Thermal Behaviour: Heat, Lithium Plating and Runaway
Why temperature is the master variable, where the heat comes from, why gradient matters more than peak — and the cold-charging failure that leaves no trace until the pack fails.
Battery Fundamentals · Part 2 — State, Health and Degradation · Chapter 6 · 18 min read
If you could control only one variable in a battery pack for its whole life, you would choose temperature. Every degradation mechanism accelerates with heat, every safety mechanism is defined by a temperature threshold, and the single most damaging routine mistake in the industry — charging a cold pack — is a temperature problem that leaves no trace until the pack fails.
15–35 °C
The sweet spot
×2 per 10 °C
Ageing rate
0 °C
Charge lockout floor
≤5 °C
Target pack gradient
6.1 — Temperature is the master variable
Chemical reaction rates follow Arrhenius kinetics, and the rough working rule that falls out of it is blunt enough to design around:
Ageing rate roughly doubles for every +10°C
A rule of thumb, not a law — but close enough to be useful for pack siting and warranty modelling.
A pack living at 45°C ages roughly four times faster than one at 25°C. Not four per cent faster — four times. That is the difference between a pack reaching its warranty term and a pack reaching half of it, and it is decided by where the pack is mounted and what colour the enclosure is, long before anybody drives it.
Cold does something different. It does not damage the cell on discharge; it simply reduces what the cell can do. At 0°C an LFP cell may deliver only 70 to 80 per cent of its rated capacity and have two to three times its normal internal resistance, which shows up as voltage sag and apparent range loss. Both recover when the pack warms.
Important
Heat causes permanent damage. Cold causes temporary limitation — except when charging, where cold causes permanent damage of the most dangerous kind. That asymmetry is why every BMS has a narrower temperature window for charge than for discharge.
6.2 — Where the heat actually comes from
Three sources, of which one dominates in almost every real application.
6.2.1 — Joule heating — I²R
P_heat = I² × R
P_heat in watts, I in amperes, R the total resistance of the current path.
The squared term is the whole story. Doubling the current quadruples the heating. This is why fast charging and hard acceleration heat a pack disproportionately, and why a resistance problem and a thermal problem accelerate one another: heat raises resistance, resistance generates more heat.
Worked example 6.1 — A bad joint on a 60 V pack
A 20S pack draws 100 A on a gradient. Every millohm of unwanted resistance in the current path — a poor crimp, a corroded terminal, a cold weld — dissipates:
P = 100² × 0.001 = 10 W
Ten watts into a connector the size of a thumbnail. That is a soldering-iron tip, sitting inside a sealed enclosure, next to cells. Three such joints and you have a 30 W heater running whenever the vehicle climbs. This is why joint resistance is measured at end-of-line and not assumed.
6.2.2 — Entropic heat
The electrochemical reaction itself absorbs or releases heat depending on state of charge. It is much smaller than Joule heating at any meaningful current, and it can be negative — some cells cool slightly over parts of their discharge. Worth knowing about, rarely worth designing around outside cell-level modelling.
6.2.3 — Ambient
In Indian conditions this is frequently the largest term, and it is present before the vehicle is switched on. A dark enclosure mounted under a three-wheeler floor, parked in direct sun at 45°C ambient, can sit well above 60°C at the cell surface having done no work at all. The pack then starts its duty cycle from there.
This is the term you fix with a decision rather than with hardware: mounting position, enclosure colour, shading and airflow are free at design time and impossible to change later.
6.3 — Managing it
Thermal management approaches and where each belongs
| Approach | How it works | Cost and complexity | Fits |
|---|---|---|---|
| Passive | Thermal mass, conductive potting, aluminium enclosure as a heatsink, cell spacing, insulation from road and motor heat, light enclosure colour, mounting out of direct sun. | Lowest. Mostly free at design time. | Low C-rate applications — e-rickshaws, 2W/3W, home storage. Adequate when done deliberately. |
| Active air | Fans forcing air through channels between cells or modules. | Low cost, but noisy and it moves dust and moisture into the pack. | Rarely worth it below the IP rating you lose. A real problem in Indian dust and monsoon. |
| Active liquid | Coolant plates in contact with cell faces, with a pump, radiator and controls. | High cost and complexity, plus a leak path next to live cells. | Cars and large stationary systems. Overkill for a three-wheeler. |
In plain English
For the duty cycles this series is mostly concerned with — a three-wheeler at 0.2 to 0.3C continuous, charging overnight at 0.2C — passive management done properly beats active management done cheaply. Most Indian pack thermal problems are mounting and layout problems, not cooling-capacity problems.
6.4 — Gradient matters more than peak
The instinct is to design for the highest temperature in the pack. The more useful target is the difference between the hottest and coldest cells.
