the battery. This effort is complemented with development of models incorporating new

physics, such as phase change and solid-electrolyte interphase growth, at Argonne.

Two exploratory projects aim to study ways to detect Li in situ during operation. NREL is pursuing the use of

microcalorimetry to detect heat signatures during plating. INL is working with Princeton University to

examine the use of acoustic methods to determine whether plating leads to a signature in the acoustic signal.

Finally, the Stanford Linear Accelerating Center (SLAC) National Acceleratory Laboratory is using

synchrotron X-ray methods to guide the cell design and charging protocols of XFC of Li-ion battery cells, and

Lawrence Berkeley National Laboratory (LBNL) is investigating the initial onset of Li plating during fast

charging as well as developing a strategy to detect it.

4

XCEL R&D: CAMP, Testing, Post-test Characterization, and Modeling

(Argonne National Laboratory)

Contributors: Alison Dunlop, Andrew Jansen, Dave Kim, Bryant Polzin, and Steve Trask

Introduction

The Cell, Analysis, Modeling and Prototyping (CAMP) Facility continues to support the XCEL Program’s

objective of identifying and mitigating causes of lithium plating at fast charge (>4C) in single-layer pouch

cells. Efforts in fiscal year 2018 (FY18) demonstrated that the choice in graphite did not significantly affect the

ability to cycle under fast charge conditions (6C charge, C/2 discharge) — all six of the selected graphites were

able to achieve 750 cycles with 80% capacity retention, using a 2-mAh/cm² graphite loading. The team made

the decision to use the SLC1506T graphite from Superior Graphite as the baseline graphite material and

NMC532 as the baseline cathode material. More than 70 single-sided, single-layer pouch cells were fabricated

in the Round 1 cell build using a 2-mAh/cm² graphite loading and were delivered to lab partners (INL,

Argonne, and NREL) for fast charge testing with a nominal capacity of 19 mAh capacity. Prescreening of

anode-cathode pairs with varying electrode capacity loading indicated that loadings over ~2.5 mAh/cm² were

not able to charge at a true 6C rate. Thus, the next pouch cell build was designed with a graphite loading of 3.0

mAh/cm². A total of 48 of these Round 2 pouch cells were delivered to Argonne, INL, and NREL for testing.

These two cell builds are now considered the baselines for the XCEL Program.

Objectives

The goal of the FY19 work is to explore methods of preventing lithium plating via modifications of the

electrode architecture. Ideally, the negative and positive electrodes should have low tortuosity to enable fast

lithium ion transport to and from the active material closest to the current collector, while maintaining low

porosity to maintain high energy density. In addition, the CAMP Facility will support the DOE-EERE Vehicle

Technologies Office Funding Opportunity Announcement and Lab Call projects and teams that are focused on

developing methods of detecting lithium plating.

Approach

In FY18, the CAMP Facility developed two single-layer pouch cells to benchmark the fast charge capabilities

of a typical lithium-ion battery, with the main difference between the cells being electrode loading. These two

electrode/cell designs will be used as the baseline designs for this program in FY19. The details of the two cell

designs are as follows:

Round 1 Pouch Cells were assembled with 14.1 cm² single-sided cathodes (0.145 grams of NMC532 per pouch

cell) and 14.9 cm² single-sided graphite anodes (SLC1506T from Superior Graphite) using the Celgard 2320

separator (20 µm, PP/PE/PP) and 0.5 mL of Tomiyama 1.2 M lithium hexafluorophosphate (LiPF6) in an

EC:EMC (3:7 wt%) “Gen2” electrolyte. The n:p ratio is between 1.12 to 1.22 for this voltage window (3.0 to

4.1 V). After assembly, the pouch cells underwent formation cycles at ~4 psi in the 3.0 to 4.1 V window as

follows: 1.5 V tap charge and hold for 15 minutes, followed by a 12-hour rest, and then three cycles at C/10,

followed by three cycles at C/2. The cells were then brought to a safe state of charge by constant voltage

charging to 3.5 V for 6 hours, and then degassed and prepared for shipping/delivery to the battery test labs. A

nominal C/3 capacity of 19 mAh was recommended for future tests.

Round 2 Pouch Cells were assembled with 14.1 cm² single-sided cathodes (0.236 grams of NMC532 per pouch

cell) and 14.9 cm² single-sided graphite anodes (SLC1506T from Superior Graphite) using the Celgard 2320

5

separator (20 µm, PP/PE/PP) and 0.615 mL of Tomiyama 1.2 M LiPF6 in an EC:EMC (3:7 wt%) “Gen2”

electrolyte for an electrolyte-to-pore volume factor of 4.20. The n:p ratio is between 1.07 to 1.16 for this

voltage window (3.0 to 4.1 V). Figure 1 details the electrode composition and design parameters for the

Round 2 electrodes. The formation process was the same as that used in making the Round 1 cells. A nominal

C/2 capacity of 32 mAh was recommended for future tests.

Figure 1. Electrode composition and design parameters for Round 2-Batch 1 pouch cell design. The Round 1 electrodes had the same composition but had less mass loading (and thickness).

Results

1. Fabrication of Transparent Pouch Cells for XRD Studies

At the last F2F meeting at INL in July, the CAMP Facility participants mentioned to the PIs that they have the

capability to make a pouch cell with a polypropylene (PP) window. Colleagues from UC-Berkeley were very

interested in having a transparent pouch cell for their X-ray diffraction (XRD) experiments. They are

developing a technique to measure temperature in the anodes (thermal expansion of copper [Cu]) and the

cathodes (thermal expansion of aluminum [Al]) separately during fast charge. Thus, they cannot have a pouch

material with any blocking metal, such as aluminum. A further suggestion was made by UC-Berkeley to use

the Kapak SealPAK pouch material, which is a polyester/polyethylene (PET/PE) laminate that can be heat-

sealed. The CAMP Facility developed sealing processes for both materials (PP and PET/PE) that still

maintained an adequate tab seal. Two pouch cell designs were developed (Figure 2):

• Fully transparent – PET/PE (Kapak SealPAK 400), and

• Windowed – PP window heat-sealed on layered pouch material.

The cells were sent to UC-Berkeley without any electrolyte. In the interest of time, the CAMP Facility did not

run any trial cycling with these cell designs. It is safe to say that these pouch cells should last several hours to

several days before drying out (or springing a leak). UC-Berkeley will vacuum-impregnate the cells with

electrolyte, complete a heat seal, and apply an abbreviated formation cycle right before the cell is placed in the

XRD setup. It is planned that the cells will be placed in a protective Kapak pouch envelope after heat-sealing

to minimize the rate of permeation and catch any leakage. If successful, these pouch cell designs may find use

in other experiments that require metal-free cell containment.

Cathode: LN3107-189-3 90 wt% Toda NMC532 5 wt% Timcal C45 5 wt% Solvay 5130 PVDF

Matched for 4.1V full cell cycling Prod:NCM-04ST, Lot#:7720301 Single-sided coating, CFF-B36 cathode

Al Foil Thickness: 20 µm Al Foil Loading: 5.39 mg/cm2 Total Electrode Thickness: 91 µm Coating Thickness: 71 µm Porosity: 35.4 % Total Coating Loading: 18.63 mg/cm2 Total Coating Density: 2.62 g/cm3

Made by CAMP Facility

Anode: LN3107-190-4A 91.83 wt% Superior Graphite SLC1506T 2 wt% Timcal C45 carbon 6 wt% Kureha 9300 PVDF Binder 0.17 wt% Oxalic Acid

Lot#: 573-824, received 03/11/2016 Single-sided coating, CFF-B36 anode

Cu Foil Thickness: 10 µm Total Electrode Thickness: 80 µm Total Coating Thickness: 70 µm Porosity: 34.5 % Total SS Coating Loading: 9.94 mg/cm2 Total SS Coating Density: 1.42 g/cm3

Made by CAMP Facility

6

Figure 2. Photo of the fully transparent (left, PET/PE [Kapak SealPAK 400]) and the windowed (right, PP window heat-sealed on layered pouch material) pouch cells

designed for brief XRD experiments.

2. Support of Electrolyte Development at INL

Lithium plating can be minimized with the use of electrolytes with high ionic conductivity. INL is exploring

several electrolyte formulations that are predicted to have ionic conductivity that is better than that of the

baseline Gen2 electrolyte. INL is currently scoping out these electrolytes in coin cells, with the plan that the

electrolytes that pass the initial tests will be explored further in pouch cells. The CAMP Facility has supported

this effort in the first quarter by fabricating 11 Round 2 pouch cells and shipping them to INL without

electrolyte added. INL will add its candidate electrolyte as needed. INL requested 18 more dry-pouch cells in

the fourth quarter to continue its studies in electrolytes. The CAMP Facility made 18 Single-Layer Pouch Cells

with the Round 2, Batch 2 Fast Charge electrodes (LN3174-73-4 [NMC 532] | LN3174-75-3 [SLC 1506T])

and shipped them to INL in August.

3. Support of Lithium Plating Detection Teams – SLAC/Stanford University

The CAMP Facility continued its support of the lithium-plating detection activities at SLAC and Stanford

University by supplying several electrode sheets, and fabricating/cycling several pouch cells in time for their

beamtime experiments at Argonne’s Advanced Photon Source (APS) in July. A few pouch cells required

opening to harvest the electrodes. A list of items supplied in this quarter include:

▪ Four sheets of Round 1 positive and 2 sheets of Round 1 negative electrodes

▪ Two sheets each of Round 2 positive and negative electrodes

▪ One single-layer pouch cell xx3450 with the Round 2 material (NMC 532 | SLC 1506T) filled with

Gen2 electrolyte and tap-charged

4. Support of Lithium Plating Detection Teams – Argonne

The CAMP Facility supported the lithium-plating detection activities at Argonne by fabricating and forming

two Round 1 pouch cells before APS beamtime scheduled for the start of October 2019.

Conclusions

The CAMP Facility successfully designed and fabricated pouch cells with no metal in the pouch material for

brief XRD experiments at UC-Berkeley. An additional 18 dry, single-layer Round 2 pouch cells were

7

fabricated to support INL’s electrolyte development effort. As in the previous quarter, the CAMP Facility

continued to support several fast charge teams this quarter by supplying requested powders, electrode sheets,

and pouch cells. Many of the cell builds were fabricated “just-in-time” for beamtime experiments at Argonne’s

APS. Technical data and electrochemical results were provided to all team members as needed to aid in their

experiments and modeling efforts.

Key Publications

T. Tanim, E. Dufek, M. Evans, C. Dickerson, A. Jansen, B. Polzin, A. Dunlop, S. Trask, R. Jackman, I.

Bloom, Z. Yang, and E. Lee, “Extreme Fast Charge Challenges for Lithium-ion Battery: Variability and

Positive Electrode Issues,” accepted in J. Electrochem. Soc.

A. M. Colclasure, A. R. Dunlop, S. E. Trask, B. J. Polzin, A. N. Jansen, and K. Smith, “Requirements for

Enabling Extreme Fast Charging of High Energy Density Li-Ion Cells while Avoiding Lithium Plating,” J.

Electrochem. Soc. 166 (8) A1412–A1424 (2019).

Acknowledgments

Key contributors to this work include: Alison Dunlop, Andrew Jansen, Dave Kim, Bryant Polzin, and Steve

Trask (all from Argonne National Laboratory).

