A migration test result describes only the article that went into the laboratory, under the simulant, temperature and contact time written on the request, so an over-limit number has to be checked against the protocol first and traced to a component second. In a glass pack the source is nearly always an organic part, such as the liner, gasket, coating or printing ink, and very rarely the glass body. The way out is to identify the substance, isolate the component that contains it, change one thing, and re-test under exactly the protocol that failed.
Overall migration and specific migration answer different questions
Two measurements sit under the single label "migration test", and a dispute with a supplier often turns out to be about the wrong one.
Overall migration is the total mass of non-volatile material that moves from the article into food or a food simulant. No substance is named; it is a sum. Laboratories report it either per unit of contact area or per kilogram of food, and those two expressions cannot be swapped, so check that the report and your specification use the same one. The European plastics measure sets its overall limit per square decimetre of contact surface, with a per-kilogram equivalent for some article types. Treat it as a screening gate. It says nothing about toxicity.
Specific migration puts a limit on a named substance or family of substances: monomers, certain additives and plasticisers, photoinitiators released from a cured ink or coating, primary aromatic amines that come from pigments. Because one substance with a low limit is a much sharper test than a total mass, this is where most fails occur. A closure can clear overall migration with room to spare and still fail on a single photoinitiator, and nobody finds out unless the specific test was ordered.
Three other documents are often offered under the same loose heading of "food contact testing". None of them replaces a migration report when a specification asks for one.
- An extraction test stresses the material harder than migration does. It is a good screen and nothing more.
- A sensory (organoleptic) assessment tells you whether the pack altered the taste or smell of the food.
- A composition statement lists what the material contains, not what comes out of it.
A supplier who sends one of these is not necessarily misleading you, but an auditor with experience will spot the substitution almost at once.
Five test conditions to fix in writing before the sample is sent
Migration is mass transfer, so it depends on time and temperature. A pack that is clean after two hours at forty degrees Celsius may not be clean after ten days at forty degrees Celsius, nor after two hours at seventy degrees Celsius. The same article can pass one protocol and fail another with no laboratory error involved, because each protocol describes a different use. That is why the result belongs to the bottle plus its intended use, never to the bottle alone, and why each of the following has to be agreed on paper beforehand.
| Condition | What has to be decided | Where buyers go wrong |
|---|---|---|
| Food simulant | The laboratory liquid that stands in for the food class: water for aqueous foods, dilute acetic acid for acidic ones, ethanol at a defined strength for alcoholic products, vegetable oil or a synthetic substitute for fatty foods, a solid porous simulant for dry foods | Assuming clearance carries across categories. An acidic hot-filled sauce and a neutral spirit at forty per cent alcohol need different simulants |
| Temperature | The regime that matches the real process: short hot fill, pasteurisation, sterilisation or long ambient storage | Testing the mild condition to save cost, then being contradicted by a harsher test later |
| Duration | Set together with temperature; the pair is meant to compress years of contact into hours or days | Describing average exposure. The protocol should cover the worst realistic combination |
| Contact geometry | The ratio of wetted surface to fill volume, and whether the article is fully immersed or contacted on one side only | Forgetting that a small bottle with a large internal surface is a harsher pack than a large one, and skipping single-sided testing because it costs more |
| Finished pack or material sample | Whether the whole assembled article is tested or a coupon of one material | Using a coupon result to support a claim about the finished pack |
The simulant is mapped from the food category, not chosen by preference, and that mapping is the most frequent point where buyer and laboratory misunderstand each other. On temperature and time, remember that a product kept for a year at room temperature and the same liquid held at seventy degrees for two hours during filling are two separate exposure profiles.
The last row deserves a decision of its own. A finished-pack test puts closure, gasket, coating and food in one vessel, which is what a retailer's technical team wants to see. A coupon is quicker and cheaper and works well as a first screen, but it cannot reproduce the geometry, the closure torque or the cure state of the ink on the real article.
