On 8 August 2026, an Amazon UK reviewer, looking at the Loopware stainless-steel food containers — silicone-free, and gripping shut with a steel-on-steel friction fit rather than an airtight silicone seal — wrote that whether you want a food container that comes with a silicone seal, or one without, is “down to personal taste.”21 That is a fair way to talk about a lid. Some people like a soft, airtight seal; others would rather do without. But is it as simple as personal taste — or is it really about whether we know what that taste costs our health and our environment? Let’s dig in and find out.
The box itself is already pointing the right way. Stainless steel is a solid everyday material for leftovers and packed lunches, and this one does without a silicone seal at all: it gives up some of that airtightness in exchange for having no silicone. The part worth sitting with is the silicone seal itself — the soft ring on so many other food containers, flasks and lunch boxes — because that is what makes them airtight and leakproof, and that useful job is why it is there, and why most of us have been handed the same comforting sentence about it: it is food-grade silicone, so it must be fine. The sentence sounds like a verdict. It is worth asking what that verdict was actually measuring — and whether choosing the silicone-sealed kind, over a silicone-free one like this, is really just taste.
From beach sand to a kitchen seal
We all “know” food-grade silicone is safe. It is everywhere in the kitchen — baking mats, spatulas, steamer lids, the seals on flasks and lunch boxes. The story often begins with sand, and that part is true. Silicone starts as silica, which is quartz sand, one of Earth’s most abundant materials. Silicon is the second most abundant element in the Earth’s crust, at about 27.7 per cent.7 Standing there, you could reasonably ask: so what could be the problem?
The honest answer is that “starts as sand” is a raw-material fact, not a safety fact. Between the sand and the gasket, the material is taken apart and rebuilt.
First the sand is smelted. Silica — chemists write it SiO₂, meaning each silicon atom is bound only to oxygen — is heated with carbon in an electric-arc furnace at about 1,500–2,000°C. The heat strips the oxygen away and leaves silicon metal, a grey industrial solid. That metal is not yet rubber, and it is not yet anything you would put on a lunch box.
The metal then meets a petrochemical called methyl chloride: a small carbon-and-chlorine compound made from fossil feedstocks. They are reacted over a copper catalyst, which is simply a helper metal that speeds the chemistry without being used up. This is the Direct process, also called the Rochow–Müller process, worked out in 1940–42. What it builds are carbon–silicon bonds that do not exist in sand. Sand is silicon and oxygen only. The finished kitchen material is an organosilicon polymer — a long, repeating chain with carbon side-groups attached.7
Those new building blocks are hydrolysed, which means they are reacted with water, and then polymerised, which means they are joined into a long chain. The chain is PDMS, short for polydimethylsiloxane: a backbone of silicon and oxygen, with two little carbon groups on each silicon. That flexible, heat-tolerant rubber is what we call silicone.
Here is the turn. The same reaction always leaves behind small leftover molecules. They are cyclic siloxanes. “Siloxane” just means they are built from the same silicon–oxygen units as the long chain. “Cyclic” means those units have closed into rings instead of stretching out into a polymer. The rings that matter here are called D4, D5 and D6 — octamethylcyclotetrasiloxane, decamethylcyclopentasiloxane and dodecamethylcyclohexasiloxane if you want their full names, which you do not need to remember. A smaller cousin, D3, appears as well. They are not extras poured in afterwards. They are leftovers of the chemical equilibrium, trapped in the finished rubber unless a maker strips them out by distillation or a long extra bake. There is no required step that specifically targets D4, D5 and D6. The food-grade volatiles cap does push down the lightest of these species, but nothing names or measures those specific cyclics, or reflects their hazard listing.8
The leftover rings also live by different rules from the rubber around them. The silicone polymer is thermally robust. Its silicon–oxygen bonds are strong, and the material holds its shape well above ordinary baking heat. The leftover rings do not share that toughness. D4 boils at about 175°C, D5 at about 210°C, and D6 at about 245°C — within or near a normal oven range. So while the gasket keeps its shape, the small rings can evaporate and move. Work on silicone baking moulds has shown that they do migrate into food, especially fatty or oily food at high heat.9 How much is dose-dependent: it depends on fat, temperature, and how many times the piece has already been heated, and it drops substantially after the first several uses. That caveat matters. It does not make the first uses disappear, and a lunch-box seal still meets warm food.
It helps, here, to name what that movement actually is. Plastic has already taught most of us that a kitchen polymer can have two different problems, and they are easy to fold into one. Chemicals can leach out of plastic — BPA is the example people know. And tiny pieces of the plastic itself can rub off. Those pieces are called microplastics when they are smaller than five millimetres, and nanoplastics when they are smaller still, down toward a billionth of a metre. A molecule is a single chemical unit, far too small to be a speck you could see. A particle is a physical fragment of the solid material: a chip of the thing itself. They are measured differently. They move differently. Finding one does not prove the other.
Silicone turns out, in principle, to have the same two.
The leftover rings — D4, D5 and D6 — are the chemical side. They are individual molecules. They leach into food and evaporate into air. Scientists weigh them as mass: micrograms per gram of food, micrograms per cubic metre of air. That is molecular migration, not particles wearing off. It is the side that has actually been studied.
