Is silicone plastic? Is it safe for food? Silicone explained

Silicone has had a very good decade. It’s the floppy muffin tray, the spatula that doesn’t melt, the “reusable” food bag, the baby teat, the collapsible travel cup. It’s sold as the grown-up alternative to plastic: heat-proof, non-stick, “made from sand.”

Some of that is true. Silicone really isn’t the same thing as polyethylene or polypropylene. But “not a conventional plastic” got rounded up to “inert,” and inert it isn’t. Bake in it with something fatty and some of it ends up in your dinner. The research on this is unusually good, and it points to a clear line: silicone is fine for some jobs and a poor swap for others.

What silicone actually is

Ordinary plastics are long chains of carbon atoms. Silicone’s chain is different: a repeating silicon-oxygen-silicon backbone, with small organic groups (mostly methyl groups) hanging off the sides. In silicone polymers, more than 90% of the main-chain links are Si-O-Si bonds. The kitchen version is mostly polydimethylsiloxane, or PDMS, cross-linked (“cured”) into a rubbery solid.

That inorganic backbone is why silicone shrugs off oven heat that would soften most plastics. It’s also why the “made from sand” line is only half the story. The silicon does start out as quartz, but turning it into silicone takes heavy industrial chemistry, and traditional ways of making silicones have driven a sharp rise in CO₂ emissions, according to a 2026 review. Sand is the starting point, not the product.

So is it plastic?

Depends who’s asking:

• Chemically, it’s a synthetic polymer, made in a factory and not found in nature. Under the LPP standard that’s what counts: we treat it as a synthetic polymer and never call a product “plastic-free” if silicone touches what you eat or drink.
• Legally, in the EU, it’s its own thing. The EU food-contact framework lists “Silicones” as a separate material group from “Plastics” and “Rubbers”. That matters more than it sounds, because the EU’s detailed plastics rules don’t apply to silicone (see the regulation section below).
• In marketing, it’s whatever sells. “Plastic-free” on a silicone product is technically arguable and misleading in practice.

The honest one-liner: silicone isn’t a conventional plastic, but it’s a synthetic polymer, and it doesn’t earn a free pass for that.

The D4, D5, D6 problem: what’s left over inside

The chemistry that matters for your food is what didn’t get built into the finished rubber. Making silicone leaves behind small, unlinked ring-shaped molecules called cyclic siloxanes. Chemists name them by ring size: D4 has four silicon-oxygen units, D5 five, D6 six, and so on up past D20.

They’re in everything. In a German study of 37 baking moulds, every one contained cyclic siloxanes with rings of 6 to 50 units, and no other migrating substances were detected. US FDA chemists found D4, D5 and D6 in every silicone baby-bottle nipple they tested, at 0.5 to 269 µg/g, and up to 7,030 µg/g of D6 in bakeware. Health Canada scientists measured total cyclic siloxanes (D4 to D16) of 680 to 4,300 µg/g in 25 bakeware products sold in Canada.

These leftovers aren’t bonded to the rubber, so they can move. Where they go depends on three things: heat, fat and how well the item was made.

What moves into food: the baking studies

Fat and heat drive migration

The headline result comes from a German food chemistry group that has spent more than fifteen years baking in silicone moulds and analysing the results. In one study, a meat loaf baked in a commercial silicone mould picked up 177 mg/kg of siloxanes. That’s nearly three times the 60 mg/kg “overall migration limit” the EU applies to plastics, which researchers use as the benchmark for silicone too. In the same study, milk-based foods picked up almost nothing: under 2.4 mg/kg, even when they were high in fat.

Cakes landed in between. Across that group’s earlier tests, siloxane migration into cakes never exceeded 21 mg/kg, depended strongly on fat content, and dropped only slightly between the first and tenth bake.

It isn’t only the oven. In a 2025 study, chocolate prepared in silicone moulds at no more than 45 °C still picked up 21.5 mg/kg of siloxanes by the third use. Fat pulls siloxanes out, and fat at moderate heat is enough.

The Canadian bakeware study used an oily food simulant baked for an hour at 177 °C and found an average of 105 µg/g of D4 to D16 in the “food”. Per kilogram of body weight, young children had the highest estimated intake.

