Insights from Safety Tests with

an On-Demand Internal Short

Circuit Device in 18650 Cells

By

Eric Darcy/NASA-JSC

Houston, TX

International Battery Seminar

Fort Lauderdale, FL

21-23 Mar 2017

Photo Credits: NREL

2

Why are Li-ion cell internal shorts still a concern?• Despite extensive QC/QA, standardized industry

safety testing, and over 26 years of manufacturing learning, major recalls have taken place and incidents still occur.

– Search “battery fire recall statistics” at CPSC website (http://www.cpsc.gov/en/Search/?query=&filters=recalls) finds:

– 28 recalls in last 12 months (May 2015-May 2016).

– The recall rate has slightly increased over the last 10 yrs.1

• Many safety incidents that take place in the field due to latent defects not detectable at the manufacturer

• These internal short incidents are estimated at 1 to 0.1 ppm probability (well beyond 6 σ) in consumer applications using cells from experienced and reputable manufacturers2

– Risk increases to 10 to 1000 ppm for certain lots of cells even from reputable manufacture

• Boeing 787• Samsung Galaxy Note 7

• This risk can’t be retired by rigorous screening

• Worldwide Li-ion battery market is valued at $20 billion and failures can cost billions

1. D. Doughty, Li-ion Cell and Battery Safety, NASA-JSC Li-ion Battery Course 2017

2. B. Barnett, TIAX, NASA Aerospace Battery Workshop, Nov 2008

Galaxy Note 7

787 battery

Sony Laptop

3

Samsung Galaxy Note 7 Lessons Learned• On 23 Jan 2017, Samsung shared

their failure investigation results– https://www.youtube.com/watch?v=Iu18C

ykEH9o

– Cell manufacturer A failures were due to poor cell design

• Not enough pouch room to prevent bending of anode tabs in upper right corner

• Separator stress causes internal short circuits with field use: cycling

– Cell manufacturer B failures were due to poor quality control

• Ultrasonic Al weld protrusions on positive tab bridge through separator to the negative with field use: cycling

– Fundamental reason was improper cell design and quality control vetting prior to highly energetic product release

• $5.3 billion cost impact to Samsung!

• Even reputable cell/battery manufacturers can occasionally produce risky cell designs and/or poor manufacturing with latent hazards

4

Examples of Other Li-ion Battery Incidents

• E-cigarettes

– https://www.youtube.com/watch?v=k1LjSuq0rk8

– https://www.youtube.com/watch?v=xu13u-jWasg

• RC Airplanes

– https://www.youtube.com/watch?v=MTFpwONHgpg

– https://www.youtube.com/watch?v=k9mcNvOGKtI

• Hover boards

– https://www.youtube.com/watch?v=yY0VdVdfWUA

– https://www.youtube.com/watch?v=3DPXyCaTMnw

Battery thermal runaway occurred

while inside user’s pant leg pocket

Links from D. Doughty presentation at NASA-JSC Battery Course 2107

5

Tragic Fatality Due Li-ion Battery Accident

• http://www.cnn.com/2017/03/14/health/pennsylvania-hoverboard-fire-kills-child/

• (CNN) - A 3-year-old girl died over the weekend in a fire ignited on March 14, 2017 by a recharging self-balancing scooter in her Harrisburg, Pennsylvania, home. Her death is believed to be the first in the nation to result from a blaze caused by the battery operated toy, some of which have previously been recalled due to their potential fire hazard.

• The blaze, which began before 8 p.m. Friday, sent six people to the hospital.

• Ashanti Hughes died in Lehigh Valley Hospital-Cedar Crest's burn unit Saturday morning, said a spokesman for Lehigh County Coroner's Office and Forensic Center.

Home fire caused by overnight

charging of hoverboard battery

going into TR

• Latency potential of metallic FOD has been demonstrated*– Small Ni particle implanted in commercial cells

in key location to produce Type 2 short• Cathode aluminum collector to anode active material

– Reassembled cell passes acceptance testing

– Requires > 50 cycles for it to develop into thermal runaway hazard

• One can not screen out all potential latent defects by acceptance testing alone– Implementation of effective manufacturing FOD

mitigation measures is key

– Periodic line audits are a must

– As are cell DPAs

Latency Potential of Metallic Defects in Cells is Real

* Barnett et. al, Power Sources Conference, Las Vegas, NV, 2012

6

Background and Prior Methods• Lithium Ion Battery Field Failures - Mechanisms

• Latent defect (i.e., built into the cell during manufacturing) gradually moves into position to create an internal short while the battery is in use.

