Limiting Oxygen Index (LOI) for Flame Retardant Plastics: What the Number Actually Tells You

Quick answer: If your compound shows a strong LOI and keeps failing V-0, the number is not lying to you — LOI and UL94 reward opposite behaviour on dripping. Molten drips carry heat away and push LOI up, while the same drips ignite the cotton and cap you at V-2. Use LOI for development and batch checks; use UL94 for the file.

We quote LOI on nearly every datasheet and run it on every project at our partner factory’s lab, so this is a number we have a stake in. It is also the number most often over-read by buyers. Here is what it does and does not tell you.

What LOI Actually Measures

A specimen is clamped vertically in a glass chimney fed with a controlled oxygen-nitrogen mixture, lit at the top, and allowed to burn downward. The oxygen fraction is adjusted across repeat runs until you find the lowest concentration at which combustion just sustains itself. That percentage is the LOI.

The single most useful reference point is that air contains 20.9% oxygen. Read every LOI value against that line:

• LOI below ~21 — the polymer sustains burning in ordinary air. It needs help.
• LOI 21–26 — marginal. Self-extinguishing in still air under ideal geometry, unreliable in a real part.
• LOI 26–32 — where most successful flame retardant compounds land.
• LOI above 32 — strongly flame retardant, and the region rail and mining specifications ask for.

What makes LOI valuable is that it is continuous. UL94 gives you three discrete rungs; LOI gives you a number that moves when you change the formulation, which is why it is the right instrument for development work.

The LOI Ladder for Common Polymers

These are the values we quote across our own material pages, gathered in one place. Neat resin first, then what a working flame retardant package achieves:

Two things fall out of this table that are worth saying explicitly. Rigid PVC at 45 needs no flame retardant to self-extinguish — when we sell into PVC, the work is smoke suppression and antimony reduction, not flammability. And POM at 15 is the hardest case in common engineering plastics, because the oxygen is inside the polymer chain where no additive can intercept it.

How Different Chemistries Move the Number

Not all flame retardants buy LOI at the same rate, and knowing the exchange rate saves a lot of pointless dosing. The underlying mechanisms explain why.

That last row is the whole argument of this article compressed into one line. The additive that does nothing for your LOI may be the one that takes you from V-2 to V-0. If you optimise on LOI alone, you will systematically under-value it, and if you are chasing a number the mineral route makes progress feel slow even when the loading is correct.

Why a High LOI Does Not Mean UL94 V-0

This is the part that costs people certification cycles, and the mechanism is genuinely counter-intuitive.

Dripping helps you in the LOI test. The specimen burns downward from the top. When molten polymer drips away it removes heat and fuel from the burning zone, so combustion is harder to sustain and the measured LOI goes up. The test rewards a material for falling apart.

Dripping is exactly what UL94 punishes. The V rating is decided largely on drips: V-0 and V-1 permit no drips that ignite the cotton indicator beneath the specimen, while V-2 is the rung materials fall to when they self-extinguish acceptably but drip flaming particles. So the same behaviour that flattered the LOI number caps the UL94 rating at V-2.

This is why an unfilled PP compound can show a respectable LOI near 28 and still be V-2, and why anti-drip additives such as PTFE appear in almost every polyolefin and polyester package. The additive does nothing for the LOI and everything for the rating.

A polypropylene job from last year makes this point better than the theory does. The customer arrived with a compound at LOI 30 and a UL94 result stuck at V-2 through four rounds of trials. Each round they had raised the flame retardant dose, which walked the LOI up towards 32 and changed the rating not at all. The oxygen index was never the problem — their specimens were dropping flaming material onto the cotton. Under 0.5% PTFE fixed the rating in a single iteration, and the LOI came down slightly in the process, which they found hard to accept until we went through the mechanism. They had spent three months optimising the wrong number, and the datasheet had been telling them they were making progress the whole time.

The general rule: LOI tells you how hard the material is to keep burning; UL94 tells you how it behaves in a specific fire scenario with a specific geometry. Neither substitutes for the other, which is the same argument that applies to UL94 versus ASTM E84, and to the electrical tests that sit alongside them.

What the Chimney Test Cannot See

Dripping is the best-known limitation but not the only one. It is worth understanding what the test geometry leaves out, because it explains why large-scale tests exist at all.

• The flame travels downward. Real fires spread upward, where rising hot gases preheat the material ahead of the flame front. Downward burning is the easy direction.
• There is no external heat. In a room fire, surfaces receive radiant heat from everything else burning. The chimney gives the specimen no such help, so a material that self-extinguishes at 28% oxygen in still air may keep burning under radiant load in air.
• The specimen is small and isolated. No joints, no corners, no stacked surfaces, none of the geometry that makes real assemblies burn differently.
• Ventilation is controlled and steady. Real fires cycle between fuel-limited and ventilation-limited, which changes smoke and toxicity far more than it changes flame spread.

This is precisely why ASTM E84 uses a 25-foot tunnel and why cable standards test bunched cables in a vertical tray rather than single wires. LOI ranks materials against each other under identical simple conditions; it does not predict what a finished assembly does in a fire. Treat it as a comparator, not a forecast.

Where LOI Is Actually Required

Here is where I would push back on how our own industry uses this number. For most plastic parts sold into North America and Europe, LOI is not the certification gate — UL94 is, or ASTM E84 for building materials, or the IEC 60332 family for cable, or CPR Euroclasses in European construction. Nobody issues a listing on the strength of an LOI value.

