Sodium-Ion Battery Charging: Charge Limits and SOC Window
Why Sodium Chemistry Needs Its Own Charge Limits
Twelve years of my career went on lithium iron phosphate packs for forklifts and stationary racks. In the last four most of my bench time has gone to sodium-ion, and the question I still get most often from integrators is simple: if the pack is 48 V, can I reuse the charger and BMS software already qualified for LFP? The honest answer is no, and the reason is two numbers: nominal cell voltage and the shape of the open circuit curve. A sodium-ion cell does not behave like a small LFP cell at the top of charge, so copying the LFP cut-off voltage leaves part of the pack idle or wears the cells out over a few hundred cycles. What follows is the charging and state of charge guidance we hand to every customer buying a sodium battery solution, written the way we actually run it on the bench.
What the Cut-Off Voltage Actually Controls
Below the cut-off voltage the pack runs in constant current, where each amp you add moves sodium ions and lifts the voltage predictably. Above it the cell holds constant voltage while the current falls away as sodium sits deeper in the host lattice. The charge ends when the current drops below the taper limit, a C-rate such as 0.05C, meaning one twentieth of rated capacity in amperes. On a 50 Ah cell that is 2.5 A.
This matters more on sodium than on LFP because the safe band at the top is narrower. LFP is forgiving: its voltage barely moves between 40 and 90 percent state of charge, so a small calibration error costs very little capacity. Layered oxide sodium climbs steadily through that same region. Stop late and you gain little energy while holding the cell at peak structural stress; stop early and you leave real capacity behind in a pack that should agree with its neighbours.
Nominal Voltage and OCV Shape by Sodium Chemistry
Sodium has two commercial cathode routes, and they need different charge settings. Layered oxide, typically a sodium nickel manganese oxide, sits at a nominal 3.00 V per cell with a working range from about 2.0 V up to a 3.90 to 4.00 V charge limit. Its open circuit voltage rises almost linearly from roughly 2.1 V at empty to 3.8 V at full, which is good news for coulomb counting, because the voltage itself tracks state of charge.
Polyanionic chemistry, the phosphate and fluorophosphate family such as NaVPO4F and the Na3V2(PO4)3 group, runs at a nominal 3.20 V with a charge limit near 3.65 to 3.75 V and a floor of 1.5 to 2.0 V. Its curve is flat across the middle, exactly like LFP. We build both in the same factory, and the BMS firmware for one cannot be flashed onto the other without rebuilding the voltage lookup table.
- Layered oxide: nominal 3.00 V, charge limit 3.90 to 4.00 V, flatness of curve poor, state of charge accuracy good.
- Polyanionic: nominal 3.20 V, charge limit 3.65 to 3.75 V, flatness of curve good, state of charge accuracy poor between 20 and 80 percent.
- Both: internal resistance around 1.0 to 2.5 milliohms on a 50 Ah prismatic cell, roughly double an LFP cell of the same size, so load spikes move the measured voltage further.
- Both: the discharge floor is lower than lithium, down to 1.5 to 2.0 V, but you should not treat that as a normal operating point.
Picking the State of Charge Window
For stationary storage we almost never charge the full 0 to 100 span. The window we qualify is 10 to 90 percent, an 80 percent depth of discharge, and the reason is calendar life rather than cycle count: cells held above 90 percent state of charge grow surface film on the anode faster than cells cycling mid range, and the loss shows up as slow capacity fade.
On a 48 V rack we shipped last year, two strings were commissioned with different limits at acceptance, one at 5 to 95 percent and one at 12 to 88 percent. After roughly six hundred equivalent full cycles the wider string showed a forty millivolt spread at rest against thirty-five for the other. Neither is dramatic, but the trend was already separating, and we now fit the wide window only where a customer needs the extra runtime and accepts a shorter warranty.
The other half of the window is the floor. Deep discharge below 20 percent leaves the negative electrode nearly empty of sodium and the voltage droops hard under load, so we disable the load contactor below 15 percent and warn the controller at 20 percent. That line of code has done more for uptime than any hardware change we have made.
Taper Current, Charge Time and Temperature
Charge time on sodium runs a little longer than on LFP because the constant voltage phase lasts longer. A 50 Ah polyanionic cell going from 20 to 90 percent state of charge at 0.5C takes about fifty-five minutes of constant current, then forty to seventy minutes of taper. Cutting the taper from 0.05C to 0.02C adds fifteen minutes for perhaps one percent more capacity, so most of our customers settle on 0.05C.
Temperature is where sodium genuinely beats lithium, and it is worth being precise, because the rule changes with chemistry. Polyatomic phosphate cells with propylene carbonate based electrolyte keep enough ionic conductivity at minus twenty degrees Celsius to accept charge, but only at low current and only below about 70 percent state of charge, because the ion movement in the host lattice slows down faster than the movement in the liquid. Layered oxide is stricter near the top of charge.