• Sony* concluded that metallic defects were the cause of its recall of 1.8-million batteries in 2006

• Inadequate design and/or off-limits operation (cycling) causes Li surface plating on anode, eventually stressing the separator

• Alternate abuse test methods are less relevant to field failures• Mechanical (crush, nail penetration, etc.)

• Cell can or pouch is breached; pressure, temperature dynamics are different

• Limited in ability to customize the type of short within the cell

• Thermal (heat to vent, thermal cycling, etc.)• Cell exposed to general overheating rather than point-specific overheating

• Cell seals can fail and/or vent activates prior to thermal runaway unlike during a JR point-specific ISC

• May not be a valid verification of “shutdown” separators

• Can only test active to active material shorts (type 1)

• Electrical (overcharge, off-limits cycling, etc.)• Induces thermal runaway transition at SOC levels not relevant to the latent-

defect–induced field failure

• Repeated off-limits cycling methods take much time, are unpredictable, and are not “on demand”

*. Nikkei Electronics, Nov. 6, 2006

7

Negative Electrode

Anode Active Material

Anode Active Material

Positive Electrode

Positive Active Material

Positive Active Material

Electrode

to

Electrode

Anode

to

Cathode

Anode

to

Electrode

SeparatorSeparator

Electrode

to

Cathode

Spiral wound battery shown – can also be applied to prismatic batteries.

Negative Electrode

Anode Active Material

Anode Active Material

Positive Electrode

Positive Active Material

Positive Active Material

Electrode

to

Electrode

Anode

to

Cathode

Anode

to

Electrode

SeparatorSeparator

Electrode

to

CathodeNegative Electrode

Anode Active Material

Anode Active Material

Positive Electrode

Positive Active Material

Positive Active Material

Electrode

to

Electrode

Anode

to

Cathode

Anode

to

Electrode

SeparatorSeparator

Electrode

to

Cathode

Spiral wound battery shown – can also be applied to prismatic batteries.

4 Types of Internal Short Circuits

None of the prior methods are

capable of replicating all 4 types

Implantable Prior Methods– Forced Internal Short Circuit Test (JIS C 08714)

• Requires opening a charged cell, unwinding JR, implanting L shaped Ni particle, and reassembling the cell (this is hazardous)

• Short is induced by pressure to the JR with a press

• Only required on certain Li-ion cells in France, Japan, Korea, and Switzerland

• Only tests active to active shorts (Type 1)

– Metallic seeding (TIAX)• Requires opening a charged cell, unwinding JR, implanting Ni particle at specific location to create Al-anode short

(Type 2) – This is hazardous

• Unreliable activation: requires time and/or charge discharge cycles

• No control of SOC when short activates

– Implanting a heater to melt the separator (SAFT)• Tungsten wire heater is implanted in dry JR between separator and active material

• Heater wires are routed through epoxy sealed openings in cell header, thus only applicable to cell designs larger than 26650 format (not applicable to most widely used industry standard size, 18650)

• Drives a thermal runaway response in large cells

– Until now, no reliable and practical method existed to create on-demand, location and type controlled, internal shorts in >1.5Ah Li-ion cells that produce a response that is relevant to the ones produced by latent defect induced field failures.

8

9

NREL/NASA Cell Internal Short Circuit Device

Wax formulation used

melts ~57C

US Patent # 9,142,829

issued in 2015

2010 Inventors:

• Matthew Keyser, Dirk

Long, and Ahmad

Pesaran at NREL

• Eric Darcy at NASA

Graphic credits: NREL

Thin (10-20 m) wax

layer is spin coated

on Al foil pad

Tomography credits: University College of London

ISC Device in 2.4Ah cell designPlaced 6 winds into the jellyroll

Active anode to cathode collector short

2016 Award Winner

ISC Device Design

Anode Active Material to Cathode Current Collector ShortType 2 – “Anode Active to Collector”

