XRF Screening for Restricted Substances in Waterproof Bags: What a Handheld Analyser Can and Cannot Prove

What a handheld XRF analyser physically measures

Most buyers meet XRF screening as a device pointed at a buckle that returns a parts-per-million number, and in restricted substances work that number is read as though it were a compliance verdict. Understanding the physics takes five minutes and changes how the number is used. The instrument irradiates a small area with X-rays, the atoms in that area emit secondary X-rays at energies characteristic of their elements, and the detector resolves those energies into a spectrum. The output is an elemental composition, weighted towards the near-surface region the X-rays reached.

Three physical facts follow, and all three constrain interpretation. It measures elements, not molecules: a reading for bromine means bromine atoms are present, not that a specific brominated flame retardant is present, and a reading for chromium says nothing about whether any of it is the hexavalent form that regulations restrict. It has an element floor: light elements below magnesium are not detected in practice, which excludes boron, carbon, nitrogen, oxygen, fluorine and sodium. And it samples a small volume: a spot of a few millimetres to a centimetre across, to a depth that depends on the matrix.

The element floor is the single most consequential limitation for this product category, and it deserves stating plainly. Fluorine is light, so a fluorinated water-repellent finish is invisible to XRF. Chlorine can be detected but is a poor proxy for PVC because chlorine has many other sources. Carbon and hydrogen are invisible, which means every organic restricted substance — phthalates, azo colourants, bisphenols, organotins, perfluorinated compounds, formaldehyde — is entirely outside the technique. A bag can be perfectly clean by XRF and fail a wet chemistry screen comprehensively.

What the instrument is genuinely excellent at is metals in dense matrices: alloys, platings, coatings with heavy pigments, and polymers loaded with metal-containing additives. Those are exactly the places where lead and cadmium turn up, and for those applications a well-run screening programme is fast, cheap and remarkably informative. The technique is published and standardised, and the underlying method descriptions are available from standards bodies such as ASTM International, which publishes the relevant practice for portable analysers.

The core judgement: XRF screens, it does not confirm

Everything else in this guide follows from one sentence, and it is the sentence buyers most often get wrong in both directions. A positive XRF result is a candidate, not a finding. A negative XRF result is a partial clearance of the elements measured, not a clearance of the product. Neither is a compliance determination, and treating either as one produces both false alarms and false comfort.

Why a positive needs confirmation is partly legal and partly analytical. Legally, the limit in most regimes is expressed against a defined test method applied to a defined material, and a screening result from a handheld device is not that method. Analytically, the screening number carries uncertainty from matrix effects, from surface roughness and curvature, from coating thickness, and from spectral overlap between elements whose X-ray energies are close. A screen that reads lead just over a threshold has a real chance of reading under it in the laboratory, and a screen just under has a real chance of reading over.

The practical way to set the threshold is to screen at a fraction of the legal limit rather than at the limit itself. If the applicable limit is one hundred parts per million, screening everything above fifty or sixty parts per million as a candidate for confirmation gives margin for that uncertainty and catches the cases that matter. Screening at the limit itself means every borderline result becomes an unbudgeted laboratory test, which is how screening programmes lose their economic advantage.

There is also a direction of error worth naming explicitly, because it is the expensive one. A false positive costs a laboratory test and some nervousness. A false negative ships product. Screening programmes should therefore be designed to be deliberately conservative, accepting more confirmations in exchange for fewer misses, and the threshold should be set with that asymmetry in mind rather than with symmetry.

What XRF cannot see, and why a negative clears so little

The clearest way to explain the limitation to a non-technical buyer is to list the restricted substances that matter in a waterproof bag and mark which are elements and which are molecules. The elements are a short list and XRF handles them well. The molecules are a much longer list and XRF cannot see any of them. Since most restricted-substance lists are dominated by organic compounds, a clean XRF result addresses a minority of the risk.

• Phthalate plasticisers in PVC films, coatings and prints: organic, invisible to XRF, restricted at 0.1 per cent by weight in several regimes.
• Azo colourants that can cleave to release listed aromatic amines: organic, invisible, and a standard parameter in textile restricted substance lists.
• Per- and polyfluoroalkyl substances used for water repellency: contain fluorine, which is below the element floor, so invisible.
• Formaldehyde and formaldehyde-releasing resins: organic, invisible.
• Organotin compounds in coated films and foams: contain tin, which XRF can detect, but the elemental reading does not identify the compound.
• Hexavalent chromium: chromium is detectable as an element, but XRF cannot distinguish the trivalent from the restricted hexavalent form.