The physical side is a different question, and it has barely been asked. Once the silicone has cured, it is a solid rubber — a crosslinked polymer, meaning the long chains have been tied together so they hold their shape. Like any solid polymer, that rubber can wear. Rubbing, scratching or steam can break off micro- and nano-sized particles of the silicone itself. Those particles are pieces of PDMS, the rubber. They are not the leftover rings. Measuring the rings in food, or even finding those ring molecules in the human body, does not tell us whether the particles are there.
What little research exists shows that shedding can happen. One study of silicone baby teats found that steam sterilising — heat near 100°C, not ordinary sucking — released silicone particles down to about 0.6 micrometres into the wash water. The authors estimated that a child could ingest more than 0.66 million of those particles by the age of one. They also said the health risks were unknown.16 Another study scratched a kitchen silicone sealant at room temperature, including a sealant that had already been in place for seven years, and found particles down to about 100 nanometres — a ten-thousandth of a millimetre. Two fingerprinting methods, Raman spectroscopy and energy-dispersive X-ray analysis, confirmed those particles were silicone.17 That is the study that shows heat is not required. It is also a small proof-of-concept on a sealant, not a spatula, a baking mat, or the gasket on a lunch box.
And that is the honest gap. Nobody has measured what a normal silicone spatula, gasket or baking mat sheds in ordinary kitchen use. Nobody has measured food-contact silicone wear particles in the human body. There is no dose-response curve, and there is no particle toxicology. Both studies say the health effect is unknown. Silicone is usually left out of microplastic counts altogether, so the particles are uncounted, not proven absent. Plausible, shown in principle, unquantified for real kitchen use, health impact unknown. That is not a hazard we can name. It is the same pattern as the rest of this piece: we are using the material before we have checked.
The comforting sentence that sometimes follows — if it never touches hot food, it must be safe — does not hold. Heat is the biggest driver of the chemical migration; fat and long contact push it higher. Cyclic siloxanes are known to migrate out of silicone bakeware into food, and that movement rises with fat, with heat, and with repeated or prolonged contact.1819 There is no clean measurement of migration into cold or refrigerated food, so the honest statement is that cold reduces the chemical movement without our being able to call it zero. And the physical rubbing that sheds particles does not depend on heat at all.
So the gasket is not “just sand.” It is a heavily synthesised polymer that can carry residual rings the long chain cannot hold, and a solid rubber that can, in principle, shed pieces of itself. Heat makes the chemical side worse. It is not what creates either side.
What science has already mapped — and what it has not
None of that, by itself, tells us the rings are a problem for people. For that we need the pathway science has already drawn, and we need to keep it honest.
In June 2018 the European Chemicals Agency’s Member State Committee identified D4, D5 and D6 as Substances of Very High Concern.1 That is a formal classification, not a warning sticker on a lunch box. It rests on two properties that are easy to say and hard to live with.
They are very persistent, which means they do not break down in any useful time in the environment. And they are very bioaccumulative, which means they build up in wildlife rather than leaving the way ordinary water-soluble things do. That classification is environmental: it is about not breaking down in nature and building up in living things outdoors, not a claim that the molecules accumulate forever inside a person. In people these siloxanes are volatile and are cleared over days to weeks. D4 is also classed as persistent, bioaccumulative and toxic. D5 and D6 meet that third test when they contain 0.1 per cent or more of D4 — a reminder that these leftovers tend to travel as a family.1
There is an animal signal, and it belongs in a careful box. In a chronic high-dose inhalation study, D5 produced a small rise in a kind of uterine tumour in rats, through a route that did not damage DNA. The proposed mechanism involves dopamine and the hormone prolactin, and researchers have argued it is specific to rats.1011 D5 is not genotoxic and is not directly oestrogenic: it does not simply mimic oestrogen. Canada treats all three as persistent and bioaccumulative, and Health Canada did not find D5 and D6 harmful to human health at the exposures it assessed.6 Molecules of this kind have also been detected in the human body. Those detections are of the leftover-ring molecules, largely explained by personal-care and inhalation exposure; they have not been traced to food seals. Food-contact silicone wear particles in the human body have not been measured. They are a different measurement, and that study has not been done.
This investigation continues below.
Want the next one in your inbox?
A mapped pathway is not proof of human harm. The absence of a long-term human causal study is not proof of safety. That missing study is the reason to pay attention, not the reason to relax. What we can already see is the shape of the concern: environmentally the rings persist and they accumulate; they can leave a silicone part into food and air; they have been found in people, largely from personal-care and inhalation exposure rather than from food seals; and one high-dose animal study has raised a bounded flag. The last link — a long-term human study that could settle everyday kitchen use — has not been done.
The physical side sits one step further back. The rings have a mapped chemical pathway. The particles have a principle — solid rubber can shed, and two studies have seen it happen — and then an empty page. Uncounted is not the same as absent, and it is not the same as proven safe. Keeping those two questions separate is part of keeping the pathway honest. The molecules in people are real. The food-contact wear particles in people are unmeasured. Neither silence is a verdict.
