Journal of The Electrochemical Society, 166 (13) A3031-A3044 (2019) A3031

A Wide Range of Testing Results on an Excellent Lithium-Ion Cell

Chemistry to be used as Benchmarks for New Battery Technologies
Jessie E. Harlow, 1,2 Xiaowei Ma,1,2 Jing Li,1,2 Eric Logan,1,2 Yulong Liu,1,2 Ning Zhang,1,2
Lin Ma,1,2 Stephen L. Glazier, 1,2 Marc M. E. Cormier,1,2 Matthew Genovese,1,2,∗
Samuel Buteau,1,2 Andrew Cameron,1,2 Jamie E. Stark,1,2 and J. R. Dahn 1,2,∗∗,z
1 Department of Physics and Atmospheric Science, Dalhousie University, Halifax, N.S B3H 4R2, Canada
2 Department of Chemistry, Dalhousie University, Halifax, N.S B3H 4R2, Canada

We present a wide range of testing results on an excellent moderate-energy-density lithium-ion pouch cell chemistry to serve
as benchmarks for academics and companies developing advanced lithium-ion and other “beyond lithium-ion” cell chemistries to
(hopefully) exceed. These results are far superior to those that have been used by researchers modelling cell failure mechanisms and as
such, these results are more representative of modern Li-ion cells and should be adopted by modellers. Up to three years of testing has
been completed for some of the tests. Tests include long-term charge-discharge cycling at 20, 40 and 55°C, long-term storage at 20, 40
and 55°C, and high precision coulometry at 40°C. Several different electrolytes are considered in this LiNi0.5 Mn0.3 Co0.2 O2 /graphite
chemistry, including those that can promote fast charging. The reasons for cell performance degradation and impedance growth
are examined using several methods. We conclude that cells of this type should be able to power an electric vehicle for over 1.6
million kilometers (1 million miles) and last at least two decades in grid energy storage. The authors acknowledge that other cell
format-dependent loss, if any, (e.g. cylindrical vs. pouch) may not be captured in these experiments.
© The Author(s) 2019. Published by ECS. This is an open access article distributed under the terms of the Creative Commons
Attribution 4.0 License (CC BY, http://creativecommons.org/licenses/by/4.0/), which permits unrestricted reuse of the work in any
medium, provided the original work is properly cited. [DOI: 10.1149/2.0981913jes]

Manuscript submitted July 30, 2019; revised manuscript received August 16, 2019. Published September 6, 2019.

Lithium-ion cells are being used in grid energy storage and in to extend battery life.” Jana et al. compare their models to data pre-
electric vehicles where long lifetime is extremely important. At this sented in the literature on NMC/graphite Li-ion cells by Ecker et al.,2
time, many of these applications are generally not as demanding as Schmitt et al.,6 Schuster et al.,7 among others and show excellent
they might seem. This is because it can be rare that cells are subjected to agreement. In 2018, Dubarry et al.,8 reviewed calendar aging studies
100% depth of discharge cycling again and again on a daily or several on a variety of Li-ion cell chemistries, including NMC/graphite and
times daily basis. For example, Pearre et al.1 show that the average LiFePO4 (LFP)/graphite, among others. Dubarry et al., use data from
daily driving range for 484 gasoline vehicles in the US was 32.6 miles references2,6,7 and9 for NMC/graphite cells, data from Naumann et
(52 km) and that the vast majority of daily range need is in the 0–50 al.,10 among others, for LFP cells, and data from Keil et al.,11 among
mile range (0 – 80 km). For an EV with a 300 or 400 km range this others, for NCA cells.
represents about 25% DOD on a daily basis. Many researchers have The cycle and calendar lifetime of NMC/graphite cells has im-
shown that the lifetime of lithium-ion cells measured in cumulative proved dramatically since the time references2,5,6,8 were published. In
charge throughput is much longer when low DOD duty cycles are addition, many researchers, OEMs and battery users believe the cy-
applied (e.g.).2,3 cle and calendar lifetime of LFP/graphite cells are superior to that of
This situation may change with the proposed introduction of “robo NMC/graphite cells. We believe it is important to disseminate state-of-
taxis”, long haul electric trucks and vehicle-to-grid applications. In the-art data for NMC/graphite cells to aid future modelling efforts and
the former, vehicles will be driving all day, much like a conventional to provide modern benchmarks for both old and new cell technologies.
taxi and undergoing nearly 100% DOD cycling. Long haul trucks will In this paper, testing results on LiNi0.5 Mn0.3 Co0.2 O2 /artificial
almost certainly run in near 100% DOD situations. Cells in vehicles graphite (NMC532/AG) cells are presented. The NMC532 used in
tethered to the grid will be racking up charge-discharge cycles even these cells is “single crystal” NMC532 as described in the papers by
when the vehicle is not moving. Clearly EVs destined for vehicle- Jing Li et al.12 and Lin Ma et al.13–15 The artificial graphite (AG) used
to-grid applications, robo taxis or long haul trucking, would favor a in these cells is provided by Kaijin (China) and is grade AML-400.
lithium-ion chemistry that could deliver many more charge-discharge Over the last 10 years the Dalhousie research group has tested a wide
cycles in a decade than an EV that was destined for typical commuter variety of Li-ion cells with various positive electrodes, negative elec-
driving where high energy density to give the longest driving range for trodes and electrolytes.12–19 Of all the cells tested, the ones with the
weekend trips might be emphasized. Electric buses represent another longest lifetime are the single crystal NMC532/AG cells with opti-
application where duty cycles approaching 100% DOD are used on a mized electrolytes reported here. Therefore, we believe it is valuable
daily basis.4 to present a comprehensive summary of this test data for others to
In order to make reasonable projections about lifetime, it is insuf- benchmark against.
ficient to simply do charge-discharge cycle testing at high rate (for The data in this paper focuses on long-term tests at 20, 40 and 55°C.
example 1C). For example, if a particular cell type can provide 1000 These temperatures were selected based on historical precedents in
100% DOD charge-discharge cycles at 1C, this could be claimed to our group. We do note that many Li-ion cell manufacturers and other
be able to power a 400 km range EV for 400,000 km. In such a test- researchers present data at 25, 30, 45, 50, 55 and 60°C, some of which
ing scenario, 10–12 cycles per day can be collected and the entire are offset from our data. In the long-term cycle testing, preference is
test would only take about 100 days. Instead careful considerations given to studies over 100% DOD which is between 3.0 and 4.3 V for
of aging during cycling and aging during storage must be made as the cells considered here. However, some testing results are given for
has been done by many researchers. For example, in 2019 Jana et al.5 lower DOD ranges such as 3.0 to 4.2 V and 3.0 to 4.1 V to show the
show how “the identification of the individual degradation contribu- advantages that charging to a lower voltage can bring in both cycling
tions enables the possibility of tailoring the charge/discharge sequence and storage.
To capture the attention of the reader, Figure 1 compares long-term
charge-discharge cycle performance of the cylindrical NMC/graphite
∗ Electrochemical Society Student Member.
cells (Sanyo UR18650E) tested by Ecker et al.2 to some of the pouch
∗∗ Electrochemical Society Fellow. cells in this work. Details of the cells used in this work will be given
z
E-mail: jeff.dahn@dal.ca later, as will a full set of performance data. The 100% DOD cycling

Downloaded on 2019-09-07 to IP 100.42.241.162 address. Redistribution subject to ECS terms of use (see ecsdl.org/site/terms_use) unless CC License in place (see abstract).
 A3032 Journal of The Electrochemical Society, 166 (13) A3031-A3044 (2019)

100 100 20oC 4.2 and 4.3 V

Capacity remaining (%)

90 90
20oC 3.6 V
80
Ecker - 0-100% 4.2V 1C 35oC 80

Capacity remaining (%)