Consider a pack where cell 1 sits at 30°C and cell 12 at 45°C. Cell 12 ages at roughly three times cell 1’s rate. Within a year the two cells have measurably different capacities and resistances — and because they are in series, the pack is limited by the worse one in both directions. A thermal gradient converts itself into a permanent capacity imbalance that no amount of balancing can undo, because balancing equalises charge, not capacity.
Important
Target ≤5°C spread across the pack under load. Achieving it is usually a question of layout — where the hot cells sit relative to airflow, busbars and the enclosure wall — rather than of adding cooling power. A well-laid-out passive pack routinely beats a badly laid-out cooled one.
The imbalance this creates, and why it cannot be corrected by the BMS, is covered from the balancing side in Chapter 2.
6.5 — Lithium plating — the cold-charging failure
Charging a lithium cell below roughly 0°C is the most damaging routine mistake made in this industry, and it is routine precisely because nothing appears to go wrong.
At low temperature, lithium ions cannot intercalate into the graphite anode fast enough to keep up with the current arriving. Rather than waiting, they deposit on the anode surface as metallic lithium. Two things follow, and both are permanent.
- •Immediate irreversible capacity loss. Plated lithium is removed from circulation. It is not going back into the cathode. Every cold charge takes a slice out of the cell’s capacity, and the slices accumulate.
- •Dendrites. The deposit grows as needle-like structures. A dendrite that reaches through the separator creates an internal short — the failure mode with no fuse, described in Chapter 3. This is how a cold-charged pack becomes a fire risk months later.
Important
A pack charged on cold mornings through a North Indian winter can lose 20 per cent of its capacity in a season and become genuinely unsafe, while showing nothing unusual at any point. The customer reports “it got worse over the winter” and the diagnosis arrives too late.
The mitigation is not a recommendation, it is a requirement: a BMS charge lockout below 0°C, set conservatively at 5°C, or a heater that brings the pack into range before charge is permitted. Note that many BMS boards ship with a charge low-temperature limit of −10°C, which is looser than every cell manufacturer allows. Closing that gap is the pack integrator’s job, not the board vendor’s — the mismatch is mapped out in Chapter 1.
In plain English
Discharging in the cold is fine. Riding the vehicle on a January morning does no harm — you simply get less range until it warms. It is only charging that plates. If you teach a fleet one thing about winter, make it: ride first, charge after.
6.6 — Thermal runaway, step by step
Once a cell is pushed past its limits — by overcharge, an internal short, a crush, or external heat — it follows a sequence that is well characterised and, past a certain point, unstoppable.
- 1~80–120°C — SEI decomposition. The passivating film breaks down exothermically, exposing fresh anode to electrolyte. The cell is now generating its own heat.
- 2~120–150°C — separator melt. Shutdown layers close the pores here if the separator has them. If it does not, or if the temperature keeps climbing, the film shrinks and internal shorts develop.
- 3~150–200°C — electrolyte decomposition. Flammable gas is generated and pressure rises until the cell vents. This is the last stage at which the outcome is still a damaged cell rather than a fire.
- 4~200°C+ for NMC, ~250°C+ for LFP — cathode breakdown. A nickel-rich cathode releases oxygen. Combustion becomes self-sustaining and cannot be smothered, because the cell is supplying its own oxidiser.
- 5Propagation. The burning cell heats its neighbours past their own onset temperature and the pack goes cell by cell. This is the stage that turns one defective cell into a destroyed vehicle.
Technical framing
LFP’s advantage is real and it is specific: the strong phosphorus–oxygen bonds in the olivine structure mean it does not release oxygen the way a layered nickel oxide does, so step 4 is far less violent and the onset is 50 to 70°C higher. It is not immunity. An LFP cell still vents flammable electrolyte gas and that gas still burns.
6.7 — Propagation resistance is a design requirement, not a cell property
This is the sentence worth taking away from the chapter. You cannot buy a cell that will not fail. Cells fail from manufacturing defects at rates that are small but not zero, and a fleet of ten thousand packs will encounter it.
What you can design is what happens next. A pack that contains a single-cell failure is an incident. A pack that propagates is a fire.
- •Cell spacing and thermal barriers. Physical separation and insulating material between cells slow heat transfer to neighbours. The cost is volumetric efficiency, which is the trade being made.
- •Directed venting. Cells vent. Design where the gas and ejected material go, and make sure it is not towards the rider, the fuel of another cell, or a sealed volume that will pressurise.
- •An enclosure that contains ejecta. Containment is not the same as sealing. A sealed box with no vent path is a pressure vessel.
- •Separation from the occupant. Physical distance and a barrier between the pack and the passenger compartment buys evacuation time, which is the only thing that matters once step 4 has started.
