Cycle life is how many charge-discharge cycles a battery goes through before it degrades to some predefined capacity level. Usually that's 80%, though I've seen specs all over the place-70% for some automotive applications, 85% for others. The threshold itself is kind of arbitrary and has more to do with warranty calculations than any fundamental property of battery chemistry.
Why batteries degrade
The SEI layer on graphite anodes is probably the most studied degradation mechanism in lithium-ion batteries. When you first charge the cell, the electrolyte decomposes at the anode surface and forms this solid film-mostly lithium carbonate, some lithium fluoride if you're using LiPF6 salt, various organic species. This layer should stabilize after a few cycles. It doesn't. It keeps growing, slowly, eating up lithium that should be cycling. I worked on a project a few years back trying to stabilize SEI with electrolyte additives. FEC (fluoroethylene carbonate) helps, VC (vinylene carbonate) helps more. Neither one solves it completely.
High-nickel cathodes crack. NMC811, which is 80% nickel, 10% manganese, 10% cobalt, undergoes massive volume changes during cycling-way more than the older NMC111 or NMC523 compositions. The particles literally fracture apart. You can see it under SEM after a few hundred cycles. CATL figured out around 2019 that if you synthesize single-crystal particles instead of the polycrystalline aggregates everyone was making, you eliminate most of the cracking because there's no grain boundaries. Sounds obvious now but getting the synthesis parameters right at scale took years. I think BYD is doing something similar with their Blade battery but they don't publish much about their cathode processing.
Temperature matters more than anything else, honestly. For every 10°C increase you roughly double the aging rate. This isn't exact-different mechanisms have different activation energies-but it's close enough for engineering estimates. A pack that runs at 35°C average will last way longer than one at 45°C. Tesla learned this early with the Roadster. Their later cars have pretty aggressive thermal management, they'll run cooling even when parked if the pack gets too hot.
Cold is weird. The degradation chemistry slows down, which is good. But you get much higher resistance and if you try to charge too fast you'll plate lithium on the anode, which is very bad. The lithium deposits are irreversible capacity loss and if they grow into dendrites you can get an internal short. Most EVs won't let you DC fast charge below 0°C for this reason.
Depth of discharge
There's this well-known effect where shallow cycling extends life. If you only use 40% of the battery's capacity instead of 100%, you might triple the cycle life. Maybe more. The mechanisms aren't totally clear. Smaller volume changes in the electrodes probably help. Avoiding extreme electrode potentials where side reactions accelerate definitely helps. But nobody's really nailed down the exact contribution of each factor.
Every EV hides some capacity from you. When your dashboard says 0% you're probably at 5% real SOC. When it says 100% you're at 95% or maybe 90%. The manufacturers don't publish these numbers. I tried to reverse-engineer it on my old Bolt by looking at the BMS data over CAN bus but the calibration tables are encrypted.
Testing timeline issues
Nobody wants to wait 3-4 years to validate a battery design. So we do accelerated testing-higher temperatures, faster cycling rates. The problem is that not all degradation mechanisms accelerate the same way. Some do, some don't. Temperature acceleration works pretty well for most chemical processes. Rate acceleration is sketchy. And there are failure modes that only show up after extended calendar time that you just can't accelerate at all.
The recent ML prediction stuff is interesting. You feed early cycle data into a model and it forecasts long-term capacity fade. Stanford published some work on this in 2019, Carnegie Mellon has done similar things. It works surprisingly well on the cells they trained on. Generalization is the problem. Change the cathode composition or electrolyte formulation and you need to retrain with new data, which kind of defeats the purpose if you're trying to predict lifetime of a new design.
Different chemistries
LFP has better cycle life than NMC, period. You can get 3000-5000 cycles easily, sometimes more. The energy density penalty is brutal though-you're giving up 30-40% compared to high-nickel NMC. Chinese manufacturers decided this tradeoff makes sense for cheaper EVs where people don't need 300+ mile range. BYD, CATL, everyone there is doing LFP for standard range vehicles. Western OEMs are slower to adopt it. Cultural difference or market difference or both, I'm not sure.
Sodium-ion is supposed to have similar cycle life to LFP based on what CATL and Natron claim, but it's only been in production for like a year so who knows what happens after 5+ years in the field.
Solid-state keeps promising better cycle life because no liquid electrolyte means more stable interfaces. But they keep not being available commercially. QuantumScape has been "two years away" for about six years now. Interface resistance, contact loss during cycling, lithium dendrites in some designs-these problems are harder than anyone thought. Maybe sulfide electrolytes will work better than oxides. The data coming out of Japan suggests maybe.
Calendar aging
This gets ignored a lot but it's huge for vehicles that don't get driven much. Your battery degrades just sitting there. Storage temperature and SOC both matter. Worst case is 100% charge at high temperature-you'll lose multiple percentage points per year doing nothing. 40% SOC at room temperature is optimal storage but obviously not practical for a car you actually use.
Grid storage is a completely different animal. You need 20+ year lifetime which means 7000+ cycles minimum, probably 10,000+ if you're doing multiple cycles daily. But you can tolerate lower energy density and you have more space for thermal management. The economics work differently-CAPEX matters more relative to OPEX when you're amortizing over decades.
I don't know enough about the new battery types to say much useful. Lithium-sulfur has terrible cycle life still because of the polysulfide shuttle, though some companies claim they've solved it. Lithium-metal anodes paired with solid electrolytes might be the path forward but the dendrite problem hasn't gone away. And even if the materials work, scaling up manufacturing of anything new takes 5-10 years minimum. So whatever comes next, we're stuck with lithium-ion variations for at least another decade.
I don't know enough about the new battery types to say much useful. Lithium-sulfur has terrible cycle life still because of the polysulfide shuttle, though some companies claim they've solved it. Lithium-metal anodes paired with solid electrolytes might be the path forward but the dendrite problem hasn't gone away. And even if the materials work, scaling up manufacturing of anything new takes 5-10 years minimum. So whatever comes next, we're stuck with lithium-ion variations for at least another decade.
Lithium polymer batteries sit somewhere in between-better packaging flexibility than cylindrical cells, cycle life comparable to standard lithium-ion if you keep temperatures reasonable. Decent option for constrained spaces but nothing revolutionary.
