Most advice on extending lithium-ion life mixes real physics with folklore, and it almost always assumes a phone. The levers that matter for a 400 kWh grid battery under a throughput warranty are not the same as the ones that matter for a laptop, even though the underlying chemistry is identical.
Two ageing processes run in parallel in every cell. Calendar ageing depends on time, temperature and the state of charge the cell sits at. Cycle ageing depends on how much charge you push through, how deep the cycles are, at what current, and again at what temperature. A cell parked in a warehouse ages. A cell cycled hard ages faster. Separating the two is the first step to knowing which lever is worth pulling.
What actually consumes capacity
Three mechanisms account for most measurable fade in commercial cells.
Film growth at the graphite anode, usually described as continued SEI formation, consumes cyclable lithium. It never stops, it accelerates with temperature, and it accelerates as the anode potential drops, which is to say as state of charge rises. This is the dominant path in cells that spend most of their life sitting still.
Lithium plating occurs when lithium ions arriving at the anode cannot intercalate fast enough and deposit as metal instead. It happens at low temperature, at high charge rate, and near full charge. Some of the plated lithium is recoverable, much is not, and the dendritic morphology it can take is a safety problem as well as a capacity problem. Waldmann and colleagues found a crossover around 25 degrees Celsius, above which film growth dominates and below which plating dominates. That crossover is why "keep it cool" is not unlimited advice.
Cathode-side degradation covers transition metal dissolution, particle cracking from repeated lattice expansion, and electrolyte oxidation at high potential. It scales with upper cut-off voltage and with nickel content.
Resistance rise usually tracks these mechanisms rather than appearing independently. It matters because a cell can pass a capacity test and still fail a power test.
How much is each lever worth?
Temperature is worth the most, followed by dwell time at high state of charge, then charge current at low temperature, then depth of discharge, with sustained discharge rate last for most applications. Here is each lever ranked by the effect size you can expect in a real system, rather than in a laboratory.
Temperature comes first. Side reaction rates follow Arrhenius behaviour, which in practice means roughly a doubling for every 10 K. A pack that averages 35 degrees Celsius over its life will not reach the calendar life of an identical pack averaging 25 degrees. This is why active thermal management is the single most consequential design decision in an EV pack, and why the location of a home battery matters more than its brand.
Dwell time at high state of charge comes second. Keil and colleagues showed calendar fade increasing with storage state of charge in steps that follow the graphite anode's voltage plateaus, with fade markedly higher above roughly 60 per cent state of charge and highest near full charge. The practical consequence for a consumer is charge to 80 per cent daily. The practical consequence for a grid operator is that a battery held at 95 per cent all summer waiting for a scarcity event is paying for that readiness in capacity.
Charge current at low temperature comes third, and it is the one that produces sudden rather than gradual loss. Below roughly 0 degrees Celsius, graphite kinetics slow enough that plating begins at quite modest rates. Every modern EV preheats the pack before DC fast charging for this reason. A cell charged at 1C at minus 10 degrees can lose more capacity in a handful of cycles than in a year of temperate use.
Depth of discharge comes fourth, and it is widely misread. Cycle life rises non-linearly as depth of discharge falls, so a cell rated for 3,000 full cycles may deliver considerably more than 6,000 half cycles. Comparing systems by cycle count without stating the depth of discharge is meaningless. Total energy throughput is the honest metric.
Sustained discharge rate comes last for most applications, and its effect is largely indirect. Ohmic heating scales with the square of current, so a high continuous rate is mostly a temperature problem wearing a different hat.
What this means by application
| Application | Dominant ageing path | The lever that pays | What to ignore |
|---|---|---|---|
| Phone, laptop | Calendar, high SoC dwell | Charge limit at 80 to 90 per cent; avoid hot cars and hot charging | Cycle counting; battery calibration rituals |
| EV, thermally managed | Mixed, plating risk on DC charging | Preconditioning before fast charge; avoid parking at 100 per cent for days | Occasional fast charging on a healthy pack |
| Grid BESS | Cycle throughput plus SoC dwell | Cell temperature setpoint; SoC management between market events | Nameplate cycle numbers without a depth of discharge reference |
| Power tool, drone | High rate, thermal | Rest before recharge; avoid recharging a hot pack | Depth of discharge limits that make the tool useless |
| Stationary backup, rarely cycled | Almost pure calendar | Float state of charge well below 100 per cent | Anything cycle related |
Does chemistry change the answer?
Chemistry changes the weighting of each lever, not the mechanisms themselves.
LFP is more tolerant of dwell at high state of charge and has a higher thermal decomposition onset, which is part of why it took over stationary storage. Its flat open-circuit voltage curve creates a different problem: state of charge cannot be inferred accurately from voltage, so the BMS relies on coulomb counting and drifts. LFP systems therefore benefit from a periodic full charge to give the BMS a known reference point, which is close to the opposite of the advice for a nickel-rich cell.
Nickel-rich NMC and NCA trade energy density for sensitivity. Upper cut-off voltage matters more, high state of charge dwell matters more, and thermal margin is smaller.
Silicon-blend anodes expand by roughly 300 per cent at the particle level on full lithiation, against about 10 per cent for graphite, which makes mechanical fatigue and continuous film reformation more significant. Cells with meaningful silicon content usually show faster early-life fade followed by a flatter region.
Reading a datasheet honestly
A cycle life figure without its test conditions tells you nothing. Look for the depth of discharge, the charge and discharge rates, the ambient temperature, the end-of-life criterion, and whether there is a rest period between charge and discharge. A cell quoted at 6,000 cycles at 80 per cent depth of discharge, 0.5C, 25 degrees Celsius, to 80 per cent capacity retention is a different product from one quoted at 6,000 cycles at 70 per cent depth of discharge, 0.3C, 23 degrees, to 70 per cent retention, even if the headline numbers look similar.
The same applies to calendar life. A ten-year figure at 25 degrees and 50 per cent state of charge does not survive a rooftop installation in Seville.
The one habit worth building
Track resistance, not only capacity. Capacity fade is what warranties are written against, but resistance rise is what users notice first as lost power, longer charge times and more heat. In diagnostic work, incremental capacity analysis and differential voltage analysis let you separate loss of lithium inventory from loss of active material, which turns "the battery is degrading" into an actionable statement about which mechanism is running.
Informational and educational content only. Not professional, financial, legal, or engineering advice.