304 vs 316 Coin Cell Case: Which Grade to Use

For most lithium-ion work below 4.3 V, a 304 case is fine, and the money you save is better spent on electrolyte purity or a second replicate. Above that, or in any electrolyte carrying chloride, sulfur, or meaningful water, the case stops behaving like inert hardware and starts behaving like an electrode you did not intend to build. This article gives you a decision rule based on the potential each shell actually sees, a 24-hour bench test to qualify the stock already sitting in your drawer, and the point where upgrading from a 304 vs 316 coin cell case stops helping and you need aluminum cladding instead.

The 60-second answer: match the grade to your upper cutoff, not to a corrosion chart

Pick the grade from what the positive shell will be held at and what is dissolved in your electrolyte. Everything else is secondary.

System / chemistry	Upper cutoff	Grade	Reason
Graphite‖LFP	≤ 3.8 V	304	Steel sits well inside its passive window; no measurable benefit from Mo
Graphite‖NMC, LCO full cells	≤ 4.3 V	304 (316 optional)	Parasitic currents are typically low; consistency matters more than grade
Half cells vs Li metal	4.3–4.5 V	316	Approaching the transpassive band; 316 buys margin, not immunity
LNMO, Li-rich, 5 V spinels	> 4.5 V	Al-clad, not 316	Wrong failure mode for Mo — see the high-voltage section below
Li–S, polysulfide catholytes	any	Al-clad or carbon-coated; 316 as fallback	Polysulfides attack bare austenitic steel chemically, not just anodically
Aqueous Zn (ZnSO₄, ZnCl₂), seawater	any	316 minimum; Ti or Ti-clad better	Chloride pitting is exactly what Mo is for
55–60 °C cycling or calendar aging in LiPF₆	any	316	HF generation scales with temperature and dwell time
LiFSI or LiTFSI-rich electrolytes	> 3.8 V	316 — avoid Al-clad	Imide salts corrode aluminum above roughly 3.7–4.0 V

Na-ion in NaPF₆ behaves much like LiPF₆ — read across from the lithium rows. If two rows apply, take the more conservative one; if none apply cleanly, run the 24-hour test described below.

What actually differs between a 304 and 316 coin cell case

One difference matters: molybdenum. 316 carries 2–3%, 304 carries none, and Mo stabilises the passive oxide film against localised breakdown — which is why 316 wins in chloride and acidified conditions.

Grade	Chromium	Nickel	Molybdenum	PREN (approx.)
304 (S30400, SUS304, 1.4301)	~18–20%	8–10.5%	none	18–20
316 (S31600, SUS316, 1.4401)	16–18%	10–14%	2.0–3.0%	24–28
316L (S31603, 1.4404)	16–18%	10–14%	2.0–3.0%	24–28

Nominal ranges. ASTM A240 and legacy AISI ranges differ slightly and carbon limits have been tightened — confirm against your mill certificate [VERIFY].

PREN is the Pitting Resistance Equivalent Number, %Cr + 3.3×%Mo + 16×%N — a ranking tool for chloride pitting, nothing more. A PREN of 24 does not mean 316 survives 5 V in a carbonate electrolyte. The two mechanisms are unrelated, and confusing them is the most common mistake in this decision.

What does not differ, and therefore should not influence you

• Electrical resistivity: about 72 µΩ·cm for 304 against 74 µΩ·cm for 316, a 3% gap. Interfacial contact resistance at the spacer–electrode–can stack is one to two orders of magnitude larger, so it swamps the difference.
• Annealed yield strength: both near 205 MPa, so most labs keep the same crimp setting. 304 work-hardens faster during drawing, so springback can differ slightly — check your first five crimps for a closed-height shift over about 0.05 mm whenever you switch lots, grade or not.

Only one shell in the set is actually at risk

The positive can is held at cathode potential and therefore anodically polarised — the condition under which steel dissolves. The negative cap sits near 0–1.5 V vs Li/Li⁺, which is textbook cathodic protection: oxidative dissolution there is suppressed whatever grade you bought.

So you are choosing a positive-shell grade and paying for the negative cap out of habit. Mixed sets — a 304 cap with a 316 or Al-clad can — are legitimate, and several suppliers ship them.

How much steel is actually touching your electrolyte

Run the CR2032 geometry. Interior diameter is about 19.4 mm after a 0.20–0.25 mm wall, giving a 2.96 cm² floor plus roughly 1.8 cm² of wall, part of it masked by the gasket — call it 3–4 cm² of wetted steel. A 12 mm cathode disc is 1.13 cm². The steel presents about three times the area of the electrode you are measuring, which is why hardware shows up in data at all.

What a 1 µA leak does to your coulombic efficiency

Take a 2 mAh cell cycled at C/10, so 200 µA for a 10-hour charge. A steady 1 µA of parasitic oxidation on the can adds 10 µAh of charge that never comes back — 0.5% of capacity, straight off your coulombic efficiency.

For high-precision coulometry, 0.5% is not noise, it is the entire signal — and it reads exactly like electrolyte oxidation on your cathode, so the obvious conclusion is that your material is unstable. Hardware artefacts are hard to catch afterwards because the data looks clean, which is the argument for screening hardware on its own.