These requirements are written into the standards that govern a 2W/3W battery business in India, particularly AIS 156 and its amendments, which tightened propagation testing substantially after the 2022 fire incidents. The certification landscape is covered in Chapter 14.
Quick check: test yourself
1.A fleet reports that packs in one depot are failing warranty two years early, while an identical fleet elsewhere is fine. Where do you look first?
Show answer
2.Your pack peaks at 42 °C under load, comfortably inside limits, but has a 14 °C spread between the hottest and coldest cells. Is this acceptable?
Show answer
3.A BMS datasheet lists a charge temperature range of −10 °C to 65 °C. The cell datasheet says 0 °C to 45 °C. Which applies?
Show answer
4.Why does “buy better cells” not solve thermal propagation?
Show answer
Chapter summary
- ✓Ageing rate roughly doubles per 10 °C. A pack at 45 °C ages about four times faster than one at 25 °C, and the decision is largely made by mounting position and enclosure colour at design time.
- ✓Heat causes permanent damage; cold causes temporary limitation — except during charging, where cold causes the worst permanent damage of all.
- ✓Joule heating is I²R, so doubling current quadruples heat. One millohm of bad joint at 100 A is a 10 W heater inside a sealed enclosure.
- ✓For low C-rate Indian applications, passive management done deliberately beats active cooling done cheaply. Most thermal problems here are layout problems.
- ✓Design for gradient, not peak. A 14 °C spread becomes permanent capacity dispersion that balancing cannot correct. Target ≤5 °C.
- ✓Charging below 0 °C plates metallic lithium — immediate irreversible capacity loss plus dendrites that can short the separator. A charge lockout at 0 °C, conservatively 5 °C, is mandatory, and BMS defaults are usually looser than the cell allows.
- ✓Thermal runaway runs SEI decomposition → separator melt → electrolyte decomposition → cathode oxygen release → propagation. LFP’s advantage is a higher onset and no oxygen release, not immunity.
- ✓Propagation resistance is a design requirement, not a cell property. You cannot buy a cell that never fails; you can design a pack where one failure stays one failure.
Frequently asked questions
Why should you never charge a lithium battery below 0°C?+
Because lithium ions cannot intercalate into the graphite anode fast enough at low temperature, so instead they deposit on its surface as metallic lithium. That permanently removes lithium from circulation — immediate irreversible capacity loss — and grows as dendrites that can pierce the separator and cause an internal short. The damage is cumulative and invisible until failure. A pack charged on cold mornings through a winter can lose 20 per cent of its capacity in a season.
Is it safe to use a battery in cold weather?+
Discharging in the cold is fine — you simply get less range until the pack warms, because at 0°C an LFP cell may deliver 70 to 80 per cent of rated capacity with two to three times the internal resistance. Both recover. It is only charging that plates lithium. If you teach a fleet one thing about winter, make it: ride first, charge after.
How much does temperature affect battery life?+
Roughly a doubling of the ageing rate for every 10°C rise, from Arrhenius kinetics. A pack living at 45°C ages approximately four times faster than one at 25°C — the difference between reaching a warranty term and reaching half of it. Much of that is decided by mounting position, shading and enclosure colour, which are free at design time and impossible to change later.
Why does a temperature gradient across a pack matter more than the peak temperature?+
Because it becomes permanent. If cell 1 runs at 30°C and cell 12 at 45°C, cell 12 ages at roughly three times the rate, and within a year the two have measurably different capacities. In a series string the pack is limited by the worse one in both directions, and balancing cannot fix it — balancing equalises charge, not capacity. Target a spread of 5°C or less, which is usually a layout question rather than a cooling-power question.
Is LFP immune to thermal runaway?+
No. LFP’s advantage is real and specific: the strong phosphorus–oxygen bonds in its olivine structure mean it does not release oxygen the way a nickel-rich cathode does, so its runaway onset is 50 to 70°C higher and the event is far less violent. But an LFP cell still vents flammable electrolyte gas, and that gas still burns.
What stops one bad cell destroying a whole pack?+
Design, not cell quality. You cannot buy a cell that never fails — defect rates are small but not zero, and a fleet of ten thousand packs will meet it. What determines whether one failure stays one failure is cell spacing, thermal barriers between cells, directed venting away from occupants, an enclosure that contains ejected material rather than sealing it, and physical separation from the passenger compartment.
Reviewed by
Battery Fundamentals is an original educational series on lithium battery technology. Threshold tables, cycle-life curves and worked examples use representative values drawn from published product specifications and widely-observed industry patterns, not measured data from a specific product. Always consult the current manufacturer datasheet before making design, purchasing, warranty or certification decisions.