- Below 0 degrees Celsius: limit charge current to 0.05 to 0.10C and stop entirely above 70 percent state of charge.
- 0 to 10 degrees Celsius: 0.15 to 0.2C, still no top balancing.
- 10 to 25 degrees Celsius: the normal 0.5C rate with no restriction.
- Above 25 degrees Celsius: hold 0.5C, and if the cell surface temperature climbs past 45 degrees Celsius the charger must back off.
Balancing a Sodium Pack
Because the internal resistance is higher, a sodium pack shows a wider voltage spread under load than an equivalent LFP pack, even when every cell is healthy. We routinely see sixty to ninety millivolts of spread at 100 amperes that collapses to under thirty millivolts at rest. The balancing algorithm has to measure at rest, not in the middle of a discharge pulse, or it will chase a ghost.
Our setting is to balance on the top balance point, once per cycle during the constant voltage phase, and only when the pack sees less than 0.03C of current. We trigger at a thirty millivolt spread and finish when every cell is inside ten millivolts of the pack average. Field data from the last two racks shows this brings the pack into agreement in about forty minutes and holds it there for the rest of the charge.
How We Verify a Sodium Pack Before Shipping
Every sodium module gets a capacity test on a cycler at 0.5C charge to the chemistry specific limit and 0.5C discharge to the floor. A healthy 50 Ah polyanionic prismatic cell returns 97 to 99 percent of its rated capacity on the first run, and a layered oxide cell about 94 to 97 percent because the last few percent of the curve needs a long taper. Anything below that goes back to the cell supplier.
The second check is a two week idle log at a stored state of charge. Sodium self discharges slowly, but a drifting cell shows up as a voltage slide below three volts. We ship the cycler report, charge limit sheet and balanced spread history with every module, because the installer needs to know what good looks like before plugging anything in.
On paper the pack ships with its charging profile described under UN38.3 for transport, IEC 62133-2 where a movable application is in scope, and IEC 62619 for stationary racks. Installed systems still need UL 9540A at the system level, and the commissioning manual carries the same charge limits we use on the bench. When a customer wants a custom battery solution around a sodium pack, the charging table is the first thing we settle, because everything downstream, from the BMS firmware to the warranty, is written against it.
What is the correct charge cut-off voltage for a sodium-ion battery?
It depends on the cathode. Charge layered oxide sodium cells to 3.90 to 4.00 V per cell and polyanionic cells to 3.65 to 3.75 V per cell, then stop when the current falls below 0.05C. Using 3.65 V on a layered oxide cell leaves real capacity unused; using 4.00 V on a polyanionic cell pushes it well past its plateau into a region it was never built for.
Can I reuse an LFP charger for a sodium-ion pack?
Only if the charger sets the limit cell by cell and limits current by temperature. An LFP charger stops at 3.65 V, which truncates a layered oxide sodium pack at perhaps 85 percent of usable capacity. A fixed profile unit cannot serve both chemistries, so you need two charging stages or a sodium specific charger.
Why is my sodium pack capacity lower than the datasheet rating?
Check where you stop the charge. Capacity is measured between the stated limits, and most customers are charging to 3.65 V on a layered oxide cell that should go to 3.90 V, or stopping the taper at 0.1C instead of 0.05C and cutting the top of the curve short. Also confirm the discharge floor, because reading capacity with a 2.5 V floor instead of the proper 1.8 to 2.0 V throws away a few percent.
At what temperature should sodium-ion charging stop?
Below freezing, keep the current at 0.05 to 0.10C and stop charging above 70 percent state of charge. Between 0 and 10 degrees Celsius allow 0.15 to 0.2C. Above 10 degrees Celsius the full 0.5C is fine, with a backup at 45 degrees Celsius cell surface temperature. The trigger is never the electrolyte freezing, it is the sodium movement inside the cathode slowing down.
How tight should the state of charge window be in a stationary rack?
We qualify 10 to 90 percent for most stationary work, and go to 5 to 95 percent only where the customer needs longer runtime and accepts a shorter warranty. Charging to 100 percent every day is the single biggest avoidable cause of capacity fade in sodium packs, and the extra runtime it buys is usually worth less than the cycles it costs.
Does sodium-ion need a different balancing strategy from lithium iron phosphate?
Yes, mainly in the measurement. Higher internal resistance widens the voltage spread under load, so balance from resting voltages, not pulse samples. Top balance during the constant voltage phase below 0.03C, trigger at a thirty millivolt spread and finish inside ten millivolts, and the pack holds together for its whole service life.
Further Reading
References