The chromium and tin rows are the ones that trip up technically competent people, because they look like XRF applications and are not. Total chromium is measured easily; the legal limit in most frameworks attaches specifically to hexavalent chromium, which requires an extraction and a wet chemical determination. A high total chromium reading is a reason to test for hexavalent chromium, not evidence of a failure, and a low total reading is a reason to conclude very little. The same logic applies to tin and to bromine.

This is why the honest summary of a clean XRF screen is narrow and specific: no lead, cadmium, mercury or total chromium above the stated detection limits, in the specific materials and colourways screened, on the specific samples submitted. That is a useful statement. It is not a statement about phthalates, azo colourants, fluorinated finishes, formaldehyde or organotins, and it should never be presented as one. The broader substance landscape is set out in our guide to chemical transparency in manufacturing.

Lead, cadmium, mercury and chromium: where the risk physically sits

Heavy metal risk in a bag has a physical address, and knowing the addresses is what converts screening from a random activity into a plan. The general pattern is that metals arrive as pigments, as stabilisers, as catalysts, as alloying elements and as plating or passivation chemistry. Each of those has a typical home in the bill of materials, and each home has a typical screening response.

The plating row deserves expansion because it is the single best-return application of handheld XRF in this category and the one most likely to find something real. A decorative finish of a few microns does not stop the analyser, which means a chrome-plated buckle can be screened and the lead content of the brass beneath it read directly. It also means the plating thickness itself can be estimated, which is useful on its own because thin plating correlates with the corrosion failures discussed in our guide to hardware corrosion testing.

The pigment row carries the most historical risk and is the reason colour matters so much in a sampling plan. Strong yellows, oranges, reds and greens are the shades historically associated with lead- and cadmium-based inorganic pigments, and they remain the sensible priority when the screening budget is limited. Black, navy and white are lower risk for metals and higher risk for certain organic colourant issues, which is a different test entirely. Choosing hardware and finishes with this in mind is part of hardware selection rather than an afterthought at inspection.

Sampling strategy: split by material and by colour

The most common mistake in a screening programme is to sample by bag. An operator takes one unit off the line, fires at half a dozen points and records a result set, which produces a tidy table and a badly biased answer. The correct unit of sampling is the material-colour-supplier combination, because that is the level at which chemistry is decided upstream. Two black fabric panels from different mills are two different materials regardless of the bag they end up on.

• Build the list from the bill of materials, not from the finished product: shell, coating, tape, print, webbing, hardware by type, foam, thread, labels.
• Split every material by colourway. Screening one black and assuming it covers a yellow is the most frequent and most consequential shortcut.
• Treat each supplier as a separate population. The same nominal specification from two mills is two screening populations.
• Screen incoming components before assembly where practical, because a hardware item is far easier to read flat than once it is sewn into a bag.
• Record the lot or batch reference with each result, so a later failure can be traced to a delivery rather than to a product line.

There is a mechanical reason to prefer components over assemblies, and it is worth stating because it improves data quality for free. The analyser wants a flat surface, close contact and enough thickness that the X-rays do not read through to whatever is behind. A buckle on a bench satisfies all three. The same buckle attached to a bag sits on a soft, curved, possibly wet substrate, with the analyser unable to seat properly, and the resulting spectrum is degraded by the material behind it. Screening loose components is faster and produces better numbers.

Sample count is where judgement replaces rules. For a homogeneous metal component from a single lot, a handful of shots across several pieces is sufficient. For a printed panel, where the pigment layer is thin and possibly uneven, more shots across more pieces are needed, and the result should be treated as indicative. For a foam or a coated fabric, the density is low enough that the reading is best treated as a screen rather than a measurement. Recording the count and the logic is what makes the result defensible later.