That said, “LOI is never a requirement” is wrong, and buyers in some sectors will rightly object:

• Rail. EN 45545-2 does specify it — for the highest hazard level, HL3, materials are required to exceed LOI 32%.
• China and GB-based supply chains. GB/T 2406 is the oxygen index method, and Chinese procurement specifications frequently name an oxygen index minimum directly.
• Cable and mining tenders. Purchase specifications in these sectors commonly set an LOI floor, often 30 or 32, alongside the IEC tests.

So the honest framing is not “LOI does not matter” but: know whether your LOI number is a contractual requirement or a development metric, because they carry completely different obligations. If a customer specification names a value, it is a requirement and must be tested and reported. If it does not, quoting LOI on a datasheet is useful information but not evidence of compliance — and it cannot go into a file where UL94 data is what the standard asks for. The difference between capability data and a recognised listing is worth reading properly in what a UL Yellow Card actually covers.

The mistake in the other direction costs more. We were once asked to quote a halogen-free package for an appliance bracket at a minimum LOI of 32 — a demanding target that pushes loading, and with it the price, a long way up. On the second call it emerged that the buyer had copied the clause out of a rail specification his company used on a different product line. The actual requirement for the bracket was UL94 V-0 at 1.6 mm, which a standard package meets at a fraction of that loading. Asking where the number came from took ten minutes and took a substantial cost out of the project.

What LOI Is Genuinely Good For

None of the above makes it a soft number. It is the most useful instrument we have for three jobs.

Formulation iteration. UL94 has three rungs, so a change that moves your compound meaningfully closer to passing often produces no visible result at all — you get V-2 before and V-2 after, with no way to tell whether you are heading in the right direction. LOI moves from 27.5 to 29.2 and tells you the change worked. Development on the discrete rating alone is guesswork.

Batch consistency. LOI on incoming lots is one of the cheapest ways to detect that something has shifted — wrong dosage, poor dispersion, a substituted grade. A batch that drops two points has a real cause worth finding before the material reaches a moulding machine. We cover this in the incoming-inspection routine in how to evaluate a flame retardant sample.

Cheap elimination. A candidate that cannot get past 24 will not pass V-0 at your wall thickness, and you learn that on a bench in an afternoon rather than in a certification lab in six weeks. We run LOI alongside UL94 at our partner factory’s burn lab for exactly this reason: pre-screening is the point.

How to Read a Supplier’s LOI Number

Four questions worth asking before you compare two datasheets.

At what specimen thickness? LOI is geometry-dependent. A thicker bar has more thermal mass and typically reports better than a thin one, so values measured at different thicknesses are not comparable. This is the same thickness sensitivity that governs thin-wall ratings — if your part is 0.8 mm, an LOI from a 4 mm bar is not describing your part.

Which method? ASTM D2863 and ISO 4589-2 are closely aligned, GB/T 2406 is the Chinese equivalent. Mixing methods across a comparison introduces scatter you will misread as a formulation difference.

Is there UL94 data next to it? A supplier quoting a strong LOI and no rating may be avoiding a drip problem. Ask for the rating at your thickness, and ask what colour it was measured in.

Is it one lot or a production average? A single flattering measurement is not a specification. What you want to know is the spread.

FAQ

What is a good LOI value for flame retardant plastic?

For most applications a working compound lands between 26 and 32. Below about 24 the material will struggle to reach V-0 at normal wall thicknesses; above 32 you are in the territory rail and mining specifications ask for, usually at a real cost in loading and mechanical properties. But “good” depends on what the part has to pass — the target should come from the specification, not from wanting a higher number.

Does a higher LOI always mean better fire performance?

No, and this is the most common misreading. LOI can be raised by behaviour that makes the UL94 rating worse, because dripping removes heat in the LOI test and disqualifies the material in UL94. Two compounds at LOI 29 can be V-0 and V-2 respectively. Always read LOI together with the rating.

What is the LOI of air?

Air is 20.9% oxygen. That is the line that gives LOI its meaning: a polymer below roughly 21 can sustain combustion in ordinary room air, while one above it theoretically cannot self-sustain in air. Theoretically is doing work in that sentence — real parts have geometry, heat sources and ignition conditions that the chimney test does not reproduce.

Which standard covers LOI testing?

ASTM D2863 in North America, ISO 4589-2 internationally, GB/T 2406 in China. These are test methods, not performance standards — they define how to measure the number, not what number you need. The requirement, when there is one, comes from a product standard such as EN 45545-2 for rail or from a purchase specification.

Can LOI replace UL94 testing?

No. They measure different things and are used for different purposes: LOI for development and consistency, UL94 for certification. Where a standard names UL94, LOI data is not an acceptable substitute in the compliance file, however good the number is.

Do you test LOI in house?

Yes — we run LOI alongside UL94 vertical burn on every project, at the wall thickness you specify, before anything is quoted. It is bench-scale screening rather than certification: it tells us whether a formulation is worth putting in front of a third-party lab, and third-party reports can be arranged through an accredited laboratory when you need them for a file.

Final Thoughts

Use LOI for what it is good at – development iteration and incoming batch checks – and never submit it where a rating is required. If your incoming lots move more than a couple of points, something has shifted in dosage or dispersion and it is worth finding before the material reaches a machine. We run LOI beside UL94 on every project; our additive systems and finished compounds both ship with the data.

—— Rectivas Materials 团队 技术工程师老陈