Which component of a glass pack carries the risk
Glass is an inorganic fired material. It is not formulated from monomers, stabilisers and antioxidants, so it does not raise the positive-list question that polymers do. What it can release is metal, which is measured by a different method against a different limit. If the budget allows one test and it is spent on the glass body, it has gone to the least informative place.
The organic parts are the reverse. Their constituents are listed and limited, and they are where overall and specific migration testing pay for themselves. From the outside in, a typical pack has:
- the glass body, whose relevant risk is metal release;
- any decoration on the body or shoulder, which is a further organic layer when it is a fired organic ink or coating instead of an oxide colour;
- the closure shell, usually coated or varnished on the food side if it is metal;
- the liner or gasket inside the shell, where most failures start, because a plasticised liner, a foam liner, a pulp or board liner and a silicone gasket are four chemistries with four failure modes;
- for dispensing formats, the internal assembly of tube, ball, spring and elastomer.
Printed parts add the question of set-off: ink components moving from a printed outer face to the food contact face of a neighbouring article when packs are stacked or nested.
Reading the detected substance back to a component
Investigate by chemistry, not by how prominent a part looks. The useful first question after a fail is which component contains the substance the laboratory found.
| Substance detected | First suspect |
|---|---|
| Photoinitiator | A cured ink or coating |
| Plasticiser | The liner |
| Metal ion | The glass, its colourant or a coated shell |
| Volatile organic compound | Possibly residual solvent in an adhesive or coating |
Sampling the suspect directly is shorter and cheaper than commissioning a wide screen, which confirms that something is present without saying where it came from.
Why failures tend to surface late
Migration tests are ordered when somebody asks for a document: a retailer onboarding a supplier, a confirmed purchase order, a shipment waiting at a border. By then the shape is fixed, the mould is cut, closures are on order, artwork is signed off and a filling trial is booked. An over-limit number at that stage is a schedule problem competing with every other schedule problem in the launch.
The second cause is structural. The person who orders the test frequently does not control the part that fails. A packaging buyer submits the glass body, gets a clean report and concludes the pack is clean, while the liner, gasket, coating and ink come from three other parties who were never asked. Much of what goes wrong after filling traces back to components that were not in the beaker.
Both causes point to the same habit: write the protocol when the specification is written, and list every food contact component by supplier.
From an over-limit result to a released batch in seven steps
The sequence below is for the buyer running the corrective action on a product that still has to ship.
- Confirm the failure is real. Check that the sample came from production and was not a prototype, that simulant and conditions matched the intended use, that the laboratory used the method named in the specification, and that the chain from filling to analysis is documented. A fair number of apparent fails are protocol mismatches, and a material change launched on one of those wastes money and time.
- Identify the migrating species. Knowing only that the overall figure was high leaves the problem open-ended. The identity of the substance reduces it to a shortlist of materials.
- Tie the species to a component and prove it. Test the suspect liner, ink or gasket alone. That turns a hypothesis into evidence.
- Choose a remedy. The set is wider than most buyers expect: a different material for the component, a different formulation or grade within the same family, a different surface treatment, different curing or drying conditions, or a reformulated product that attacks the pack less. The last is often possible and seldom raised, because packaging and product development report to different departments.
- Implement under change control. For a food contact material the change creates a fresh conformity position and often a fresh declaration. Revising a drawing is not enough.
- Re-test. Use the new configuration, samples drawn from production, and the identical protocol. Where the margin was narrow, run the harshest realistic condition instead of the minimum, so the retailer's next question already has an answer.
- Decide on release. A passed test does not release anything by itself. The decision has to cover stock already made in the old configuration, material still in the pipeline, and a written record of what changed and why.