NMP used to remove active material

10

Cathode Active layer < 76 microns

Aluminum ISC Pad 76 microns

Cu Puck 50 microns

Separator 20 microns

Copper ISC Pad 25 microns

Anode Active Layer X microns

Cathode Active layer <76 microns

Anode Active Layer X microns

Wax layer ~20 microns

Cathode Current Collector

Note: Trials with 25 micron Cu puck produces frequent activation duds

11

Implantation into an 18650 Cell• NREL fabricated the ISC

devices

• Partnered with E-one MoliEnergy (Maple Ridge, BC) for the implantation into their 2.4Ah cells

• Moli performed cycling and activation tests

• NASA-JSC performed activation tests

Photo credits:

Moli Energy

a) 3D representation of the individual components of the ISC device, where the positive electrode material is

etched away to expose the aluminum current collector surface to the aluminum disk of the device.

b) 3D illustration showing where the ISC components are inserted into the spiral wound jelly roll of an 18650 cell.

Graphic credits: NREL

12

2013 CT of ISC Device Inside a MoliJ 18650

• Device implanted 3 winds into

the jellyroll at mid height

• Jellyroll length did not have to

be trimmed to fit into the can

2013 Iteration

• Compared separators

• Tri-layer shutdown separator

• Prolypropylene non-shutdown

• Compared ISC types

• collector-collector (type 4)

• Anode active – aluminum (type 2)

Image credits:

E-one Moli EnergyM. Shoesmith et.al, “Cylindrical Li-ion Cell Response to Induced Internal Short” NASA Battery Workshop, Huntsville, AL, Nov 2013

13

M. Shoesmith et.al, “Cylindrical Li-ion Cell Response to Induced Internal Short” NASA Battery Workshop, Huntsville, AL, Nov 2013

14

Type 2 ISC Device in 18650 Cell

Cell assembled with non-shutdown separator – Designed to fail

14

15

Example of Shutdown Separator Working

DPA of Type 4 hard short that shutdown benignly

• Obvious short at activation, OCV recovery is not real due to CID activation

• Separator is shutdown and adhered to cathode

• Areas of fully charged anode are visible consistent with shutdown operation

M. Shoesmith et.al, “Cylindrical Li-ion Cell Response to Induced Internal Short” NASA Battery Workshop, Huntsville, AL, Nov 2013

16

Shutdown Feature of Separator Confirmed

Separator Porosity

Edge Band << Control

Shutdown confirmed

Separator sample from cell that

benignly shutdown after a hard

ISC device short

Control separator taken from wound

jellyroll that was never wetted

16

16

M. Shoesmith et.al, “Cylindrical Li-ion Cell Response to Induced Internal Short” NASA Battery Workshop, Huntsville, AL, Nov 2013

17

Summary of 2013 Activations with 2.4Ah MoliJ Cell

• Runs in early CY13 with non-shutdown separator

– Type 4• 6 out of 10 driven into TR

• 4 duds (soft shorts)

– Type 2• 6 out of 10 driven into TR

• 4 duds (soft shorts)

• DPAs revealed possible reasons for the duds– Active material hole made too small for the device to

achieve good contact

– Alignment of ISC device may have been altered during rewind of JR

• Subsequent CY13 implantations with shutdown separator

– Type 4• 7 of 10 hard shorts

– 6 were shutdown by separator

– 1 was driven into TR

• 3 duds (soft shorts)

– Type 2• 8 of 10 hard shorts

– 2 were shutdown by separator

– 6 were driven into TR

• 2 duds (soft shorts)

• Shutdown separator does not work against Type 2 shorts

– Single TR with Type 4 might be due to misalignment of ISC device which could have turned it into a Type 2 short

M. Shoesmith et.al, “Cylindrical Li-ion Cell Response to Induced Internal Short” NASA Battery Workshop, Huntsville, AL, Nov 2013

18

Electrolyte Study in 20 Ah Cell

18

Non-flammable

electrolyte showed

no improvement over

the control

electrolyte in

preventing fire.