Ecker - 25-75% 1C 35oC
70 Ecker - 40-60% 1C 35oC 40oC 4.2 V
100
This work - 0-100% 4.3V C/3 40oC 40oC 4.3 V
60 This work - 0-100% 4.3V 1C 20oC
90 45 oC 3
.6 V
50 Black - Naumann et al. (LFP/graphite)
80 Green - This work (NMC532/graphite)
40
0 1000 2000 3000 4000 5000 100
Equivalent Full Cycles

Figure 1. Long-term cycling data plotted as percent initial capacity versus 90

equivalent full cycles for NMC/graphite cells as described in the legend. The 55oC 4.2 V
data from this work for 100% DOD cycling was collected to an upper cutoff 55oC 4.3 V
potential of 4.3 V. The data from Ecker et al.,2 used 4.2 V as 100% state of 80 60 oC 3
.6 V
charge. The purple and green data (this work) should be compared to the black
data (Ecker et al.). Data for restricted range cycling (i.e. 25 - 75% SOC and 0 200 400 600 800
40 -60% SOC) for the cells in this work is not available but is expected to be Storage time (days)
far better than the data shown for 0 – 100% DOD cycling by analogy with the
cells tested by Ecker et al.
Figure 3. Percent capacity remaining versus storage time for LFP/graphite
cells (Sony US26650FTC1) from Naumann et al.10 and for the cells in this work
as determined by periodic reference performance testing (RPT). The voltages
of the cells in the present work is far superior to that of the Sanyo cells and temperatures at which the cells were stored are given in the legends.
even though the upper cutoff potential was 4.3 V compared to 4.2 V.
Figure 2 shows long-term storage data of NMC/graphite cells stud-
ied by Ecker et al.2 and Schmitt et al.6 to the cells of this work. The ing (RPT) was performed periodically to determine the remaining cell
cells studied by Schmitt et al.6 were cylindrical model US18650V3 capacity as a function of storage time. Figure 2 shows that the pouch
from Sony Energy Devices Corporation. Reference performance test- cells in this work far outperform the cylindrical cells tested by Ecker
et al. and Schmitt et al. Researchers and users of Li-ion cells should
be aware of this vast performance increase that is possible.
Figure 3 shows long-term storage results for the NMC/graphite
100 20oC 4.3 V cells of this work compared to results for cylindrical LFP/graphite
20oC 4.2 V cells from Naumann et al.10 The cells tested by Naumann et al. were
90 US26650FTC1 cells from Sony. Figure 3 shows that the long-term
storage results for the NMC/graphite pouch cells of the present work
80 are at least as good as the results for these LFP/graphite cells even
though the storage voltages at 100% (4.3 V) and 90% SOC (4.2) for
Capacity remaining (%)

70 the NMC cells are much higher than that of the LFP cells (3.6 V) at
100% SOC. Researchers and users of Li-ion cells should be aware
100
40oC 4.3 V that NMC/graphite cells of an appropriate design, to be described in
this paper, can have longer storage lifetimes than LFP/graphite cells.
90 Since NMC/graphite cells have much larger volumetric and specific
Black - Schmidt et al. 45oC 4.2 V energy densities than LFP/graphite cells20 one does not need to make
80 Blue - Ecker et al. a trade-off between energy density and lifetime anymore.
Green - This work It is the intention of Figures 1, 2 and 3 to have whetted the appetite
70 of the reader to learn more about the cells of this work to be described
100 below.

90 55oC 4.2 V Experimental

50oC 3.87 V
Pouch cells.—Dry (no electrolyte) and sealed
80
o
Li[Ni0.5 Mn0.3 Co0.2 ]O2 (NMC532)/graphite pouch cells were ob-
50oC 4.2 V 50 C 4.07 V tained from Li-FUN Technology (Xinma Industry Zone, Golden
70 Dragon Road, Tianyuan District, Zhuzhou City, Hunan Province,
0 200 400 600
Storage time (days) China, 412000). These were coated NMC532/artificial graphite
(240 mAh at 4.3 V). The NMC532 used in these cells was single
Figure 2. Capacity remaining versus storage time for NMC/graphite cells as crystal NMC532 as described by Li et al.12 with a Ti-based coating
determined by reference performance testing every several months. The data as described by Ma et al.14 The artificial graphite used was Kaijin
from Ecker et al.2 and Schmitt et al.6 are for Sanyo UR18650E and Sony AML-400 from Kaijin, China. The single side active material
US18650V3 cells, respectively. The voltages and temperatures at which the electrode loadings were 21.1 mg/cm2 for the positive electrode
cells were stored are given in the legends. and 12.2 mg/cm2 for the graphite negative electrode. The positive

Downloaded on 2019-09-07 to IP 100.42.241.162 address. Redistribution subject to ECS terms of use (see ecsdl.org/site/terms_use) unless CC License in place (see abstract).
 Journal of The Electrochemical Society, 166 (13) A3031-A3044 (2019) A3033

SC-532 c
3 m

a 5 m

Figure 4. a) SEM image of the single crystal NMC532 powder (SC-532);

b) Top view SEM image of the AML-400 negative electrode surface after Figure 5. Solvents and additives used in the electrolytes incorporated into
compression; c) shows an image of one of the 402035-size pouch cells next to the pouch cells used in this work. EC = ethylene carbonate (CAS# = 96-49-
a Canadian $2 coin (a “toonie”). 1), DMC = dimethyl carbonate (CAS# = 616-38-6), EMC = ethyl methyl
carbonate (CAS# = 623-53-0), FEC = fluoroethylene carbonate (CAS# =
114435-02-8), DTD = ethylene sulfate (CAS# = 1072-53-3), VC = vinylene
and negative electrodes consisted of 94 wt.% and 95.4 wt.% active carbonate (CAS# = CAS No.872-36-6), LiPO2 F2 (called LFO here) = lithium
difluorophosphate (CAS# = 24389-25-1).
materials, respectively. The positive electrode was compressed to 3.5
g/cm3 and negative electrode was compressed to 1.55 g/cm3 . Prior to
filling with the designed electrolytes, pouch cells were cut open and
dried at 100°C under vacuum for 14 h to remove residual moisture. of 62.5 g, a volume of 24.3 cm3 and a capacity of 3.75 Ahr when
Afterwards, pouch cells were filled with 0.85 mL electrolyte in an charged to 4.3 V (which is the design upper cutoff voltage for the
Ar-filled glove box and sealed with a pouch sealer (MSK-115A 402035 pouch cells used here). Assuming an average voltage of 3.75
Vacuum Sealing Machine) under vacuum at a pressure of −90 kPa V, this corresponds to a specific energy of 225 Wh/kg and an energy
(relative to atmospheric pressure). Figure 4a shows an SEM image density of 580 Wh/L. The energy density of this 21700 cell is 72.5%
of the single crystal NMC532 powder while Figure 4b shows a top of the stack energy of the cell design used in this paper (see Table Ib).
view SEM image of the negative electrode surface. Figure 4c shows All of the cells in this work were bar coded so that their results
an image of one of the 402035-size pouch cells next to a Canadian $2 could be included in laboratory data bases. Although unconventional,
coin (a “toonie”). we have included the barcodes for the cells in many of the legends
The cells in this work have a designed capacity of about 240 mAh of the graphs of results. This is because many of the cells discussed
at C/20. Table Ia shows the actual specific energy and energy density here are still running ongoing tests. Interested readers can contact the
of the pouch cells used in this work. Even though these are very small authors to obtain updates on one or more cells by referring to barcode
cells where the area of the pouch is much larger than the area of the numbers.
jelly roll, the specific energy is still 200 Wh/kg. The energy density
is relatively low compared to that of a larger cell. To give readers Electrolytes used.—There were several electrolytes used over the
some idea of the energy density that could be obtained in a larger cell, course of this work. The chemicals used in the electrolytes are shown
Table Ib shows the “stack energy density” for three designs which in Figure 5. All these chemicals, plus methyl acetate (MA) and LiPF6
mirror what was used in the 402035 cells but with differing electrode (not shown in Figure 5), were battery grade, obtained from Shenzhen
thicknesses. The stack energy density accounts for everything except Capchem (China) and used as received.
the additional volume due to the cell casing, dead space and excess The electrolyte solvent blends used in this work were EC:EMC 3:7
electrolyte. Typically real energy densities can be 75% of the stack (by weight), EC:DMC 3:7 (by weight), EC:EMC:DMC 25:5:75 (by
energy density for cells of a few Ah capacity. For example a 21700- volume) and EC:EMC:MA 3:5:2 (by weight). All electrolytes used
size NMC532/graphite cell from a reputable manufacturer has a mass either 1.0, 1.2, 1.5 or 1.8 M LiPF6 salt sourced from either BASF or

