Factors Governing Life of High-Energy Lithium-Ion Cells

Factors Governing Life of High-Energy Lithium-Ion Cells D.P. Abraham IBA 2013 March 11, 2013 Barcelona, Spain

Research sponsors are both Government and Private Sector 2

Diagnostics Overview Use of characterization tools (electrochemical, physicochemical, mechanical, acoustic) to explain the electrochemical behavior of materials, electrodes, and cells Every part of the cell is examined electrodes (active and inactive components in coating, current collector, tabs), electrolyte, separator, etc. To identify constituents and mechanisms responsible for cell performance and performance degradation To recommend solutions that improve performance and minimize performance degradation of materials, electrodes, and cells 3

Electrochemical couples examined ATD-1 LiNi 0.8 Co 0.2 O 2 /MCMB-6 graphite HEV ATD-2 LiNi 0.8 Co 0.15 Al 0.05 (Al 0.1 )O 2 /Mag-10 graphite HEV 150 Wh/kg ATD-3 Li 1.04 Ni 1/3 Mn 1/3 Co 1/3 O 2 /MCMB-10 graphite HEV ABR-1 Li 1.2 Ni 0.15 Mn 0.55 Co 0.1 O 2 /A-12 graphite PHEV 300 Wh/kg 0.5Li 2 MnO 3 0.5LiNi 0.375 Mn 0.375 Co 0.25 O 2 Integration and interconnection of LiMO 2 -like (rhombohedral) and Li 2 MnO 3 (monoclinic) structures at atomic level EC:EMC (3:7 by wt.) + 1.2M LiPF 6 electrolyte 4

LiMO 2 (M=Ni,Co,Al) Crystal Structure TM O Li O TM Li atoms enter and exit through edge planes (2-D motion) c c a Li 2 MnO 3 = Li(Li 1/3 Mn 2/3 )O 2 Z-contrast STEM images that show Li ordering in TM planes of Li 2 MnO 3 D.P. Abraham, Advanced Materials 22 (2010) 1122-1127 5

Electronic conductivity of ABR-1 oxide powder is more than 3 orders of magnitude lower than that of NCA, and about 2 orders of magnitude lower than that of NMC 1.E-03 3.1 x 10-3 Oxide Composition Conductivity, S/cm 1.E-04 1.E-05 1.E-06 1.3 x 10-5 NCA LiNi 0.8 Co 0.15 Al 0.05 O 2 NMC333 Li 1.04 Ni 1/3 Mn 1/3 Co 1/3 O 2 NMC992 Li 1+x Ni 0.45 Mn 0.45 Co 0.1 O 2 HE5050 Li 1.2 Ni 0.15 Mn 0.55 Co 0.1 O 2 5 x 10-7 2 x 10-7 1.E-07 NCA NMC333 NMC992 HE5050 W. Lu, Argonne Electrode preparation with HE5050 oxide powder is expected to be more challenging than that for NCA and NMC333 oxide powders 6

Chemistry of ABR-1 cells Li 1.2 Ni 0.15 Mn 0.55 Co 0.1 O 2 = 0.5Li 2 MnO 3 0.5LiNi 0.375 Mn 0.375 Co 0.25 O 2 ABR-1S(+) Positive Electrode: 86%wt Li 1.2 Ni 0.15 Mn 0.55 Co 0.1 O 2 8%wt Solvay 5130 PVDF binder 4%wt Timcal SFG-6 graphite 2%wt Timcal Super P ABR-1S(-) Negative Electrode: 89.8 %wt ConocoPhillips A12 graphite 6%wt KF-9300 Kureha PVDF binder 4 %wt Timcal Super P 0.17 %wt Oxalic Acid 6.64 mg/cm 2 active-material loading density 5.61 mg/cm 2 active-material loading density 37.1% electrode porosity 26% electrode porosity 35-µm-thick coating 40-µm-thick coating 15-µm-thick Al current collector 1 µm 10-µm-thick Cu current collector 5 µm B. Polzin, A. Jansen Argonne Oxide primary particles consist of randomly oriented plate-like grains A12 graphite particles, potato-shaped morphology; surface-treated D. Miller, Argonne 7

Identifying sources of performance degradation is the first step to designing long-life cells Significant impedance rise during cell cycling/aging Arises mainly from the positive electrode Significant capacity fade during cell cycling/aging Originates at the positive electrode Manifests itself at the negative electrode thick negative electrode SEI Voltage profile changes observed during cell cycling/aging Arises from crystal structure changes 8

Positive electrode has to be cycled over a wider voltage window to obtain higher capacities 4.8 4.4 ABR-1 Cell oxide activation 268 mah/goxide Cell Voltage, V 4.0 3.6 3.2 62% 90 mah/goxide 92% 2.8 57 mah/goxide 2.4 2.0 3 4.1V, 1 st cycle 2 4.6V, 1 st cycle 0 0.4 0.8 1.2 1.6 2 Cell Capacity, mah/cm 2 248 mah/goxide NCA(+)/Gr(-) cells yield a discharge capacity of 165 mah/g (3-4.1V) NMC333(+)/Gr(-) cells yield a discharge capacity of 140 mah/g (3-4.1V) Ref: Li et al., J. Electrochem. Soc. 160 (2013) A3006 9