Metal hardware and plating: the best use of the tool

If a programme can only afford to screen one thing, screen the hardware. Metal components are where metals are, the matrix is ideal for X-ray fluorescence, the surfaces can usually be presented flat, and a finding is actionable in a way that a fabric finding often is not — a buckle with leaded brass can be replaced at modest cost, whereas a non-compliant coating may require requalifying the whole shell supply.

• Screen by hardware type and by finish: a black-coated slider and a bright chrome slider are different materials.
• Screen the substrate through the plating, and record plating thickness where the instrument reports it.
• Screen the smallest components separately. Rivets, eyelets and small D-rings are frequently bought from a different source than the main hardware and are frequently the worst offenders.
• Screen incoming lots, not finished bags, and keep the results against the lot reference.
• Re-screen whenever the hardware supplier or finish changes, which is the highest-risk substitution in the whole bill of materials.

The small-component point is worth emphasising because it is where programmes consistently get caught. The main buckles and sliders are usually specified and usually sourced from a named supplier with documentation. The rivets, eyelets, washers and end caps are bought by the piece, often from a local hardware market, and they are exactly the items where leaded brass and unrecorded plating are most common. They are also small enough that a spot measurement needs a collimator to be meaningful, which is a cheap accessory and a frequent omission.

One caution applies even here. A reading on a small, curved or rough surface carries considerably more uncertainty than a reading on a flat one, and the instrument will often report this as a wider error or fail its own quality check. Those results should be recorded as inconclusive rather than quietly discarded, because a discarded inconclusive reading is the mechanism by which a screening programme develops a blind spot. Where hardware carries a legal limit directly, confirmation testing is the answer, and the wider framework is set out in our guide to REACH and CPSIA compliance testing.

Coatings, prints and pigments: where false negatives hide

Coated and printed surfaces are the second priority and the place where the technique is weakest, which is an uncomfortable combination. The pigment is where lead and cadmium sit, but the pigment layer is thin, sits on a polymer or textile substrate of quite different density, and is often applied unevenly or textured. All of that degrades the measurement, and a degraded measurement in a screening context tends to bias towards a lower reading rather than a higher one.

• A thin layer over a low-density substrate can be partially transparent to the instrument, so the reading reflects a mixture rather than the layer of interest.
• Textured, embossed or curved print cannot seat against the analyser window, and the air gap attenuates the signal, particularly for the lower-energy lines.
• A print on a coated fabric means two unknown layers, and the instrument reports whatever combination it saw.
• Multilayer laminates and transfer prints can put the pigment of interest beneath a clear film that further attenuates the signal.

The practical mitigations are straightforward and cheap. Use a small-spot collimator so the measurement covers print rather than the substrate beside it. Increase the count time, since detection improves with the square root of measurement time and a doubling of time buys a meaningful improvement. Take multiple shots across the printed area and treat the highest defensible value as the screening result rather than the average. And where the print is a strong yellow, orange, red or green on a product with a legal limit, skip the inference and send it to the laboratory.

There is a second reason prints deserve laboratory attention regardless of the metal result. Printing inks are the component most likely to contain plasticisers and solvents, and it is applied after any fabric certification was issued, which puts it outside the scope of the certificate on the shell. The metal screen and the organic screen are answering different questions about the same panel, and doing only the cheaper one leaves the larger exposure open. The adhesion and durability side of printed layers is covered in our guide to print ink adhesion on coated fabrics.

How screening maps onto REACH, CPSIA and RoHS limits

Screening only makes sense against a limit, and the limit depends on the regulation and the product. Three frameworks come up repeatedly for bags, and each expresses its limits differently, against different materials, with different test methods. Mapping them onto what the analyser measures is the step that determines whether a screening number means anything.

Two structural mismatches recur and both matter. The first is total versus extractable or migratory: several limits attach to what comes out of the material under a defined extraction, not to what is in it, and XRF measures what is in it. The second is speciation: chromium, tin and bromine readings are elemental, while the restrictions attach to particular compounds or oxidation states. In both cases the screen is a triage tool that tells you where to spend laboratory budget, not a determination.

There is one place where a screen is close to determinative, and it is worth using confidently: lead in a homogeneous metal alloy. A clean non-detect on a brass buckle, with a stated detection limit well under the applicable threshold, on a flat surface, with an instrument that passed its calibration check, is strong evidence. Everywhere else, treat the result as triage. The current statutory text and the accepted laboratory framework are published by the US Consumer Product Safety Commission, and the European substance restrictions are published by the European Chemicals Agency.

