Na-ion

Overview

Status as of September 6, 2026. This page is organized around NFPP — the polyanion cathode route that currently carries the volume — with layered oxide (NFM) and Prussian blue covered for contrast.

Sodium-ion is lithium-ion's structural twin: the same rocking-chair cell, with Na⁺ shuttling instead of Li⁺. Two component swaps define the chemistry. First, the anode must be hard carbon — sodium ions cannot usefully intercalate into graphite, so every sodium-ion cell pays the anode premium that lithium cells never do. Second, the anode-side current collector switches from copper to aluminum, because sodium does not alloy with aluminum — a structural BOM saving that requires no commodity forecast.

The chemistry exists for one reason: lithium is scarce where batteries are made. Sodium is everywhere, the cathode can be built from iron, manganese and phosphate, and the cell keeps usable performance at −40°C with thermal-runaway margins that exceed most lithium systems. The fact card above states the family's energy-density band (~100–160 Wh/kg) — the spread is the point, because "sodium-ion" is three different cathode routes chasing three different targets, not one material.

The honest scale picture: global sodium-ion cathode shipments were roughly 20,000 tonnes in 2025, which at NFPP's 3,543 t/GWh converts to about 5.6 GWh of cells — a rounding error next to a lithium industry shipping more than 2,200 GWh a year. The direction is what matters. Per SMM, China's sodium-ion cathode output rose 71% year-on-year in August 2026, the market tipped into undersupply with orders being deferred, and polyanion-route materials made up 87% of that output. CATL has put its Naxtra platform into mass production; the Datang Hubei 100 MW/200 MWh sodium ESS is grid-connected. This page therefore treats NFPP as the reference route and reads NFM and Prussian blue against it.

My take: Sodium-ion's story in 2026 is not "cheaper than LFP everywhere" — it is a chemistry with a structural cost floor, currently buying its way down that floor with scale. The numbers on this page are the state of that purchase.

Chemistry

Sodium-ion cathodes split into three families, and the choice of family is really a choice of which problem to inherit:

NFPP: sodium iron (pyro)phosphate, the reference route

NFPP is Na₄Fe₃(PO₄)₂P₂O₇ — a three-dimensional framework of FeO₆ octahedra bridged by PO₄ tetrahedra and P₂O₇ groups, with open Na⁺ channels threaded through it. The redox chemistry is Fe²⁺/Fe³⁺ only: no nickel, no cobalt, no vanadium. Discharge shows two flat plateaus at roughly 3.4 V and 3.1 V versus Na⁺/Na, which lands the full-cell average against hard carbon at about 3.0 V — the convention used across this site's conversion models.

The capacity ladder, from theory to shipment: theoretical capacity is about 129 mAh/g; the site's engineering models anchor practical capacity at 98 mAh/g (→ 3,543 t/GWh with the standard 96% yield correction, or 2.8 GWh per 10,000 tonnes of cathode); the leading commercial material (Huzhou Yingna's third generation) reaches 110 mAh/g at 2.3 g/cm³ compaction; vanadium/nickel doping roadmaps point toward 130 mAh/g with compaction above 2.5 g/cm³. In other words, NFPP today ships at roughly three-quarters of its theoretical ceiling — a wider gap than LFP's, and the main headroom left in the chemistry.

Everything else about NFPP rhymes with LFP, deliberately. The phosphate framework is an electronic insulator, so the powder needs nano-sizing and carbon coating — the same fixes LFP industrialized a decade ago, on largely the same equipment. The same rigidity delivers the safety case: thermal runaway above 300°C and nail-penetration pass reports, against a layered-oxide class that stays harder to certify. And unlike layered oxides, whose surfaces react with ambient moisture and CO₂ to form Na₂CO₃ that degrades slurry and electrode quality, NFPP powder is air-stable — the cathode stage does not mandate a full dry room.

The polyanion family has two notable siblings. NFS (Na₂Fe₂(SO₄)₃) trades capacity for voltage — 95 mAh/g at a 3.5 V full-cell average, now in first-year mass production at Zhongna Energy's Meishan base. NVPF (Na₃V₂(PO₄)₂F₃) offers the family's highest voltage but drags vanadium cost into a chemistry whose whole purpose is cheap — it stays a niche. Neither competes with NFPP's position; both confirm it: the polyanion route wins by being boring.

