How every number on this page is set
Category 5 carries two native bases and the compiler is configured never to mix them. A ready-to-drink beverage is certified in the as-consumed liquid basis at single strength, and a concentrate is assessed as reconstituted to single strength, which is the basis EU 2023/915 and Codex CXS 193 already write into their juice cells and Hong Kong Cap 132V Note 11 writes into its beverage entries. A dry leaf, dry botanical, roasted bean or beverage powder is certified in the as-sold dry-commodity basis, because that is the article the applicant packages and the lot the auditor samples. No brewing, extraction, dilution or reconstitution factor is invented in either direction: the tea-leaf lead maximum level is never applied to a brewed cup, and a ready-to-drink beverage maximum level is never applied to a dry leaf. Where a government instrument writes the product-form split itself, HMTc preserves it rather than collapsing it: Hong Kong Cap 132V sets Tea, Green, Black at 5000 ug/kg as sold against tea beverages made from it at 200 ug/kg as consumed, and coffee beans at 500 ug/kg as sold against coffee beverages at 200 ug/kg as consumed. Two consequences of the method are disclosed rather than engineered away. The rice-based drink row is LOOSER on inorganic arsenic (30 ug/kg, EU 2023/915 section 3.4.1.6) than the juice rows (20 ug/kg, and 10 ug/kg for apple-containing juice), because the EU assessed that exact commodity and set that exact number. And several formulated-beverage rows carry a lead ceiling of 200 ug/kg because Hong Kong Cap 132V is the only credible sovereign that regulates those forms at all, while the flavoured-water row carries 5 ug/kg because the packaged-drinking-water quality standards do reach a product that is water plus flavour. Those spreads are applicability judgements, each recorded on the face of the cell, and only the endogenous Standards Ratchet in Governance Policy section 12 may lower any of them.
Master Limit Table
All values in ppb (µg/kg), as-consumed basis. 5 of 11 analytes carry a firm ceiling; the remainder are governed by mandatory control or reflex-speciation screening with no standalone number, itemised in the derivation below. A brand meeting every firm value on a like-for-like basis, and passing every control, qualifies for the mark.
| Analyte | Tier | HMTc limit | Binding basis |
|---|---|---|---|
| Lead (Pb) | Tier 1 | 20 | government read-across (Hong Kong) |
| Inorganic arsenic (iAs) | Tier 1 | 20 | government read-across (EU) |
| Total arsenic (tAs) | Tier 2 | control | mandatory control — no government ML; screened every lot (see note) |
| Mercury, total (tHg) | Tier 1 | 10 | government read-across (Hong Kong) |
| Methylmercury (MeHg) | Tier 1 | n/m | not material — controlled analyte (see note) |
| Cadmium (Cd) | Tier 1 | control | mandatory control — no government ML; screened every lot (see note) |
| Hexavalent chromium (Cr-VI) | Tier 2 | control | mandatory control — no government ML; screened every lot (see note) |
| Nickel (Ni) | Tier 2 | 250 | government read-across (EU) |
| Tin (Sn) | Tier 2 | 100000 | strictest government ML (EU) |
| Aluminium (Al) | Tier 2 | control | mandatory control — no government ML; screened every lot (see note) |
| Total chromium (Cr) | Tier 2 | control | mandatory control — no government ML; screened every lot (see note) |
Tier 1 (Pb, iAs, Hg, MeHg, Cd) are zero-tolerance toxics with no established safe threshold. Tier 2 (tAs, Cr-VI, Ni, Sn, Al, Cr) carry a 150% transitional allowance under the program’s Status A–E framework.
Derivation — every government’s number, side by side
The full derivation for Plant milks, soy-based. Each row lists every government maximum level in force worldwide, converted to the as-consumed basis; the HMTc limit is the strictest of them, and the last column names which sovereign law binds. A blank cell means that government sets no limit for this analyte in this product.
| Analyte | HMTc | Binds |
|---|---|---|
| Lead (Pb) | 20² | read-across |
| Inorganic arsenic (iAs) | 20² | read-across |
| Total arsenic (tAs) | control&sup4; | screened |
| Mercury, total (tHg) | 10² | read-across |
| Methylmercury (MeHg) | n/m³ | controlled |
| Cadmium (Cd) | control&sup4; | screened |
| Hexavalent chromium (Cr-VI) | control&sup4; | screened |
| Nickel (Ni) | 250² | read-across |
| Tin (Sn) | 100000 | EU |
| Aluminium (Al) | control&sup4; | screened |
| Total chromium (Cr) | control&sup4; | screened |
All values µg/kg (ppb), as-consumed basis. ² no government regulates this analyte for this exact product form: the value is a disclosed read-across from the nearest applicable government maximum, and the derivation names the instrument it comes from. It is never taken from occurrence data. After publication the standards ratchet may tighten it, using certified-lot results only. ³ not material in this matrix: controlled by reflex speciation or packaging control, not a standalone number. &sup4; no sovereign maximum exists for this analyte in this product, but a credible exposure pathway does — it is screened on every lot under ALARA against the tightest analogous reference, not assigned an invented number.