Reading a result: detection limits, matrices and interference

A screening result without its detection limit is not interpretable, and this is the most common defect in the spreadsheets suppliers send. A non-detect at a detection limit of five hundred parts per million tells you almost nothing about a hundred parts per million threshold. A non-detect at ten parts per million tells you a great deal. The number that matters as much as the result is the limit the result was measured against, and it should be recorded in the same cell.

• Detection limit depends on count time: longer measurement improves it, roughly with the square root of time.
• Detection limit depends on matrix: heavy elements in a light polymer matrix are harder at low levels than in a metal alloy.
• Spectral overlap is real: some element lines sit close together, and the instrument’s deconvolution can misattribute counts, most famously where arsenic and lead lines interfere.
• Calibration matters: the analyser should be checked against a certified reference material at the start of a session and the check recorded.
• Surface condition matters: paint, lacquer, dirt, moisture and oxidation all change what the instrument sees.

The overlap point is worth one concrete example because it is the one that generates false positives in practice. Where a spectrum contains arsenic, an instrument can attribute some of that signal to lead, because their characteristic energies are close. A screening result showing modest lead on a material that also shows arsenic should be treated as suspect and confirmed rather than escalated, and a competent operator will flag it. Instruments differ in how well they deconvolve, which is one reason the same sample can give different numbers on two devices.

The operational conclusion is that a screening record should contain more than a number. Element, result, detection limit, count time, material description, colourway, supplier, lot reference, instrument model, calibration check reference, operator and date. That is ten fields, it takes seconds to capture, and it is the difference between a record that survives scrutiny and a spreadsheet that cannot be defended. Accreditation and what it implies for laboratory work generally is discussed in our review of third-party testing laboratories.

Designing a screening plan: how many shots and where

A screening plan is a budget allocation problem. There is always more to screen than there is time to screen it, so the plan has to rank. The ranking that works in this category is by consequence and by likelihood: how bad is a positive in this component, and how likely is one. A leaded rivet on a children’s product is high on both. A foam insert is low on likelihood and moderate on consequence. A white shell fabric is low on both.

• Rank hardware first, especially small components and anything plated over brass.
• Rank strong pigment colours next: yellows, oranges, reds and greens in coatings, prints and films.
• Rank PVC and PVC-blend films next, but remember the plasticiser risk needs a different test.
• Rank fabric and webbing low for metals, and do not let a clean metal result stand in for organic screening.
• Allocate confirmations in advance: budget for a fixed number of laboratory tests per season so that a positive does not become a crisis.

Shot counts follow from homogeneity. For a homogeneous metal part from one lot, five to ten shots across three to five pieces is a reasonable default and will find a contaminated lot. For a printed or coated textile, ten to twenty shots across more pieces, with the highest defensible reading taken as the screen result. For anything where the instrument reports a quality or matrix warning, discard the reading and either rescreen properly or send the item to the laboratory rather than averaging it in.

Timing is the other half of the plan. Screening at incoming inspection is far more useful than screening at pre-shipment, because a positive found on a component delivery can be resolved by rejecting that delivery, whereas a positive found on finished goods means either scrapping or shipping with a known issue. Programmes that screen incoming also build the supplier-specific history that makes later screening cheaper, because a supplier with a clean record over several lots needs less attention than a new one. That record-keeping discipline is the same one described in our pre-shipment inspection checklist, applied earlier in the process.

What happens after a positive screen

A positive screen starts a defined sequence, and having the sequence written down before the first positive is what prevents both panic and complacency. The first step is to confirm the reading is real rather than instrumental: rescreen the same spot, rescreen another piece from the same lot, check the calibration record, and check whether an interfering element is present. A large fraction of positives do not survive this step.

• Rescreen and check the instrument state before escalating. Do not send a single unverified reading to a laboratory.
• If the reading survives, quarantine the lot and identify its scope: which components, which colourways, which orders.
• Submit the same material to an accredited laboratory for confirmation by the applicable standard method.
• Decide disposition before the result arrives: what happens to finished goods, work in progress and incoming stock if confirmed.
• Record the root cause, because a positive is usually a purchasing event rather than a manufacturing one.

