Most lithium salt comparisons rank salts by ionic conductivity and stop there. That approach won't explain why an NMC811 pouch cell swelled at 45 °C, or why the aluminum beneath the cathode pitted after formation. This guide compares the seven salts most commonly quoted for lithium-ion electrolytes across the four variables that decide the choice in practice — aluminum passivation, hydrolytic stability, cost per mole of lithium, and purity specification — along with two protocols that can be run within a month.

Lithium Salt Comparison at a Glance

Salt	MW (g/mol)	Conductivity(1 M, EC/DMC, 25 °C)	Aluminum Passivation	Hydrolytic Stability	Typical Role
LiPF6	151.9	~10.7 mS/cm	Excellent (forms AlF3)	Poor — hydrolyzes to HF	Main salt, most commercial cells
LiFSI	187.1	~10–12 mS/cm	Poor when dilute and alone	Good	Co-salt at 0.1–0.3 M; main salt in premium cells
LiTFSI	287.1	~9.0 mS/cm	Poor above ~3.7 V	Excellent	Polymer/solid electrolytes, Li–S, Li metal
LiBF4	93.7	~4.9 mS/cm	Good	Moderate	Wide-temperature and co-salt use
LiBOB	193.8	~7.5 mS/cm (0.8 M limit)	Excellent — inhibits corrosion	Good	Film former, 0.5–2 wt%
LiDFOB	143.8	~7–8 mS/cm	Good	Good	Film former / co-salt, 0.5–2 wt%
LiPO2F2 (LiDFP)	107.9	Low solubility, not a conductor	Neutral	Good	Additive, 0.5–1 wt%

 

The conductivity figures for LiPF6, LiTFSI, LiBF4, and LiClO4 are the widely cited EC/DMC (1:1, 25 °C) values from Kang Xu's 2004 Chemical Reviews survey. LiFSI and LiDFOB values vary considerably between studies and solvent blends; confirm these figures with in-house conductivity measurements in the target solvent system before formulating around them.

LiClO4 (106.4 g/mol, ~8.4 mS/cm) is excluded deliberately — a strong oxidizer, suitable for coin-cell screening only. LiAsF6 performs well but is obsolete on arsenic-toxicity grounds.

Why Conductivity Is the Least Useful Column in a Lithium Salt Comparison

Bulk ionic conductivity rarely limits a modern cell. The spread between the best- and worst-performing salts above is roughly 2×, while the interphase resistance those same salts generate can differ by 5–10× after a few hundred cycles at 45 °C.

A 15% conductivity advantage at 25 °C is meaningless if the anion strips the aluminum current collector at 4.2 V. Conductivity should be treated as a filter, not a decision criterion: anything above roughly 6 mS/cm in a carbonate blend will not be the limiting factor. Aluminum stability, HF generation, and interphase quality — the three properties that actually limit cell performance — appear on no conductivity table.

Variable 1: Aluminum Passivation Is the Pass/Fail Gate

Aluminum current collectors survive above 3 V only because a fluoride-rich passive layer forms on their surface. LiPF6 supplies that fluoride readily; imide salts do not — the primary reason LiFSI has never simply replaced LiPF6.

In cyclic voltammetry of Al/Li half-cells, LiPF6 and LiBF4 show an anodic current that plateaus around 20–30 µA/cm² above 3.8 V vs. Li/Li⁺, then decays cycle by cycle to below 1 µA/cm² by cycle 30 as an AlF3 layer seals the surface. LiFSI behaves differently: the current does not decay in the same way, according to a 2025 study in the Journal of The Electrochemical Society.

Two nuances that generic comparisons tend to miss:

• Concentration flips the result. Above roughly 3 M, LiFSI passivates aluminum reasonably well, as fluoride impurities and reduced free solvent change the interfacial chemistry. Dilute-LiFSI data does not transfer to high-concentration systems.
• Purity often matters more than the anion. Residual chloride and fluorosulfonate in LiFSI drive corrosion significantly, so two lots both labeled "99.9%" can behave very differently. Request the chloride specification on the certificate of analysis.

A Three-Day Aluminum Screening Protocol

This test should precede any long cycling program — it eliminates more candidate formulations than any single subsequent measurement.

1. Prepare the aluminum discs. Punch from production-line foil (not lab foil), 12–15 µm, 15–16 mm diameter. Degrease in DMC and air-dry in the antechamber. Do not polish, as this removes the native oxide under study.
2. Build the cells. Assemble CR2032 Li/Al half-cells with 80 µL electrolyte and a standard trilayer separator. Rest for 6 hours at open circuit.
3. Run cyclic voltammetry. Sweep 2.5–4.5 V vs. Li/Li⁺ at 1 mV/s for 30 cycles at 25 °C (approximately 33 hours). Record the cycle-30 current density at 4.5 V: below ~1 µA/cm² indicates passivation; tens of µA/cm² indicates corrosion.
4. Run a potentiostatic hold. Hold at the upper cutoff plus 0.1 V for 10 hours at 45 °C and integrate the charge passed. Temperature is not optional — LiFSI-driven corrosion accelerates sharply above 40 °C.
5. Image the surface. Disassemble, rinse in DMC, and image the aluminum at 1,000× by SEM. Pitting typically appears before leakage current rises appreciably, so current data alone can be misleading.