8

XCEL R&D: Extreme Fast Charging (National Renewable Energy Laboratory)

Contributors: Matthew Keyser, Aron Saxon, Kandler Smith, Tom Bethel, Andrew Colclasure, Weijie

Mai, Francois Usseglio-Viretta, Donal Finegan, and Shriram Santhanagopalan

Background

This report summarizes NREL’s fourth-quarter FY19 work in five areas:

1. Macro-homogeneous electrochemical modeling

2. Electrode architecture optimization with secondary pore network (SPN)

3. Measurement of graphite electrode heterogeneities

4. Fast charge algorithms

5. Electrolyte development and characterization

Through electrochemical transport/reaction modeling at several different length scales, NREL is exploring how

to best design electrodes for fast electrolyte transport and to minimize degradation and lithium plating. Macro-

scale models rank the relative benefits of how much charge batteries employing different design/thermal

control strategies can accept at the 6C constant current rate. One of the compelling strategies is to introduce

secondary pores into the electrode to provide fast channels for through-plane electrolyte transport. A 2D meso-

scale model explores this design space. Microstructure models explore the role of different microstructures in

suppressing the Li plating side reactions by reducing tortuosity and heterogeneities. Data analysis is being

performed on previous diagnostic experiments that measured graphite electrode state-of-charge (SOC)

heterogeneity in operando at the 6C rate, as well as microscopy that resolved the polycrystalline architecture of

nickel-manganese-cobalt (NMC) particles responsible for fracture under high C-rates. In previous quarters’

electrolyte development efforts, we identified several candidate solvent molecules to provide the requisite

solubility while retaining good transport properties (viscosity, dielectric constant). This quarter we tested

several of these solvent recipes in graphite/NMC cells and showed improved 5C fast charge acceptance

compared to the standard Gen2 electrolyte.

Results

1. Macro-homogeneous Model Updates

NREL continues to refine its macro-homogeneous electrochemical model as new data become available from

the XCEL team. The addition of multi-phase graphite transport and kinetics is underway. Initial comparisons

have been made to beamline XRD data that resolve total lithium and phase-specific (LiC6, LiC12, etc.)

concentrations across a 100-μm-thick electrode during operando 6C charging. Initial comparisons are

reasonable; however, secondary effects are present, and the team is working to further understand them and

decouple resultant signals in the data. For instance, lithium plated during the first 6C charge event appears to

block (de)intercalation on subsequent discharge and charge events of the electrode.

Separately, NREL has added a Li plating reaction to its model and is working with LBNL to compare

predictions of plated Li to their Round 1 half-cell coin cell experiments. In those experiments, LBNL followed

electrochemical cycling with destructive tests in which de-lithiated graphite electrodes are injected with water

in an air-free environment, and the amount of hydrogen gas evolved is measured via mass spectroscopy and

correlated with dead metallic lithium. The lithium plating is first modeled as an irreversible reaction. Thus, no

lithium stripping is initially considered. Figure 3 shows a comparison between titration results after three

cycles (blue dots) and model predictions with different exchange current densities for lithium plating. The

relatively thin 47-micron graphite electrodes plate lithium because cycling consists of lithiation to graphite’s

full theoretical capacity of 372 mAh/g; graphite is lithiated ~20% more than in standard full cell cycling. As

expected, the amount of plated lithium exponentially increases with charge rate. An exchange current density

9

for irreversible lithium plating around 0.2 A/m2 best fits the measured data. At 4C, the 2500 nmoles of lithium

measured corresponds to 3% of the total graphite capacity. As part of the newly formed heterogeneity

workgroup, NREL is working with Oak Ridge National Laboratory to develop a more detailed lithium

plating/stripping model considering:

• Lithium stripping

• Electrical isolation

• Solid electrolyte interface (SEI) effects/lithium reactions with electrolyte

• Ion transport effects

• Local condition effects on reversibility

Figure 3. Comparison of gas-titration results from LBNL and macro-homogeneous model predictions for lithium plated during three cycles vs. charge rate. Half-cells were fully lithiated to graphite’s

theoretical capacity of 372 mAh/g-graphite and then discharged to an upper voltage of 1.5V vs Li.

2. Electrode Architecture Optimization

NREL submitted its fourth quarter milestone, “Propose design guidelines for active material particle size,

morphology, and composite electrode tortuosity, as well as advanced strategies/architectures that can help

batteries achieve 10-minute charge. Parameterize life model versus aging data for application to battery

systems and EEMS studies.” For the guidelines portion of this milestone, NREL combined its existing NREL

microstructure-scale electrochemical model with a newly developed microstructure generation stochastic in-

house algorithm.

• The model predicts a linear correlation between specific surface area and lithium plating, with higher

𝑆𝑝 delaying lithium plating due to lower local C-rate. Because higher surface area also reduces

calendar life due to faster SEI layer growth, it makes sense to utilize higher surface area material

mainly at the front of the electrode.

• The model indicates that bimodal size distribution is slightly detrimental in terms of lithium plating

(2.5% anode SOC earlier), not resulting from the tortuosity difference but from higher lithiation

heterogeneities (cf. Figure 4a). Interestingly, depending on the electrode particles, bimodal size

distribution can have either a positive (NMC case) or a negative (SLC1506T Round 1 case) impact on

the tortuosity factor.

• For ellipsoid or platelet-like particles, alignment can delay lithium plating compared with spherical

particles (4% anode SOC later) if correctly aligned, that is, with their long diameters parallel with

10

electrode thickness (cf., Figure 4b). In contrast, particle misalignment triggers earlier lithium plating

and should be avoided. The tortuosity factor – correlated with particle alignment and aspect ratio –

explains this differences in lithium plating. Macroscopic isotropy coupled with local anisotropy

(randomly oriented particles) also triggers earlier plating.

Regarding electrode architectures, design guidelines for advanced architectures tailored for fast charging have

been provided for two distinct designs: (1) a dual-coated architecture, and (2) a secondary/dual pore network

(SPN) architecture, relying on a newly developed analytical MATLAB model complemented with a two-

dimensional (2D) electrochemical model.

• The model predicts that a dual-coated architecture with higher porosity (i.e., a lower tortuosity factor)

and smaller particles (i.e., higher specific surface area and thus lower local C-rate) for the layer near

the separator delays lithium plating by 7% anode SOC compared with the non-graded reference

electrode (cf. Figure 5a). Adequately tuned, graded electrodes also reduce electrolyte gradient and

homogenize intercalation reaction along the electrode thickness.

• An analytical transport model indicates that electrodes with higher initial tortuosity benefit more from

SPN (cf. Figure 5b) and that a combination of high through-plane tortuosity and low in-plane

tortuosity is a good fit for SPN. The latter case corresponds to some battery electrodes, such as A12

graphite. Being extremely fast (a 20-million ase in 10s with a laptop), the analytical model is useful

for a large design space.

The analytical SPN model will be released as open source in FY20, together with publication of a two-part

journal article from NREL on the optimal SPN architecture for fast charge.

(a) (b) Figure 4. (a) Lithiation heterogeneity induced by particle size distribution: large particle lithiation is lagging due to the solid-sate diffusion limitation. (b) The anode state of charge associated with the onset of lithium

plating calculated for different particle alignment and compared with spherical reference case.

11

Figure 5. (a) Anode state of charge associated with the onset of lithium plating calculated for different dual-

coating architectures and compared with non-graded reference case, (b) schematic of a secondary pore network, and (c) higher overall diffusion gain achieved for a higher tortuosity coefficient 𝜶.

(b) (c)

(a)

12

3. Measurement of Graphite Electrode Heterogeneities

Understanding the limitations of coated graphite electrodes during fast charging requires a spatial and temporal

measurement of lithiation and Li plating. Utilizing the state-of-the-art capabilities of the European Synchrotron

(ESRF), a high-resolution picture of lithiation gradients and Li plating within a graphite electrode was

constructed via high-speed XRD measurements (Figure 6). The measurements provided a live view of how a

graphite electrode lithiates throughout its depth, from separator to current collector, and revealed conditions

that lead to the onset of Li plating. New insights into the function of graphite electrodes under fast charge

conditions were revealed. For example, the onset of Li plating was quantified in real time and shown to occur

only over the first 20 µm of depth into the electrode. Li plating was also shown to increase and then decrease

when lower charge currents during the constant voltage step were reached, indicating that some Li plating is

reversible. Furthermore, regions where Li plating was detected did not fully delithiate upon discharge

(Figure 7), and the residual lithiated graphite existed as LiC6, which represents the first time that the co-

existence of delithiated graphite and fully lithiated LiC6 was observed. This result raises questions about how

the behavior of graphite changes when Li plates on its surface and how this might affect the propensity of that

graphite to further plate and affect the safety and performance of the cell, questions that are guiding NREL’s

modeling pursuits.

Figure 6. (a) Illustration showing the cell design used for high-speed XRD experiments. (b) An illustration showing the depth profiling XRD measurements carried out on the graphite electrode. (c) A plot showing the

quantified state of lithiation (LixC6) as a function of depth and time, as measured with XRD.

13

Figure 7. (a) Illustration showing the depth of the electrode and corresponding depth-dependent colors for the following plots. The region where Li plating was detected is highlighted. (b) The state of lithiation over

time for each depth (color code) during 6C charge, and (c) 6C discharge.

4. Charge Algorithm Development

XFC is a highly desirable feature for EVs. However, the fast-charging capability of lithium-ion batteries

(LIBs) is limited by safety concerns, which includes excessive heat generation and lithium plating. While the

development of new active materials and electrolytes with improved properties is the key to tackling this

problem, charging protocol also influences the fast-charging capability of LIBs. Using the existing Pseudo2D

battery model, NREL has screened hundreds of different charging protocols and investigated their effects on

fast-charging capabilities in terms of charge capacity; minimum phis-phie (an indicator for lithium plating);

average cell temperature; and maximum cell temperature in 10 mins, which corresponds to 6C charging.

Figure 8 shows the minimum phis-phie as a function of charge capacity for different charging protocols, with

the colormap representing the average cell temperature. The simulation results show that there is a strong

correlation between charge capacity, minimum phis-phie, and average cell temperature. Higher charge capacity

inevitably leads to smaller phis-phie, indicating a higher driving force for lithium plating. None of the

conventional charging protocols we investigated are significantly superior to others. Performance metrics

different from minimum phis-phie need to be developed to differentiate different charging protocols. Other

novel charging protocols are being investigated, aiming at achieving high charge capacity in 10 minutes

without triggering lithium plating.

Figure 8. Effects of charging protocol on charge capacity and minimum (phis-phie) in 10 mins: (left) two steps

and three steps constant current charging, and (right) positive pulse current charging. 5. Electrolyte Screening for Extreme Fast Charge

14

In previous quarters, we identified several candidate solvent molecules to provide the requisite solubility while

retaining good transport properties (e.g., viscosity, dielectric constant). The work this quarter was comprised of

experimentally evaluating these solvent recipes. For the first round of screening, we used ethylene carbonate,

ethyl formate, methyl difluoro acetate, and propionitrile in the ratio 10:25:25:40 with 3% vinylene carbonate

and another with ethylene carbonate, dimethyl carbonate, methyl difluoroacetate, and trimethyl phosphate in

the ratio 15:45:20:20. The salt concentration was maintained at 1.2M LiPF6. These candidates were selected as

a baseline to compare against previous reports. The formate- and acetate-based recipes showed some adhesion

issues, in particular involving the anode active material adhering onto the copper current collector. This result

is likely due to the binders used in the electrode formulation (CMC). Following these results, we screened a

second batch, including some non-flat cyclic compounds as co-solvents (30% by weight) to Gen2. The

electrochemical stability is comparable to Gen 2 electrolyte (Figure 9) and the discharge capacities after a 5C

charge are reasonable (Figure 10).