Table 1. a) Measured properties of the NMC532/graphite 402035 (40 mm x 20 mm x 3.5 mm thick) pouch cells used here. The positive electrode
was 94% active material, the loading was 21.1 mg/cm2 (target was 21.3) and the electrode density was 3.5 g/cm3 . The negative electrode was
95.4% active material, the loading was 12.2 mg/cm2 (target was 11.8) and the electrode density was 1.55 g/cm3 . b) Stack energy density of the
NMC532/graphite couple for several electrode thicknesses. b) Stack energy density calculations – gives values for the electrode stack (negative
coating/copper/negative coating/separator/positive coating/aluminum/positive coating/separator). Assumptions – copper foil = 8 μm, aluminum
foil = 15 μm, separator = 16 μm, N/P capacity ratio = 1.1 at 4.3 V, average cell voltage = 3.75 V. The highlighted row represents the design used
in this work.

a)
Specific Energy Positive areal
Cell Average Capacity Mass Volume Energy Density capacity
designation Voltage (mAh) (g) (cm3 ) (Wh/kg) (Wh/L) (mAh/cm2 )

403025 3.75 240 4.5 2.3 200 390 3.5

b)
Cathode Cathode Anode Anode Stack
Cathode Cathode active Cathode reversible irreversible Anode Anode Anode reversible irreversible Energy
loading material density capacity capacity loading active material density capacity capacity Density
(mg/cm2 ) fraction (g/cm3 ) (mAh/g)∗ (%) (mg/cm2 ) fraction (g/cm3 ) (mAh/g) (%) (Wh/L)

21.3 94 3.5 175 10 11.8 95.4 1.55 350 8 795

25.0 94 3.5 175 10 13.85 95.4 1.55 350 8 815
30.0 94 3.5 175 10 16.6 95.4 1.55 350 8 836

∗ measured at 30°C and C/10 between 3.0 and 4.3 V.

Downloaded on 2019-09-07 to IP 100.42.241.162 address. Redistribution subject to ECS terms of use (see ecsdl.org/site/terms_use) unless CC License in place (see abstract).
 A3034 Journal of The Electrochemical Society, 166 (13) A3031-A3044 (2019)

Shenzhen Capchem. We have seen no penalty associated with lifetime from the variation of the 3C cell capacity versus cycle number in the
reduction using the first three solvent blends, unless high rate charge rate maps.
(> 1C at 20°C) is attempted. The DMC containing blends are pre-
ferred due to higher ionic conductivity for high rate charge.21 If even Cross-Sectional scanning electron microscopy of cycled positive
higher rates of charge are desired, then methyl acetate can be added as electrodes.—Cross sectional SEM images were taken of positive elec-
in the fourth solvent blend. Figure 6 shows the conductivity of several trodes from cycled pouch cells. Cells after long duration testing were
electrolyte systems as a function of LiPF6 content at 20°C and also charged to 4.3 V and opened carefully in an argon-filled glove box.
their temperature dependence at 1 molal. Figure 6 shows the advan- Circular discs of 1.1 cm diameter were punched from the positive
tage of EC:DMC over EC:EMC and the advantage of adding MA. electrode. Electrodes were rinsed with dimethyl carbonate twice be-
Electrolytes with DMC are preferred over those with EMC when high fore ion-milling with IB-19530CP cross-section polisher. Electrodes
rate charge or discharge is desired. Additions of MA cause a lifetime were milled with an Ar-ion beam for a 50 minute and 6 kV coarse
reduction for NMC/graphite cells22 so one must realize that a trade-off step followed by a 5 minute and 6 kV fine step. SEM images were
between lifetime and higher rate capability is being made when MA taken using a Hitachi S-4700 field emission electron microscope with
is introduced. This trade-off is explored in this article. a secondary electron detector. The images were obtained using an
The cells in this study used several electrolyte additives which have accelerating voltage of 5 kV and current of 15 uA.
been shown previously to dramatically increase lifetimes. The combi-
nation of 2% VC + 1% DTD promoted by Li et al.,12 the combination
of 2% FEC + 1% LFO promoted by Ma et al.16 and 1% LFO alone Results and Discussion
promoted by Ma et al.,13 were used in the cells in this work. Charge-discharge cycling results.—Figures 8 and 9 provide ca-
pacity retention versus cycle number data and normalized V versus
cycle number data for the NMC532/graphite cells testing between 3.0
Testing protocols used..—Long-term storage was performed by and 4.3 V (0 – 100% SOC, Figure 8) and between 3.0 and 4.2 V (0
placing cells at various voltages in convection ovens at 20, 40 or 55°C – 90% SOC, Figure 9). The cells tested at C/3 used 1.0 M LiPF6 in
for extended periods of time. Periodically, cells were extracted and EC:EMC 30:70 electrolyte with the selected additives indicated in the
subjected to a reference performance test (RPT) as shown in the left Figure legend while the cells tested at 1C used 1.2 M (2FEC + 1LFO)
panel of Figure 7. The C/3 capacity, measured between 3.0 and 4.1 V or 1.5 M (2VC+1DTD) LiPF6 in EC:EMC:DMC 25:5:70 electrolyte.
at 20°C was monitored. After the RPT test, the cells were charged to The duration of the testing is indicated in the legend. The appropri-
the original storage voltage at C/3 CCCV (C/100 current limit) and ate data in Figure 8 was used in the comparison provided to Ecker
placed back in their respective ovens. et al’s.2 data as Figure 1. The reader should note the extremely ex-
Long-term cycle testing was performed using battery testers from panded vertical scales in Figures 8 and 9, which vary from panel to
Neware (Shenzhen, China). Cycling was normally between 3.0 and panel.
4.3 V using a C/3 rate at 20, 40 and 55°C. Charging used CCCV mode Figure 8 shows that the additive blend 2% FEC + 1% LFO provides
with a C/20 current cutoff. Some testing was made at 20°C using 1C a substantial advantage at 20°C compared to 2%VC + 1% DTD. How-
charge and discharge with CCCV charge to a C/20 current limit. The ever, at elevated temperature, the situation is reversed. So the choice of
1C testing also included a “rate map” with discharge currents of C/20, additives could be application specific. In any case, the data in Figure 8
C/2, C, 2C and 3C where the charging currents were C/20 for the C/20 is far better than any literature data we have seen for NMC/graphite
discharge cycle and 1C (with a CCCV to C/20) for all the other cycles cells tested between 0 and 100% SOC. Comparing Figures 8 and 9
in the rate map. The rate map protocol is shown in the center panel of reveals that the main benefit of lowering the upper cutoff potential is
Figure 7. better control of impedance growth at 20 and 40°C.
The difference between the average charge voltage and the average Figure 10 shows capacity versus cycle number for
discharge voltage, V, was measured and used as a measure of cell NMC532/graphite cells tested at 1C with rate maps and at 20°C.
internal resistance. V was calculated as shown in the right panel Figures 10a and 10b show that the cycling of these cells is incred-
of Figure 7. Graphs of V versus cycle number or time are used to ibly impressive between 3.0 and 4.1 V with no measurable loss
monitor the stability of cell internal resistance. This can also be gleaned in 3C capacity. Figure 10c shows that when the additive mixture