Where 316 earns its premium, and where it does not

Situations where molybdenum genuinely pays

• Aqueous chemistries. Zn-ion cells in ZnSO₄, or worse ZnCl₂, are a chloride pitting environment — exactly what PREN ranks. 316 is the floor; titanium or Ti-clad is better.
• Elevated temperature, or electrolyte you cannot fully dry. LiPF₆ hydrolyses to HF with trace water, and the rate climbs sharply with temperature. Above roughly 20 ppm water, or at 60 °C, HF becomes a design consideration rather than a footnote.
• Sulfur and polysulfide systems, as a fallback only. Polysulfides attack austenitic steel chemically; 316 lasts longer than 304, but carbon-coated or Al-clad hardware is the real answer.

The high-voltage mistake: 316 is not a substitute for aluminum cladding

This is where most published advice goes wrong. Above roughly 4.5 V vs Li/Li⁺, stainless fails by transpassive anodic dissolution: the passive film is driven past its stability window and iron goes into solution. Molybdenum was never designed for that — it resists chloride-induced pitting, a different, localised mechanism.

So upgrading to 316 for a 4.9 V LNMO study buys a modest delay and a larger invoice. Aluminum-clad cases solve it properly: in LiPF₆ the aluminum passivates as AlF₃, thin enough to pass electrons but stable enough to block Al³⁺ dissolution. Suppliers typically specify a ~1 µm aluminum layer on the positive can and rate the parts above 4.5 V, several quoting 5.0–5.5 V.

The trade-off nobody flags: aluminum is corroded by imide salts. If your electrolyte is LiFSI- or LiTFSI-rich rather than LiPF₆-rich, Al-clad hardware can become the weak point above roughly 3.7–4.0 V, and 316 is the better choice despite the lower nominal voltage rating. Blended salt systems sit in between — test, do not assume.

Reported onset potentials for anodic current on bare stainless in LiPF₆/carbonate electrolytes vary widely with surface finish, water content, hold time and temperature — treat any single quoted number as conditional on someone else's setup [VERIFY against a specific study before citing a figure].

Walkthrough 1: a 24-hour blank-cell test to qualify your case stock

Before committing 100 cells to a campaign, spend a day and a dozen case sets finding out whether your hardware is clean at your operating voltage.

1. Build three cells per grade with no cathode. Stack: positive can, two layers of separator, 60–80 µL of your working electrolyte, lithium metal disc, spacer, spring, negative cap. The positive can floor is your working electrode.
2. Crimp at your normal setting and rest at open circuit for 6 hours. If OCV drifts more than about 50 mV in the final hour, the cell has not settled — do not proceed.
3. Hold potentiostatically at your intended upper cutoff — 4.3 V, 4.5 V, whatever you actually use — at 30 °C for 24 hours, with the cycler current limit an order of magnitude above your expected leak so it does not clip. Repeat at 55 °C if any part of your programme runs hot; cases that pass at 30 °C routinely fail at 55 °C.
4. Record current at 1 h, 6 h and 24 h. A passivating surface decays monotonically. Something that plateaus or climbs is dissolving.
5. Apply thresholds as a working rule of thumb: below ~0.5 µA at 24 h is clean; 0.5–2 µA is usable if your capacity is comfortably above 2 mAh; above ~5 µA, change hardware [VERIFY these bands against your own baseline — they scale with wetted area and temperature].
6. Disassemble in the glovebox and photograph the can floor at 20–40×. Look for loss of specular finish, dark speckling near the gasket line, or brown haze on the polypropylene gasket — iron staining on a white gasket is the easiest tell in the workflow.

What goes wrong with this test, from experience:

• Comparing case lots from different orders. Surface finish and passivation vary batch to batch more than 304 varies from 316 — test one variable at a time.
• Glovebox water above 1 ppm, or aged electrolyte for one grade and fresh for the other. HF content dominates everything here; aliquot from the same bottle on the same day.
• Reading the current too early. Most of the divergence appears between hour 6 and hour 24. And run two replicates per grade — the lithium counter electrode is not inert either, and you need a noise floor.

Walkthrough 2: verifying you actually received 316

Grade substitution is a real risk in research consumables, where identical-looking parts arrive in unlabelled bags. Three checks, escalating in cost:

1. Magnet test, 30 seconds, free. Both grades are nominally non-magnetic when annealed, but deep drawing cold-works them heavily: 304 transforms partly to strain-induced martensite and becomes noticeably magnetic, while 316's higher nickel and molybdenum stabilise the austenite. Hold a small neodymium magnet against a known 304 case and a suspect one; the difference in pull is usually obvious by hand. Indicative, not proof.
2. Handheld XRF, about 30 seconds per part. Any alloy analyser — Niton XL2, Olympus Vanta and similar — reports molybdenum directly. You want 2.0–3.0%; below 0.5% you have 304 in a 316 bag. Most university materials departments have one and will run a sample as a favour.
3. Request an EN 10204 Type 3.1 material test report tied to the actual coil, not a generic datasheet. A supplier who can produce one knows what they shipped.