Conclusions

The macro-homogeneous model was updated with multi-phase graphite transport/kinetic behavior and a Li-

plating side reaction. Initial comparisons were made to data, and a Li-stripping reaction will be added in the

next quarter. A series of two journal articles are being prepared on SPNs. Combining elevated charging

temperature (45℃) with optimal SPN design, the model shows that the volumetric discharge energy density of

a 3-mAh/cm2 cell can reach 270 Wh/L after a 6C constant current charging. Further, the macro-homogeneous

model was used to perform an initial screening of charge protocols reported in the literature. To date, the

model predicts limited improvement in charge capacity and reduction in lithium plating. NREL is investigating

novel protocols driven by this theory.

A 3D microstructure model explored how graphite electrode particle size, morphology, and graded and dual-

coated electrodes either suppress or accelerate the onset of the undesired Li plating side reaction. Compelling

designs for suppressing Li plating include (1) single-coated electrodes with uniform particle size and particles

aligned in the in-plane direction, and (2) dual-coated electrodes with higher-porosity and smaller particles near

the separator. NREL is working with Argonne to determine how to fabricate such electrodes.

Figure 9. Verifying the electrochemical stability of Round-2 solvents against Li/Li+.

Figure 10. Full cell discharge capacity measured at C/3 using NMC-622 and

graphite electrodes from Targray following a 5C charge to 4.2 V.

15

For XFC electrolyte development, NREL screened a second batch of novel co-solvents to a standard Gen2

electrolyte. These novel formulations showed reasonable stability and, in some cases, significantly enhanced

the high rate charge capability of graphite/NMC 622 cells. Novel operando, in-situ XRD results were presented

for 100-micron-thick graphite electrodes during XFC. Measurements were performed as a function of depth

within the electrode with ~2-micron resolution. As expected, the electrode is used preferentially near the

separator interface and lithium plates near this region. XRD results are being compared with model predictions

and will be written into an upcoming journal article.

Milestones and Deliverables

NREL FY19Q4 Milestone to DOE, “Propose design guidelines for active material particle size, morphology,

and composite electrode tortuosity, as well as advanced strategies/architectures that can help batteries achieve

10-minute charge. Parameterize life model versus aging data for application to battery systems and EEMS

studies,” – September 30, 2019.

Publications

1. A. M. Colclasure, A. R. Dunlop, S. E. Trask, B. J. Polzin, A. N. Jansen, and K. Smith, “Requirements for

Enabling Extreme Fast Charging of High Energy Density Li-Ion Cells while Avoiding Lithium Plating,” J.

Electrochem. Soc. 166 (8) A1412–A1424 (2019).

2. D. Finegan, et al., “Quantifying Heterogeneities in State of Charge and Li Plating during Fast Charging,”

in preparation.

3. Quinn, et al., “The Application of Electron Back Scatter Diffraction for Quantifying Grain Orientations

and Transport Pathways in Li-ion Electrodes,” in preparation.

4. F. Usseglio-Viretta, W. Mai, A. Colclasure, and K. Smith, “Enabling Fast Charging of Lithium Ion Battery

through Secondary/Dual Pore Network: Part I - Analytical Model,” in preparation.

5. W. Mai, F. Usseglio-Viretta, A. Colclasure, and K. Smith, “Enabling Fast Charging of Lithium Ion Battery

through Secondary/Dual Pore Network: Part II - Numerical Model,” in preparation.

6. F. L. E. Usseglio-Viretta, D. P. Finegan, A. Colclasure, T. M. M. Heenan, D. Abraham, and K. Smith,

“Quantitative Relationships between Pore Tortuosity, Pore Topology, and Solid Particle Morphology

Using a Novel Discrete Particle Size Algorithm,” in preparation.

7. J. Allen, J. Chang, F. L. E. Usseglio-Viretta, P. Graf, and K. Smith, “Accelerating a Segregated Approach

for Li-Ion Battery Modeling using Block Preconditioners,” submitted.

8. A. Mistry, F. L. E. Usseglio-Viretta, A. Colclasure, K. Smith, and P. P. Mukherjee, “Fingerprinting

Distributed Activity of Inhomogeneous Electrodes during Extreme Fast Charging,” submitted.

16

XCEL R&D: Electrochemical, Atomistic, and Techno-Economic (BatPaC)

(Argonne National Laboratory)

Contributors: Dennis Dees, Hakim Iddir, Juan Garcia, Alison Dunlap, Andrew Jansen, and Shabbir

Ahmed (Argonne National Laboratory)

Background

Electrochemical modeling uses continuum-based transport equations combined with kinetic and

thermodynamic expressions to allow the potential, concentration, and current distributions to be determined

throughout the cell. The recent focus of the electrochemical modeling effort is to improve and better quantify

diffusion and phase change in graphite-active materials based on previous modeling efforts [1]. The previous

model treats graphite-active materials as multiple phases, also referred to as stages for graphite, where the

well-known Avrami equation was introduced to describe the phase changes as a function of lithium

concentration. Further, the model effectively correlated lithium diffusion and phase change during

galvanostatic intermittent titration technique (GITT) studies. However, based on limited half-cell

(i.e., graphite//lithium metal cell) data with a mesocarbon microbead (MCMB) graphite electrode, the model

tended to underestimate the performance of the graphite at high current rates. Work in this program has

identified that the diffusion coefficient increases significantly with applied current rate for all phases.

In order to examine an increasing diffusion coefficient with charging rate, both the Li-C phases and Li

diffusion in graphite are modeled at the atomic scale to characterize the structure of the starting material and its

changes during fast charging. Further, bulk defects, surface, and edge effects on Li diffusion will be

investigated for select conditions.

The Battery Performance and Cost (BatPaC) model was developed for lithium-ion battery packs used in

automotive transportation. The model designs the battery for a specified power, energy, and type of vehicle

battery. The cost of the designed battery is then calculated by accounting for every step in the lithium-ion

battery manufacturing process.

Results

1. Electrochemical Modeling

As introduced in the previous quarter, it is believed that electrode wetting of the electrolyte can be attributed to

the non-uniform distribution of lithium plating during extreme fast charging. Similar to a liquid wetting a

capillary tube, a liquid is drawn into a porous material through capillary pressure (Pc). Instead of the liquid

rising to a single height as in a capillary, in a porous material the pores go from fully saturated (i.e., s = 1) to

partially saturated (i.e., s < 1) to empty (i.e., s = 0) over a significant length scale (e.g., tens of centimeters).

The distance depends on a number of factors (e.g., porosity, pore size distribution, liquid surface tension); but

generally, the fractional level of saturation (s) vs. height has a characteristic S-shaped curve [2]. Leverett

developed an expression (Equation 1) relating capillary pressure to the S-shaped curve referred to as the

Leverett- or J-function:

(1),

where is the liquid surface tension, is the porosity, and is the permeability (i.e., as in Darcy flow). The J-

function is generally determined experimentally.

𝑃𝑐 = 𝛾 (𝜀

𝛫)

1 2⁄

J(s)

17

Measuring the J-function directly for both electrodes and the separator would be a challenge for a number of

reasons (e.g., foil backing, thin layers). However, we identified a company, MPM&P Research, Inc., that uses

the method of standard contact porosimetry to measure the wetting characteristics of thin layers [3]. In this

method, the unknown layer is placed in contact with a sample of known wetting characteristics.

Thermodynamically, the capillary pressure in both samples equilibrate over time. The changes in weight of the

individual samples are occasionally measured as the liquid is slowly removed. Although the ideal approach

would have been to use the electrolyte as the wetting fluid in this experiment, that proved to be problematic.

The experiment is conducted by initially wetting the sample completely under a full vacuum and then using a

partial vacuum to remove the fluid. Two fluids, octane and propylene carbonate (PC) were chosen to study the

electrodes and separator characteristics. Octane is considered a universal wetting fluid (i.e., wetting angle is

equal to zero), as most small straight-chain hydrocarbons are, and has a vapor pressure in a good range to

enable conducting the experiment within a reasonable timeframe. PC was chosen as a fluid closer to the

wetting characteristics of the electrolyte. The Round 2 cathode, anode, and separator were submitted for

evaluation.

Not only does this porosimetry technique yield key information needed to determine the J-function, it is also

very useful in establishing the overall pore volume, pore size distribution, and internal surface area. The pore

size distributions of the cathode, anode, and the separator for both fluids are given in Figure 11. Normally, the

difference between fluids for each sample can be used to establish the average wetting angle () of the PC

(i.e., cos() = 1 for octane). However, two issues make this treatment problematic. First, the vapor pressure for

PC on the separator sample was too small at 22ºC, where all of the other samples were measured, and had to

be conducted at 60ºC. Second, apparently the binder adsorbed enough PC to cause the electrodes to swell

slightly, resulting in a higher maximum pore volume, as can be seen in Figure 11 for the electrodes.

Figure 11. Contact porosimetry pore size distributions of Round 2 cathode, anode, and separator using octane and PC.

The results in Figure 11 can be easily converted to capillary pressure as a function of saturation, as shown in

Figure 12 (left) for octane and (right) for PC. Recalling that one atmosphere is approximately 106 dyne/cm2,

the figure shows that the capillary pressure is well above one atmosphere at less than 50% saturation, and in

general, the PC samples have a higher capillary pressure. While more evident in the octane data, the capillary

pressure for the separator is higher than that for the electrodes. To convert the capillary pressure to a J-

function, the permeability for each layer needs to be determined. While ideally this quantity would be

measured, it is possible to estimate it from the porosimetry results using the Kozeny-Carman equation [2]. Of

18

note, the permeability of the separator is approximately two orders of magnitude lower than that of the

electrodes, which is in contrast to its higher driving force for wetting. As shown in Figure 13 (left) for octane,

converting the capillary pressure to a J-function significantly reduces the variation between layers, despite the

separator, cathode, and anode having obviously different properties. There is more variation in the PC results

as shown Figure 13 (right). Initially for this wetting study, the fitted J-function shown in Figure 13 (left) is

used for all three layers. While fitting individual layers may result in more accuracy, it would not likely affect

the conclusions drawn from this effort.

Figure 12. Capillary pressure as a function of saturation for Round 2 cathode, anode, and separator using (left) octane and (right) PC.

Figure 13. J-function vs. saturation for Round 2 cathode, anode, and separator using: (left) octane and (right) PC.

19

The model describing the electrolyte wetting of the layers was developed decades ago and is still readily used

today [2, 4]. Assuming an isothermal system with no evaporation, the electrolyte (denoted l for liquid) will

displace the air (denoted g for gas). Darcy flow is assumed in the pores for each phase in Equations 2 and 3.

Mass balances (Equations 4 and 5) follow the movement of each phase. The difference between the liquid and

gas pressure (Equation 6) is the capillary pressure (Equation 1), as follows:

(2)

(3)

(4)

(5)

(6),

where is the density, µ is the viscosity, and ĸ is the relative permeability. The Advanced Electrolyte Model

was utilized to establish the electrolyte properties [5]. The relative permeability, like the permeability, should

be measured but must vary from zero to one and is assumed to follow the cube of the fraction of each phase

[2]. The above model can be applied to each layer.