Occurrence and derivation rationale
For each analyte on the panel: what the peer-reviewed and survey literature measured in this category, cited one way to the Heavy Metal Index, and the Method v2.0 reasoning behind the certification limit — including, where no government sets a number, why the program sets one anyway. Occurrence figures are literature findings; the derivation is the certifier's.
The corpus holds 4 measurements of cadmium in Plant milks, soy-based (189 samples reported across those studies), which the determination below is tested against: Good 2026, Zverina 2025, Andrade 2023, Redan 2023.
HMTc sets no standalone cadmium limit for Plant milks, soy-based. No credible sovereign sets a cadmium maximum level for a soy-based plant drink. EU 2023/915, Codex CXS 193, FSANZ Schedule 19, Health Canada and Hong Kong Cap 132V set no cadmium maximum level for a plant-based drink, and none of them sets a cadmium cell for dairy milk either, so the dairy-milk analog relied on for lead and total mercury in this row supplies nothing for cadmium. The sovereign cadmium cells for the dry source seeds, grains and nuts (for example soy beans at 200 ug/kg) are as-sold dry-commodity values and cannot be carried into a ready-to-drink liquid without an invented dilution factor. No near-neighbour read-across is supportable, so the fail-safe disposition is a control: cadmium testing is mandatory on every certified lot and no ceiling is invented from occurrence data. The analyte is controlled by Cadmium by ICP-MS on every certified lot, reported on the certificate. A result above 10 ug/kg in a ready-to-drink form, or above 100 ug/kg in a dry form, escalates to an ingredient-origin and process-water investigation and is a Standards Ratchet surveillance input under Governance Policy section 12, not a standalone number. Why cadmium is limited →
1 study in the corpus measured nickel in Plant milks, soy-based (32 samples reported across those studies): Marques 2021.
HMTc sets 250 ppb by read-across: No credible sovereign has written a nickel maximum level for a plant-based drink. EU 2024/1987's fruit and vegetable juices and nectars cell (250 ug/kg) is the only sovereign nickel value written for a ready-to-drink beverage and is adopted by disclosed read-across as the nearest applicable value in the identical liquid basis. Why nickel is limited →
No citable primary occurrence measurement is routed to this cell in the corpus.
HMTc sets no standalone aluminium limit for Plant milks, soy-based. No credible sovereign sets an aluminium maximum level for a soy-based plant drink. The only sovereign aluminium values in a dietary basis are the packaged-drinking-water quality standards (US 21 CFR 165.110 secondary allowable level 200 ug/kg, EU 2020/2184, Health Canada 2900 ug/kg), which govern water sold as water and are Category 8 instruments, and the FSANZ infant-formula aluminium limits, which are Category 1 and are excluded from this category's anchor set by the taxonomy boundary rule. In this form aluminium is a packaging and processing signal (aluminium cans and can linings, closures, filter aids, additive carry-over) rather than a raw-material contaminant. Plant-based drinks additionally carry mineral-fortification premixes and stabiliser systems, so an aluminium result is investigated against the fortificant and stabiliser specifications as well as the packaging. The analyte is controlled by Total aluminium by ICP-MS on every certified lot, reported on the certificate. A result that departs materially from the applicant's own documented process baseline escalates to a packaging, can-lining and processing-aid investigation. This is a process-control tripwire, not a numeric standard: no ceiling is published because no credible sovereign maximum exists for this cell, not a standalone number. Why aluminium is limited →
Cite this standard
Pendergrass, K. “Plant milks, soy-based.” In Heavy Metal Tested & Certified (HMTc) Beverages standard, version 1.0. Institute of Contaminant Standards (ICS), 2026. DOI: registration pending.
The literature baseline for every analyte is maintained independently at the Heavy Metal Index. This certification standard applies those findings; the two are kept editorially separate by design.