When a formulation fails this screen, the fix is almost always a boron-oxalate co-salt: 0.5–1 wt% LiDFOB or LiBOB rebuilds passivation via an AlF3/B2O3 interphase. A small LiPF6 fraction achieves the same result more economically.

Variable 2: Hydrolytic Stability, or What LiPF6 Really Costs in Process Control

LiPF6's weakness is not its electrochemistry — it is its equilibrium with PF5. In solution, LiPF6 dissociates into LiF and PF5; PF5 then reacts with trace water to form POF3 and HF, generating two HF molecules per water molecule consumed. This is why battery-grade electrolyte specifications cluster around ≤20 ppm water and ≤50 ppm free acid (as HF).

That specification carries a recurring operational cost that salt price comparisons typically omit: a dry room at −40 °C dew point or lower for filling, Karl Fischer titration on every incoming and mixed batch, cold storage at 2–8 °C (since LiPF6 electrolytes age at room temperature), and HDPE or PFA containers, since HF etches borosilicate glass.

LiFSI and LiTFSI are genuinely more forgiving: LiTFSI is stable in water, and LiFSI hydrolyzes far more slowly. For a lab struggling with moisture control, this robustness can outweigh any advantage shown in cycling data.

Variable 3: Cost Per Mole of Lithium, Not Cost Per Kilogram

This is where most published comparisons go wrong. Salts are quoted in $/tonne, but cells are formulated in mol/L — and LiFSI's molar mass is 23% higher than LiPF6's, so a kilogram-basis price understates the true switching cost. The correct metric is:

Cost per mole of Li = (MW in g/mol ÷ 1,000) × price in $/kg

At illustrative prices — battery-grade LiPF6 near $12/kg and LiFSI near $13.70/kg, broadly in line with late-2025/Q1-2026 assessments:

Salt	MW	Price ($/kg)	Cost per mol Li
LiPF6	151.9	12.00	$1.82
LiFSI	187.1	13.70	$2.56

 

Verify before publishing: LiPF6 assessments diverged widely in Q1 2026, ranging from roughly $8,200 to $14,400/tonne FOB China. Confirm current pricing against Fastmarkets MB-FLU-0023 or SMM on the quote date.

The headline price gap is 14%. The gap that actually matters is 41%. That correction alone can reverse a build-versus-buy conclusion.

Worked Example: What a LiFSI Co-Salt Costs Per kWh

Assuming an NMC pouch cell at 0.45 kg electrolyte per kWh (0.4–0.5 kg/kWh is typical for NMC; LFP runs 0.6–0.8 kg/kWh):

• Baseline — 1.0 M LiPF6: 151.9 g/L at ~1.20 kg/L electrolyte density = 12.7 wt% salt, consistent with the industry rule of thumb that LiPF6 constitutes 12–13% of electrolyte mass. Salt cost: 57 g/kWh → $0.69/kWh.
• Co-salt — 0.85 M LiPF6 + 0.15 M LiFSI: 129.1 g/L plus 28.1 g/L = 13.1 wt%. Salt cost: $0.74/kWh.
• Delta: $0.05/kWh — approximately $3.75 on a 75 kWh pack.

Full substitution to 1.0 M LiFSI lands near $0.92/kWh, or roughly $17 per pack above baseline. The chemistry itself is inexpensive; qualification is not. Twelve weeks of 45 °C pouch cycling, abuse testing, and dry-room revalidation dwarf the material-cost delta.

One important caveat: this analysis holds only at current prices. LiPF6 pricing spiked roughly fivefold in 2021–2022, and lithium carbonate prices rose 52% between December 31, 2025, and January 28, 2026 (Benchmark Mineral Intelligence). This calculation should be rerun quarterly.

Variable 4: The Purity Specification to Buy Against

"99.9% battery grade" is not a meaningful specification on its own — impurities at single-digit ppm levels determine performance. Fastmarkets' battery-grade LiPF6 assessment specifies Cl ≤ 5 ppm, SO4 ≤ 20 ppm, Na ≤ 5 ppm, K ≤ 5 ppm, Fe ≤ 5 ppm, Cu ≤ 2 ppm, and Pb ≤ 2 ppm — a reasonable floor for any quote under consideration.

The impurity to prioritize, by salt:

• LiPF6 — free acid as HF (≤50 ppm) and water (≤20 ppm), tested on the shipped lot rather than a type-test certificate.
• LiFSI — chloride above all, plus residual fluoride and sulfate. This is the impurity profile that determines whether the aluminum survives.
• LiBOB — moisture and insolubles. Its ~0.8 M solubility ceiling in carbonates means a cold shipping container can precipitate salt out of a premixed electrolyte.

A Five-Step Protocol for Building a Co-Salt Formulation

Most production electrolytes are salt systems: a main salt, a co-salt, and one or two film formers. This sequence minimizes wasted cells.