The interaction of each layer depends greatly on the process used to wet the cell. For this program, the single-

layer pouch cells are fabricated with the pouch open at the bottom (i.e., on the opposite side from the tabs).

Approximately four times the minimum amount of electrolyte needed to wet all the porous components is

added to the cell. The excess factor is set high enough to account for the estimated loss of usable electrolyte

resulting from electrolyte clinging to non-electrode coating components within the pouch. The electrolyte is

worked into the cell through a gentle massage of the pouch by hand, followed by pulling a partial vacuum, and

then further massaging. The pouch is massaged in various directions to distribute the electrolyte down the

sides into the cell layers and to help push the air out of the cell. This process takes a few minutes, and then the

pouch is sealed. After sealing, the cell is mounted horizontally under several psi pressure, and the forming

process starts.

It is assumed that some of the electrolyte is forced between the layers during the massaging and comes in

direct contact with the “face” of the electrodes and separator. The time constant for the electrolyte soaking into

the porous components follows the square of the characteristic length, so entering through the face obviously

occurs much more quickly than through the edge of the cell. However, it is difficult to determine exactly how

much electrolyte soaks into the cell during this process. The model can be used to examine a partial wetting of

the components. Focusing on the negative electrode face, we assume that a two-millimeter strip of electrode

comes in contact with an excess of electrolyte at a given time (i.e., at time = 1s for this simulation). The

change in saturation level of the electrode can be followed under the strip and another millimeter outside the

strip. Figure 14 gives the change in the average saturation of the portion of the electrode under the strip. It is

evident that the saturation level reaches more than 90% within a few tenths of a second. After the pores

become mostly saturated, the electrolyte starts soaking into the electrode area that is not wetted by electrolyte.

The insets in Figure 14 show the saturation distribution (red > orange > yellow > green > blue > indigo >

violet) of electrolyte in the pores for one millimeter of wetted surface area (left side) and one millimeter of dry

surface area (right side) for the indicated times after the electrolyte is applied.

𝑣𝑙 = 𝛫𝜅𝑙

𝜇𝑙

(𝛻𝑃𝑙 − 𝜌𝑙�⃗�)

𝑣𝑔 = 𝛫𝜅𝑔

𝜇𝑔

(𝛻𝑃𝑔 − 𝜌𝑔�⃗�)

𝜕(𝜌𝑙𝜖𝑠)

𝜕𝑡= −𝛻 ∙ (𝜌𝑙𝑣𝑙)

𝜕 (𝜌𝑔𝜖(𝑠 − 1))

𝜕𝑡= −𝛻 ∙ (𝜌𝑔𝑣𝑔)

𝑃𝑐 = 𝑃𝑔 − 𝑃𝑙

20

Overall, the capillary pressure for the electrolyte wetting the cell components is sufficient that, given enough

time, the air will be displaced and the pores should be completely saturated. Clearly, the process of the

electrolyte soaking into the negative electrode from the face takes place relatively quickly when compared to

its pace when traveling along the length of the layer. Although it is possible that the process of filling with

electrolyte and sealing the pouch cells will result in fully saturated pores, it is much more likely that the pores

will be only partially saturated at the time that the cells begin the formation process. During the formation

process, the electrolyte must soak into the cell from the edges, and it is much more likely that air will be

trapped in the pores. In the next quarter, the model described above will be applied to all three components

sandwiched together and wetted with electrolyte along the edge. The thinness of the layers when compared to

the length and width of the cells suggests that the time constant for the capillary pressure to equilibrate across

the layers is relatively fast when compared to the movement of the electrolyte along the layers. Assuming the

capillary pressure across the layers at any given point has equilibrated, greatly simplifies the calculation.

Figure 14. The change in average saturation level of a two-millimeter strip of the negative electrode

when excess electrolyte is applied to the face at one second. The insets show the saturation distribution

(red > orange > yellow > green > blue > indigo > violet) of electrolyte in the pores for one millimeter of wetted surface area (left side)

and one millimeter of dry surface area (right side) for the indicated times after the electrolyte is applied.

2. Atomistic-Level Modeling

Atomistic simulations have been carried out using a combination of spin-polarized density functional theory

(DFT) calculations and the Nudged Elastic Band (NEB) method as implemented in the Vienna Ab Initio

Simulation Package (VASP) to compute total energies and Li migration barriers [6, 7]. The exchange-

correlation potentials have been treated by the Generalized Gradient Approximation (GGA) parametrized by

Perdew, Burke, and Ernzerholf (PBE) [8]. The interaction between valence electrons and ion cores were

described by the Projected Augmented Wave (PAW) method [9]. All of the ions were allowed to relax until the

total energy differences were no more than 0.003eV. To account for the interplanar Van der Waals (VdW)

interactions between the graphene planes, we use a dispersion correcting scheme, D2 [10]. Calculations with

extra charge were performed using an effective uniform background compensation charge as implemented in

VASP [11, 12]. The applied first-order correction to the energy is given by the expression (𝑒2𝑞2𝛼/𝐿𝜖), where

q is the net charge of the system, α the Madelung constant of a point charge q placed in a homogeneous

background charge -q, and ε the dielectric constant of the system.

21

Using the migration barriers computed for several configurations of Li environments, a Kinetic Monte Carlo

(KMC) model was developed to evolve the system in time and compute a diffusion coefficient corresponding

to possible fast charge scenarios and hence closer to experimental conditions. In order to describe the system, a

hexagonal lattice is built. Figure 15 shows a schematic representation of such a lattice. This lattice restricts the

movement of the Li ions. They are allowed to hop through the carbon-carbon bonds only (which was shown to

be more favorable than along the C-C bond). Hence, for each Li ion, there are a maximum of six possible

hoping events (west, east, northeast, southeast, northwest, or southwest). Each hoping event is allowed if the

neighboring site is empty or the Li-ion occupying the neighboring site is also moving simultaneously

(concerted diffusion). Figure 15(a) represent an initial configuration of ions where Li ions are accumulated at

the polarized graphite surface. The red arrows in Figure 15(a–b) represent possible Li-ion hops. Figure 15(b)

represents a random intermediate state, and Figure 15(c) represents the final state, where the local composition

LiC6 is everywhere. The horizontal boundaries are set as periodic to emulate a large region of the surface. The

vertical boundaries are closed (diffusion is not allowed beyond those borders). However, the left vertical

Figure 15. Schematic representation of the hexagonal lattice used in the KMC model: (a) initial state, (b) intermediate state, and (c) final state. The blue spheres represent Li-occupied sites, and the red arrows represent potential Li hops.

boundary could be open (grand canonical ensemble), allowing the simulation of Li interchange with a Li

reservoir (electrolyte). That could allow the simulation of Li plating conditions. Furthermore, multiple layers

of the lattice can be arranged in a tridimensional network to simulate the staging and the effect of interlayer

spacing.

For each Li-ion, the algorithm identifies empty and occupied neighboring sites. Thereafter, all possible events

are identified. An event in the KMC algorithm is one or a combination of several Li-ion hops (concerted hop

events are allowed). All of the possible Li hop energy barriers are required to feed the KMC model, including

single Li hops or concerted Li hops. In previous reports, we have shown DFT-level energy barriers computed

for several different cases. Table 1 shows some selected energy barriers for a few potential events (Cases a–j).

Figure 16 is a schematic of movements for Cases a–j. Subsequently, each potential event rate (Wi) is computed

as:

22

where 𝜈 is a pre-exponential factor, Δ𝐸 is the DFT-computed energy barrier, 𝑘𝐵 is the Boltzmann constant, and

T is the temperature. The state of the model system is evolved by choosing one event (Wk) stochastically with

the expression:

where 1 is a random number evenly distributed over the range [0,1], and Nhops is the total number of possible

events. Finally, the time is updated according to the following equation:

The described procedure is based on the Markov approximation, the use of which implies that the next process

can be determined from the current state and that all processes take place independently of one another because

any memory of the system is erased after each step [13].

Table-1: Li-ion diffusion in graphite energy barriers

Configuration

Energy

barrier

(eV)

Case a: 1 Li-ion (dilute system) 0.34

Case b: 2 Li-ion-cluster perpendicular 0.22

Case c: 2 Li-ion-cluster parallel with vacancy 0.32

Case d: Linear cluster parallel displacement 1.10

Case e: Linear cluster perpendicular displacement 0.32

Case f: Linear cluster perpendicular displacement, 1st Li-ion 0.04

Case g: Linear cluster moving from high concentration front 0.11

Case h: Second linear cluster front moving 0.39

Case i: Li-ion moving back to high concentration front 1.36

Case j: Li-ion second step after moving away from cluster 0.26

Wi = * exp−E i

kBT

∑ 𝑊𝑖 < 𝜉1𝑊 ≤ ∑ 𝑊𝑖

𝑘

𝑖=0

𝑘−1

𝑖=1

𝑊 = ∑ 𝑊𝑖

𝑁ℎ𝑜𝑝𝑠

𝑖=0

Δ𝑡 = −1

𝑊log 𝜉2

23

Finally, the diffusivity will be computed exploiting the random-walk theory using the expression:

where D is the diffusivity, t is the simulation time, and 𝑟(𝑡 + Δ𝑡) − 𝑟(𝑡) represents the change in position after

a time interval, Δt. The square brackets indicate an expectation value, which is obtained by averaging over

long-time intervals.

Given the peculiarities of the system under study, we decided to write the software for the KMC algorithm

instead of using some of the packages already available [14, 15]. The software is under development using the

Python programming language (some critical routines in C/C++ are needed for scaling).

Conclusions

A series of wetting studies on the cell components were initiated to further examine the cell electrolyte wetting

process. Overall, the capillary pressure of the electrolyte wetting the cell components is sufficient that, given

enough time, the air will be displaced and the pores should be completely saturated. Clearly, the process of the

electrolyte soaking into the components from the face takes place relatively quickly when compared to its pace

when traveling along the length of the layers. Although it is possible that the process of filling with electrolyte

and sealing the pouch cells will result in fully saturated pores, it is much more likely that the pores will be only

partially saturated at the time that the cells begin the formation process. Atomistic-level studies are developing

a framework for obtaining the lithium diffusion coefficient through graphite under fast charge conditions.

Milestones and Deliverables

Determine model parameters: parameter set for Superior Graphite-based graphite negative electrodes will be

developed. – Completed.

Examine rate-dependent diffusion: possible mechanisms for apparent increase of lithium diffusion coefficient

in graphite at higher rates will be examined. – Initial studies completed.

D = − lim𝑡→∞

1

6𝑡[∑⟨𝑟(𝑡 + Δ𝑡) − 𝑟(𝑡)⟩

∞

𝑡=0

2

]

24

References

1. K. G. Gallagher, D. W. Dees, A. N. Jansen, D. P. Abraham, and S.-H. Kang, Journal of the

Electrochemical Society, 159 (12) A2029–A2037 (2012).

2. A. E. Scheidegger, The Physics of Flow Through Porous Media, The Macmillan Company, 1960.

3. Y. M. Volfkovich, V. E. Sosenkin, and V. S. Bagotsky, Journal of Power Supplies, 195, 5429–5441

(2010).

4. N. Susarla, S. Ahmed, and D. W. Dees, Journal of Power Supplies, 378, 660–670 (2018).

5. E. R. Logan, E. M. Tonita, K. L. Gering, and J. R. Dahn, Journal of The Electrochemical Society, 165 (14)

A3350–A3359 (2018).