Your kitchen air gets some too

The smallest siloxanes, like D5, are volatile. In the 2025 German study, about 98% of the D5 in a non-post-cured mould evaporated into the air over five bakes, and only about 1% went into the food. The Canadian study measured one-hour average air levels during baking of up to 646 µg/m³, falling quickly once baking stopped. A German study of 14 silicone moulds found that four of them pushed air levels above Germany’s precautionary indoor guide value for cyclic siloxanes during baking. In the cakes themselves, that study found larger rings such as D9, D12, D15 and D18 at higher levels at the edges than in the centre, while D3 to D5 were barely present.

So the picture is: small, volatile rings mostly go up the extractor hood, and bigger, heavier rings end up in the food, especially the fatty crust touching the mould.

Consumer testing tells the same story

In 2022 the European consumer federation BEUC published a test of 44 silicone baking moulds bought by nine national consumer groups, including Denmark’s Forbrugerrådet Tænk. This is a consumer-group lab test, not a peer-reviewed study, but its results line up with the research. Ten moulds (23%) got the worst rating, either because they released more than 30 mg/kg of volatile compounds or because releases rose over three test runs. Two moulds bought on Amazon went over 60 mg/kg. Only seven (16%) came out clean. Thirty-six released at least one “substance of concern” at low levels, including D4, benzophenone and the phthalate dibutyl phthalate.

A Danish study for Norway’s food authority found much the same pattern in 2016. DTU’s National Food Institute tested 49 silicone food-contact items from the Norwegian market. Twenty-four contained more siloxanes than proposed action limits would allow, but migration estimates suggested none would actually exceed those limits in use. The proposed limits included 12 mg/kg of food for adults and just 2 mg/kg for children (for D3 to D8), plus 60 mg/kg for a wider group of siloxanes. That gap between “what’s in it” and “what comes out” is why migration testing matters more than content.

“Silicone” doesn’t guarantee siloxanes are the only thing in there

Silicone doesn’t need plasticisers the way soft PVC does; that’s partly why EU toy rules exclude silicone rubber from their definition of “plasticised material,” as we cover in our toys and teethers explainer. But some silicone kitchen items have turned up with phthalates anyway. A Japanese survey found packings and spatulas containing the phthalate DEHP at 60 to 5,830 µg/g, although the nipples in the same survey contained no additives. A 2026 study of air-fryer moulds found dibutyl phthalate migrating above its legal limit, alongside cyclic siloxanes above 60 mg/kg in a fatty simulant. (More on these chemicals in our phthalates explainer.)

Post-curing: the factory step that changes everything

Here’s the most useful fact in this whole article. Most of the small siloxanes can be driven out before the product ever reaches you, by a long, hot “post-cure” (also called tempering) at the factory.

The effect is big. In the 2025 German study, smaller siloxanes (D4 to D13) migrating into shortcrust pastry fell from 153 mg/kg from a non-post-cured mould to 9.5 mg/kg from a post-cured one. That’s a sixteenfold drop from one manufacturing step. In the meat-loaf study, non-tempered moulds with 1.1% volatile content migrated “considerably” more than tempered ones.

Two catches:

1. It doesn’t fix everything. In the same 2025 study, migration of the biggest rings (above D20) was almost unaffected by post-curing, and total siloxanes in pastry still reached 167 mg/kg by the third bake.
2. You can’t see whether it was done. Post-curing costs energy and time, and BEUC’s report says manufacturers don’t consistently do it. Nothing on a typical label tells you.

Use helps, and silicone doesn’t seem to break down into more siloxanes

The good news for anyone with silicone already in the drawer: migration falls with use. In a commercial pizza bakery, moulds used up to about 1,700 times went from 11 to 18 mg/kg of migration down to below detection (under 1 mg/kg). The researchers saw no sign of the rubber breaking down to form new migrating siloxanes, and the Canadian study likewise found migration and air emissions declined steadily with repeated baking. One caveat: the moulds soaked up fat as they went, up to 8 g per kg of silicone.

Also, the decline isn’t universal. Six of BEUC’s 44 moulds released more on the third run than the first.

Cold drinks, milk and baby teats

The picture is much calmer away from heat and fat.

FDA chemists put milk and infant formula in direct contact with silicone baking-sheet plaques for six hours at 40 °C and detected no siloxanes at all. Even in alcohol-water simulants over up to 72 hours, migration was tiny: at most 155 ng/mL of D6. The Japanese survey found nothing released into a 20% alcohol simulant at 60 °C for 30 minutes. Milk-based foods in the meat-loaf study were under 2.4 mg/kg. Teats put through 100 simulated feeds with milk and microwave sterilisation kept working normally, with little change in their siloxane content because milk is a weak extractor.

