How to Choose a Lithium Battery Chemistry: Duty Cycle First

Start with the constraint, not the chemistry

Every battery program has one constraint that binds before the others: the pack is too heavy, or it will not survive the cycle count, or it cannot recharge in the window the duty cycle allows, or it cannot charge where it sits in January, or it costs too much per delivered kilowatt-hour. Chemistry selection is the exercise of naming that constraint honestly and then reading the datasheet rows that speak to it.

This is the framework PBC applies across its own five production chemistries: LFP when cycle life and safety outrank weight, NMC or NCA when energy density is the binding constraint, LTO when the duty is defined by power, fast charge, or extreme temperature, and sodium-ion when material cost dominates and the application is stationary. The sections below put datasheet numbers behind each branch.

What the datasheets say, once conditions are attached

The table below is built only from documents we could verify: manufacturer-published datasheets and spec pages, quoted with their conditions. Where a manufacturer publishes nothing, the cell says so, because the absence is information.

*EVE does not print an energy density row; 165 Wh/kg is our arithmetic from the sheet's own values (3.2 V × 280 Ah = 896 Wh, at 5.42 kg). Molicel, Panasonic and Toshiba figures are the documents' own stated values. The bottom row is the honest state of sodium-ion disclosure, covered below.

Two readings of that table do most of the selection work. First, the energy column: to carry 10 kWh you need about 41 kg of NMC or NCA cells, 60 kg of LFP, 94 kg of LTO, or 57 kg of sodium-ion if the claimed density holds (our arithmetic, cells only, pack overhead excluded). Second, the cycle column reads roughly inverse to the energy column, and it only means anything with conditions attached. EVE's own sheet is the cleanest illustration: the same cell, same 0.5C protocol, same 80% endpoint, is rated 6,000 cycles at 25 °C and 2,500 at 45 °C. Ambient temperature alone cut the rating by more than half. We covered the same conditionality inside one chemistry in our LTO deep dive.

The voltage column carries a quieter cost. A 48 V bus takes 15 LFP cells in series (48.0 V) but 21 LTO cells at Toshiba's 2.3 V nominal (48.3 V; other LTO makers rate 2.2–2.4 V, which shifts the count); NMC lands off-grid at 13 in series (46.8 V) or 14 (50.4 V). Every extra series element is another monitoring channel, another balancing circuit, another interconnect. At high voltage the effect compounds: PBC's own 800 V-class LTO product runs 348 cells in series at 800.4 V nominal, where an LFP string reaches 800.0 V with 250, a 39% difference in series elements (our arithmetic).

Temperature picks before you do

Cold applications are decided by one datasheet row: the charge temperature floor. Graphite-anode lithium-ion cells cannot accept charge below it because, as Battery University's temperature reference states, "plating of metallic lithium occurs on the anode during a sub-freezing charge that leads to a permanent degradation in performance and safety." The floors on current documents: 0 °C for EVE's LF280K (LFP), 0 °C for both Molicel NMC cells, and +10 °C for Panasonic's NCR18650B (NCA).

Discharge is the misleadingly generous row. Molicel rates discharge down to −40 °C on the same sheet that stops charge at 0 °C, and EVE discharges to −20 °C. Equipment that only discharges in the cold has options; equipment that must recharge where it sits does not, and the standard workaround is pack heating, which costs energy, mass and BMS logic. LTO removes the mechanism instead: its anode sits about 1.55 V above lithium plating potential (per the Frontiers in Materials review). Toshiba states its 10 Ah cell "can be charged and discharged at temperatures below the freezing point" while publishing no numeric floor, and Leclanché publishes a −20 to +55 °C range for the LT34 covering charge and discharge. Confirm the rated cold-charge current on the controlled datasheet for your specific cell, and note that charge-current limits can also derate between the floor and room temperature, so ask for the derate table too. CATL states an operating range of −40 to +70 °C for Naxtra and, in its February 2026 release, over 90% capacity retention at −40 °C; no charge window is published on either page, so treat cold-climate sodium designs as datasheet-pending.

When does NMC or NCA beat LFP?

When the energy column binds. Molicel states 242 Wh/kg and 643 Wh/L for the P45B, and Panasonic's NCR18650B datasheet carries 243 Wh/kg and 676 Wh/L; the LF280K-class storage cells common in industrial packs derive to about 165 Wh/kg. (Newer EV-grade LFP advertises higher figures; we compare against the cell whose full specification we can verify.) Where every kilogram displaces payload (mobile robots on a mass budget, aerial systems, handheld and wearable equipment), that 46–47% density advantage is usually decisive, and it is the reason nickel chemistries own those categories.

LFP answers with three rows of its own. Cycle life: EVE's ≥6,000-cycle rating against, on the nickel side, sheets that publish no numeric rating at all (Molicel's cycle chart simply ends at 500 cycles). That absence is a property of the public power-cell documents, not necessarily of NMC as a class; energy-optimized NMC in larger formats carries conditioned cycle ratings on controlled datasheets that are rarely public, so make the conditioned number a required deliverable in any RFQ. Price: BloombergNEF's 2025 survey puts average LFP pack prices at $81/kWh versus $128/kWh for NMC, with the all-chemistry average at $108/kWh. Charge window: EVE charges to 55 °C and Molicel's P45B to 60 °C, but Panasonic's NCA floor starts at +10 °C, the narrowest window in the table.

For industrial equipment that returns to a charger daily for years, the cycle and price rows dominate the mass row, which is why LFP is the default answer to "LFP or NMC for industrial equipment" unless the application is mass-limited. State the exception honestly and the rule survives it.

Which chemistry for forklifts and AGVs?