6. G. Kresse and J. Furthmüller, Efficiency of Ab-Initio Total Energy Calculations for Metals and

Semiconductors Using a Plane-Wave Basis Set. Comput. Mater. Sci. 1996, 6 (1), 15–50.

https://doi.org/10.1016/0927-0256(96)00008-0.

7. G. Kresse and J. Hafner, Ab Initio Molecular Dynamics for Liquid Metals. Phys. Rev. B 1993, 47 (1),

558–561. https://doi.org/10.1103/PhysRevB.47.558.

8. J. P. Perdew, K. Burke, and M. Ernzerhof, Generalized Gradient Approximation Made Simple. Phys. Rev.

Lett. 1996, 77 (18), 3865–3868. https://doi.org/10.1103/PhysRevLett.77.3865.

9. P. E. Blöchl, Projector Augmented-Wave Method. Phys. Rev. B 1994, 50 (24), 17953–17979.

https://doi.org/10.1103/PhysRevB.50.17953.

10. S. Grimme, Semiempirical GGA-type density functional constructed with a long-range dispersion

correction. J. Comput. Chem. 2006, 27 (15), 1787–1799. https://doi.org/10.1002/jcc.20495.

11. G. Makov, M. C. Payne, Periodic Boundary Conditions in Ab Initio Calculations. Phys. Rev. B 1995, 51

(7), 4014–4022. https://doi.org/10.1103/PhysRevB.51.4014.

12. J. Neugebauer and M. Scheffler, Adsorbate-Substrate and Adsorbate-Adsorbate Interactions of Na and K

Adlayers on Al(111). Phys. Rev. B 1992, 46 (24), 16067–16080.

https://doi.org/10.1103/PhysRevB.46.16067.

13. C. C. Battaile, The Kinetic Monte Carlo Method: Foundation, Implementation, and Application. Comput.

Methods Appl. Mech. Eng. 2008, 197 (41), 3386–3398. https://doi.org/10.1016/j.cma.2008.03.010.

14. M. J. Hoffmann, S. Matera, and K. Reuter, Kmos: A Lattice Kinetic Monte Carlo Framework. Comput.

Phys. Commun. 2014, 185 (7), 2138–2150. https://doi.org/10.1016/j.cpc.2014.04.003.

15. M. Leetmaa and N. V. Skorodumova, KMCLib: A General Framework for Lattice Kinetic Monte Carlo

(KMC) Simulations. Comput. Phys. Commun. 2014, 185 (9), 2340–2349.

https://doi.org/10.1016/j.cpc.2014.04.017.

25

XCEL R&D: Performance Characterization and Post-Test Analysis

(Argonne National Laboratory)

Contributors: Zhenzhen Yang, Seoung-Bum Son, David Robertson, Nancy Dietz Rago, LeRoy Flores,

Alison Dunlop, Stephen Trask, Bryant Polzin, Andrew Jansen, and Ira Bloom

Background

One of the objectives of these projects is to determine how small, pouch cells, which contain selected graphite

anodes and oxide cathodes, respond to extreme fast charging. One of the key points here is to detect lithium

plating as soon as it occurs using electrochemical methods. Another aspect is to provide materials

characterization data, such as X-ray diffraction, Raman spectra, and diffusion coefficient measurements, which

can be used to understand the observed performance response and for modeling. Another objective is to

determine the physical and chemical changes that extreme fast charging caused. These changes will be

characterized by post-test analysis of the fast-charged cells. A suite of materials and surface characterization

techniques will be used.

Results

1. Fast-Charge-Driven Li Plating on Anode and Structural Change of NMC Cathode The objective of this work is to quantify the amount of Li plating on the graphite anode and to study the

structural changes of the NMC cathodes due to fast charging conditions. First, cells were cycled 10, 20, 50,

and 100 times in the Li(Ni0.5Mn0.3Co0.2)O2 cathode (NMC532)/Graphite 2032-type coin cell configuration.

Charging rates of 1-C or 6-C and a discharging rate of 0.5 C were used, and the ambient temperature was

fixed at 30C. The Gen2 electrolyte (1.2M LiPF6 in EC: EMC [3:7 by weight]) was used. Figure 17 shows the

difference in cycling performance of cells cycled at the 1-C and 6-C charging rates. The 1-C-rate-charged cell

shows an initial capacity of 115 mAh g-1 and 90 mAh g-1 after 100 cycles, whereas the 6-C-charged cell

showed only an initial capacity of 88 mAh g-1 and 32 mAh g-1 after 100 cycles. We assumed that the

differences in cycling performance of those cells could be caused by kinetic limitations, Li plating on anodes,

and/or structural failure of the cathodes.

Then the cells were

disassembled in the glove box and

washed two times in excess

dimethylcarbonate (DMC), swirling for

1 minute each. Inductively coupled

plasma mass spectrometry (ICP-MS) was used to quantify the amount of Li on the graphite anode surface.

Anodes were first hydrolyzed with high-performance liquid chromatography (HPLC)-grade water, and the

amounts of Li and F were measured using ICP-MS and an ion-selective electrode, respectively. It was assumed

that F came from LiF, which is a major SEI component, and that the amount of Li found in the 1-C cells

Figure 17. Cycling performance of NMC532/graphite coin cells.

26

represented the amount that would normally be found in the graphite anode. Figure 18 shows the corrected

amount of the Li plating at the 6-C charging rate versus cycle count. Here we show the initial values in

µmole/ml (right y-axis) and their capacity equivalents (mAh, left y-axis). The ICP-MS results show gradual

increases from 10 to 100 cycles, indicating that the amount of plated Li on anodes is increasing as the cycle

number increases. The plated Li on the anode, that is, removed from the NMC532 cathode, might cause

structural changes in the NMC532 cathode.

XRD was performed on the cycled NMC532 cathodes to study the structural variation and lattice parameter

changes based on cycle numbers and charging condition. Measured XRD patterns in Figure 18 are µmole/ml

(right y-axis) and their capacity equivalents (mAh, left y-axis). The ICP-MS results show gradual increases

from 10 to 100 cycles, indicating that the amount of plated Li on anodes is increasing as the cycle number

increases from 10 to 100 cycles. The plated Li on the anode, that is, removed from the NMC532 cathode,

might cause structural changes in the NMC532 cathode.

Measured XRD patterns were also in good agreement with the NaFeO2 structure with an R-3m space group.

Figure 19a–b show the XRD results of 1-C- and 6-C-charged NMC532 focusing on (003) peaks after 10, 20,

50, and 100 cycles. We saw that (003) peaks gradually shift to the lower values of two theta as the cycle

number increases, indicating an increase of the c-axis lattice parameter with cycles. In general, the increase of

the c-axis lattice parameter is attributed to the loss of active Li during cycling. The c-axis lattice parameter was

carefully calculated and displayed in Figure 19c. We observed there was no remarkable difference between 1-

C- and 6-C-charged NMC532 in all cycle numbers measured. The c-axis parameter after 100 cycles was found

to be larger for 1-C-charged than 6-C-charged NMC532. This result implies that structural damage to

NMC532 that results from the fast charging condition is not critical; although this conclusion cannot rule out

the possibility of localized structural and mechanical damage to NMC532 given that the XRD results represent

general and bulk characteristics of the materials.

Figure 18. Quantification of Li plating at 6-C with different cycle numbers. At least

three cells were used for each condition, and average values were used.

27

To better understand the structural changes of NMC532 caused by the fast charging condition, the data in

Figure 17 were replotted as shown in Figure 20. In addition to Figure 17, Figure 20 contains Li plating at the

6-C charging condition and total Li deficiency from NMC532, which simply combines Li plating and cycling

performance achieved at 6-C. Although fast charging with the 6-C rate generates a notable amount of Li

plating on the anode, total Li deficiency at the 6-C charging rate is still lower than the specific capacity

achieved at the 1-C charging rate for all of the cycle numbers we studied. Considering that we have attributed

the increase in the c-axis lattice parameter to the loss of active Li during cycling, the results shown in

Figure 20 make sense with what our XRD analysis revealed, which showed negligible lattice parameter

variations between the 1-C and 6-C charging rates.

2. Diffusion Coefficient Measurement

The diffusion coefficient versus state-of-charge for 1506T is being measured in ‘T’-shaped cells (external

lithium reference electrode) at temperatures in the range of 20–50°C. These data can be used as input to the

modeling effort.

Figure 19. XRD results focusing on (003) peaks at (a) 6-C and (b) 1-C. (c) C-axis lattice parameter variation with cycle numbers.

Figure 20. Cycling performance of NMC532/ graphite coin cells with Li plating

and total Li deficiency from NMC532 at 6-C.

28

In the last quarterly report, we showed our diffusion coefficient results from charging at the C/20 and 4-C

rates. These experiments were performed in full cells, and the anode voltage limits were 1.5 to 0.1 V, corrected

for iR. At the C/20-rate, the current was on for 15 min and off for 90 min. The current-on time was reduced to

45 sec at the 4-C rate, keeping the amount of capacity removed during each pulse approximately constant. We

continued the investigation, adding the 1-C charge rate to determine the trend in the diffusion coefficient with

rate. The titration curves for C/20, 1-C, and 4-C are given in Figure 21.

Figure 21. Anode potential vs. Li/Li+ vs. specific capacity at the C/20, 1-C, and 4-C charge rates at 30°C. The sloping part of the curve, labeled 2, is the single-phase region of interest.

The diffusion coefficients were estimated using Huggins’s method from the data in the single-phase region

shown in Figure 21. The estimated diffusion coefficients are shown as a function of capacity density in

Figure 22. It was not expected that the diffusion coefficient should increase so much at the 1-C rate and then

decrease precipitously at the 4-C rate. This trend could be part of the reason for lithium plating at high charge

rates.

As a point of comparison, the diffusion coefficient was estimated using the depolarization method. These

results are shown in Figure 23 for the C/20 charge rate. It was interesting to see that the values of the diffusion

coefficient calculated by the depolarization method were much smaller than those calculated using Huggins’s

method. Based on the data shown in Figure 22, Huggins’s method produced values on the order of 10-10 cm2/s,

whereas those from the depolarization method were on the order of 10-14 cm2/s.

0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

0 50 100 150 200 250 300

Po

ten

tial

, V v

s. L

i/Li

+

Capacity, mAh/g

C/20

1-C

4-C

1

23

29

Figure 22. Diffusion coefficient, D, as a function of capacity density and charge rate. These values were calculated in region 2 of Figure

Figure 23. Diffusion coefficient, D, as a function of capacity density

at the C/20 rate calculated using the depolarization method.

0

1E-09

2E-09

3E-09

4E-09

5E-09

6E-09

7E-09

8E-09

9E-09

0 20 40 60 80 100 120 140

D,

cm2/s

Capacity, mAh/g

C/20

1-C

4-C

0.0E+00

1.0E-14

2.0E-14

3.0E-14

4.0E-14

5.0E-14

6.0E-14

7.0E-14

8.0E-14

9.0E-14

0 20 40 60 80 100 120

D, c

m2/s

Capacity, mAh/g

30

Conclusions

The amount of lithium plating on the 1506T anode is sensitive to cycle count at 30°C. After ~20 cycles, the

amount of plated lithium seems to depend linearly on cycle count. No obvious changes to the crystal structure

of NMC532 were observed after 100 fast-charge cycles.

The diffusion coefficient on 1506T was calculated using Huggins’s method at three charge rates: C/20, 1-C,

and 4-C. There was a significant increase in the value of the diffusion coefficient between C/20 and 1-C. An

unexpected result was that the value of the diffusion coefficient decreased between 1-C and 4-C.