1. Fix a baseline; never move two variables at once. 1.0 M LiPF6 in EC/EMC 3:7 with 2 wt% VC is a defensible starting point. Change the salt or the solvent — not both.
2. Substitute molar-for-molar, not additively. 0.85 M LiPF6 + 0.15 M LiFSI holds total Li⁺ at 1.0 M, isolating the anion effect from concentration effects.
3. Anchor the aluminum first. Add 0.5–1 wt% LiDFOB or LiBOB before running any long-duration test. This is inexpensive insurance.
4. Screen in sequence: aluminum CV and chronoamperometry → conductivity at −20, 25, and 60 °C → coin-cell DCIR by HPPC at −20 °C and 50% SOC → 45 °C pouch cycling for 8–12 weeks.
5. Watch for two common false positives: room-temperature conductivity gains that vanish at −20 °C, and early-week improvements that reverse by week eight. Both are common with imide co-salts.

Five Handling Mistakes That Invalidate a Lithium Salt Comparison

These are the most frequent failure modes observed in submitted samples and customer data:

1. Weighing LiPF6 outside a glovebox. It absorbs moisture within seconds. Caked powder or straw-colored electrolyte indicates HF is already forming and the comparison is compromised.
2. Dissolving too quickly. Dissolution is exothermic. Add the salt in portions with the flask in a chilled bath and hold below 30 °C — overheating generates PF5 and discolors the batch.
3. Storing in borosilicate glass. HF etches it, silicon then appears in surface analysis, and the resulting contamination is time-consuming to trace.
4. Testing only at 25 °C for 50 cycles. Salt differences typically emerge at 45 °C and after roughly six weeks. A short room-temperature run produces false confidence.
5. Comparing at fixed molarity when costing. Fixed molarity correctly isolates anion effects but misrepresents cost. Report both mol/L and wt%.

Where This Comparison Does Not Apply

The aluminum-first logic above assumes a liquid carbonate electrolyte in a 4.0–4.4 V cell. Outside that envelope, the ranking changes:

• Solid polymer electrolytes (PEO): LiTFSI is standard and LiPF6 is essentially unusable. Operating below 4 V also removes most aluminum-related pressure.
• Sulfide and oxide solid electrolytes: No dissolved salt exists in the separator layer, so this comparison does not apply.
• Lithium metal anodes: High-concentration and localized high-concentration LiFSI electrolytes are the design target; dilute-electrolyte aluminum data is misleading here.
• LFP and LMFP: Charging to ~3.65 V stresses the aluminum far less — a genuine reason LFP storage cells adopted LiFSI ahead of high-nickel EV cells.
• Sodium-ion: NaPF6 vs. NaFSI follows similar logic, but the specific figures do not transfer.

Frequently Asked Questions

Is LiFSI better than LiPF6?

LiFSI outperforms on thermal and hydrolytic stability and low-temperature conductivity, and underperforms on aluminum passivation and cost per mole of lithium. It is not a drop-in replacement. In production cells, it is typically used as a co-salt at 0.1–0.3 M alongside LiPF6, capturing most of the benefit without the corrosion risk.

Why is LiPF6 still the industry standard if it is thermally unstable?

Because it passivates aluminum. LiPF6 releases fluoride that builds a protective AlF3 layer on the cathode current collector — something most alternative anions cannot do. It also offers a well-balanced conductivity, transference number, and SEI profile, backed by two decades of manufacturing infrastructure and quality data.

What water content is required in the electrolyte?

Target below 20 ppm water and below 50 ppm free acid (as HF) for LiPF6-based electrolytes. Above these thresholds, hydrolysis to HF accelerates and attacks the cathode surface. Measure by Karl Fischer titration on every batch, and again after mixing, since both solvents and salt contribute moisture.

Can LiPF6 be replaced entirely with LiFSI?

Only with aluminum protection in place. Full LiFSI substitution requires a boron-oxalate co-salt (such as LiDFOB or LiBOB), a small residual LiPF6 fraction, a high salt concentration above roughly 3 M, or a coated current collector. Without one of these, aluminum pitting should be expected during high-voltage holds at elevated temperature.

What is the difference between LiBOB, LiDFOB, and LiPO2F2?

All three are film formers used at 0.5–2 wt%, not conducting salts. LiBOB produces the most robust interphase but is limited to roughly 0.8 M solubility in carbonates. LiDFOB is more soluble and combines oxalate and fluoride chemistry. LiPO2F2 primarily suppresses impedance growth during cycling and storage.

How much does switching lithium salts cost per kWh?

At early-2026 prices, adding a 0.15 M LiFSI co-salt costs roughly $0.05/kWh — about $4 on a 75 kWh pack. Full substitution to 1.0 M LiFSI adds around $0.23/kWh. The material cost is minor; the qualification program, dry-room revalidation, and abuse testing dominate the real cost of switching.

Conclusion

Rank candidates on aluminum passivation first, hydrolytic stability second, cost per mole of lithium third, and impurity specification fourth. Conductivity is a filter, not a decision criterion. Plan a salt system from the outset rather than searching for a single perfect anion.

Next step: Run the three-day Li/Al coin-cell screen on the two leading formulations before committing to a full cycling program. It costs a handful of coin cells and routinely eliminates one candidate outright.