Remediation options by source of the failure
Use this table for planning only. Limits, simulants and methods depend on the destination market and are amended over time, so confirm the actual figures for your market at the time of shipment.
| Source | Typical cause | Ways to fix it | Re-test scope | Interim release of existing stock |
|---|---|---|---|---|
| Closure liner | Formulation heavy in plasticiser or additives, or a grade picked on cost for an aggressive fill | Same-size liner from another polymer family; foam or pulp liner if seal and barrier permit | Finished pack with the new liner at the harshest realistic condition, with seal integrity and torque checked as well | Possible when one substance is involved, the market accepts a documented corrective action plan while stock moves, and the retailer agrees in writing |
| Printing ink or coating on the pack | Ink not fully dried or cured, a photoinitiator system that does not suit the substrate, or set-off from stacked printed surfaces | New ink chemistry or photoinitiator package; revised curing or drying; an inert over-varnish that takes the printed layer out of food contact | Migration or set-off on the finished decorated article, and a colour match against the approved standard | Seldom clean. With set-off the defect sits on the food contact face, so the retailer's written position comes first |
| Decoration on the glass body | Fired organic decoration, applied coating or heavily pigmented layer near the filling line where an inert oxide colour would have served | Shift the decoration out of the food contact zone; replace it with an inert ceramic or oxide colour; remove it from the affected area | The decorated area that touches food, the whole article where appropriate, and appearance approval against the original sample | Only if evidence, not assertion, shows the decoration lies outside the food contact zone |
| Gasket or elastomer in a dispensing closure | Silicone or rubber grade whose cure system or additive does not suit the fill, or a part with no food contact position of its own | A grade with a documented food contact position; a different cure system; a redesign that keeps the elastomer away from the product | The complete closure in operation, for migration and function together | Not advisable, since product stays in contact through the dip tube and outlet path |
| Metal shell coating or internal varnish | Film chosen for looks or corrosion resistance and never assessed for migration, or under-cured on the inside | A coating with a food contact position; more cure; a liner that fully separates metal from product, making the shell a non-contact part | The closure as supplied, liner fitted, at the worst realistic condition | Possible where the liner is a complete, verifiable barrier and the fail is traced to the shell face |
| Glass body colourant | Colour produced by a metal-bearing oxide system, or a raw material lot with an unexpected metal load | Reduce or change the colourant package; go to a lighter shade; keep the colour and tighten raw material control | Metal release from the finished article by the method for glass hollowware, not overall migration | Only with a market-specific assessment and an agreed sampling plan, because exposure in most uses is cumulative, not acute |
For any closure change, remember that a safer material that leaks has not solved the problem.
Five mistakes that turn remediation into a loop
Projects that drag on usually do so through lapses in discipline, not through hard chemistry.
Changing two variables at once
Swap liner and ink together and a pass leaves you unsure whether one change was needed, while a fail leaves you unsure which to revisit. Change one suspect per cycle, even though it means one more test.
Re-testing under a different protocol
This creeps in when a second laboratory is used for speed and applies its own standard regime. The result reads as a pass and says nothing about the original fail. Hold the protocol constant, and if it must change, justify that as a separate decision.
Treating a substitute as like for like
Two liners can look the same and fit the same shell yet migrate quite differently, because the difference is in the formulation and not on the drawing. Without a fresh conformity position from the component supplier, a substitution is an unverified change.
Relying on an old declaration
A declaration covers a defined article, use and market. One written for a liner in a neutral beverage does not extend to that liner under sixty per cent alcohol, and one issued years ago may describe a formulation since revised. Request the current document, addressed to the article and use on your purchase order.
Shipping before the corrective action closes
The pressure to ship is real and usually justified. A release based on a supplier's verbal assurance, though, creates a liability that lasts longer than the shipment. If stock must move, use a documented conditional release naming the destination, the shelf life and the action to take if the re-test fails, and obtain that agreement in writing from the customer, not the supplier.
What a pass report must show before you release
A report counts as evidence only if someone outside the conversation can read it. At a minimum it names:
- the article tested and the material composition it represents;
- the production batch or sample date;
- the simulant, temperature and contact time;
- the method applied;
- the laboratory and its accreditation.
Leave one out and you cannot show that the test describes the goods on the purchase order. A reviewer will record the gap as a finding.