Future work

We will continue the work to characterize the factors that affect lithium deposition, namely, time and

temperature. We will also continue to characterize cathode cracking using the materials from the lithium

plating work, as part of a more general aging project. The question here is how does fast charging affect

cathode capacity?

We plan to do two more diffusion coefficient measurements, one at 2-C and one at either 3-C or 6-C. These

results should help define the trends we observed this quarter and described above.

31

XCEL R&D: Extreme Fast Charging R&D: Battery Testing Activities (Idaho National

Laboratory)

Contributors: Eric Dufek (INL), Tanvir Tanim (INL), Kevin Gering (INL), and Daniel Steingart (Princeton)

Background

XFC of Li-ion batteries can create battery life and safety issues, specifically, shortened battery life due to

enhanced loss of lithium inventory and electrolyte degradation and elevated safety concerns due to the

potential for shorts created by Li dendrites. The detection and monitoring of the onset and evolution of Li

plating over aging is a significant challenge. In operando detection schemes to understand the dynamics of Li

plating and the role that aging has on Li plating are vital to enabling fast charging of specific energy cells.

Sensitively detecting Li plating is difficult without performing a destructive post-test analysis. The objective of

the proposed work at INL and Princeton is to overcome these limitations using electrochemical analysis and

nondestructive ultrasonic acoustic methods that can be applied directly in operando to understand the onset

and growth of Li plating during XFC. The team is closely coordinating with other efforts at NREL and

Argonne to understand the key limitations that enhance the probability of Li plating. A key to these efforts is

understanding the interplay between materials, electrode structure, and use conditions. The ability to

understand the interplay will be distinctly aided in this project through the joint use of electrochemical and

ultrasonic tools, which ultimately will aid in the scientific understanding required to facilitate development of

XFC batteries for electric vehicles.

Electrochemical methods that can be used to identify Li plating include the use of differential capacity

(dQ/dV) and quantitative analysis of the cells’ charge and discharge profiles. These tools yield pertinent

information associated with both the kinetic and thermodynamic processes that occur in batteries, and as such

provide the direct ability to better understand how variation in materials and electrodes have an impact across

the life of a battery. There are limitations to the types of techniques that can be used to complement

electrochemistry in operando using standard cell formats. One method that is showing promise is the use of

ultrasonic measurements. Ultrasonic measurements rely on acoustic waves propagating through a structure,

such as an electrode, which are modulated by its properties and encode structure/property relationship data.

These relationships are directly tied to the material and mechanical changes that occur in a cell during cycling

and can be used to characterize change in a nondestructive, real-time manner. Coupling ultrasonic and

electrochemical measurements will enable a more complete evaluation of the impacts of XFC to be

understood. In particular, both methods are expected to produce distinct and complementary signals that will

indicate the deposition of Li on the negative electrode of a battery during aggressive charging conditions.

Tracking changes with the coupled ultrasonic and electrochemical methods over battery life and as batteries

age will provide pertinent information related to the distinct conditions that drive Li plating during XFC.

Results

The use of acoustic methods including ultrasound provides the opportunity to clearly identify signatures from

lithium plating in operando. During FY19, the team led by Daniel Steingart observed a strong correlation in

both shifts in the acoustic time of flight (ATOF) as well as acoustic transmittance (AT) with deposition of

lithium metal on the anode of lithium-ion batteries. These correlations held for single–layer, high-rate test cells

designed and fabricated at Argonne’s CAMP Facility. In separate activities, the utility of using acoustics to

detect plating in multilayer, commercial-off-the-shelf cells was also confirmed. Currently, both efforts are in

the publication process.

During the early stages of the work, the team realized that to achieve detection in single-layer pouch cells,

detection sensitivity was going to be a critical activity. To increase sensitivity, the team advanced multiple

areas, including tight temperature control to minimize experimental drift, as well as changes to the frequency

32

of transponders used for the characterization effort. The impact of these modifications was the observance of a

shift in the ATOF during the later stages of a fast-charging event at rates above 6C, in cells with a loading of

3.0 mAh/cm2. Plating of lithium in these cells and at the rates used had previously been confirmed by other

XCEL team members, including those at INL and Argonne. Using lower rate cycling (below the threshold for

lithium plating), no observable shift in ATOF or AT was identified. Currently, future work is planned to

quantify the detection limits of the acoustic method in conjunction with electrochemical detection methods.

In research areas outside of the acoustic detection of Li metal plating, the team continued to refine analysis and

charge methods on higher loading cells (3.0 mAh/g) that had undergone different levels of fast charging. At

3.0 mAh/cm2, rates as high (9C) as used at lower levels were deemed not attainable based on transport

overvoltage analysis. For this reason, fewer protocols were used and a base case of 4C was also investigated.

Figure 24 contains data related to two of the conditions analyzed, 4C and a 5-step protocol, which started at 9C

and progressively switched to lower currents based on cell overvoltage. As with the earlier round of cells, a

distinct increase in cell-to-cell aging was observed as more aggressive protocols were used; indeed, this aging

was observed even at 6C (data not shown). Analysis on these cells is continuing, and samples have been shared

with several entities for post-cycling characterization.

Figure 24. Capacity fade and cycle-by-cycle coulombic efficiency for 4C CC-CV (a, c) and 9C MS5 (b, d) charging protocols. Here, coulombic efficiency is based

on fast charge and slow (C/2) discharge capacities.

Conclusions

33

During Q4, it was identified that a distinct shift in ATOF and AT can be observed during fast charging when

using an experimental acoustic set-up. Both signatures align with the range expected for lithium plating, and

the presence of lithium was confirmed visually after cycling. Additional work to examine the characterization

of higher negative electrode cells (3.0 mAh/cm2) has provided additional information on transport limitations

within the cell and distinct heterogeneity, which emerges as cells are cycled at rates above 4C.

Milestones and Deliverables

Q1 – Quantitate single-layer pouch cell failure modes – Complete

Q2 – Identify enhanced transport electrolytes – Complete

Q3 – Characterize electrochemical Li reversibility – Complete

Q4 – Correlate electrochemical and acoustic signals – Complete

Publications

T. R. Tanim, et. al, "Electrochemical Quantification of Lithium Plating: Challenges and Considerations,"

Journal of the Electrochemical Society, 166 (12), A2689–A2696 (2019).

XCEL R&D: SLAC National Accelerator Laboratory

Contributors: Hans-Georg Steinrück, Chuntian Cao, and Michael Toney

Approach

34

Our approach has been to exploit the nondestructive nature and fast acquisition times of characterization

techniques at synchrotron sources to understand parasitic Li plating in LIBs under fast charging conditions and

across multiple length scales. These findings will be used as inputs in simulations to guide cell design and

charging protocols of extreme fast-charging LIBs. The specific aim in this quarter was to build on the mm-

scale XRD studies conducted in the previous quarters by examining other length scales as well as

complementary imaging techniques. This approach also allowed exploitation of the nondestructive nature of

characterization of full cells using mm-scale XRD in the fully assembled conditions.

The characterization techniques were expanded in two main directions focused on complementing the

structural information obtained by diffraction with some morphological information from imaging-based

techniques. First, X-ray-based micro-computer tomography (microCT) was used to obtain information on the

porosity of the anode after XFC charging, specifically to enable comparison between spots that did and did not

show significant Li plating in the mm-scale XRD. Second, neutron-based imaging was conducted on the same

areas observed using the X-ray microtomography, with the main aim of characterizing metallic Li plated on the

anode. Using X-ray tomography alone cannot distinguish between graphitic carbon and metallic Li; however,

neutrons can make this distinction, which is a huge advantage while studying Li plating on the graphite anode.

The plan of action was to select specific “interesting areas” on the cells based on the prior mm-scale XRD

results. Toward this end, areas on the anode that showed low and high intensities for metallic Li were

identified and characterized further. One caveat was that the geometry of these other techniques forced the

tearing open of the cells to enable creating sample geometries better suited for that particular technique.

However, this step enabled taking optical images of the anodes to provide a base comparison for the spatial

maps of Li graphite intensities obtained

through mm-scale XRD, as shown in

Figure 25 for one representative cell cycled

450 times with the 6C CCCV (constant

current, constant voltage) protocol, where

the Li intensity map corresponds very well

to “bright” regions seen in the optical image.

Finally, all cells presented herewith are in

the fully discharged state at a nominal SOC

of 0% (3.0V) after 450 XFC cycles. The

cycling rate and protocols vary from cell to

cell. The cells are single-layer pouch cells

assembled at CAMP, with an NMC 532 cathode and graphite anode and separated by a Celgard 2320

separator. The cells have a capacity of ~3.0 mAh/cm2 and a nominal area of ~14 cm2. The Results section

contains one key finding outcome from the mm-scale XRD that connects the amount of plated Li calculated

from XRD to the capacity loss of the cell. Then, X-ray microCT imaging of portions of the disassembled

anode are shown, followed by a novel technique of using simultaneous neutron and X-ray-based imaging on

the same region of the cut-up anode.

Figure 25. Comparison of optical and XRD spatial map of the anode, showing Li intensities across the electrode.

35

Results

1. Relating the Capacity Fade in Cells to the Amount of Dead Li

Figure 26 shows the amount of plated Li in all the cells, normalized by the total amount of Li in the pristine

cell, as a function of the capacity fade of the cell after 450 cycles. The total amount of plated Li is calculated

after 450 cycles at 0% SOC, thus any Li on the anode is considered ‘dead’ Li. The data points of the cells are

colored by the charging rate. In general, the higher the capacity fade of the cell, the higher is the intensity of

plated Li. Note that we have previously shown that the graphite near the Li metal deposits is lithiated in LiC6

or LiC12 and this some of the Li loss is accounted for within the graphite. The exception is the cells colored in

red, that are both charged at a 4C rate, that show negligible plating, irrespective of the capacity fade. This

result showing direct correlation between capacity fade and plating differs from earlier works that report that

the amount of Li plating (dead Li) is primarily a function of the C-rate. Our result also shows that cells cycled

under the same charging protocols and rates, for the same number of cycles, show a significant heterogeneity

in performance (in terms of capacity fade) as well as amount of Li plating over the cell. Therefore, in order to

ensure the wider deployment of XFC enabled batteries commercially, addressing the root cause for this

heterogeneity of behavior between similar cells is of paramount importance.

Figure 26. Correlation

between mass of plated Li (as a % of initial mass of Li before cycling) and

capacity fade in Round 2 cells. While cells charged at 4C (red) show

negligible plating independent of the capacity fade, the other cells (#3–#7)

show a direct correlation between the amount of plating and capacity fade.