Control electrolyte IR image

when ISC activated

Non-Flammable

electrolyte IR

and visual image

when ISC is

activated

19

Single Cell TR – Moli 2.4Ah with ISC Device

Open air test with cell charged to 4.2V and with TCs welded to cell side wall (2) and bottom (1)

Tomography credits: University College of London

2014 iteration with type 2 device 6 winds into JR and PP sep

20

CT Images of ISC Device (2015 implantation)Clearly shows that active material hole

boundaries are much wider than the device

Cu puck

Al pad removed for clarity

Images courtesy of D. Finegan, UCL

21

CT images (cont.)

Misalignment of Cu and

Al pads creates stress

zones on the separator

and could explain the

damage initiation at the

ISC device edge in

some videos

Image picks up tweezer

marks during fabrication

on the Cu puck

Images courtesy of D. Finegan, UCL

22

Milliseconds after short initiation

a) Before activation

b) 5ms later, short

initiation is visible

on periphery of

device

c) Snapshots

showing the

nearby electrode

layers fluidization

and delamination

with gas

generation“Characterising Thermal Runaway by Inducing and Monotoring Internal Short Circuit within Li-ion Cells” D. Finegan, et.al., EE-

ART-02-2017-000385 accepted for publication in the Journal of Energy & Environmental Science, Mar, 2017.

23

2.4Ah 18650 with ISC device video courtesy of D. Finegan, UCL

Extremely high speed X-ray videos of cell with ISC device taken at the European Synchrotron Research Facility in Grenoble, France

24

Lessons Learned with 2.4Ah MoliJ - ISC

• In all cell designs tested, device has negligible impact on performance

– Capacity performance at low and high rates (1C) over 20 cycles unaffected• Other cells tested include 2 pouch cell designs

• 18650 cell designed to fail on-demand is maturing well

– After 3 iterations of implantations, Moli achieved >90% success rate in getting hard shorts

• With anode-aluminum ISC device and non-shutdown separator

• This success rate makes it suitable for single cell TR battery hazard testing

• Insights from the ISC device implantations• Shutdown separator appears effective against collector-collector shorts

• Anode to Al shorts produce the worst case response

• Shorts involving cathode active material are benign suggesting that manufacturing measures to prevent metallic FOD from getting imbedded into the cathode are very important– This implies that manufacturers should be most vigilant during active material mixing, coating and baking

onto collectors, and calendaring to final coating thickness

– Shorts to aluminum collector must be rigorously prevented

25

Current Li-ion Spacesuit Battery

Used on over 27 spacewalks for far

Battery

26

Spacesuit Battery Heat Sinks

0.5mm cell spacing, Al 6061T6

Sink A Sink A

Sink A

Sink B Sink BSink C

No corner cells - Every cell has at least 3 adjacent cells

27

Attempt to Drive TR with Bottom Heater While in Al HS

40

30

20

10

0

He

ate

r P

ow

er,

W

500040003000200010000

Time, s

100

80

60

40

Te

mp

era

ture

, C

HeaterW TC_1 TC_2 TC_3 TC_4 TC_5 TC_7 TC_8

Bottom of Cell Heater Test with Al Heat Sink

TCs 1-7

TC 8

TC 8

Heater fails at 48W

Can’t get trigger cell > 100C

after > 1hr and 3 attempts

Bottom surface heater

28

Spacesuit Battery Brick: Thermal Runaway Test

Trigger cell is 2.4Ah cell with Type 2 ISC device

Result - No TR propagation to other cells

29

Metallic Interstitial Heat Sink is Effective

• Cell can isolated

with mica paper

sleeves and very

small air gap

• Heat sink

spreads heat

more quickly

through multiple

layers than

through mica and

onto cells

• Heat from trigger

cell is quickly

dispersed and

shared among

more cells

Graphic and analysis courtesy of Paul Coman

30

Safer, Higher Performing Spacesuit Battery Design

65-Battery Brick

Features

• 65 High Specific Energy Cell Design 3.4Ah (13P-5S)

• 37Ah and 686 Wh at BOL (in 16-20.5V window)

