The analytical method decides the number, not the metal. Fire assay collects gold, silver and the platinum group into a lead button, cupels away the base metals and weighs the bead; ICP-OES measures a digested solution; XRF reads the surface without destroying it. On the same lot the three can disagree by several percent, and that disagreement is the single most common reason a seller and a refiner fail to agree on a settlement. Choose the method by what the lot is, not by what the laboratory prefers: plated and coated material needs a total-content method, refractory platinum group scrap needs fusion before any solution technique, and a mixed consignment needs sorting before anything is assayed at all.
Fire assay. A weighed sub-sample is fused with litharge and fluxes at roughly 1,000 to 1,200 degrees Celsius. The precious metals report into a lead button, the button is cupelled in a magnesia cupel where the lead oxidises away, and a precious metal bead remains. Gold and silver are then parted or weighed directly; for platinum group metals the bead is dissolved and finished by ICP-OES or AAS. Because the whole sub-sample is dissolved into the collector, fire assay is far less sensitive to nugget effect and surface condition than any spot method, which is why it is the method contracts fall back on.
ICP-OES. A digested sample is aspirated into an argon plasma and the emission lines are measured against calibrations. It is fast, covers many elements in one run, and is the workhorse for base metals and for platinum group metals after fusion. Its weakness is upstream of the instrument: if the digestion is incomplete, the result is low and there is no warning on the printout. Iridium, rhodium and ruthenium are the usual casualties, and they need fusion with sodium peroxide or a similar flux rather than an acid attack.
XRF. An X-ray beam excites the surface and the returning fluorescence gives an elemental spectrum. It is non-destructive, takes seconds, needs almost no sample preparation and is unbeatable for sorting. Its limit is physical: the beam penetrates micrometers to a few millimetres depending on energy and matrix, so on gold-plated brass it measures the plating, and on a mixed heap it measures whatever happens to be under the aperture.
| Question | Fire assay | ICP-OES | XRF |
|---|---|---|---|
| What it reports | Total content of precious metals in the sub-sample | Element concentrations in the digested portion | Surface composition, penetration depth depends on matrix |
| Destroyive | Yes | Yes | No |
| Typical use | Settlement reference, umpire assay | Base metals, PGM after fusion, multi-element panels | Sorting, grade verification, incoming inspection |
| Strength | Insensitive to surface condition and nugget effect | Many elements per run, good throughput | Seconds per reading, no sample preparation |
| Main failure mode | Longer turnaround, needs a skilled cupellation bench | Incomplete digestion reads low, silently | Reads plating not core, matrix effects on light elements |
| Fits best when | The lot is heterogeneous or the value is high | The metal suite is wide and the matrix is known | A decision is needed on the loading dock |
Turnaround and cost are deliberately left out of this table: they vary by laboratory and by element suite, and any figure published here would be out of date before it was read. Ask the laboratory for its own scope, detection limits and reporting time.
The variance of an assay result is the sum of sampling variance and analytical variance, and on scrap the first term usually dominates by an order of magnitude. A laboratory that reports gold to two decimal places on a sub-sample that was never representative has produced a precise number about the wrong material.
The governing relationship is that the minimum representative mass rises with the cube of the top particle size and with the heterogeneity of the lot. In practice that means three rules. Crush and riffle in stages rather than grabbing a scoop from the top of a drum. Record the mass retained at each stage, because the reduction ratio is the first thing an umpire laboratory will ask about. And treat any lot with visible mixed materials as a sorting job before it is an assay job, which is exactly the point made in the PGM sampling and assay note.
Moisture belongs in the same conversation. A grade reported on an as-received basis and a grade reported on a dried basis differ by the moisture content, and on catalyst, sludge and filter media that difference is not small. The contract has to name the basis, and the moisture determination has to be done on a sub-sample that was sealed at the point of sampling rather than after the material sat in an open drum.
Four failure modes account for most settlement disputes, and each has a specific counter.
Method mismatch on plated material. XRF reads the gold or silver layer, the refiner assays total content after stripping, and the two figures differ because one is a thickness and the other is a mass. The counter is to name a total-content method in the contract and to use XRF only for sorting. The gold-plated settlement walk-through works through the arithmetic.