Route trade-offs at a glance

My take: NFPP is the LFP of sodium — same bargain, same playbook. It trades headline energy density for cycle life, temperature window, safety and a cathode cost curve that scales down with iron-phosphate chemistry. Layered oxide will own the sodium-EV headline numbers; NFPP will own the installed base.

Energy Density

The numbers below follow the definitions fixed in What 10,000 Tonnes of Cathode Is Actually Worth in GWh: practical half-cell capacity × full-cell average discharge voltage (against hard carbon for sodium systems), with a 96% manufacturing-yield correction for the t/GWh column. Material-level Wh/kg is not cell-level Wh/kg — pack integration, electrolyte and hard carbon all take their share — but the material-level ranking survives into the cell.

Prussian blue sits outside this table because its cell results span an enormous range depending on the trade made. CATL's first-generation sodium cell (2021) used Prussian white and reached 160 Wh/kg — at the time, the family's proof that sodium could approach LFP gravimetrically. Natron Energy's power-optimized Prussian blue cells sat near 50–60 Wh/kg with extreme cycle life for data-center UPS — a legitimate product, until funding ran out in September 2025. The PBA route's lattice-water and tap-density problems are exactly what push designs toward either the high-energy or the high-power corner, and neither corner currently has a commercial champion.

What NFPP gives back below its headline number: cycle life claimed at 6,000–8,000 cycles for commercial polyanion cells versus roughly 3,000 for layered oxide; 90% capacity retention at −40°C versus a reported ~30% decay for NFM at merely −20°C; and a flat discharge plateau that keeps power electronics happy across the SOC window. A 122 Wh/kg NFPP pack (72 V 20 Ah, Yingna/Chilwee) is already in two-wheeler service — the pack-level number matters more than the cell number for exactly these vehicles.

My take: For storage duty, energy density is the wrong single ranking metric. Rank by lifetime throughput: a 105 Wh/kg cell that cycles 8,000 times delivers more kWh over its life per kilogram of cathode than a 160 Wh/kg cell that cycles 3,000 times. NFPP loses the datasheet line and wins the TCO line — which is why the ESS segment is where its volume actually is.

Cost

Site cost models are CNY-denominated; USD figures convert at a simplified fixed CNY/USD 6.8, consistent with the research articles. The crossover logic below is developed fully in The Sodium Crossover Is a Scale Problem.

At the cathode-material level, NFPP's model band of $3,676–4,044/t (¥25–27.5k/t) undercuts layered-oxide NFM at roughly ¥45–55k/t by around half — the direct dividend of an iron-phosphate supply chain with no nickel. The live NFPP CAM model series prints ¥25,523/t as of September 6, 2026, sitting inside the crossover model's 1–25GWh cathode band. Prussian white's raw materials are cheaper still, but lattice-water removal, purification yield and low tap density eat the advantage in rework — and after Natron's shutdown there is no volume PBA producer left to prove otherwise.

The full-cell BOM is where scale does its work. The modeled NFPP route:

Three readings of that table. First, the aluminum-for-copper switch is worth a fixed $6.8/kWh — the one saving already banked at any scale. Second, hard carbon and finished electrolyte carry more than half of the 25GWh BOM; they are the scale tests. On current evidence the anode test is passing early: the live hard carbon series prints ¥35,650–40,150/t (power and ESS grades, September 4, 2026) — inside the model's 50GWh band while cathode volume still sits near the 1–10GWh rows. Third, the gap between model and reality is the execution burden: actual NFPP cells were quoted at ¥0.5–0.6/Wh in mid-2025 against the model's ¥0.317/Wh at 25GWh.

Against LFP, the comparison moves with lithium. At the August 20, 2026 anchor ($22,206/t), the LFP line costs $53.7/kWh against the 25GWh NFPP case's $46.7 — a $7.1/kWh gap. Lithium's slide since then (the carbonate series printed ¥140,500/t on September 4) narrows that gap to roughly $6/kWh; the no-tax crossover sits at $11,029/t. China's consumption tax tilts the field the other way: the 2% rate effective September 1, 2026 lowers the 25GWh crossover to $8,971/t, the 4% rate from September 2027 to $7,941/t, and sodium-ion remains exempt through end-2028. Meanwhile the spot market is doing its part — SMM reports NFPP cathode supply tight through early Q4 2026 with orders deferred, iron phosphate above ¥15,000/t squeezing producer margins, and NFPP quotes holding stable despite it.

