20.3 Sampling, Fire Assay, Cupellation, Parting & Assay Calculations

An assay result describes the submitted sample, not automatically the entire orebody, stockpile or shipment. Sampling error commonly exceeds analytical error because valuable minerals are heterogeneous and gold can occur as rare coarse particles. The chain must preserve representativeness, identity and mass.

Sampling and Preparation

A defensible protocol defines lot, increment selection, sample mass, top size, frequency, tools, moisture, and compositing. Avoid grab samples for grade estimation where they favor visible mineralization. Channel samples use consistent geometry; drill core is split according to orientation and recovery controls; moving streams are sampled across the full stream where possible.

Laboratory preparation typically includes:

1. receive, inventory and dry as specified;
2. crush to a controlled passing size;
3. homogenize and split with a riffle or rotary divider;
4. pulverize an aliquot to required fineness;
5. split the assay charge; and
6. retain coarse reject and pulp under controlled storage.

Clean equipment between samples and order batches to detect carryover. A fine blank after high-grade material tests contamination.

Fire-Assay Fusion

For gold and silver, a weighed sample charge is mixed with flux and fused in a crucible. The objective is two liquid phases:

• a dense lead collector that scavenges precious metals; and
• a fluid slag that captures gangue and base-metal oxides.

Flux formulation responds to ore chemistry:

Too much reductant creates an oversized button; too little can produce inadequate collection. Sulfide-rich samples may need roasting or oxidizing flux. Fusion must avoid boilover and ensure clean separation when poured into a mold.

Cupellation

The cleaned lead button is placed in a porous cupel and heated in an oxidizing furnace. Lead oxidizes to litharge; much is absorbed into the cupel and some volatilizes under controlled ventilation. Base metals oxidize while gold and silver remain as a doré bead.

Cupellation temperature matters. Too cold can freeze litharge or retain lead; too hot can increase precious-metal loss, cupel attack or volatilization. “Feathering” of litharge and the brightening stage are operational observations, but final quality is confirmed by method controls. Lead fumes and contaminated cupels require stringent occupational and waste controls.

Parting and Calculation

For a gravimetric gold finish, weigh the cleaned combined gold-silver bead, flatten or roll it, and dissolve silver in nitric acid under the method conditions. Wash, anneal and weigh the remaining gold. Difference gives silver after applicable blank and correction.

If a 30.00-g sample yields 0.0060 g gold after parting:

$Grade=\frac{0.0060\ g}{30.00\ g}\times10^6=200\ g/t$

Because 1 g/t equals 1 microgram per gram, multiplying the mass fraction by one million gives g/t. Such a high result should trigger the high-grade protocol and check for weighing, transcription or contamination.

An instrumental finish may dissolve the bead and measure gold by atomic absorption or ICP, improving sensitivity for low grades. State the finish, charge mass and detection range.

QA/QC

• Certified reference material: tests accuracy/bias near relevant grades and matrices.
• Blank: tests contamination and sample carryover.
• Coarse duplicate: tests crushing and splitting plus analysis.
• Pulp duplicate: tests pulverized-sample and analytical precision.
• Field duplicate: includes field heterogeneity and sampling.
• Umpire assay: independent laboratory comparison.

Plot control results in sequence. A standard outside acceptance limits requires investigation of the affected batch, not deletion of the standard. Duplicate scatter may reflect coarse gold; use appropriate protocols such as larger charge or screen fire assay when justified.

Common Errors

Mislabeling, wet/dry-basis confusion, loss of fines, contamination, incomplete fusion, prill loss, cupellation loss, balance drift, wrong dilution, transposed units and result clipping can bias grade. Chain of custody preserves identity but does not prove representativeness.

Exam Trap

The analytical method cannot correct a biased sample. Choose the answer that controls the entire chain—lot definition, sampling, preparation, fusion, measurement, calculation and QA/QC.

Reporting Basis

State whether a result is g/t, mg/kg, percent, troy ounce per short ton, or another unit, and whether the mass basis is dry. One g/t equals one mg/kg. Never treat ppm in water automatically as ppm in solids without recognizing the mass/volume convention. Preserve unrounded bead and sample weights through calculation, then round to method precision and report detection or quantitation limits.