• Cell design likely to side wall rupture, but supported

Fusible link

Assembly tab

Removed after welding

Aluminum interstitial

heat sink protects

adjacent cells from side

wall ruptures during TR

and dissipates heat very

effectively

Full scale battery

31

Cell Brick Assembly > 180 Wh/kg

• With 12.41 Wh/cell, cell brick

assembly achieves 191 Wh/kg• Assuming 12.41Wh per cell

• Design has 1.4 parasitic mass

factor

– Cell mass x 1.4 = Brick mass

Cells

Heat sinksMica sleeves

Capture plates

Ceramic bushings

Ni-201 bussing

Other

Mass Distribution

Cells Heat sinks Mica sleeves Capture plates Ceramic bushings Ni-201 bussing

Mass Categories g %

LG MJ1 Cells 3012.75 71.3%

Heat sinks 824.95 19.5%

Mica sleeves 182.31 4.3%

Capture plates 115.81 2.7%

Ceramic bushings 60.15 1.4%

Ni-201 bussing 29.71 0.7%

Total 4225.7

32Spacesuit Prototype Battery Test Summary

• Al Heat Sink Tests– 4 attempts to drive > 250Wh/kg cell into TR – All failures

• 2 with Panasonics, 2 with LGs, all with home made bottom heaters

– 5 attempts with 2.4Ah ISC device cells – No propagation of TR• 1 dud and 4 success with the 2.4Ah ISC cell driven into TR

– 2 heat to vent tests with 5 fully charged 3.4Ah cells each• No side wall ruptures in areas supported by the sink

• LLB2 brick tests (All six 2.4Ah ISC cells successfully driven to TR)– 3 no-Ni bussing brick tests

• No TR propagation and no OCV changes to adjacent cells with excellent temp margins– Interior cell trigger T ~ 19C (one run)

– Edge cell trigger T ~ 42C (two runs)

• Interior cell trigger are less vulnerable than edge cells based on temperature rise (max-onset T) on adjacent cells

– 3 Ni bussing (13P5S)• No propagation of TR, no impact on adjacent cell OCVs

• Very good temperature margins (vs onset of TR temperature)– Interior cell trigger: T ~ 30C (one run)

– Edge cell trigger T ~ 48C (one valid run)

• LLB2 full scale tests (4 runs – 2 w/ 2.4Ah, 2 with 3.4Ah ISC device implanted cells)– No propagation of TR (even with side wall rupture of trigger cell in 1st test w/ 3.4Ah trigger cell)

– Maximum adjacent cell temperature rise with 2.4Ah trigger cell was 55-58C

– Maximum adjacent cell temperature rise with 3.4Ah trigger cell was 94C w/ side wall rupture and 46C with bottom rupture

– Screened vents were demonstrated as a successful flame arresting solution

Pre-testPost-test

33

ISC Device Enable Verification of Vaporizing Heat Sink TR Shields

• Tests of Thermal Runaway Shields from KULR Technology were made possible with ISC device inside 3.3Ah cell design– Single trigger cell cylindrical wall covered with

2 TR shields

– Only top and bottom surface of cell is exposed to apply heater to activate ISC device

– Adjacent Al cylinders see < 20C temperature rise from onset of thermal runaway of trigger cell

– Fully populated battery module tests to come next

34

3.5Ah LG ISC Device Implantation

• LG ISC device implantation went

exceedingly well

– 36/37 cells have successfully

completed formation

– 5/5 cells activated at 100% SoC so

far and both were driven into TR

after measuring 60-70C on cell

skin prior to onset of TR

– 3/3 cells activated at 0% SoC with

hard short but no TR

– Delivery = 28 cells with ISC device

• All cells arrived at 0% SoC and > 2.9V

CT images of ISC device 3 winds into the JR

353.5Ah LG Cell with ISC Device Video

Image and video

courtesy of D. Finegan,

University College of

London

Extremely high speed X-ray videos of cell with ISC device taken at the Diamond Synchrotron Facility near Oxford, UK

36

3.5Ah LG Cell with ISC Device – Post Test

• JR ejected

• Top edge of crimp shows reflow steel

• Side wall breach in neck of crimp is clocked with ISC device

• Smaller breach in can wall is slightly off the ISC device clocking and above it

• ISC device suggests that cell enclosure design needs improvement

Image courtesy of D. Finegan, UCL

37

ISC Device Location Reveals Side Wall Rupture Risk

• 3.5Ah LG cell can thickness– 155-165 microns

– No bottom vent

– Cell design not intended for EVs

• Unsupported oven heating test– No side wall ruptures (30 cells)

