Heavy metals are the least visible failure mode in a supplement. Nothing looks wrong, the COA arrives on time, and the product still ships. The problem is that contamination is a property of the raw material's origin and the process around it — not of the formula on paper. Here is where it actually comes from, and what a control program has to do to catch it.

The four elements that matter

Regulatory and pharmacopoeial thinking converges on four: arsenic, cadmium, lead and mercury. USP general chapter <2232> on elemental contaminants in dietary supplements describes them as the four major elements of toxicological concern, and that framing is a useful default for a specification even when your market's rules differ.

Where contamination enters

  • Soil and geography. A botanical takes up what the ground and air give it. Rice-derived and seaweed ingredients are the classic cases; the same species grown in two regions can differ by an order of magnitude.
  • Mineral ingredients. Calcium carbonate, clays, sea moss and some mineral salts are mined or harvested materials. Lead in a limestone-derived excipient is a known pathway.
  • Processing. Drying on open surfaces, milling with worn equipment, and reusing solvent drums all add metal contact. So does cheaper equipment with brass or leaded fittings.
  • Adulteration. Heavy, cheap fillers are one reason a botanical extract tests "on assay" while carrying lead.
  • Packaging. Less common, but ink and recycled board can migrate.

"It passed last year" is not a control

A single annual test tells you what one lot looked like. Raw material lots change with harvest, supplier and region. The control that matters is lot-level: every incoming batch tested or covered by a verified supplier program, with results trended so a drift shows up before a customer finds it.

How the testing actually works

The standard approach is digestion of the sample followed by inductively coupled plasma mass spectrometry (ICP-MS), which is what the published survey work on dietary supplements uses for arsenic, cadmium, mercury and lead. Two details decide whether a result is meaningful:

  • Total versus speciated arsenic. Total arsenic includes organic forms that are far less toxic than inorganic arsenic. A supplier reporting only "arsenic: pass" without saying which is being measured is telling you very little.
  • Method and limit of detection. A result below a limit is only useful if the detection limit sits well under the specification.

Limits come from your specification

There is no single global number that applies to every ingredient and market. Limits come from the target market's rules, the pharmacopoeial chapter you have adopted, or the customer's own specification. What matters commercially is that the limit is written down, is lower than the regulatory ceiling, and is applied per lot rather than per supplier relationship.

What a defensible program looks like

  • Supplier qualification with the origin of each botanical or mineral documented.
  • A COA per lot, with the method named and the result against a stated limit.
  • Independent verification on a defined frequency — every lot for high-risk materials, skip-lot for low-risk ones.
  • Retention samples kept for the shelf life, so a complaint can be traced back to a batch.
  • Trend charts per material, reviewed — not filed.

Red flags in an answer

Three replies should end the conversation: a COA with no lot number, a COA where the specification and the result are identical text, and "we test once a year, it is always fine". Each describes a document, not a control.

Blending down is not a fix

Mixing a hot lot with clean material until the average passes is technically a way to make a number disappear, and it is the wrong one. It spreads a known contaminant across a larger quantity of product, destroys the traceability that a recall depends on, and leaves you unable to tell a customer which batches were affected. Reject the lot instead.

What to put in your spec sheet

Name the four elements, give a limit for each, state the method family, require a per-lot COA, and require that out-of-spec material is not shipped. That single paragraph does more for product safety than any claim on the front of the label.

How we handle it

Elemental contaminants are part of the specification before sampling starts, results come with every batch COA, and out-of-spec material is rejected rather than blended down. Samples in 3-5 days, production in 7 days, MOQ 500 units.