Particles are a separate question. A 2022 study in Nature Nanotechnology found that steam disinfection etched the surface of silicone teats and released micro- and nanoparticles of 0.6 to 332 µm. The authors estimated a baby could swallow more than 0.66 million micro-sized particles by age one, and said plainly that the health risks are “as yet unknown.” That’s one study using a new imaging method, with modelled intake. A 2023 study found that silicone pacifiers released nanoparticles when shredded with an immersion blender, more so after boiling. That’s an extreme abrasion test, not a model of a baby sucking a dummy. We cover teats and bottles in detail, including what to do without breaking formula safety rules, in our baby bottles article.

Is any of this actually harmful?

This is where the evidence goes thin, and we’d rather say so than fake confidence.

• D4, D5 and D6 are an environmental problem first. EU member states identified all three as substances of very high concern in June 2018. D4 meets the criteria for persistent, bioaccumulative and toxic (PBT) and very persistent, very bioaccumulative (vPvB); D5 and D6 meet the vPvB criteria. That listing is about what they do in rivers and wildlife, not about a dose in your muffin.
• The dietary risk from D4 and D5 looks small. A 2026 Chinese study of 30 silicone food-contact items under worst-case conditions rated D4 and D5 negligible risk. But it flagged D3 as a potential concern and found 14 smaller siloxanes, including D6, above a screening threshold called the Threshold of Toxicological Concern. Exceeding that threshold means “this needs proper toxicity data,” not “this is proven harmful.”
• Most of what migrates has barely been studied. The bigger rings (D7 and up) dominate what ends up in baked food, according to the Canadian study, and there’s little toxicity data for them.

So: we know a lot about how much siloxane moves into food, and little about what it does at those doses. That’s an argument for avoiding the easy exposures, not for panic.

What the rules say

Europe: a patchwork

The EU has detailed food-contact rules for plastics, and no detailed rules for silicone. National rules and guidance fill some of the gap. BEUC counts only five member states with silicone-specific rules. The most influential is Germany’s BfR Recommendation XV. It isn’t law, but German inspectors treat it as the benchmark for good manufacturing practice, and it sets a 0.5% guide value for volatile organic compounds released from finished silicone. The air-fryer moulds tested in 2026 exceeded that value. Spain applies the plastics-style 60 mg/kg limit to silicone.

The D4, D5, D6 restriction (and what it doesn’t cover)

You may have seen headlines that the EU “banned” D4, D5 and D6. Here’s what Regulation (EU) 2024/1328 actually does:

• Since 31 January 2020: D4 and D5 limited to under 0.1% in wash-off cosmetics like shampoo.
• From 6 June 2026: each of D4, D5 and D6 limited to under 0.1% in substances and mixtures placed on the market generally, and banned as dry-cleaning solvents. Exemptions and later dates apply to some uses.
• From 6 June 2027: the limit extends to leave-on and other cosmetics.
• From 6 June 2031: medical devices and medicines.

What it doesn’t do is cover silicone articles, the solid things you bake in. The restriction targets substances and mixtures, and it exempts D4, D5 and D6 present as constituents of a silicone polymer, as well as their supply for use in making articles. Your muffin tray isn’t touched by it. So when a seller says their silicone is “REACH compliant,” that tells you very little about migration into food.

The US

The FDA lists silicone elastomers among the rubbers allowed for repeated food contact under 21 CFR 177.2600. The test is an extraction limit: for fatty foods, total extractives in hexane can’t exceed 175 mg per square inch over the first seven hours and 4 mg per square inch over the next two. “FDA compliant” means a product meets that test, if anyone ran it. It’s not an FDA approval of the specific item.

Labels decoded: food-grade, platinum-cured, medical-grade

“Food-grade” or “FDA-approved.” Food-grade has no single legal definition. The useful version is a named standard: compliance with 21 CFR 177.2600 in the US, or with Germany’s BfR Recommendation XV in Europe. Ask which one, and whether the finished item (not just the raw material) was tested. “FDA-approved” is usually wrong: the FDA doesn’t approve individual silicone products.

