Silicone

What actually migrates out of silicone is not a plasticiser or a monomer like BPA. It is low-molecular-weight cyclic siloxanes - D4, D5, D6 and heavier homologues - left over from manufacture, and they are present in the material at real levels: 0.5 to 269 ug/g across bottle nipples in an FDA survey, and 680 to 4300 ug/g of total cyclic siloxanes across 25 bakeware products in a 2025 Canadian study. How much of that reaches you depends almost entirely on what the silicone touches. The same FDA survey found no siloxanes at all in milk or infant formula after six hours of contact, and at most 155 ng/mL in an aggressive 95% ethanol simulant. Germany's BfR Recommendation XV, the benchmark most food-grade silicone is tested against, caps volatile organic components and extractables at 0.5% each, requires that finished parts do not test positive for residual peroxides, and limits platinum catalyst residue to 50 mg per kg of finished product; FDA covers the same parts under its repeated-use rubber rule, 21 CFR 177.2600. The honest verdict has two halves. For cold or watery contact - a jar gasket, a straw, a pacifier - measured migration is very low and the usual worry is misplaced. For hot, fatty, prolonged contact the numbers stop being negligible. And the deeper objection is not toxicological at all: silicone is a synthetic polymer that does not biodegrade, which is why we score it in its own tier rather than counting it as plastic or giving it a pass.

Silicone is a thermoset, so it does not melt, soften or warp anywhere in the normal kitchen range - freezer to oven to dishwasher to boiling water. The problem is chemical rather than structural, and it scales with temperature, fat and time. An hour of baking at 177C transferred an average of 105 ug/g of cyclic siloxanes into a fatty food simulant. A German study measuring real foods found 177 mg/kg of siloxanes in a meat loaf baked in a commercial mould, above the 60 mg/kg overall migration limit, while milk-based bakes in the same moulds stayed under 2.4 mg/kg. That study is also the case for post-curing: an untreated mould still carrying 1.1% volatile organics migrated far more than a properly tempered equivalent. Run a new silicone mould empty through a hot oven once or twice before you use it, and keep silicone out of the hottest, greasiest jobs, where metal or glass does the work better anyway.

Silicone carries no resin code - the numbered triangles are for thermoplastics - so a soft part with no code, a springy feel and no ability to hold a crease is either silicone or a silicone lookalike. The lookalike is TPE, a thermoplastic elastomer used for cheap straws, gaskets and lids because it moulds like plastic and feels like rubber; it goes tacky and deforms with heat, where silicone does not. Ignore the pinch test. Twisting silicone until it turns white shows only that the compound uses precipitated rather than fumed silica as its filler. It is strain whitening, not a purity test, and FDA-compliant food-grade silicone will happily turn white. The questions worth putting to a brand are which cure chemistry was used (platinum/addition or peroxide), whether the part was post-cured, and whether the compound is certified to FDA 21 CFR 177.2600 or BfR Recommendation XV - then ask for the report, because "food grade" and "medical grade" on their own are marketing phrases, not certifications.

Silicone does not perish the way natural rubber does - ozone, oxygen and sunlight leave it alone - so a gasket normally outlasts the appliance it seals. It is weak in tension and abrasion, though, so it tears from a nick, takes cuts from knives and scrapers, and holds onto grease and smells; a hot dishwasher cycle or a spell in the oven usually resets that. Replace it when it goes sticky, gritty or permanently flattened, not on a schedule.

Silicone is a thermoset: it cannot be melted down and reformed, so no curbside programme takes it, and putting it in one only means it gets pulled back out as a contaminant. A few specialist take-back schemes exist; the realistic destination is landfill, where it is inert but does not biodegrade. Its long service life is the real argument in its favour - one gasket that lasts a decade beats ten that do not.