2. Insights into Porosity of Anode and Amount of Plated Li

X-ray microCT was conducted at Beamline 2BM at Argonne’s APS. This technique offers a superior spatial

resolution of ~0.5μm, which is several orders of magnitude finer than the mm-scale XRD, albeit with a

proportionally smaller field of view. Therefore, this technique is ideal for studying small volumes of the

electrode. In addition, microCT provides an avenue to study the heterogeneity of various species across the

depth of the anode, from the separator to the current collector. The ideal geometry for such imaging

characterizations involves cutting the anode into thin strips about 1–2 mm wide. These anode strips were

selected based on the mm-scale XRD maps to cover regions with a large amount of Li, those with no Li, and

the regions in between. Reconstruction and segmentation of the data (Figure 27) were performed by the team

at INL led by Francois Viretta, and revealed a porosity of ~35% in the anode, which agrees with the nominal

porosity of the pristine anode. This finding points toward the regions of the anode where there is plated Li that

does not have a large change in porosity as compared to regions without plating. In addition, the deposited Li

is observed to be purely on the separator side, with a thickness of the order of 10s of μm. During the creation

of anode strips from the anode sheet, the plated Li was converted to lithium hydroxide (LiOH), which

enhanced the X-ray contrast between the graphite and Li, thus allowing for such characterization. Furthermore,

calculations were performed on the thickness of each species, with the graphite showing a fairly uniform

thickness of ~80 μm and the Li layer showing very uneven thickness. Additional calculations are ongoing, with

the aim of calculating the tortuosity factor, Brugemann exponent, and diffusion co-efficient ratio, as well as the

36

anisotropy of each of these factors based on the direction of the calculations from tomography (in-plane versus

through-plane directions).

Figure 27. microCT of a

Round 2 anode cycled at the 9C charging rate for 450 cycles.

(a) Projection of the tomographic reconstruction showing anode (gold) and Li

in the form of LiOH (purple). (b) Segmentation, with the Li layer on top and the graphite layer below, showing the pore structure in each layer.

3. Simultaneous Neutron and X-ray Tomography (NeXT)

Neutrons and X-ray-based, dual-mode tomography are advantageous for improving the estimation of the

sample composition due to the complementary interaction of the two imaging modalities with the sample. A

combination of X-ray and neutron imaging is useful because X-rays are sensitive to graphite in the anode,

whereas neutrons are sensitive to Li (see the schematic in Figure 28). The difference in the absorption cross-

section of neutrons for carbon (0.0035) and the natural abundance of lithium (70.5) is fairly high, making

neutron-based imaging a good candidate for directly detecting Li metal. The spatial resolution achieved by this

technique is ~15 μm, thus providing an intermediate step between the mm-scale XRD and microCT. We

performed NeXT experiments at BT-2 imaging beamline at the National Institute of Standards and Technology

Center for Neutron Research. The main cells characterized were the anode strips used for X-ray microCT, as

described earlier. In addition, as a trial, a full pouch cell (Round 2 electrodes, assembled at CAMP) was also

scanned in discharged condition and after formation cycles. The data processing is currently underway, with

the aim of obtaining the total volume of plated Li on the anode. In the future, this result will be considered a

substitute for X-ray microCT, especially at coarser length scales (on the order of 15 μm) due to the fewer

constraints it places on sample preparation.

37

Figure 28: Left: schematic depicting a graphite anode strip with different regions of lithium plating indicated by bright regions.

The strip was vacuum sealed in a glass capillary in a kapton tube. Right: top-view

diagram of the perpendicular neutron and X-ray beams in NeXT with the rotation

angle (yaw).

38

XCEL R&D: SLAC National Accelerator Laboratory

Contributors: Yi Cui, William Chueh, and Mike Toney

Approach

In recent years, lithium-ion batteries have become the battery technology of choice for portable devices, electric vehicles, and grid storage. While increasing numbers of car manufacturers are introducing electrified models into the market, range anxiety and the length of time required to recharge the batteries are still common concerns. The high currents needed to accelerate the charging process have been known to reduce energy efficiency and cause accelerated capacity and power fade. Among many of the fading mechanisms during fast charging, lithium plating is known to deleteriously affect the cycle life and safety of the lithium ion battery. Nondestructive operando diagnosis of lithium plating is of great significance for onboard application, as well as a technique to assist scientific study of fast charging. In the XCEL program, we have developed a platform based on uniaxial mechanical pressure measurement combining with electrochemical analysis, which enables us to detect lithium plating at a very early stage (the seeding stage) in real time. Graphite electrodes have a drastically different change of thickness during lithiation vs. during lithium plating. Graphite has a 10% volume change (3% uniaxial length change) after lithiation. The Round 2A electrodes have a coating thickness of 70 um with an area capacity of 2.84 mAh/cm2, which has only about a 2-um change in thickness after full lithiation. In contrast, if the same capacity is delivered by lithium plating, the thickness change will be 14.2 um without considering the mossy structure. The cathode side also experiences a small volume change during cycling; however, the change is negligible (~3% for NMC) compared to the anode side. Therefore, the volume change in the cell is dominated by the anode. The lithium-ion battery is a fixed-volume system; any volume change in the battery material will eventually translate to pressure change. Here we describe the new method we have developed: by analyzing the pressure change per charge (dP/dQ), lithium plating can be identified easily. The operando uniaxial pressure profile of a constrained pouch cell is measured by a load cell, as shown in Figure 29a. Figure 29b shows the typical operando pressure and the dP/dQ profiles versus the voltage curve of a 70-mAh NMC/Graphite 5-layer pouch cell during 0.5 C charging and discharging. The plateau on the pressure curve and the valley on the dP/dQ curve indicate the 2L->2 phase transition of graphite, during which there is minimum volume change. The data from the slow (0.5 C) charging reveal the upper boundary of dP/dQ (pink dashed line) when graphite undergoes intercalation chemistry. In contrast, when lithium plating happens during fast charging, dP/dQ quickly passes the threshold as shown in Figure 29d. Scanning electron microscopy (SEM) images of the graphite anode after different lengths of time after passing the threshold show that lithium metal first appears as droplets on the surface and then evolves to a dendritic structure (Figure 29c). This image reveals that this nondestructive technique is capable of detecting lithium plating in real time at a very early stage.

39

Figure 29. (a) The configuration of operando pressure measurement: a multilayer pouch cell is stacked with a metal plate and load cell and contained in a bench vise to define a fixed thickness. (b) A typical voltage profile

versus pressure and dP/d|Q| profiles are shown for a 70-mAh multilayer pouch cell under slow charging/discharging (0.5 C). (c) SEM images of the top surface of graphite electrodes under different charging

condition. From left to right: after 50 cycles under 0.5 C, the graphite surfaces are smooth and clean; after 1 min, the dP/dQ curve passes the threshold (SH) and lithium droplets nucleate on the graphite surface; after 3 min,

passing the SH, lithium dendrites can be clearly observed. (d) The dP/dQ curve during cycling in which slow 0.5 C cycles are followed by a fast-charging (3C) cycle. The blue and orange dots indicate the time points where the

cells were open for SEM imaging.

40

XCEL R&D: Detecting Lithium Plating during Fast Charging, and Graphite Anodes with

Directional Pore Channels Made by Freeze Tape Casting (Lawrence Berkeley National

Laboratory)

Contributors: Nitash Balsara, Vince Battaglia, Marca Doeff, Bryan McCloskey, Guoying Chen, Robert

Kostecki, Wei Tong, and Ravi Prasher

Background

The goal of the LBNL team is to detect the initial nucleation event using data available in typical battery

management systems. Identifying the initial onset of Li plating during fast charging is a multifaceted,

challenging “needle-in-a-haystack” problem. First, there may be a barely perceptible chemical signature of this

event. For example, the presence of lithium may lead to the sudden formation of a small amount of gases due

to irreversible reactions between the plated lithium and the electrolyte-soaked SEI. Second, identifying the

particular particle where the overlithiation and Li plating occur can be challenging at the immediate onset of Li

plating. Clearly, the particular particle that is overlithated first will depend on the state-of-charge of the

electrode, as it will be governed by the amount of lithium intercalated at that time and the lithium

concentration in the surrounding liquid electrolyte. The third and final challenge is identifying electrical

signatures of the overlithation event so that our work can be used in practical battery management systems. We

note here that the signatures may be different at different states-of-charge and that the electrical signature may

be inherently nonlinear, and thus not clearly evident in standard electrochemical measurements. Our objective

is to address all three of these challenges.

Characterization of the onset of lithium plating and its mechanism is, in practical graphitic lithium-ion battery

anodes, complicated by the three-dimensional porous composite nature of the anode and its interaction with an

electrolyte whose lithium concentration at the interface with the graphite varies as a function of depth within

the electrode during fast charging. We aim to determine some of the fundamental factors at play by performing

experiments using planar, highly oriented pyrolytic graphite (HOPG) samples. HOPG is as close to

monocrystalline graphite as is commercially available, and consists of a stack of graphite planes, with in-plane

grain sizes on the order of 0.1–1 mm. It is therefore possible to study the difference in lithium plating behavior

on edge and basal planes as a function of current density on these samples, initially and as a function of aging.

In addition, graphite anode architectures with low–tortuosity, open pores parallel to the ion transport direction

during cycling will be developed by a freeze tape casting method. Various types of commercial graphite,

carbon, and binder will be tested for slurry stability. The down-selected combinations will be used to produce

graphite anodes with controlled porosity, pore size, and layer thickness. The final product may serve as a drop-

in replacement for the conventional graphite anode, but with a goal of improved charge acceptance capability.

Modeling efforts of the Kandler group (NREL) will provide guidance to the desired overall porosity and pore

structure. Tomography data will be shared with NREL to improve model accuracy. The Jansen group

(Argonne) will aid with cycling the produced freeze tape cast graphite anodes to compare with conventional

counterparts.

Results

The Balsara group constructed Li metal/Superior Graphite SLC1520P(D50 = 16.94 µm) microtomography

cells with a Whatman QMA separator soaked in 1.0 M LiPF6 1:1 wt% EC:EMC. In an initial study, a micro-

cell was imaged on the X-ray microtomography beamline at LBNL before and after three formation cycles of

C/10 intercalation and C/5 deintercalation. The results in Figure 30a show the presence of lithium

microstructures at the interface between the separator and Li metal after formation cycles. Figure 30b shows

the formation of moss-like lithium in the same cell after 20 cycles of 6C intercalation and C/2 deintercalation.

41

In future work, we will increase pressure via a spring within the cell and replace the Whatman QMA with

Celgard 2500 to help suppress these moss-like formations. Furthermore, cells made with free-standing graphite

electrodes will be imaged with tomography and electrochemically tested within the micro-cell.

The Tong group fabricated freestanding graphite electrodes without the use of copper current collector. These

customized graphite electrodes are specifically developed to eliminate the strong X-ray signal from copper in

X-ray tomography in collaboration with the Balsara group. A previous protocol was developed for the graphite

(SLC1506T, from Superior Graphite) with an average particle size of 7 µm. Detecting metallic lithium in such

thin graphite films (~35 µm) with a relatively small particle size has proven to be challenging. In this quarter, a

second graphite (SLC1520T, from Super Graphite) with a relatively large particle size (~17 µm) was made.

However, the previous electrode fabrication procedure resulted in severe film cracking. A variety of

parameters, including slurry/electrode formulation and film thickness, were tested to obtain a physically robust

film for an ~17-µm graphite. Films with a high content of acetylene black exhibit a high initial lithiation

capacity due to the contribution of acetylene black (Figure 31a). Further efforts were made to produce good

freestanding graphite film using 17-µm graphite by reducing the content of the acetylene black and plasticizer.

Films obtained at an optimal procedure result in a reversible delithiation capacity of 356 mAh/g between 1.5

and 0.01 V (Figure 31b). In next quarter, these physically robust graphite films will be subjected to designated

electrochemical testing for X-ray tomography characterization.

a

)

Figure 30. (a) Slice of an X-ray tomogram of micro-cell after three formation cycles (C/10 intercalation, C/5 deintercalation), and (b) slice of an X-ray tomogram of the same cell after 20 cycles (6C intercalation,

C/2 deintercalation).

b

)

Figure 31. Voltage profiles of a free-standing graphite electrode with a formulation of (a) 60% graphite (~17 µm), 20% acetylene black and 20% PVdF and (b) 57% graphite (~17 µm), 12% acetylene black and 31% PVdF. The inset

shows the free-standing tape prepared at an optimal procedure. All half cells with graphite films of ~100 µm thickness are cycled between 1.5 and 0.01 V at C/20.