While you are asking, confirm the cases were passivated after stamping, per ASTM A967 or equivalent. Free iron smeared on by the tooling causes rust spots that look like corrosion failure but are a cleaning problem — that is what orange specks straight from the bag usually mean.

The parts everyone forgets: spacers, springs and gaskets

A "316 case set" often contains 304 spacers and springs. Ask, because in some stack configurations those parts sit at cathode potential too.

• A 15.8 mm spacer presents about 1.96 cm² per face. In the standard half-cell stack it sits on the lithium side and is cathodically protected, so grade is irrelevant. In full cells, or wherever you add a spacer on the cathode side for stack pressure, that disc is at cathode potential and rivals the can floor in area — specify its grade.
• The gasket, not the steel, sets your practical temperature ceiling. Polypropylene creeps at sustained 60 °C and above; PFA or PTFE is the fix. Buying 316 for a hot-box study while keeping a PP gasket solves the wrong problem.

What a 304 vs 316 coin cell case upgrade actually costs per study

Small enough that cost should almost never decide it. On 2025–2026 list pricing from research suppliers, 316 and 316L CR2032 case sets sit around US$65–80 per 100 sets, with 304 typically 20–40% below and Al-clad above [VERIFY with your supplier — 60-set and 100-set packs are both common, which makes headline prices hard to compare].

On a 100-cell study that is a US$20–40 difference, against weeks of instrument time if it has to be re-run. Two conclusions follow, pointing in opposite directions:

• If your chemistry sits in a 316 row of the table above, buy 316 — the saving is not worth the risk.
• If it does not, 316 buys nothing measurable. A 4.2 V graphite‖NMC screening campaign gives the same data in either grade, and three cells per condition beats one cell in nicer hardware.

When this advice does not apply

• Never change grade mid-dataset. Switching halfway through a 200-cell campaign introduces a step change you cannot deconvolute later. Same for a lab with years of published 304 data — validate any change in parallel first.
• Fix assembly variability first. If your cell-to-cell capacity spread is 5%, case grade is a second-order term — crimp pressure, electrolyte volume and electrode alignment dominate. Get your coefficient of variation under about 2% before optimising metallurgy.
• Solid-state and sulfide electrolytes. Coin cells are usually the wrong format entirely — stack pressure matters more than case chemistry, and sulfides attack both grades.
• Teaching and high-throughput screening. When the readout is whether a material works at all rather than coulombic efficiency to three decimals, 304 wins on cost alone.

Frequently asked questions

Is 316 always better than 304 for coin cell cases?

No. The 2–3% molybdenum in 316 resists chloride pitting and acidic attack, but that advantage only appears in aqueous, chloride, sulfur or high-HF environments. In a dry LiPF₆ carbonate electrolyte below 4.3 V, the two grades perform indistinguishably and the premium buys nothing measurable.

Can I use 316 cases for high-voltage cathodes above 4.5 V?

Not as a real solution. Above roughly 4.5 V vs Li/Li⁺, stainless fails by transpassive anodic dissolution, which molybdenum does not prevent. Use aluminum-clad positive cases instead: they passivate as AlF₃ in LiPF₆ and suppliers rate them above 4.5 V, commonly to 5.0–5.5 V.

Is 316L better than 316 for coin cell cases?

Not for this application. The L means lower carbon (≤0.030%), which prevents chromium carbide precipitation during welding. Coin cell cases are stamped and drawn, never welded, so that benefit never materialises. Buy whichever is cheaper or better documented, and do not pay a premium for the L.

Do I need to change crimping pressure when switching from 304 to 316?

Usually not — both have annealed yield strength around 205 MPa. 304 work-hardens faster during drawing, so springback can differ slightly. Verify empirically: crimp five cells, measure closed height with a micrometer, and confirm you are within about 0.05 mm of your 304 baseline.

How can I tell whether my cases are really 316?

Start with a magnet: deep-drawn 304 becomes noticeably magnetic from strain-induced martensite, while drawn 316 stays weakly magnetic. That is indicative only. Confirm with handheld XRF, which reads molybdenum directly — expect 2.0–3.0% for 316 and essentially none for 304 — then request an EN 10204 3.1 certificate.

Does case grade affect measured capacity or only cycle life?

Both, indirectly. Case corrosion adds a parasitic oxidation current that inflates charge capacity and depresses coulombic efficiency — a steady 1 µA in a 2 mAh cell at C/10 costs about 0.5% CE. Dissolved iron can also deposit on the anode, so first-cycle data can look normal while cycle life suffers.

Conclusion

Choose on potential and electrolyte chemistry: 304 for lithium-ion below about 4.3 V; 316 for chloride, sulfur, water-tolerant or hot systems, and for imide-salt electrolytes where aluminum is not an option; aluminum-clad above 4.5 V, because that failure mode is transpassive dissolution and molybdenum does not address it. And remember you are really only buying a positive shell.

Next step: run the 24-hour blank-cell hold from Walkthrough 1 on the cases already in your drawer, at your actual upper cutoff. One day and a dozen case sets will tell you more than any datasheet.