Cells show capacity loss on aging After 30 cycling in the 2.5-4.4V voltage window up to 1500 cycles Cell capacity decreases on cycling, even at 30 C 4.8 4.4 Formation Initial 100 cycles 400 cycles 800 cycles 1200 cycles 1500 cycles 4 Full Cell Voltage, V 3.6 3.2 2.8 245 mah/goxide 2.4 95 mah/goxide Capacity Loss Offset 2 0 0.4 0.8 1.2 1.6 Cell Capacity, mah/cm 2 Performance degradation is faster at higher upper-cutoff voltages, at higher temperatures, and for wider voltage windows Ref: Li et al., J. Electrochem. Soc. 160 (2013) A3006 10

Cells show impedance rise on aging After 30 cycling in the 2.5-4.4V voltage window up to 1500 cycles 50 45 FULL 40 1500 -Z(Im), Ω-cm 2 35 30 25 20 0 100 400 800 1200 15 10 5 0 0 50 100 150 200 250 Z(Re), Ω-cm 2 3.75V, 30 C 100KHz-10mHz Mid-frequency arc increases when the UCV is higher (4.6V for example) Capacity data obtained at high rates is strongly affected by cell impedance Ref: Li et al., J. Electrochem. Soc. 160 (2013) A3006 11

Full cell impedance increase is mainly from the positive electrode D. Dees, A. Jansen Argonne S P N RE 50 45 Oxide-carbon 50 45 40 1 khz 40 -Z(Im), Ω-cm 2 35 30 25 20 15 10 0 100 400 800 1200 Oxideelectrolyte 0.6 Hz 1500 35 30 25 20 15 10 5 10 khz POS 5 NEG 0 0 50 100 150 200 Z(Re), Ω-cm 2 0 0 50 100 150 200 Z(Re), Ω-cm 2 The positive electrode impedance rise arises from the oxide-carbon (highfrequency arc) and oxide-electrolyte (mid-frequency arc) interfaces. Ref: Li et al., J. Electrochem. Soc. 160 (2013) A3006 12

Data from harvested electrodes (vs. Li) shows that positive electrode contribution to true capacity fade is small 5.0 4.5 2 4.7V vs. Li, 30 C, 7.5 ma/g 4.0 Voltage, V 3.5 3.0 Voltage Fade 2.5 2.0 Fresh Characterization 300 cycles 1500 cycles 0 50 100 150 200 250 300 Capacity, mah/g The small capacity loss could be due to oxide particle isolation that may result from loss of electronic connectivity (loss of oxide-carbon contact) or ionic connectivity (particle surface films or surface structure changes) Ref: Li et al., J. Electrochem. Soc. 160 (2013) A3006 13

Data from harvested negative electrodes (vs. Li) shows that the graphite bulk is not damaged by cycling/aging 1.8 1.6 1.4 1.2 Fresh Characterization 300 cycles 1500 cycles 0.30 0.25 2 0V vs. Li, 30 C, 6.8 ma/g enlarged Voltage, V 1.0 0.8 0.6 0.4 0.20 0.15 0.10 0.05 0.00 0 50 100 150 200 250 300 350 400 0.2 0.0 0 50 100 150 200 250 300 350 400 Capacity, mah/g The small capacity loss could be due to active particle isolation that may result from thick SEI films Ref: Li et al., J. Electrochem. Soc. 160 (2013) A3006 14

Cells containing the Li 4 Ti 5 O 12 (-) electrodes (coupled with the baseline positive electrode) show negligible capacity fade on cycling: 0.75 3.15V, 30 C 1.8 1.7 100 1.6 ~235 mah/g Capacity, mah/cm 2 1.5 1.4 1.3 1.2 1.1 ~180 mah/g C/10 ~C/2 80 60 40 Coulombic Efficiency, % 1 0.9 0.8 ABR-1S(+)//LTO(-) Discharge Capacity Pos. cycling ~2.3 4.7V vs. Li Coulombic Efficiency 0 0 100 200 300 400 500 Cycle Number These data confirm that the graphite-based negative electrode is the main contributor to capacity fade in our baseline cells 20 Ref: Li et al., J. Electrochem. Soc. 160 (2013) A3006 15

SEI formation/dissolution/reformation reactions during cell cycling results in Li trapping 1 µm Fresh 1 µm 1500 cycles SuperP graphite Transition metals (Mn, Ni, Co) from the oxide(+) electrode accumulate at the graphite(-) electrode and are believed to accelerate capacity fade Ref: Li et al., J. Electrochem. Soc. 160 (2013) A3006 16