My take: The fact card's "Very Low" cost is a destination, not a spot price. NFPP's cost advantage is real but conditional — it clears the LFP benchmark only at 25GWh-plus of qualified scale, with hard carbon and electrolyte delivering on schedule. At today's 1–10GWh reality it is still a premium chemistry buying its way down. The supply chain running at a loss on hard carbon right now is, bluntly, that purchase in progress.

Manufacturing

Cathode: LFP's factory, re-run. NFPP is made by solid-state synthesis — iron phosphate, a sodium source and a carbon coating precursor, milled, spray-dried and sintered, then carbon-coated. The equipment chain is close enough to LFP production that the fastest capacity adds have come from toll processing: retrofitting existing lines rather than building new ones (SMM, August 2026). That reuse works in both directions — NFPP inherits LFP's process maturity and its input exposure, with iron phosphate above ¥15,000/t and tight. Quality control centers on impurity phases (spray-plus-freeze-drying routes hold them below 3%), first-cycle efficiency and the carbon layer; a group standard released in Q3 2025 formalized compaction and impurity specs for the first time.

The dry-room question. Layered-oxide NFM powder reacts with ambient moisture and CO₂ — its surface grows Na₂CO₃ that degrades slurry stability and electrode quality — so the cathode stage inherits ternary-style moisture control: dry rooms, humidity-logistics, expensive rework. NFPP is air-stable through the cathode stage. That is a real capex and OPEX delta per GWh, and one reason polyanion capacity is ramping faster than layered-oxide capacity at equal capital.

Anode: the actual bottleneck. Hard carbon has no lithium-era incumbent to copy. Biomass routes (the early volume source, ~$9,559/t in the model) are being displaced toward pitch- and coal-based routes ($5,882–7,353/t at scale) for cost and consistency. The market structure per SMM: tier-one producers ship several hundred tonnes a month, second tier ~100 t/month, and the industry is broadly "losing money on every tonne" as downstream customers push price — two-wheeler small-power grade is the volume driver, with high-rate and low-temperature grades holding the price ceiling. Supply stays adequate only while demand is small; the model's 25GWh case needs the pitch/coal routes to land at ¥50k/t-class.

Cell assembly is conventional prismatic/wound practice on LFP-like lines, with formation and grading as the final quality gate. Design innovation is aimed at power duty: full-tab architectures (BYD has announced full-tab NFPP cells targeting up to 45C discharge for start-stop and similar roles) attack internal resistance where NFPP's flat plateaus are already an advantage.

My take: The manufacturing story is the scale ladder in motion — announced lines converting into qualified output, toll processing bridging the gap, and a cathode that sells out while the anode subsidizes demand. The binding constraint into early Q4 2026 is not enthusiasm or even capital; it is qualified tonnes.

Major Producers

The producer map as of September 2026, by chain position. Polyanion routes took 87% of sodium-ion cathode output in August 2026 (SMM), up from ~70% in 2025 — NFPP is the bulk of that share.

The market state behind the table: mainstream NFPP producers ran at full capacity through August 2026 with monthly orders above the thousand-tonne level at the leaders, deliveries covering less than 90% of demand, and new-client scheduling stretched — SMM expects tight supply to persist into early Q4 and September cathode production to rise ~160% year-on-year. Capacity is being added on two tracks (self-built bases and toll processing), but trial-production ramps mean no rapid relief. Hard carbon producers, in contrast, ship more every month at falling prices and negative margins.

My take: Watch two things from here. Whether polyanion's 87% output share holds as CATL's Naxtra ramps the layered-oxide EV channel — the two routes are still complementary, but they will compete for the same hard carbon and electrolyte scale. And whether today's cathode undersupply survives contact with the Q4 capacity wave; a seller's market for qualified NFPP is the best possible negotiating position for the material suppliers building this chemistry's cost curve.