– Slow external heating to TR

• Unsupported circumferential heater test – No side wall ruptures (5 cells) at ~30W

– 1 of 3 side wall rupture at ~60W

• With ISC device (11 tested so far)– 8 sidewall ruptures

• 5 unsupported

• 3 supported by Al interstitial heat sink

– 1 bottom rupture• Supported by Al interstitial heat sink

– 2 vented through header• Supported by Fe tubes

Photo credit: D. Finegan, University College of London

ISC device in 3rd windCircumferential heater

near bottom of can wall

38

Orion Battery 14-cell Block

UPPER CAPTURE PLATE

G10 FR4 FIBERGLASS

COMP

MACOR VENT

TUBES

SYNTACTIC

FOAM LINER18650 CELL

304 Stainless

Steel Sleeve –

9 mil wall

thickness

LOWER HEAT-SINK

CAPTURE PLATE

6061-T651 ALUM

Orion 14P-8S

Superbrick

Draw cell heat generation

through cell bottom

How Effective Are Steel Tubes?

• Fully charged 3.4Ah ISC device cells in positions 1 (corner) and 8 (interior) clocked towards adjacent cells

• Block heated to > 60C to activate ISC devices

• Corner cell wrapped with 0.015” (381 m) SS tube experienced side wall rupture outside of tube– Dissection of tube found

no cell can side wall ruptures inside tube area

• Interior cell wrapped with 0.009” (229 m)– No side wall ruptures

outside or inside tube

1

8

18

Corner cell 1

Interior cell 8

Orion 14-cell assembly with cell,

tubes, foam

40

Cell Level Benefits

• ISC device enable unique insight into cell thermal runaway mechanism that replicates field failure responses and conditions– It’s consistent and reliable, safe to implant, and can be

activated on demand

– Can be used to test all 4 types of shorts, at any state of charge, and without compromising cell enclosure

• We’ve confirmed that the anode-to-Al short is most hazardous

• Predisposition of cells to experience side wall ruptures can be assessed fairly with device

– Can be used to test numerous cell safety features and find out their limitations and greatly improving safety

• non flammable electrolytes, advanced separators, internal fusible links, bottom vents, thicker can walls, etc

• The insights from high speed X-ray videos are shedding new light on cell failure and are guiding the development of safer commercial cell designs

41

Battery Level Benefits• ISC device enables critical battery safety

verification– Recent NASA studies show that maximizing heat

dissipation between cells is best way to achieve high performance and best protect adjacent cells from TR propagation

– With the aluminum interstitial heat sink between the cells, normal trigger cells can’t be driven into TR without excessive temperature bias of adjacent cells

– ISC device in trigger cells enabled the verification of the spacesuit battery to be passively TR propagation resistant

• Spacesuit battery brick design achieves > 190 Wh/kg (vs 120 Wh/kg for current spacesuit battery design)

– ISC device in trigger cells enable verification that the steel sleeves on each cell effectively mitigates side wall rupture hazard – critical for Orion CM battery

• Replaces the catastrophic hazard of the previous large cell battery design

42

ISC Device is Maturing Very Well• Maturity

– Achieving 100% activation into thermal runaway response with >3.3Ah 18650 cell designs and Al-anode short

• Despite ceramic coated separators

– Cells with device sought by numerous researchers (including University College of London) and battery developers (ex, Navy applications, X-57 electric airplane)

– Device has been successfully implanted in more than 5 commercial cells designs to date

– Device has enabled the safety verification of the spacesuit, small experiment, and Orion batteries, and with many more to come

– Licensing agreements are currently in negotiations with >3 parties (cell manufacturers, battery heat sink developer, and consumer electronic company)

X-57 Electric Airplane

Orion CM Battery

43

Acknowledgements

•Matthew Keyser, NREL, for designing and fabricating all the

ISC devices

•Mark Shoesmith, E-one Moli Energy, for successfully

implanting the ISC device in one of their 2.4Ah cell design

•Jee young Park, et. al, LG Chem, for successfully implanting

the ISC device in one of their 3.5Ah cell design

•Donal Finegan, University College of London, for tomography

and high speed X-ray videos