Incomplete digestion of refractory metals. Rhodium, iridium and ruthenium resist acid attack. A solution result that looks clean and low is the classic signature. The counter is to require fusion for any lot where those metals are payable, and to check that the certificate names the decomposition route, not just the instrument.
Head grade versus payable content. A certificate says 3.2% gold; the settlement says 2.8% payable. Both can be correct, because the second figure has had smelting loss and the agreed deduction applied. The counter is to state in the offer which of the two the quote is based on, so the two sides are not comparing different quantities.
Umpire assay without a rule. A dispute goes to a third laboratory, which returns a figure outside both original results, and nobody agreed in advance what that means. The counter is a tolerance band and a defined consequence: inside the band the original stands, outside it the umpire governs and the losing side pays. Without that clause the umpire result becomes a third opinion rather than a decision.
A workable assay clause fits in one paragraph and answers seven questions: which method, on what sample mass, on which moisture basis, at which laboratory, with what tolerance band, with which umpire and at whose cost, and for how long the reserve sample is kept. Everything else in the clause is commentary. Two of those seven, sample mass and moisture basis, are the ones most often left out and the ones that cause most of the arguments.
One further point on timing. An assay certificate is only useful while the benchmark it will be multiplied against is still current. A three-week turnaround on a lot whose price was fixed at the offer date leaves the seller exposed to every move in between, which on a day like 11 November 2026, when the domestic board fell 1.70% on gold while the international board rose 0.28%, is a larger exposure than the assay dispute itself. Fix the benchmark timestamp and the assay turnaround in the same clause.
Feiyan runs sampling, XRF screening and laboratory assay as one chain, so the method used for screening and the method used for settlement are chosen together rather than by default. Details of the XRF screening service and the pricing mechanism behind a final payout are published separately, and precious metal products and raw materials can be quoted against the current domestic benchmark.
Prepared by the Feiyan industry research team. Method descriptions are standard practice for precious and refractory metal assay; operating parameters, detection limits, turnaround and fees differ between laboratories and should be confirmed with the laboratory's own scope before being written into a contract. No price data is quoted in this article. For industry reference only, not investment advice. Liaocheng Feiyan New Material Technology Co., Ltd., Liaocheng, Shandong.
Fire assay with a gravimetric finish is the reference method for gold and silver in settlement, because it collects the metal into a lead button, cupels away the base metals and weighs the resulting bead. It handles heterogeneity better than surface methods and is normally the method named in a contract when the two sides disagree.
XRF is the right choice for fast identification, sorting mixed scrap into alloy families and verifying that a consignment matches its declared grade. It is a surface measurement, so it reads plating thickness rather than total content, and it should not be the only method behind a settlement figure on plated or coated material.
Usually for four reasons: the sub-sample was not representative of the lot, moisture was measured on a different basis, refractory metals such as iridium, rhodium and ruthenium were not fully dissolved before ICP-OES, or the reported figure is a head grade rather than a payable content after losses.
It depends on top size and heterogeneity rather than on the value of the lot. A few hundred grams of homogeneous powder can be representative, while a two-tonne load of mixed connectors cannot be represented by a handful of pieces; the sample must be crushed and riffled down in stages, and the mass retained at each stage recorded.
It should name the method, the sample mass, the moisture basis, the laboratory, the tolerance band that triggers a repeat assay, the umpire laboratory and who pays for it, and the retention period for the reserve sample. Naming a method without naming the sample mass leaves the largest source of disagreement unaddressed.
No. A head grade reported by a surface method can be higher than the payable content after smelting losses, and a grade quoted on a wet basis is lower once moisture is deducted. The number that matters is payable metal after the deductions written in the contract, not the highest figure on any certificate.
Liaocheng Feiyan New Material Technology Co., Ltd.
Address: Room 8-1503, 15F, Building 9, Xingguang International Financial Center, Dongchang East Road, Dongcheng Sub-district, Liaocheng Economic and Technological Development Area, Shandong, China
Tel: 0635-2112001 / 19963594228 | Email: fykj425@gmail.com
Copyright © Liaocheng Feiyan New Material Technology Co., Ltd.. All Rights Reserved.