The charge regime decides. A lead-acid conversion that charges overnight or between shifts lands on LFP in most PBC programs: the voltage windows are compatible with 12/24/48 V legacy systems, the cycle rating covers years of daily use, and LFP carries the lowest chemistry-average pack price in BloombergNEF's 2025 release. That is the standard lead-acid replacement case.

Opportunity charging changes the arithmetic. Take an AGV that consumes 2 kWh per operating hour and sees one 10-minute dwell window per hour: returning 2 kWh in a sixth of an hour requires 12 kW of charge acceptance. On LFP rated 1C continuous (EVE's row), accepting 12 kW takes roughly a 12 kWh pack, six times the energy the vehicle actually consumes between windows. A 6C-capable LTO pack (Leclanché rates the LT34's maximum charge at 204 A, 6C, distinct from its 20-second 10C pulse rating) covers the same 12 kW at about 2 kWh. This is our own illustrative arithmetic, ignoring charge taper, efficiency and margin, but the six-fold sizing gap is the structural point: in opportunity-charged duty you size LFP on power and buy energy you do not need, or you pay LTO's premium per kWh on a much smaller pack. Cold stores push the same direction, since the 0 °C LFP and NMC charge floors in Fig. 3 sit at or above cold-room temperatures. The full trade, including LTO's energy and cost penalties, is in our LTO article; PBC builds LTO packs specifically for this duty class. (Fast-charge LFP variants exist as well; if a supplier offers one, get the continuous charge rating at your operating temperature in writing, not the peak.)

Where sodium-ion actually fits in 2026

Sodium-ion is real, shipping, and thinly documented. CATL states its Naxtra line reaches 175 Wh/kg, "comparable to LFP batteries," with "over 10,000 cycles," and a production passenger vehicle announced for mid-2026:

"CATL's Naxtra passenger EV Battery achieves an energy density of 175Wh/kg, the highest among sodium-ion batteries worldwide, and comparable to LFP batteries. It offers a 500-kilometer range and can achieve over 10,000 cycles, which significantly reduces maintenance costs."CATL press release, 21 April 2025

What CATL's public pages do not state matters as much: no cell voltage, no capacity or format, no indication whether 175 Wh/kg is cell-level or pack-level, and no conditions (rate, temperature, depth of discharge, endpoint) behind the 10,000-cycle figure. HiNa Battery announced in October 2025 that four sodium-ion product lines had "entered the mass-production and sales phase," including a 240 Ah storage cell with two years of production validation, and likewise publishes no conditioned ratings on its English pages. On price, BloombergNEF's published 2025 price-survey release contains no sodium-ion figure at all; the nearest institutional estimate, from Argonne National Laboratory in January 2024, put prospective sodium-ion cost at "one-third less than a lithium-ion one," and it predates the 2025 price declines BloombergNEF recorded, which moved the target.

The selection posture that follows: sodium-ion earns a quote today where the duty is stationary, cost-driven, or cold (among CATL's first Naxtra products is a 24 V heavy-truck battery positioned against lead acid, with claimed −40 °C starting), and where the program can absorb a young supply base. Demand conditioned data in the RFQ: rated cycle life with endpoint and temperature, the charge window, and the voltage curve your power electronics will actually see. If the numbers arrive and hold, the material-cost logic is sound; until they arrive, LFP remains the benchmark it is being sold against. The full sodium picture, the three cathode families, vendor by vendor, and what each actually replaces, is in our sodium-ion deep dive.

A chemistry change later is a new certification program

Chemistry choice is sticky in a way the spec-sheet comparison does not show. The UN Manual of Tests and Criteria defines a tested type as "a particular electrochemical system and physical design of cells or batteries," and the note to its §38.3.2.2 lists "a change in the material of the anode, the cathode, the separator or the electrolyte" first among changes that may make a cell or battery a new type subject to the full test sequence. US regulation follows the same document: 49 CFR 173.185 requires each lithium cell or battery to be "of the type proven to meet the criteria in part III, sub-section 38.3 of the UN Manual of Tests and Criteria."

In practice: swapping LFP for sodium-ion, or NMC for LTO, should be planned as a new type with a fresh transport test campaign, and it usually reopens the safety-standard work (UL, IEC) built on the same design. The destroyed-sample economics are covered in our UN 38.3 cost breakdown, and the schedule coupling in our program timeline article. For sodium-ion the rules are in motion: PHMSA's February 2026 proposed rule would adopt the UN's sodium-ion battery entries (UN 3551 and 3552) into the US hazmat regulations, stating that the UN sub-committee has adopted provisions to regulate sodium-ion batteries in the same manner as lithium-ion, while the operative text of 49 CFR 173.185 remains lithium-only as of this writing. Settle transport classification with your lab and carrier at program start, not at first shipment. Chemistry decisions deserve the certification path drawn next to them before anything is ordered.

The decision rule

Name the binding constraint, then apply the table. Mass or volume: NMC or NCA, checking the charge floor on the specific sheet (Panasonic's NCA floor is +10 °C). Daily cycling for years at sane temperatures: LFP, and derate the cycle rating if the environment runs hot. Charge windows measured in minutes, charging below 0 °C, or cycle counts past 15,000: LTO, sized on power, priced per delivered cycle. Stationary and cost-driven: run sodium-ion against LFP and make the supplier condition every claim. Then, whatever the shortlist says, check the charge window against the coldest charging location and draw the certification path before committing. That last check is cheap on day one and expensive every day after.

PBC engineers custom packs across all five of these chemistries, so the duty-cycle numbers, not chemistry loyalty, pick the cell.

Frequently asked questions