The Chemistries: LFP, NMC, NCA, LTO and Sodium-ion

7.1 — The comparison table

Figures below are representative ranges from published product specifications, not values from any single cell. Use them to narrow a choice, then work from the datasheet of the specific cell you intend to buy.

Two older chemistries are worth naming even though they rarely appear alone now. LMO (lithium manganese oxide) has a spinel structure with excellent power and safety but poor life at high temperature, and is now mostly blended with NMC. LCO (lithium cobalt oxide) was the original commercial lithium chemistry — high energy, poor safety, expensive — and now appears essentially only in consumer electronics.

7.2 — What the letters actually stand for

• •LFP — lithium iron phosphate, LiFePO₄. An olivine structure whose strong phosphorus–oxygen bonds resist releasing oxygen. That is the entire origin of its safety advantage, and it is structural rather than a manufacturing quality anyone can add elsewhere.
• •NMC — nickel manganese cobalt, written with ratios: NMC111, NMC532, NMC622, NMC811. The digits are the Ni:Mn:Co proportions.
• •NCA — nickel cobalt aluminium. The highest practical energy density, and the tightest process control required to make it safely.
• •LTO — lithium titanate. This is an anode material replacing graphite, not a cathode, even though it is always listed alongside the cathode chemistries. It gives extraordinary life and fast charging at the cost of a much lower cell voltage, which is why it is expensive per kWh.
• •Na-ion — sodium-ion. No lithium and no cobalt at all.
• •SLA / VRLA — sealed lead acid, also called valve-regulated; includes AGM and gel variants. Included in the table because it is what most Indian e-rickshaw packs are being replaced from, and the comparison is worked through in our three-year cost post.

7.3 — The nickel trade-off

Within the NMC family the ratios are not marketing variants; they are a single dial with three consequences, all moving together.

This is why the industry did not simply converge on the highest-nickel option available. Each step up the nickel ladder moves engineering effort and cost from the cell into the pack, and for applications where mass is not the binding constraint that trade is a bad one.

7.4 — LFP’s flat curve and what it costs you

LFP has an unusually flat discharge plateau. From roughly 20 to 90 per cent state of charge the cell sits between about 3.20 V and 3.30 V — a window of around 100 millivolts covering 70 per cent of the capacity.

For the cell this is a virtue: the vehicle sees near-constant voltage almost all the way down, so performance does not fade as the pack empties the way a lead-acid pack’s does.

For the fuel gauge it is a serious problem. You cannot estimate LFP state of charge from voltage with any useful accuracy, because a measurement error of 20 mV — well within what sensing tolerances and temperature drift produce — moves your answer by tens of percentage points.

• •Consequence one: LFP packs need coulomb counting — integrating current over time — rather than voltage lookup. The mechanics of that are in Chapter 1.
• •Consequence two: coulomb counting drifts, so it must be recalibrated at the ends of the curve where voltage does move — a full charge, and occasionally a deep discharge. A fleet that only ever charges to 80 per cent never gives the algorithm a reference point.
• •Consequence three: a poorly implemented LFP gauge sits at “60 per cent” for an hour and then collapses. That is not a faulty pack, and diagnosing it as one wastes a service visit.

NMC, by contrast, has a sloped curve and is far more forgiving of voltage-based estimation. If you are moving a product from NMC to LFP, the fuel gauge is the part that will surprise you. The state-metric side of this is developed in Chapter 5.

7.5 — Sodium-ion, and where it might land

Sodium-ion uses the same rocking-chair principle with sodium in place of lithium. Sodium is abundant and cheap, there is no cobalt and no lithium in the cell, and it performs unusually well in the cold — which is a real weakness of lithium chemistries.

The costs are lower energy density and a technology that is still maturing, with far less field data behind it than LFP has. Treat published cycle-life figures with more caution than you would for an established chemistry, because there are fewer cells that have actually lived that long.

7.6 — Choosing a chemistry, in order

Ask these five questions in this sequence. Most disagreements about chemistry turn out to be people answering them in different orders.

1. 1What is the duty cycle? Deep daily cycling — an e-rickshaw doing a full discharge every day, a storage system cycling on a tariff — points hard at LFP, because cycle life is the dominant cost term. Occasional deep cycling with a weight premium points at NMC.
2. 2What is the ambient temperature? Indian summers at 45°C ambient, with pack surfaces well above that, strongly favour LFP — both for the runaway margin and because high-temperature cycle life is where LFP’s advantage widens.
3. 3Is weight or volume the binding constraint? If genuinely yes, NMC or NCA. On a three-wheeler, in a stationary cabinet, or under a rickshaw floor, it usually is not — and paying a nickel premium for mass you were not short of is the most common specification error in this market.
4. 4What is the safety and liability exposure? Passenger vehicles, shared charging areas, unattended overnight charging in residential buildings — all push towards LFP, and increasingly towards what the certification regime will accept without argument.
5. 5What does the warranty model assume? A five-year warranty written against a chemistry good for 1,500 cycles, on a vehicle that cycles daily, is a loss-making promise made at signing. Cycle life at the actual DOD and temperature must exceed the warranty term’s expected cycles with margin, or the commercial model is wrong regardless of how good the cell is.