Figure 6. a) Conductivity at 20°C versus molality in LiPF6 EC:DMC:MA 30:50:20, EC:DMC and EC:EMC electrolytes. b) Conductivity versus temperature for
1.0 m LiPF6 in the same solvent blends. The solid lines are the predictions of Gering’s Advanced Electrolyte Model (AEM).23

Downloaded on 2019-09-07 to IP 100.42.241.162 address. Redistribution subject to ECS terms of use (see ecsdl.org/site/terms_use) unless CC License in place (see abstract).
 Journal of The Electrochemical Society, 166 (13) A3031-A3044 (2019) A3035

4.4 RPT: 20°C, 3.0-4.1V Storage V 4.4 Rate map: 3.0-UCV 4.4 V calculation Fresh
V = Vav,c - Vav,d

Potential (V)

Potential (V)
Potential (V)

4 4 4

3.6 3.6 3.6

3.2
3.2 3.2
Qc, Qd
80 400 2.8
C/3 CCCV C/20 C/3 CCCV C/100 1C CCCV C/20 4.4 Charge Aged
Discharge
40 C/20

Current (mA)
Current (mA)

Potential (V)
0 4
C/20 C/20 C/20 C/2
0 1C
OCV 3.6
C/20 1C:1C 1C:1C
-400 2C
-40 3.2
OCV Vav,c = Area under the V-Q curve/Qc
C/3 3C Vav,d = Area under the V-Q curve/Qd
-80 -800 2.8
0 20 40 60 80 0 20 40 60 80 0 50 100 150 200
Time (hours) Time (hours) Capacity (mAh)

Figure 7. Left panel) Showing the reference performance test protocol used to periodically monitor the cells under storage testing; center panel) Protocol used in
the rate maps collected every 100 cycles of cells cycled at 1C at 20°C; right panel) Schematically indicating how V is calculated.

2%VC + 1%DTD is used there is a loss in 3C capacity during testing Readers may wonder why the C/20 capacity versus cycle number
between 3.0 and 4.3 V. Figure 10d shows that when the additive in Figure 10 slowly decays with cycle number while the 1C capacity
mixture 2% FEC + 1% LFO is used there is virtually no 3C capacity decays more slowly. This is explained in the text surrounding Figures 6
loss even after 5400 cycles between 3.0 and 4.3 V. Figures 10e and in Reference 16.
10f show that the 3C capacity is even better maintained when higher Figure 11 shows the voltage - capacity curves of the discharge
LiPF6 concentrations are used for cells with 2% FEC + 1% LFO of the cells described by Figure 10 measured during the rate maps.
tested between 3.0 and 4.3 V. The 8 curves of each color were taken at equally spaced intervals

Figure 8. Fractional capacity and normalized V plotted versus cycle number Figure 9. Fractional capacity and normalized V plotted versus cycle number
for NMC532/graphite cells tested between 3.0 and 4.3 V under the conditions for NMC532/graphite cells tested between 3.0 and 4.2 V under the conditions
shown in the legends. The duration of testing is shown for the most mature shown in the legends. The duration of testing is shown for the most mature
cells. cells.

Downloaded on 2019-09-07 to IP 100.42.241.162 address. Redistribution subject to ECS terms of use (see ecsdl.org/site/terms_use) unless CC License in place (see abstract).
 A3036 Journal of The Electrochemical Society, 166 (13) A3031-A3044 (2019)

after 1100 cycles its capacity and V were close to those of the cell
which had been cycled continuously at 20°C. The capacity of the cell
after moving from 55 to 20°C (shown in Figure 12c) is lower than the
cell which was tested entirely at 20°C because of lithium inventory
loss during the 1100 cycles at 55°C. The results in Figure 12 demon-
strate this cell chemistry is extremely tolerant to extended periods at
elevated temperatures.

Long-term storage testing.—Figure 13 shows the results of long-

term storage testing at 20, 40 and 55°C on the NMC532/graphite pouch
cells containing electrolyte with 2%VC + 1% DTD additives. Fig-
ure 13 shows these cells behave in an outstanding manner during stor-
age testing with little loss of capacity and minor increases in V. Only
the cell stored at 55°C and 4.3 V for 1.3 years shows signs of trou-
blesome capacity loss and impedance growth. The combination of the
additives 2% VC + 1% DTD with this single crystal NMC/artificial
graphite chemistry is simply outstanding in storage.
Figure 14 shows the results of long-term storage testing at 20, 40
and 55°C on the NMC532/graphite pouch cells containing electrolyte
with 1% LFO as the additive. Figure 14 shows that these cells have
an inferior storage behavior at 40 and 55°C compared to the cells
with the additive blend 2%VC + 1% DTD shown in Figure 13. It
is believed that the reason for this inferior behavior is the lack of an
Figure 10. Results of 1C testing of NMC532/graphite cells at 20°C with rate effective negative electrode SEI former like VC or FEC. Unfortunately,
maps every 100 cycles. The C/20, C/2, 1C, 2C and 3C markers show the given the excellent performance of cells with 2%FEC + 1% LFO in
discharge capacity of the corresponding discharge at the indicated rate. a) 1.5 Figures 8, 10 and 11, storage data for cells with 2%FEC + 1%LFO
M LiPF6 in EC:EMC:DMC 25:5:70 with 2% VC + 1% DTD tested between is not mature enough to report at this time. The reason for showing
3.0 and 4.1 V; b) 1.2 M LiPF6 in EC:EMC:DMC 25:5:70 with 2% FEC + 1% the data in Figure 14 is to highlight that electrolyte additives make a
LFO tested between 3.0 and 4.1 V; c) 1.5 M LiPF6 in EC:EMC:DMC 25:5:70 huge different to cell performance. The data presented in the reports
with 2% VC + 1% DTD tested between 3.0 and 4.3 V; d) 1.2 M LiPF6 in on commercial cells2,5–8,10 used for lifetime modelling do not report
EC:EMC:DMC 25:5:70 with 2% FEC + 1% LFO tested between 3.0 and 4.3
V; e) 1.5 M LiPF6 in EC:EMC:DMC 25:5:70 with 2% FEC + 1% LFO tested
the details of cell chemistry, including electrolyte additives, as we do
between 3.0 and 4.3 V; f) 1.8 M LiPF6 in EC:EMC:DMC 25:5:70 with 2% here. We examine later below the reasons for the inferior performance
FEC + 1% LFO tested between 3.0 and 4.3 V. Data for cells that use the 2% of the cells with 1% LFO.
VC + 1% DTD additive blend are shown in black while data for cells that use Figure 15 shows the drop in voltage of the cells during 3-month
the 2% FEC + 1% LFO additive blend are shown in blue. open circuit storage periods plotted as a function of the storage voltage
for the cells in Figures 13 and 14. The average of the voltage drop
during the last four 3-month storage periods (between 0.5 and 1.3 years
along the capacity versus cycle number curves. Figures 11a and 11b in Figures 13 and 14) is shown. Figure 15 shows that the voltage drops
show that there is virtually no impedance growth in the cells cycled during the storage periods are very small but that they increase with
to 4.1 V as can be determined from the virtual overlap of the 3C both storage voltage and temperature. Cells with 2% VC + 1%DTD
discharge curves for 5200 (Figure 11a) and 3100 (Figure 11b) cycles. show smaller voltage drops (less self-discharge) than cells with 1%
By contrast, Figures 11c and 11d do show evidence of impedance LFO at both 40 and 55°C. The fitted lines to the data in Figure 15 are
increase when cells are charged to 4.3 V because the 3C discharge exponential in voltage. Figure 15b shows that the data appears to be
curves shift down sequentially during the cycle testing. Figures 11e exponentially activated in voltage based on the linearity of the data
and 11f show that increasing the LiPF6 concentration in the presence when plotted with a logarithmic y-axis.
of the 2% FEC + 1% LFO additive virtually eliminates impedance Figure 16 shows the results of storage and charge-discharge cycling
growth during cycling to 4.3 V at 3C even over 4100 cycles. tests for NMC532/graphite cells with 2%VC + 1% DTD plotted on
Figure 11 also can be used to determine that major cause of C/20 the same graph. This is a useful comparison to see the impact of
capacity loss in these cells is lithium inventory loss. Careful exami- exercising the cells on lifetime. The 55°C data show similar results
nation of the eight C/20 discharge curves in each of Figures 11a–11f for cells that were cycled or stored for the 1.3 years where the test
shows that the portion of the curves above 3.6 V are unchanged over results for both are available. This suggests that the major mechanism
the duration of the cycles. This indicates no active material loss. The for capacity loss at 55°C is lithium inventory loss due to a thickening
loss of capacity occurs only in the lower voltage region. This is caused negative electrode SEI. At 20 and 40°C the cells undergoing charge-
by a shift of the negative electrode voltage-capacity curve relative to discharge testing lose slightly more capacity per unit time than the
the positive electrode voltage-capacity curve due to lithium inven- corresponding cells undergoing storage testing, however their rates
tory loss. It will be shown later, using differential voltage (dV/dQ vs. of impedance growth, as monitored by V/Vo , are very similar.
Q) analysis that the only capacity loss mechanisms in these cells are As will be shown later using differential voltage analysis (dV/dQ vs.
lithium inventory loss and impedance growth, in cases where the latter Q) the major mechanism for capacity loss in cells tested at 40°C is
occurs. lithium inventory loss due to a thickening negative electrode SEI. This
We wondered if the increase of V with cycle number at 40°C accelerates during cycling due to the approx. 10% volume changes of
and 55°C shown in Figures 8 and 9 would affect subsequent opera- the graphite particles during lithiation and delithiation.3 The cells at
tion at 20°C. Figure 12 shows the capacity versus cycle number for 20°C have lost only 4% capacity after 2.25 years and 3400 cycles.
NMC532/graphite cells that operated for about 1 year at 40°C or 55°C Differential voltage analysis does not have the resolution to determine
and were then moved to 20°C for continued testing for another year the capacity loss mechanisms with such small capacity loss. However,
or so. Figures 12a and 12b show that when the cell tested at 40°C was by analogy to the 55°C and 40°C results it is believed the major loss
moved to 20°C after 1000 cycles its capacity and V matched those mechanism is Li inventory loss at the negative electrode SEI for the
of the cell which had been cycled continuously at 20°C. Figures 12c cells that are cycling. This is consistent with the discussion of the C/20
and 12d show that when the cell tested at 55°C was moved to 20°C capacity loss of the cells described by Figures 10 and 11.