Look at the margin as well as the verdict. A figure just under the limit passes on paper and is a risk in practice, since production and raw material lots vary. When the result is close, widen the margin with a material change, add a routine incoming check on the sensitive component, or lock the tested material to one supplier and grade. A thin margin combined with a material source that moves between suppliers will come back to you.
Finally, check where the samples came from. Laboratory-made coupons and hand-assembled prototypes do not represent production, and for closures the difference is practical: applied torque, liner compression and ink cure all change the outcome. Sample from the line at commercial settings and record those settings so a later change is visible.
What to send us, and where related questions are covered
When we help define a protocol or respond to a fail, we need three things to draft a test configuration and a remediation path for your product, with a note on the component most likely to be the source:
- The contents: alcohol strength, acidity, fat content and any heat treatment.
- The use conditions: fill temperature, storage temperature and intended shelf life.
- The destination market, since the limit, the simulant and the required document all change with it.
If a closure, liner or decoration may need to change, add the current component specification so an alternative can be compared against it.
A protocol is written against a specification, and a specification that is still moving cannot be tested, so buyers who have not settled on a shape should first choose from the glass bottle and jar range. Two neighbouring subjects have their own guides. Which declaration each market requires, who may sign it, how long it remains usable and how the file is assembled for an onboarding review is covered under the food contact declaration for glass bottles. Lead and cadmium release from the glass, from colouring oxides or from fired decoration is covered under heavy metal release testing. For packs with a decorated closure, sleeve or capsule, the reference on wine bottle capsules deals with that component class.
Frequently asked questions
How do overall and specific migration testing differ?
Overall migration is a total: all non-volatile material transferred to the food or simulant, stated per unit of contact area or per kilogram of food, with no substance named. Specific migration measures one named substance or family, such as a monomer, a plasticiser or a photoinitiator from cured ink, against a limit of its own. Passing the first does not rule out failing the second, so an overall figure alone is incomplete evidence.
How should I pick the simulant and the test conditions?
Start from the food category and the most demanding use the product will really meet. Aqueous, acidic, alcoholic, fatty and dry foods each have a defined liquid or solid substitute. Temperature and time follow from the toughest of hot fill, pasteurisation or sterilisation, or long ambient storage. Put every condition in writing before sampling and do not alter any of them for a re-test.
Does the glass body need a migration test?
Metal release is normally the more informative test for the glass, carried out by the method written for hollowware. Migration risk concentrates in the organic components: liner, gasket, coating, printing ink and any fired organic decoration. Spend a limited budget there first.
The liner failed. Can the cap stay and only the liner change?
In many cases yes, and it tends to be the least costly fix available. The new liner has to match the old dimensions, seal on the same neck finish at the same application torque, and exist in a grade with its own food contact position. You still need a new conformity position from the component supplier and a seal integrity check on the finished pack, because a compliant liner that cannot hold pressure or protect the product fixes nothing.
How long does remediation take?
It is set per project by the material with the longest qualification path, not by the laboratory work. Replacing a liner in an existing shell is normally the shortest route. An ink or coating change takes longer because of the qualification run and, often, colour re-approval. A colourant or decoration change on the glass is the longest, since it involves a mould trial and the brand. Allow for one full test cycle after the change at the least, and treat any schedule as indicative until both the material and the laboratory are fixed.
Can stock made before the corrective action be shipped?
Only on a documented decision. You can hold and re-test it, ship it under a conditional release with a defined destination and the customer's written agreement, or rework it where the faulty component can be replaced. A verbal assurance about the replacement material is not a defensible basis, because liability rests with the brand owner who placed the article on the market and not with the component supplier.
What does a re-test report need in order to close the corrective action?
A precise description of the new configuration, the article and its production batch or sampling date, the simulant, temperature, contact time and method, and a laboratory whose competence can be verified. It should give the numerical result, not just pass or fail, so the margin can be followed over later batches. If the original fail was a specific migration finding, the report must cover that same substance so the two results can be compared directly.