Positive and Negative Electrodes were reconstituted by altering oxide/carbon/binder ratios and calendaring conditions Electrodes were prepared at CFF in 2012 Electrodes were prepared at CFF in 2012 L7(C45+) A002 (CFF) Positive Electrode: Negative Electrode: 92 wt% Li 1.2 Ni 0.15 Mn 0.55 Co 0.1 O 2 (HE5050) 91.8 %wt ConocoPhillips: CGP-A12 graphite 4 wt% Solvay 5130 PVDF binder 6%wt KF-9300 Kureha PVDF binder 4 wt% Timcal C45 2 %wt Timcal C45 0.17 %wt Oxalic Acid 5.89 mg/cm 2 active-material loading density 5.16 mg/cm 2 A12 graphite loading density 36.1% electrode porosity 38.8% electrode porosity 26-µm-thick coating 43-µm-thick coating 20-µm-thick Al current collector 10-µm-thick Cu current collector Only 4wt% carbons and 4 wt% binder Only 2 wt% carbons Electrolyte: EC:EMC (3:7 by wt.) + 1.2M LiPF 6 Separator: Celgard 2325 B. Polzin, S. Trask, A. Jansen, Argonne 17

Related to information from previous slide By modifying electrode constitution and by using electrode additives we can dramatically reduce cell impedance rise -Z(Im), ASI 20 18 16 14 12 10 8 Initial 50X 100X 300X 500X 700X 900X 1000X Full Cell: L7(C45)+/A002-2.5 4.4V, 30 C ~C/2, 1000 cycles 2 wt% LiDFOB added to baseline electrolyte 6 4 2 0 0 5 10 15 20 25 30 35 Z(Re), ASI Cell impedance rise is reduced by an order of magnitude (relative to baseline cell chemistry) 18

XPS data show that LiDFOB generates surface films on the positive electrode that inhibits cell impedance rise pristine cathode Formed_Gen2 Aged_Gen2 Formed_2wt% LiDFOB Aged_2wt% LiDFOB C1s O1s Pristine cathode Formed_Gen2 Aged_Gen2 Formed_2wt% LiDFOB Aged_2wt%LiDFOB oxide C-F (PVdF) 294 293 292 291 290 289 288 287 Binding Energy, ev 286 285 284 283 282 537 536 535 534 533 532 531 Binding Energy, ev 530 529 528 527 Our data indicate that LiDFOBalso reduces on the graphite negative electrode, enhances the negative electrode SEI, and inhibits cell capacity fade Ref: Zhu et al., J. Electrochem. Soc. 159 (2012) A2109 19

Alumina-coating of positive electrode and/or alumina addition to positive electrode improves cell capacity retention 1.05 1 0.995 Normalized Capacity 0.95 0.9 0.85 0.8 no coating ABR-1S(+)/(-) 0.4nm coating on positive electrode 0.96 0.93 0.86 0.8 ALD-coated electrodes Electrode coating thickness is an estimate based on the thickness measured on a planar Si substrate 0.75 0.7 1.0nm coating on positive electrode 3.4nm coating on positive electrode 5wt.% Al2O3 added to positive electrode Full Cells, 2.2-4.6V 30 C, ~C/3 rate 0 10 20 30 40 50 60 70 Cycle Number Alumina reduces dissolution of Mn, Ni, and Co from the positive electrode by acting as a HF-getter. Incorporation of Al-bearing species may further help stabilize the SEI. Ref: Bettge et al., J. Power Sources 233 (2013) 346 20

Positive electrode potential fades on cycling Open Circuit Voltag, V vs. Li/Li+ 4.8 4.4 4 3.6 3.2 2.8 2.4 GITT study @ ~C/250, 0.16mA (charge 10 min., rest 100 mins.) ABR-1(+) vs. Li, 2 4.7V, 20 ma/g 2-4.7V, after 2X 2-4.7V, after 80X 0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2 2.2 Capacity, mah/cm2 Voltage Fade not a kinetic phenomenon Voltage Fade (VF) is non-uniform VF on charge is greater than that on discharge Voltage fade is accelerated by higher temperature/upper cutoff voltage 21

Voltage profile changes on cycling caused by intermixing of Li and TM atoms Fresh sample Li(Li0.2Mn0.4Co0.4)O2 Cycled sample Transition Metal planes TM in Li planes Li-Mn ordering in TM planes results in Li2MnO3-like and LiCoO2-like areas Periodic intensity in Li layers is characteristic of spinel structure Z-contrast STEM 22 Abraham, Chem. Mater. 23 (2011) 2039-2050

Approaches to designing long-life cells Significant impedance rise during cell cycling/aging Can be reduced by reformulating positive electrode constitution, modifying oxide surface through coatings (pre-treatment), using electrolyte additives (in-situ), and by altering the cycling window Significant capacity fade during cell cycling/aging Can be minimized by fixing problems at the positive electrode, such as transition metal dissolution from the oxide Voltage profile changes observed during cell cycling/aging Solutions may include oxide surface modification, alternative synthesis techniques, oxide composition modification 23

Acknowledgements Support for this work is from DOE-EERE, Office of Vehicle Technologies Peter Faguy, Dave Howell, Tien Duong Argonne colleagues ABR collaborators National Lab colleagues & University Collaborators The presentation has been created by UChicagoArgonne, LLC, Operator of Argonne National Laboratory ( Argonne ). Argonne, a U.S. Department of Energy Office of Science Laboratory, is operated under Contract No. DE-AC02-06CH11357. 24