Downloaded on 2019-09-07 to IP 100.42.241.162 address. Redistribution subject to ECS terms of use (see ecsdl.org/site/terms_use) unless CC License in place (see abstract).
 Journal of The Electrochemical Society, 166 (13) A3031-A3044 (2019) A3037

Figure 11. Voltage versus capacity for the C/20, C/2, 1C, 2C and 3C cycles of the discharge rate maps for the cells in Figure 10. The 8 curves of each color were
taken at equally spaced intervals along the capacity versus cycle number curves. a) 1.5 M LiPF6 in EC:EMC:DMC 25:5:70 with 2% VC + 1% DTD tested between
3.0 and 4.1 V; b) 1.2 M LiPF6 in EC:EMC:DMC 25:5:70 with 2% FEC + 1% LFO tested between 3.0 and 4.1 V; c) 1.5 M LiPF6 in EC:EMC:DMC 25:5:70 with
2% VC + 1% DTD tested between 3.0 and 4.3 V; d) 1.2 M LiPF6 in EC:EMC:DMC 25:5:70 with 2% FEC + 1% LFO tested between 3.0 and 4.3 V; e) 1.5 M
LiPF6 in EC:EMC:DMC 25:5:70 with 2% FEC + 1% LFO tested between 3.0 and 4.3 V; f) 1.8 M LiPF6 in EC:EMC:DMC 25:5:70 with 2% FEC + 1% LFO
tested between 3.0 and 4.3 V.

Figure 12. Capacity versus cycle number and V versus cycle number for NMC532/graphite cells which were first cycled at elevated temperature, then moved
to 20°C compared to cells which had been testing at 20°C continuously (black). a) Cell tested at 40°C then moved to 20°C (red) compared to a cell which had been
at 20°C continuously; b) V for the cells described by panel a); c) Cell tested at 55°C then moved to 20°C (green) compared to a cell which had been at 20°C
continuously (black); d) V for the cells described by panel c). All cells were tested at C/3 CCCV between 3.0 and 4.3 V. Electrolytes in these cells were 1.0 M
LiPF6 in EC:EMC 30:70 with additives listed in the figure legend.

Downloaded on 2019-09-07 to IP 100.42.241.162 address. Redistribution subject to ECS terms of use (see ecsdl.org/site/terms_use) unless CC License in place (see abstract).
 A3038 Journal of The Electrochemical Society, 166 (13) A3031-A3044 (2019)

Figure 13. Long-term storage results for

NMC532/graphite cells at 20, 40 and 55°C
as indicated in the legend. Storage voltages
were 3.67, 4.0, 4.1, 4.2 and 4.3 V. Normalized
capacity and V/Vo results were obtained
from RPT performed at 20°C at the times in-
dicated. These cells contained 2% VC + 1%
DTD as the additive mixture.

Figure 14. Long-term storage results for

NMC532/graphite cells at 20, 40 and 55°C
as indicated in the legend. Storage voltages
were 3.67, 4.0, 4.1, 4.2 and 4.3 V. Normalized
capacity and V/Vo results were obtained
from RPT performed at 20°C at the times in-
dicated. These cells contained 1% LFO as the
additive.

Downloaded on 2019-09-07 to IP 100.42.241.162 address. Redistribution subject to ECS terms of use (see ecsdl.org/site/terms_use) unless CC License in place (see abstract).
 Journal of The Electrochemical Society, 166 (13) A3031-A3044 (2019) A3039

cells will find the use of MA compelling. They should be aware that
2VC + 1DTD 20oC b) using MA will lead to lifetime penalties.
Voltage drop after 3 months (V)

2VC + 1DTD 40oC

2VC + 1DTD 55oC Ultra high precision charger testing.—The Ultra High Precision
0.1
1LFO 20oC Charger (UHPC) at Dalhousie University24,25 was used to measure
1LFO 40oC the coulombic efficiency and charge endpoint capacity slippage of
1LFO 55oC fresh and aged NMC532/graphite cells that contained 2%VC (or 1%
VC) + 1%DTD or 2%FEC + 1%DTD additives. The UHPC results
were collected at 40°C between 3.0 and 4.3 V using currents corre-
sponding to C/20. Figures 18a shows zeroed charge endpoint capacity,
18b shows discharge capacity, 18c shows coulombic efficiency (CE)
and 18d shows V, all plotted versus cycle number. Cells 54271 and
54275 were cycled 1000 times at 40°C between 3.0 and 4.3 V using
currents corresponding to C/3 prior to being moved to UHPC testing.
Cell 54275 was previously highlighted in Figure 12a.
0.01 Figure 18a shows that the aged cells have virtually no charge end-
2VC + 1DTD 20oC a) point capacity slippage. This means there are no species being oxidized
Voltage drop after 3 months (V)