42

Previously, the McCloskey group described an ex situ differential electrochemical mass spectrometry (DEMS)

titration for quantifying plated Li and an in-operando impedance technique to detect plating. Work using the

titrations to cross-validate plating detection in techniques such as impedance and XRD (in collaboration with

Weker/Toney, SLAC) is ongoing. Here we introduce a differential voltage technique that appears promising

for detecting the onset of in-situ Li plating after fast charge. When Li plating occurs on the surface of lithiated

graphite (LixC6), the measured potential will be a mixed potential of the Li and LixC6. During cell relaxation

after fast charge, small amounts of plated Li will chemically intercalate into graphite, resulting in a decay of

the mixed Li/LixC6 potential into a pure LixC6 potential. The presence of this decay feature in the open-circuit

voltage (OCV) has been shown to indicate Li plating. A differential analysis of the OCV with respect to time

shows a more pronounced plating signature. Our differential OCV technique is distinct from differential

capacity work by Bloom et al. (Argonne) and Tanim et al. (INL) because it analyzes the OCV after charge

instead of the voltage profile during discharge. Progressive coin cell cycling enables us to use this differential

OCV technique to estimate the graphite SOC at which plating begins in graphite/Li cells. Figure 32a shows the

cycling protocol used to generate abundant OCV data corresponding to different C-rates and the final SOC of

the graphite. Each cycle has four steps: constant-current (CC) intercalation to a varied SOC, 30 min OCV rest,

CC deintercalation at C/5 to 1.5 V, and a final 30 min OCV before starting the next cycle. For the first seven

cycles (Figure 32a, black), the graphite intercalation was at 1C with SOC cutoffs of 10% for cycle 1, 20% for

cycle 2, etc., up to 70% for cycle 7. The next seven cycles are identical except with an intercalation rate of 2C

(Figure 32a, blue). Although plating is not expected at the 1C and 2C intercalation rates, those cycles do

provide a baseline of OCV profiles for comparison with the 6C cycles.

Comparing the OCV data extracted from the cycling in Figure 32 can show the SOC onset of Li plating. All

OCV profiles and derivatives are grouped by SOC and plotted in Figure 33. At 20–30% SOC (left), the

differential OCV profiles are similar for all c-rates, which indicates minimal plating at 6C after intercalation to

30% SOC. At 40% SOC, we attribute a shift in the 6C differential OCV to the onset of Li plating. This feature

becomes more prominent at higher SOC when we expect increased Li plating. Coulombic efficiencies for all

cycles, defined as a ratio of the deintercalation capacity to the intercalation capacity, support our hypothesis

that this feature corresponds to Li plating (Figure 32b). At 40% SOC, where the differential OCV suggests

plating has occurred, we also see a slight drop in the coulombic efficiency that becomes larger drop at higher

SOC. In future work, we plan to cross-validate this plating feature with our ex situ electrode titrations and to

verify that this technique is applicable for full graphite/NMC cells.

Figure 32. (a) Graphite/Li cell cycling to generate OCV data after fast charge for various SOC and c-rates. (b) Coulombic efficiencies (CEs) for these half-cell cycles. We believe the drop in CEs at 6C and 40% SOC

corresponds to the onset of Li plating.

a

) b

)

43

The Chen group carried out in situ microscopic characterization of graphite lithiation and delithiation. An

optical cell with a graphite working electrode (SLC1605T, Argonne), Li metal counter and reference

electrodes, and a polyvinylidene difluoride (PVDF) membrane separator soaked with the Gen2 electrolyte

(1.2 M LiPF6 in EC: EMC 3:7 wt. %) was used for the study. Details on the cell design were described in the

previous quarterly report. Figure 34 shows the preliminary results obtained during the first charge and

discharge at a C/10 rate. The voltage profile of the lithiation process (Figure 34a) is consistent with what was

reported on typical graphite half-cells in the literature, confirming that the optical cell is suitable for the in situ

studies. As the lithiation process proceeded, the graphite electrode experienced a progression in colors that

ranged from grey, blue, red, and gold, corresponding to the different lithiated stages of C, LiC18, LiC12, and

LiC6, respectively. Figure 34b and c compare the in situ optical images taken from the graphite electrode

before and after the first lithiation, as indicated on the voltage profile shown in Figure 34a. At the pristine

state, large graphite particles with an average particle size of ~10–15 µm were seen uniformly in grey color. At

the end of the first lithiation process at 0.01 V (vs. Li/Li+), gold-colored LiC6 particles along with red-colored

LiC12 were broadly visible. Li dendrites were not observed under current conditions.

The optical cell suffered large resistance during the delithiation process, which prevented the graphite

electrode from full extraction of the intercalated Li+ (Figure 34a). This was mostly a result of electrolyte loss

from the current cell. In future work, we will optimize the various fabrication steps, especially electrolyte

soaking and epoxy sealing to prevent electrolyte leaking and improve total charge and discharge capacities, as

well as the couloumbic efficiency of the optical cell. Studies using various current densities will be performed

to investigate the correlation between charging rate and Li dendrite formation and propagation under operando

conditions.

Figure 33. OCVs (top row) and differential OCVs (bottom row) during relaxation after graphite intercalation at 1C, 2C, and 6C. Each window shows the OCVs after intercalation to the designated SOC.

44

The Prasher group is working on thermal analysis of fast-charging full cells. Figure 35a shows that the

interface thermal resistance and total thermal resistance increases by 183% and 50% after 120 cycles charged

and discharged at C/3, respectively. The total thermal resistance increases rapidly in the first 60 cycles and

tends to be stable after that. To understand the evolution of thermal resistances, we performed ex situ

measurements of interface resistance between the separator and electrode (see Figure 35b). Preliminary results

show that the thermal interface resistance levels between the separator and dry electrodes are 4.1×10-5 m2K/W

Figure 35. (a) The evolution of thermal resistance in 100+ cycles, (b) 3w sensor on separator for ex situ measurements, and (c) the impact of external pressure on thermal contact resistance.

Figure 34. (a) First cycle charge/discharge voltage profiles of the in situ optical cell when cycled at a C/10 rate between 0.01 and 3 V, (b) and (c) in situ optical images of the graphite electrode

before and after the first lithiation process.

(b)

25 µm

0 50 100 150 200 250 300 350

0.0

0.5

1.0

1.5

2.0

2.5

3.0

Capacity (mAh g-1)

Po

ten

tia

l (V

vs. L

i/L

i+)

1st lithiation

1st delithiation

(a)

(c)

25 µm

b

c

45

and 3.63±1.06 × 10-5 m2K/W for the cathode and anode, respectively. These measurements were performed

with an external pressure of 3 psi. In addition, we studied the impact of external pressure on the thermal

interface resistance (see Figure 35c). The thermal contact resistance decreases 53% from 1.35±0.25 × 10-4

m2K/W to 6.38±0.38 × 10-5 m2K/W as the pressure varies from 0 to 3 psi. In future work, thermal

measurements at different temperatures and fast charge rates will be performed using pouch cells with new

thermal sensors (see Figure 35c).

To enable electrochemical experiments on HOPG that involve only the basal plane, the Kostecki group utilized

a beaker cell with an O-ring. A fresh surface was created on a piece of HOPG by exfoliation with tape, and the

HOPG was sealed against the O-ring and measured against a Li counter and reference electrodes. Two CV

cycles were performed in the range 0.01–1.5 V vs. Li in order to achieve an SEI-coated graphite, which is the

system we wish to interrogate. Then, a current ramp was applied to determine the current density that can be

sustained above 0 V. The results are shown in Figure 36. The CV formation cycles exhibit a pronounced peak

at about 0.5 V vs. Li, which is indicative of carbonate electrolyte reduction. The lack of a peak at lower

voltages, and the absence of an anodic current during the sweep back to 1.5 V, indicates that no reversible

process such as Li intercalation is occurring at significant current densities. It can also be seen that electrolyte

reduction and the concomitant SEI formation is incomplete after the two CV cycles, and continues during the

current ramp. This process cannot sustain more than 46 µA/cm2; and above this current density, the potential

drops below 0 V vs. Li, at which point plating cannot be ruled out.

Figure 36. Left: CV formation of the basal HOPG/Li half-cell. Right: Current ramp behavior of basal HOPG/Li half-cell after CV formation.

Similarly, the edge plane was examined by immersing a piece of HOPG in a beaker cell with a Li counter and

reference electrodes. Although the basal areas were not isolated in this case, it is evident that the currents are

so much higher that the basal contribution is negligible. In this case shown in Figure 37, we applied two CV

formation cycles and a current ramp (black), then three CV cycles and a current ramp (red), and three more CV

cycles and a current ramp (green). The CV cycling reveals a cathodic electrolyte reduction peak at 0.5 V–0.6 V

vs. Li, which disappears by the final CV cycling step, indicating that the SEI stabilized. The feature visible at

that voltage during the first current ramp is similarly absent in subsequent current ramps. More importantly, a

large cathodic peak occurs at low voltage, and a concomitant anodic peak appears on sweeping from low to

high voltage. From the first CV procedure, we can see that Li intercalation occurs at up to 14 mA/cm2 during

the first cycle and at about 6 mA/cm2 during subsequent cycles. Current ramps sustain about 4 mA/cm2 before

leading to negative voltages where plating can occur, in reasonable agreement with the CV result. Finally, all

these effects appear to stabilize after the second CV cycle, and no deterioration of sustained lithiation current

due to the — admittedly mild — aging was observed. We have found that basal and edge graphite can sustain

about 46 µA/cm2 and 4000 µA/cm2, respectively, while remaining in the plating-free regime above 0 V vs. Li,

although only the latter is clearly an intercalation current. In the next quarter, we will focus on reproducing

these results and investigating the effect of cycling/aging in greater detail.

Basal: V<0 at 46 µA/cm2

46

Figure 37. Top: CV cycles before first current ramp (left), before second current ramp (middle), and before third

current ramp (right). Bottom: Cycling procedure (left), current ramps (middle), and image of cell (right).

The Doeff group investigated dimethyl sulfoxide (DMSO) as the freeze tape casting solvent. PVDF binder was

used, and superior wetting/dispersion of graphite/carbon were observed in the organic solvent. It was possible

to formulate 50 vol.% solid loading slurry, 5 vol.% less, compared to what could be reached in water-based

slurries (55 vol.%). However, the high melting point of DMSO (19°C) complicated the casting process,

resulting in premature freezing of the slurry before entering the freezing bed. A heating lamp was used to raise

the temperature of the slurry; yet once the initial part of the slurry froze, the freezing spread to the entire

sample. Higher temperature and better temperature control in terms of uniformity may be necessary for freeze

tape casting of a DMSO solvent to work. Water/DMSO mix-solvents were tested as well, but the addition of

water, even at small fractions (<5% of solvent), exacerbated the dispersion of graphite.

In parallel, freeze tape casting of water-based 1506T graphite slurries were conducted to produce samples for

preliminary electrochemical cycling. This will help us pre-screen the performance of the produced samples

before sending them to Argonne.