The root cause point is the one that pays for the whole programme. A positive on a metal component almost always traces to a purchase: a different alloy, a different supplier, a lot bought outside the specification. Once that is known, the control is a purchasing rule rather than a test — the named supplier and alloy go into the specification, and any change requires re-screening. Programmes that convert positives into purchasing rules stop finding the same positive repeatedly; programmes that only quarantine the lot find it again next season.

There is a commercial step that should happen in parallel and often does not. If the affected goods have already shipped, the question is whether the exposure is legal, contractual or reputational, and that depends on the market and the customer. A goods-inward screen that finds lead in hardware destined for an adult general-purpose market is a different problem from the same finding on a children’s product, and the response should be proportionate. The full set of obligations for regulated markets is outlined in our overview of safety and regulatory compliance.

The limited generality of any test report

Every analytical report, screening or confirmatory, applies to the sample that was submitted. This is printed on the reports, it is universally true, and it is routinely ignored. A screening result on five buckles from one lot says nothing about the next lot, and a laboratory report on a fabric swatch says nothing about the rolls that were cut three weeks later. The generality people want is not a property of the report; it is a property of the sampling plan and the change control behind it.

• A report covers a sample, not a product line, and not a supplier.
• Generality comes from repetition: many lots, over time, with recorded results, builds a statistically defensible picture.
• Generality dies on substitution: a material change invalidates the history built up before it.
• A report without a sample description, a date and a lot reference cannot be connected to anything, which makes it decoration.
• A screening programme is a way of buying generality cheaply, which is its real economic argument.

The economic argument is worth stating explicitly because it justifies the whole activity. A single laboratory determination costs a meaningful amount and covers one sample. A screening programme costs a device and some operator time and covers hundreds of samples per season, triaging them so that laboratory budget is spent where the risk actually is. The screening programme does not replace the laboratory; it makes the laboratory spend informative rather than random.

The failure mode of this argument is to let screening become a substitute for evidence in front of a customer or a regulator. It is not, and presenting a screening result as though it were an accredited laboratory determination is a misrepresentation regardless of how good the instrument was. The two live in different places in the file: screening drives internal decisions, laboratory reports support external claims. Keeping that boundary clear is what makes both useful. Our companion guide to OEKO-TEX STANDARD 100 for waterproof bags shows how a certificate sits in the same file with a different and narrower job.

Writing the requirement so a number is not the whole answer

The way a screening requirement is written determines what comes back. Ask for "XRF test results" and you get a spreadsheet of numbers with no detection limits, no material descriptions and no lot references, which cannot be used for anything. Ask for a defined record and you get something that supports decisions. The difference is six lines on a specification.

• Name the elements to be screened and the screening threshold for each, set below the applicable legal limit rather than at it.
• Name the materials and colourways in scope, taken from the bill of materials rather than from the finished bag.
• Require the detection limit to be reported alongside every result, with the count time.
• Require the material, colourway, supplier and lot reference for every reading.
• Require instrument identification and the calibration check reference for each session.
• State the confirmation rule: which results go to an accredited laboratory, by which method, and who pays.

The final line is the one that prevents arguments. Without a stated confirmation rule, a borderline positive becomes a negotiation about who pays for the laboratory test, and the negotiation takes longer than the test. With one, the rule executes automatically: results above the screening threshold go to the laboratory, the cost sits where the contract says it sits, and the programme keeps moving.

One further line is worth adding for programmes that sell into regulated markets: state that screening does not satisfy any legal testing obligation. That single sentence prevents the most common and most dangerous misuse, which is a supplier presenting a clean screen as compliance documentation. It costs one line and it closes a gap that otherwise stays open for years.

If you want this applied to a specific programme, send the bill of materials with the colourways and the markets you sell into, and a screening plan can be written against those three before the first components arrive. You can see how a programme moves from first enquiry through sampling into bulk production; every style starts at 500 pieces minimum, with samples in 6–10 working days and bulk in 35–50 days, quoted FOB Xiamen.

Frequently Asked Questions

Q1. What does a handheld XRF analyser detect?