2VC + 1DTD 40oC at the positive electrode. Figure 18b shows that aged cells have virtu-
0.16 2VC + 1DTD 55oC ally no capacity fade over 10 cycles in agreement with the long-term
1LFO 20oC testing data. Figure 18c shows that the aged cells have coulombic ef-
1LFO 40oC ficiencies that very closely approach the perfect value of 1.00000….
0.12 1LFO 55oC Aged cell 54275 has an average CE between cycle 6 and 9 of 0.99985
± 0.00003 suggesting the rate of parasitic reactions has become ex-
tremely small in this cell after aging. For comparison, UHPC studies
0.08 of aged LiCoO2 /graphite (LCO/graphite) cells have been reported by
Fathi et al.26 Cells that had been testing continuously for 11.7 years at
C/6 charge and C/150 discharge between 3.0 and only 4.075 V at 37°C
0.04 had accumulated 750 cycles and had 80% capacity remaining. These
were then characterized by UHPC. The CE of the aged LCO/graphite
cells tested at 40°C between 3.0 and 4.075 and with currents corre-
0 sponding to C/20 was 0.99979 ± 0.00002.26 The fact that the aged
3.6 3.8 4 4.2 4.4 NMC532/graphite cells tested between 3.0 and 4.3 V had a CE higher
Storage voltage (V) than those of the aged LCO/graphite cells suggests the lifetime of the
NMC532/graphite cells will be extremely long.
Figure 15. Voltage drop during the three month OCV storage periods for the
cells in Figures 13 and 14 plotted versus storage voltage. The data plotted are Degradation mechanisms of the NMC532/graphite cells.—Dif-
the averages of the voltage drops during the last four 3-month storage periods in
ferential voltage analysis (dV/dQ vs. Q) has been shown by Bloom et
Figures 13 and 14. Electrolyte additives and storage temperatures are indicated
in the legend. al. and Honkura et al.27–29 to be very effective in elucidating degra-
dation mechanisms in Li-ion cells. Our laboratory has implemented
dV/dQ analysis in a user-friendly software package30 which was uti-
lized here. Five NMC532/graphite cells with 2% VC + 1% DTD
Impact of methyl acetate on lifetime.—There is a large empha- additives were cycled 0, 200, 400, 700 and 900 cycles between 3.0
sis at the current time on lithium-ion cells with fast charge capability and 4.3 V at 40°C at C/3. After those cycles, cells were moved to the
that do not sacrifice energy density. Modifications to the electrolyte UHPC where high fidelity measurements of dV/dQ vs. Q and dQ/dV
that improve Li+ diffusion can increase charging rates before the on- vs. V were collected. Positive and negative electrode reference curves
set of lithium plating, however, there are lifetime penalties.21,22 Fig- required for dV/dQ analysis were collected in half coin cells using
ure 17 shows the results of storage and charge-discharge cycling for electrodes extracted from new, dry (no electrolyte) pouch cells.
NMC532/graphite cells with or without MA. At 20°C, there is little or Figure 19 shows the results of dV/dQ analysis on these cells. Fig-
no penalty in storage when MA is added, but capacity retention during ure 19a shows the full cell voltage curves and the voltage versus ca-
cycling is compromised, especially for cells with 20% MA charged pacity curves of the positive and negative electrodes versus lithium.
to 4.3 V, where rollover failure is observed after about one year of The charge axis in Figure 19a is zero at the point where the full cell
testing or 1700 C/3 cycles. Cells with MA cycled to 4.2 V show early is fully discharged. Figures 19b and 19c show the excellent agree-
indications of failure beginning after about 1.8 years or 2500 cycles. ment between the calculated and measured values of dV/dQ vs. Q and
At 40°C, there are only comparative results available for cycle testing dQ/dV vs. V, respectively. Figures 19d and 19e show the cells capacity
to 4.3 V. After two years of cycle testing, cells with 20% MA lost an and V/Vo , respectively, versus cycle number. Figures 19f, 19g and
additional 3% capacity compared to cells with no MA. Storage test- 19h show the relative electrode shift or slippage (due lithium inven-
ing at 40°C shows that cells with MA lose capacity slightly faster, by tory loss), the positive electrode capacity and the negative electrode
about an extra 1% per year, compared to cells without MA. capacity, respectively, for the five cells plotted versus cycle number.
The impact of MA on lifetime can be considered minor or major Figure 19d shows that over the 900 cycles the cells lost about 11 mAh
depending on the intended application of the cell. In a situation where of capacity. Figure 19e shows that over the same period the shift loss,
high rate charge and discharge is required and intermittent use is envis- due to Li inventory loss was about 9 mAh, in good agreement. Fig-
aged (cells spend a large fraction of time in storage) the incorporation ures 19f and 19g show no evidence for loss of positive or negative
of MA would lead to a minor penalty at most. In addition, if the in- electrode capacity. Based on these experiments one concludes that Li
corporation of MA eliminates Li plating during high rate charge, then inventory loss, due to thickening of the negative electrode SEI, is the
MA would lead to a lifetime enhancement by removing this serious major degradation mechanism for these cells. Figure 19e also shows
degradation pathway. On the other hand, if cells are intended to be cy- that there is a small impedance growth in the cells.
cled continuously between 3.0 and 4.3 V for years, Figure 17 suggests Practitioners of dV/dQ analysis will appreciate that it is hard to be
a major penalty associated with the incorporation of MA. Researchers absolutely certain of dV/dQ analysis results when a cell has only lost a
contemplating the use of very thick electrodes even for low-rate Li-ion few percent of capacity as in Figure 19. Therefore, dV/dQ analysis was

Downloaded on 2019-09-07 to IP 100.42.241.162 address. Redistribution subject to ECS terms of use (see ecsdl.org/site/terms_use) unless CC License in place (see abstract).
 A3040 Journal of The Electrochemical Society, 166 (13) A3031-A3044 (2019)

Figure 16. Normalized capacity and V/Vo plotted versus time for NMC532/graphite cells undergoing cycling or storage testing at 20, 40 and 55°C. The
charge-discharge cycling was performed at C/3 between 3.0 and 4.3 V (blue circles) or between 3.0 and 4.2 V (red circles). The storage testing was done at 4.3 V
(blue crosses) or at 4.2 V (red crosses). Normalized capacity for the storage testing was measured with the RPT protocol (Figure 7) at 20°C. The cells contained
2% VC + 1% DTD as additives.

Blue - 2VC+1DTD Blue - 2VC+1DTD

Red - 2VC+1DTD+20MA Red - 2VC+1DTD+20MA
3000 3.67 V
cycles
4.0 V
1 1 4.1 V
Normalized capacity

Normalized capacity

4.2 V
0.95 0.95 4.3 V
3.67 V
Storage data

3000
4.0 V cycles
0.9
4.1 V 0.9
4.2 V
4.3 V
0.85 Cycling C/3 3.0-4.3 V (circle) 0.85
3.0 - 4.2 V (cross) Cycling C/3 3.0-4.3 V
CCCV, 200C CCCV, 400C
0.8 0.8
0 4000 8000 12000 16000 20000 0 1 2 3
Time (hours) Time (years)

Figure 17. Normalized capacity plotted versus time for NMC532/graphite cells undergoing cycling or storage testing at 20 and 40°C. The charge-discharge cycling
was performed at C/3 between 3.0 and 4.3 V (circles) or between 3.0 and 4.2 V (crosses). The storage testing was done at the voltages indicated in the legend.
Results for cells that had 20% MA in the electrolyte are shown in red and results for cells that had no MA are shown in blue. Normalized capacity for the storage
testing was measured with the RPT protocol (Figure 7) at 20°C. The cells contained 2% VC + 1% DTD as additives.

Downloaded on 2019-09-07 to IP 100.42.241.162 address. Redistribution subject to ECS terms of use (see ecsdl.org/site/terms_use) unless CC License in place (see abstract).
 Journal of The Electrochemical Society, 166 (13) A3031-A3044 (2019) A3041

Figure 18. Ultra high precision charger (UHPC) results for fresh and aged NMC532/graphite cells containing the electrolyte additives listed in the legend. The
UHPC results were collected at 40°C between 3.0 and 4.3 V using currents corresponding to C/20. a) zeroed charge endpoint capacity; b) discharge capacity; c)
coulombic efficiency (CE); and d) V. Cells 54271 and 54275 were cycled 1000 times at 40°C between 3.0 and 4.3 V using currents corresponding to C/3 prior
to being moved to UHPC testing. Cell 54275 was previously highlighted in Figure 12a.