“Platinum-cured.” Silicone is cured, or cross-linked, one of two common ways: with a platinum catalyst (addition curing), or with organic peroxides. Peroxide curing with one common initiator breaks down into dichlorobenzoic acid, dichlorobenzene and specific PCBs (PCB-47, -51 and -68), which have been found in silicone factory workers’ blood and, around German silicone plants, in local food crops. That’s a real factory and environmental issue. We found no study measuring these by-products in finished kitchenware, so we can’t tell you it matters at the table. And “platinum-cured” doesn’t guarantee low migration: the moulds in the 2025 pastry study were platinum-cured, and the one that skipped post-curing still released 153 mg/kg. Platinum-cured and post-cured is the combination that counts.

“Medical-grade.” It isn’t a food-contact standard: neither the FDA rule nor the European rules above use the term, and we found no study showing medical-grade silicone migrates less into food. Treat it as marketing unless the maker also names a food-contact standard.

The pinch test

The internet’s favourite home test: pinch and twist a piece of silicone, and if white shows through, it’s supposedly been bulked out with cheap “filler.” We searched for a study that tests this and found none. Nobody has shown that whitening on a pinch tells you anything about what migrates into food, which is the thing you actually care about. A pass doesn’t mean post-cured, and a fail doesn’t mean dangerous. Skip it, and ask the seller for test documentation instead.

When you’re done with it

Silicone lasts a long time, which is its best environmental feature. The pizza-bakery moulds kept working through around 1,700 uses. But when it’s finished, it’s awkward. Silicone isn’t one of the plastics that household recycling systems are set up to sort, so don’t put it in your recycling bin unless your local council says it accepts silicone. Industrial and chemical recycling exists, but a 2025 paper in Science described silicone recycling as one that still “remains in its infancy”. For more on why “recyclable” claims mislead, see our recycling labels explainer.

The practical upshot: if you already own silicone, use it for years. Buying silicone to replace plastic you were going to keep using anyway isn’t a win.

What to do, ranked by evidence

Strong evidence

• Don’t bake or roast fatty food in silicone. Meat loaf, pastry, roast vegetables in oil and greasy air-fryer liners are the worst cases. Metal, glass, enamel and ceramic tins do the job without the migration. (Check glazed ceramics and enamel for lead and cadmium compliance; our standards page explains why.) This is why our standard never accepts silicone for baking or cooking.
• Brand-new silicone releases the most. If you’re going to use a new silicone item hot, the first few uses carry the most siloxanes. Migration falls with repeated use, and a lot of the early release goes into the air, so ventilate: open a window or run the extractor.
• Buy post-cured, if you buy at all. It’s the single biggest factor in how much moves into food. Ask for it in writing.

Moderate evidence

• Cold and room-temperature uses are much lower risk. Milk and formula at 40 °C picked up nothing detectable, and mild alcohol-water simulants very little, in the studies we found. A silicone seal on a water bottle or storage-jar lid is a small concern. We still name it on any listing, and it caps a product at “plastic-minimal” under our standard.
• Oil and spirits are different. Fatty and alcoholic liquids do extract siloxanes, so don’t store cooking oil, infused oils or spirits in silicone.
• Follow the maker’s temperature limit. The German researchers who measured kitchen air recommend precleaning moulds and sticking strictly to the stated temperature.

Thin evidence (precaution, not proof)

• Pre-baking a new mould empty in a ventilated oven may drive off some volatile siloxanes, since that’s what factory post-curing does at a larger scale. Nobody has tested whether a home pre-bake matches a factory post-cure, and it won’t touch the bigger rings that don’t evaporate.
• Teats and steam. One study found particle release from steam-disinfected silicone teats. Health effects are unknown. Don’t stop sterilising when guidance says to; see our baby bottles article for options.
• Hot drinks through silicone. No study we found measured migration or particle release from small lid gaskets in tea or coffee. Our standard treats them as a named, minor seal. We’d avoid cups, kettles and infusers where silicone is the main hot surface, but that’s precaution.
• Ignore the pinch test. It’s not validated.

Silicone is a reasonable material for an ice-cube tray, a seal on a cold jar or a lid that never meets hot food. It’s a poor choice for the job it’s most often sold for: baking. For the bigger kitchen picture, see our plastic-free kitchen swaps, and for why “non-stick” often means something worse, our nonstick pans article.