Why the rules meant to protect us do not close this gap
If the pathway is visible, why does a gasket still arrive labelled as if the question were settled?
Because “food-grade” does not mean what it sounds like. In Europe the de facto benchmark is a German recommendation from the Federal Institute for Risk Assessment, BfR Recommendation XV on silicones, revised on 1 February 2023. It says silicone elastomers — the rubbery form used in gaskets and moulds — must hold no more than 0.5 per cent volatile organic components, a residual-content cap measured as weight loss, and must release no more than 0.5 per cent extractable components.2 Extractable simply means whatever comes out of the rubber when it is soaked or heated in a test liquid, added up by weight. In the United States the matching rule is FDA 21 CFR 177.2600, written for rubber articles intended for repeated use. It sets generic total-extractables limits: 20 milligrams per square inch in water for the first seven hours, and 175 milligrams per square inch in hexane, a stand-in for fatty food, for the same period.3 The US FDA rule, 21 CFR 177.2600, does not name D4, D5 or D6. The German BfR recommendation does list cyclic siloxanes as permitted starting materials, but it sets no D4-, D5- or D6-specific migration limit and never mentions the very-high-concern listing, even though it was revised years after that 2018 finding.
So food-grade means the part passed a generic total-weight extraction test. It does not mean anyone measured the leftover rings.
The chemicals regulator did act. On 27 June 2018 those three molecules were formally identified as Substances of Very High Concern. In May 2024, Regulation (EU) 2024/1328 amended REACH, the EU’s main chemicals rulebook. It restricted them in certain releasing uses — mixtures such as cosmetics and cleaning products — for environmental reasons. The main restriction applies after 6 June 2026, with later dates for a few specialised uses.1 A cured food gasket is a different case, and it sits outside that restriction. The regulation’s stated reason is environmental — these substances persist and build up once released from products into water and air — not a finding that they harm people through food.
Food-contact law never took that very-high-concern identification into account at all. There is still no EU-wide, silicone-specific food-contact measure. The framework regulation, (EC) 1935/2004, says food-contact materials must not transfer their constituents in amounts that endanger health, then leaves the detail to national rules.4 Two rulebooks, written by different people for different harms, were never reconciled. One side can name a Substance of Very High Concern. The other can keep calling a gasket food-grade because a mixed 0.5 per cent of extractables came in under the line, without acknowledging that finding.
That split is difficult, and it is not a quirk of one lunch box. Different institutions were asked different questions. Chemicals law was looking at what happens when these rings escape into water and air and stay there. Food-contact law was looking at how many milligrams of mixed stuff come out of a rubber in a laboratory soak. Nobody was assigned to join those answers, so the kitchen label never had to hear the very-high-concern finding.
The physical side has no home in either book. The food-contact tests weigh a mixed extract. They are not a particle count. And the European rule written for microplastics does not catch wear. Regulation (EU) 2023/2055 restricts only microplastics that are intentionally added to a product.20 Particles that rub off an article — silicone or plastic — sit outside that rule entirely. The gasket’s leftover rings are unnamed by food-grade. The gasket’s wear particles are unnamed by the microplastic restriction. Two silences, stacked.
Safety numbers in this field also move only after years of being told a material is fine. Take BPA, a different chemical used for years in some plastics and can linings. It is not silicone, and D5 is not BPA; the useful comparison is how official comfort gets revised. In 2006 the European Food Safety Authority set a tolerable daily intake — the amount thought safe every day over a lifetime — at 50 micrograms per kilogram of body weight. In 2015 that figure became 4. On 19 April 2023 it became 0.2 nanograms per kilogram, twenty thousand times lower than the 2015 number.5 Typical dietary exposure, the Authority said, now exceeds the new safe level. Other official bodies still disagree with that revision. The fact that remains is simpler: the official safe number moved by orders of magnitude after decades of reassurance.
Microplastics tell a similar story about knowledge, not about chemistry. They were long assumed inert. They have since been documented in human placenta, blood, brain and arterial plaque.12131415 Harm at those levels is not established. The arterial finding is an association — two things seen together — not proof that one caused the other, and it is contested. The pattern is the point. “We have no long-term human proof” is exactly how comfort keeps getting sold while a pathway sits in plain sight. Silicone particles are not part of that body of evidence. They are usually left out of the counting. The honest reading is not that they have been cleared. It is that the experiment is still running, unmeasured.
A choice that does not need the final study
Come back to the lunch box. The reviewer who called the choice personal taste was doing what most of us do: trusting that food-grade silicone is safe because the label says so. That trust was never earned. The food-grade stamp never measured the leftover rings already listed as substances of very high concern, and nobody has studied what a gasket sheds when you close it, wipe it and stack it. Behind the word there are more open questions than answers.
You do not have to wait for the finished study to act on that. A box already exists that simply takes the material out, and once you can see it, the choice is no longer a preference. It is the obvious one.
Keep the metal. Choose the silicone-free container when you can — a steel friction-fit like this one, which grips shut steel on steel and is not airtight, or a seal that is not silicone. The silicone ring is what makes the other kind leakproof. You give up a little of that airtightness, and you leave the open questions behind.