Figure 19. Differential voltage analysis of NMC532/graphite cells tested at C/3 and 40°C between 3.0 and 4.3 V. a) the full cell voltage curves and the voltage
versus capacity curves of the positive and negative electrodes versus lithium for five cells cycled 0, 200, 400, 700 and 900 times. The charge axis in Figure 18a is
zero at the point where the full cell is fully discharged; b) the calculated and measured values of dV/dQ vs. Q; c) the calculated and measured values of dQ/dV vs.
V; d) the cell capacity versus cycle number; e) V/Vo versus cycle number; f) the relative electrode shift or slippage (due to lithium inventory loss) versus cycle
number; g) the positive electrode capacity vs. cycle number; h) the negative electrode capacity versus cycle number.

Downloaded on 2019-09-07 to IP 100.42.241.162 address. Redistribution subject to ECS terms of use (see ecsdl.org/site/terms_use) unless CC License in place (see abstract).
 A3042 Journal of The Electrochemical Society, 166 (13) A3031-A3044 (2019)

performed on a more heavily cycled NMC532/graphite cell containing gas generated in NMC532/graphite cells during long-term cycling and
2% VC + 1% DTD which was tested at 55°C (C/3 between 3.0 and storage at 20, 40 and 55°C for cells with 2%VC + 1%DTD and cells
4.3 V) for 20 months. High fidelity data for dQ/dV vs V and dV/dQ with 1% LFO as additives. As shown in Table I, the volumes of the
vs Q was available after 0, 100 and 2500 cycles. This cell was from cells themselves are 2.3 mL. Figure 21 shows that gassing is not an
a different build compared to the cell considered in Figure 19. It had issue for cells cycled or stored at 20°C. Figure 21 shows that cells with
more excess negative electrode so it could be charged to 4.5 V without 2%VC + 1%DTD undergoing continuous cycling at 40°C generate
Li plating, if desired. Figure 20 shows the results of dV/dQ analysis on minimal gas even after 2.5 years of 100% DOD cycling. Figure 21
the cell cycled at 55°C. Figure 20 has the same format as Figure 19. shows that cells with 1%LFO generate more gas than cells with 2%VC
Figure 20d shows that over the 2500 cycles the cell lost about 45 + 1%DTD under all conditions and untenable amounts of gas at 55°C.
mAh of capacity. Figure 20e shows that over the same period the shift Unfortunately, comprehensive data for cells with the additive system
loss, due to Li inventory loss, was about 42 mAh, in good agreement. 2%FEC + 1% LFO are not available yet. Gas generation in the cells
Figures 20f and 20g show evidence for a small loss of positive (6 with 2%VC + 1%DTD cycled or stored continuously at 55°C may
mAh) and negative electrode (4 mAh) capacity, both about an order of also be problematic. It is not known what volume of gas produced in
magnitude smaller than the shift loss and hence very small. Just as the these small pouch cells would correspond to the amount needed to
experiment considered in Figure 19, one concludes that Li inventory open the pressure activated vent in a cylindrical or prismatic cell.
loss, due to thickening of the negative electrode SEI, is the major The increased gassing in cells with 1% LFO stored at 40 and 55°C
degradation mechanism for this cell. Figure 20e also shows that there compared to cells with 2% VC + 1% DTD correlates well with the
is a significant impedance growth in this cell presumably due to the differences in their storage performance shown in Figures 13 and 14.
high temperature of operation and the 20 month duration of the test. Gas generation is caused by reactions between the charged electrode
The volumes of gas generated in many of the cells under long- materials and the electrolyte. Therefore one can roughly say that when
term testing were measured using Archimedes principle as described there is little gassing there are few parasitic reactions occurring in
in the experimental section of reference.22 The experiments on the the cells. The gas volume data at 20°C suggests very few parasitic
cylindrical cells reported in the literature2,5–9 give no indication of gas reactions and correlates well with the excellent long-term cycling of
generated during cycling or storage. Figure 21 shows the volume of these cells at 20°C (Figures 8, 9 and 10) with little impedance growth.

Figure 20. Differential voltage analysis of a NMC532/graphite cell tested at C/3 and 55°C between 3.0 and 4.3 V. a) the full cell voltage curves and the voltage
versus capacity curves of the positive and negative electrodes versus lithium for the same cell at cycle 0, 100, and 2500. The charge axis in Figure 18a is zero at the
point where the full cell is fully discharged; b) the calculated and measured values of dV/dQ vs. Q; c) the calculated and measured values of dQ/dV vs. V; d) the
cell capacity versus cycle number; e) V/Vo versus cycle number; f) the relative electrode shift or slippage (due to lithium inventory loss) versus cycle number;
g) the positive electrode capacity vs. cycle number; h) the negative electrode capacity versus cycle number.

Downloaded on 2019-09-07 to IP 100.42.241.162 address. Redistribution subject to ECS terms of use (see ecsdl.org/site/terms_use) unless CC License in place (see abstract).
 Journal of The Electrochemical Society, 166 (13) A3031-A3044 (2019) A3043

the electrolyte additives 2% VC + 1%DTD was tested for 5300 cycles

at 20°C using currents corresponding to 1C. The voltage limits were
3.0 to 4.1 V. The cell had 97% of its initial capacity remaining much
like the results for the cell shown in Figure 10a which was tested under
the same conditions. After 5300 cycles the cell was charged to 4.3 V
at C/20 and then opened in a glove box. The positive electrode was
cross-sectioned and imaged as described in the experimental section.
Figure 22 shows representative cross sectional SEM images of the
electrode particles after 5300 cycles. No evidence of micro-cracking
can be observed in particles that were initially less than 3 micrometers
in size, demonstrating again the advantages of single crystal materials.
This why there is no loss of positive electrode active mass in these cells
during cycling. Figure 22 suggests that the cell in Figure 10a would
have an incredibly long cycle life. The paper by Li et al.31 introduces
the advantages of single crystal materials compared to conventional
polycrystalline positive electrode materials.

What might be the lifetime of these NMC532/graphite cells?—It

is interesting to consider the possible lifetime of these cells in EV and
Figure 21. Volume of gas generated during cycling and storage in the grid storage applications. Complex models are not used here. Consider
NMC532/graphite cells with electrolyte additives listed in the legend. The the data in Figures 10, 11 and 16. At 20°C cells can be cycled at either
storage potential or the upper cutoff potential during cycling is indicated along 1C or C/3 for 4000 100% DOD cycles with about 4% capacity loss
the x-axis for the three temperatures considered: 55, 40 and 20°C. The gas and less than 10% impedance growth. The fractional capacity loss
bags in these cells can hold about 2.5 ml of gas before the gas begins to be rate is therefore about 0.00001/cycle. At 20°C cells can be stored
compressed above atmospheric pressure. This means gas volumes measured at full charge for 1.3 years with about 0% loss and no impedance
by Archimedes principle that are greater than 2.5 mL would be even greater if
the gas were at 1 atm pressure.
change and at 90% charge with 0% loss and no impedance change.
The capacity loss rate is 0.0/year. At 40°C, cells can be cycled at C/3 for
3700 100% DOD cycles with about 12% capacity loss and about 15%
impedance growth. The fractional capacity loss rate at 40°C is about
The small gas volumes generated at 20°C also correlate well with
0.000032/cycle. At 40°C, cells can be stored at full charge for 1.3 years
the excellent storage behavior of these cells at 20°C with virtually no
with about 3% loss and 10% impedance change or at 90% charge with
capacity loss nor impedance growth (Figures 13 and 14).
about 1.5% loss and 8% impedance change. The fractional capacity
loss rate during 100% DOD storage at 40°C is about 0.02/year.
Virtually no micro-cracking in the single crystal NMC532 elec- Assume a use scenario that blends storage (worst case scenario
trode particles after 5000 cycles.—A cell (barcode 82306) containing at top of charge) and cycling (worst case scenario 100% DOD) and

Figure 22. Cross sectional SEM of a single crystal NMC532 positive electrode taken from a cell tested exactly under the conditions of the cell in Figure 10a. The
cell had 97% capacity retention after 5300 cycles. Notice that there are virtually no microcracks in any of the electrode particles. This is why these cells show no
loss of positive electrode active mass during cycling.