Elements, roughly from magnesium upwards, in the near-surface volume it irradiates. That covers lead, cadmium, mercury, total chromium, bromine, antimony and tin, and excludes every organic compound and every element lighter than magnesium, including fluorine.

Q2. Can XRF prove a bag complies with a legal limit?

No. It screens. A positive is a candidate for confirmation, and a negative is a partial clearance of the elements measured, reported against a detection limit, on the sample submitted.

Q3. Why does a positive XRF result need wet chemistry?

Because limits attach to defined methods applied to defined materials, and because the screening number carries uncertainty from matrix, surface condition and spectral overlap. Only an accredited laboratory determination is a finding.

Q4. Can XRF detect phthalates?

No. Phthalates are organic molecules and contain no element above the detection floor. They require solvent extraction and instrumental analysis in a laboratory.

Q5. Can XRF detect PFAS or fluorinated finishes?

No. Fluorine is lighter than magnesium and is outside the technique entirely. Fluorinated finishes need a total fluorine screen or a targeted organic analysis.

Q6. If chromium is detected, does that mean hexavalent chromium is present?

Not necessarily. XRF measures total chromium and cannot distinguish oxidation states. A high total reading is a reason to test for the hexavalent form, not evidence of a failure.

Q7. Is bromine a reliable indicator of brominated flame retardants?

It is a marker, not proof. Bromine can come from other sources, and a bromine reading justifies targeted laboratory testing rather than a conclusion.

Q8. Which components should be screened first?

Metal hardware, especially small parts such as rivets and eyelets and anything plated over brass, followed by strongly pigmented coatings and prints, then PVC or PVC-blend films.

Q9. Why does colour matter in a screening plan?

Because pigments are the usual source of lead and cadmium. Strong yellows, oranges, reds and greens carry higher historical risk, so screening one colour and assuming it covers the others is a serious shortcut.

Q10. How many measurements should I take?

For a homogeneous metal part, five to ten shots across three to five pieces. For printed or coated textiles, ten to twenty across more pieces, taking the highest defensible reading as the screen result.

Q11. Why should components be screened before assembly?

Because the analyser needs flat contact and sufficient thickness. A buckle on a bench reads well; the same buckle sewn into a curved, soft bag gives a degraded spectrum.

Q12. What is a detection limit and why must it be reported?

It is the lowest level the measurement could reliably detect in that matrix with that count time. A non-detect is meaningless without it, since a non-detect at a high limit says nothing about a low threshold.

Q13. Does a longer measurement time improve results?

Yes. Detection improves roughly with the square root of measurement time, so doubling the count buys a meaningful improvement, which matters most for low-level readings in light matrices.

Q14. Can XRF read through metal plating?

Yes, and this is one of its best applications. A thin decorative layer does not stop the analyser, so leaded brass beneath chrome plating can be detected and the plating thickness estimated.

Q15. Does a test report cover my whole product line?

No. It covers the sample submitted. Generality comes from repeating the test across lots and suppliers over time, and it is destroyed by any material substitution.

Q16. What should I do when a screen is positive?

Rescreen and check the instrument first, since many positives do not survive. If it holds, quarantine the lot, confirm with an accredited laboratory, and trace the root cause, which is usually a purchasing event.

Q17. Is screening cheaper than laboratory testing?

Per sample, very much so, and that is its purpose: triage hundreds of samples so laboratory budget goes where the risk is. It does not replace accredited testing and should never be presented as if it did.

People Also Ask

What is XRF screening?

Using a handheld X-ray fluorescence analyser to detect elements such as lead, cadmium, mercury and chromium in a material, quickly and without destroying it.

Can XRF prove compliance with lead limits?

No. It screens. A positive needs wet chemistry confirmation by an accredited laboratory against the standard method.

What can XRF not detect?

All organic substances, including phthalates, azo colourants, PFAS and formaldehyde, plus elements lighter than magnesium such as fluorine.

Which parts of a bag should be screened first?

Metal hardware and small components, then strongly pigmented coatings and prints, then PVC films and coatings.

Does an XRF test report cover all my products?

No. It applies to the sample submitted. Generality comes from repeat testing across lots and dies with any material substitution.

Why does colour matter when screening?

Because pigments are the main source of lead and cadmium, and strong yellows, oranges, reds and greens carry the highest historical risk.