Downloaded on 2019-09-07 to IP 100.42.241.162 address. Redistribution subject to ECS terms of use (see ecsdl.org/site/terms_use) unless CC License in place (see abstract).
 A3044 Journal of The Electrochemical Society, 166 (13) A3031-A3044 (2019)

Kilometers (km) that lifetime models for NMC532/graphite cells consider the data pre-
0 1000000 2000000 3000000 sented here.
1 Full details of these cells including electrode compositions, elec-
trode loadings, electrolyte compositions, additives used, etc. have been
provided in contrast to literature reports using commercial cells. This
20oC has been done so that others can re-create these cells and use them as
benchmarks for their own R+D efforts be they in the spaces of Li-ion
Fractional Capacity

cells or “beyond Li-ion cells”.

0.9
Projected lifetime The cells described here have two dominant failure mechanisms at
1 100% DOD cycle per day 20, 40 and 55°C. These are Li inventory loss and impedance growth.
Assume 350 km per cycle We find virtually no evidence for loss of active material in these cells
during cycle testing under the conditions used here.
The numeric data used in Figures with cell barcode labels indicated
0.8 can be made available to interested readers. Please contact the authors.
40oC

Acknowledgments
The work done here was supported by NSERC and Tesla Canada
0.7 under the auspices of the Industrial Chairs program. Many of the
0 5 10 15 20 25
authors have received scholarship support from NSERC, the China
Years
Scholarship Council and the Nova Scotia Graduate Scholarship pro-
gram and that support is gratefully acknowledged.
Figure 23. Worst-case scenario lifetime and total driving range projections
for the NMC532/graphite cells with 2% VC + 1% DTD at 20 and 40°C.
Assumptions made were one 6 hour 100% DOD cycle per day and 350 km initial
driving range per cycle. Equation 1, with M = 1, n = 3, Lc = 0.00001/cycle ORCID
(20°C), Lc = 0.000032/cycle (40°C), Ls = 0.0/year (20°C) and Ls = 0.02/yr
(40°C) was used to calculate Q/Qo . Jessie E. Harlow https://orcid.org/0000-0002-3928-2573
Stephen L. Glazier https://orcid.org/0000-0003-2961-3049
J. R. Dahn https://orcid.org/0000-0002-6997-2436
cycling in a mode where there are M 100% DOD cycles per day at an
average C/n rate. That means the time of cycling each day would be 2
M n hours. The time of storage would be 24 – 2 M n. The fractional References
capacity as a function of time in years would be: 1. N. S. Pearre, W. Kempton, R. L. Guensler, and V. V. Elango, Transp. Res. Part C
Q/Qo = 1 − Lc M 365 t − Ls (24 − 2 M n) /24 t [1] Emerg. Technol., 19, 1171 (2011).
2. M. Ecker et al., J. Power Sources, 248, 839 (2014).
where Lc is the fractional loss per cycle, Ls is the fractional loss per 3. R. D. Deshpande and D. M. Bernardi, J. Electrochem. Soc., 164, A461 (2017).
4. Z. Gao et al., Energy, 122, 588 (2017).
year during storage and t is the time in years. This assumes that all 5. A. Jana, G. M. Shaver, and R. E. García, J. Power Sources, 422, 185 (2019).
degradation processes are linear and do not slow with time. Again, 6. J. Schmitt, A. Maheshwari, M. Heck, S. Lux, and M. Vetter, J. Power Sources, 353,
this is a worst case scenario. 183 (2017).
Figure 23 shows the projected fractional capacity of these 7. S. F. Schuster et al., J. Energy Storage, 1, 44 (2015).
8. M. Dubarry, N. Qin, and P. Brooker, Curr. Opin. Electrochem., 9, 106 (2018).
NMC532/graphite cells as a function of time in years in a scenario 9. P. Keil et al., J. Electrochem. Soc., 163, A1872 (2016).
where the cells are cycled once per day (100% DOD to 4.3 V) in a 10. M. Naumann, M. Schimpe, P. Keil, H. C. Hesse, and A. Jossen, J. Energy Storage,
cycling event that takes 6 hours. It is also assumed that during the 17, 153 (2018).
remaining time, cells are stored at full state of charge. It is clear from 11. P. Keil and A. Jossen, J. Electrochem. Soc., 164, A6066 (2017).
12. J. Li, H. Li, W. Stone, S. Glazier, and J. R. Dahn, J. Electrochem. Soc., 165, A626
Figure 23 that these cells would provide an exceptionally long total (2018).
driving range in an EV if the cells were maintained at an average tem- 13. L. Ma et al., J. Electrochem. Soc., 165, A891 (2018).
perature of 20°C. Even if the cells were continually at 40°C, 10 years 14. L. Ma et al., ACS Appl. Energy Mater., 1, 7052 (2018).
of lifetime to 70% capacity and a total driven distance of 1,200,000 km 15. L. Ma, L. Ellis, S. L. Glazier, X. Ma, and J. R. Dahn, J. Electrochem. Soc., 165, A1718
(2018).
is projected. It is worth noting that only 3650 cycles would be required 16. X. Ma et al., J. Electrochem. Soc., 166, A711 (2019).
for this total driven distance and 3700 cycles have been demonstrated 17. D. Y. Wang et al., J. Electrochem. Soc., 161, A1818 (2014).
in Figure 16. 18. D. Y. Wang, N. N. Sinha, R. Petibon, J. C. Burns, and J. R. Dahn, J. Power Sources,
Most important to realize is that Figure 23 assumes 100% DOD 251, 311 (2014).
19. L. Ma et al., J. Electrochem. Soc., 161, A1261 (2014).
cycling on every cycle and storage at full state of charge. If the reader 20. J. R. Dahn, G. Erlich, and D. Doughty, Linden’s Handbook of Batteries, Chapter 17
reviews the literature data in Figures 1 and 2, the reader will realize K. Beard, Editor, 5th ed., p. 755, McGraw-Hill, (2019).
that the lifetime will be much better in situations where the DOD 21. E. R. Logan et al., J. Electrochem. Soc., 165, A21 (2018).
is limited and in situations where cells are stored at lower states of 22. J. Li et al., J. Electrochem. Soc., 165, A1027 (2018).
23. K. L. Gering, Electrochim. Acta, 225, 175 (2017).
charge. Admittedly, the projections in Figure 23 use the incredibly 24. T. M. Bond, J. C. Burns, D. A. Stevens, H. M. Dahn, and J. R. Dahn, J. Electrochem.
simple model described by equation 1. It is our opinion that more Soc., 160, A521 (2013).
sophisticated models will lead to even longer lifetime projections. 25. A. J. Smith, J. C. Burns, S. Trussler, and J. R. Dahn, J. Electrochem. Soc., 157, A196
(2009).
26. R. Fathi et al., J. Electrochem. Soc., 161, A1572 (2014).
Conclusions 27. I. Bloom, J. P. Christophersen, D. P. Abraham, and K. L. Gering, J. Power Sources,
157, 537 (2006).
Single crystal NMC532/graphite cells with exceptional lifetime 28. I. Bloom et al., J. Power Sources, 139, 295 (2005).
have been developed. Storage and cycle testing up to 3 years in du- 29. K. Honkura, H. Honbo, Y. Koishikawa, and T. Horiba, ECS Transactions, 13, 61
(2008).
ration has been presented at 20, 40 and 55°C. The lifetime of these 30. H. M. Dahn, A. J. Smith, J. C. Burns, D. A. Stevens, and J. R. Dahn, J. Electrochem.
cells far exceeds that of other NMC/graphite cells reported in the liter- Soc., 159, A1405 (2012).
ature and which have been used for lifetime modelling. It is suggested 31. J. Li et al., J. Electrochem. Soc., 164, 1534 (2017).

Downloaded on 2019-09-07 to IP 100.42.241.162 address. Redistribution subject to ECS terms of use (see ecsdl.org/site/terms_use) unless CC License in place (see abstract).