Geochemical Analysis Methods Compared: XRF, Fusion, Assay

Geochemical analysis is how an exploration team turns a bag of rock or soil into numbers it can map: parts per million of copper, parts per billion of gold, percent of silica and iron oxide. The number on the certificate depends as much on the method as on the rock. The same sample can report two different zirconium values from two digestions, or “below detection” for copper on a handheld XRF and 30 ppm in the lab. Choosing the right geochemical analysis for each sample type is therefore a technical decision with money attached, not a line item to leave to the lab’s default package. Designing the soil and stream program that produces those samples is covered in our geochemical survey design guide.

Try it: soil and rock sampling programme planner →

This guide covers the four method families that carry most exploration budgets: portable and laboratory XRF, whole-rock fusion, acid digestions with ICP finishes, and fire assay for gold. Every detection limit quoted here comes from a published method document or lab schedule, and we give the source so you can check whether it has changed. For how the samples are drilled in the first place, see our guide to exploration drilling methods and costs.

“A 30 g fire assay can report gold down to 0.002 ppm, or two parts per billion.”

What geochemical analysis measures, and what it does not

Every method in this article answers one question: how much of an element is present in the part of the sample the method can reach. That last clause is where most misreadings start. An aqua regia digestion reaches sulphides and elements stuck to clays and iron oxides, but the lab’s own procedure notes say it “will not dissolve significant quantities of the silicate and alumino-silicate minerals.” A four-acid digestion goes further and is often called near-total, yet the same document warns that not all elements are quantitatively extracted. A fusion dissolves almost everything, including resistant minerals such as zircon, at the cost of higher detection limits for some trace elements (chromium, for example, reports from 10 ppm by fusion ICP-MS against 1 ppm by four-acid in the same schedule).

So a geochemical analysis result is always a pair: the value and the method code. Report one without the other and the number is close to meaningless. The JORC Code makes the same point for public reports, listing “methods of analysis” among the details a report of exploration results must include (Clause 19 of the 2012 edition).

Three questions to answer before you submit samples

1. What are you looking for? Pathfinder elements at ppb levels in soil need a different package from ore-grade copper in drill core.
2. Where does the element live? Gold in coarse free grains, tin in cassiterite, zirconium in zircon and chromium in chromite all resist weak digestions.
3. What decision will the number support? A field screen that picks the next soil line tolerates more error than an assay that goes into a resource database.

Geochemical analysis of soil: planning the grid

Soil sampling is usually the first ground geochemistry on a new licence. It is cheap per sample, fast to collect and good at outlining a target a few hundred metres across. The main design choices are line spacing, station spacing along each line, the soil horizon you collect, and the size fraction you send to the lab.

For a sense of scale at the other extreme, the U.S. Geological Survey’s national soil survey (Data Series 801) sampled 4,857 sites across the conterminous United States, about one site per 1,600 km². At each site it took three samples: the top 0 to 5 cm, a composite of the A horizon, and the C horizon (or roughly 80 to 100 cm depth where the C horizon was deeper than a metre). Each was analysed for 45 major and trace elements by methods that give total or near-total content. That density maps regional baselines. It would miss almost any single orebody, which is why exploration grids are measured in tens of metres rather than tens of kilometres.

Choosing a digestion for soils

Soils are where low detection limits matter most, because the anomaly is often a few times background. The Super Trace aqua regia methods in the NT Geological Survey copy of the lab schedule use a nominal 0.5 g sample and ICP-MS with collision/reaction cell technology. The procedure note says these methods are “particularly useful for exploration in soils or sediments” and “not appropriate for mineralized samples.” It also warns that coarse native gold, silver, platinum and palladium are not represented properly in a sample that small.

That warning shapes the planner below. The digestion you pick sets how many grams of pulp each analysis consumes, and the insertion rate you choose for standards, blanks and duplicates sets how many extra bags go in the box.

Most licences are too big to grid completely on day one. Screening the whole area from satellite imagery first lets you put the soil lines where alteration and structure suggest they will count. You can draw your licence on mining.farmonaut.com (Map Your Mining Site) to start, or send a boundary through our mining query form for a quote.

Geochemical analysis of rock samples: from field bag to pulp

Rock chips, channel samples and diamond drill core go through the same preparation chain before any instrument sees them: dry, crush, split, pulverise. The chain matters because each digestion uses only a small scoop of pulp. If the pulp is not well mixed, the scoop is not the rock.

How small is small? The procedure sheets lodged with the Northern Territory Geological Survey give the nominal sample weights. A lithium borate fusion for whole-rock oxides uses 0.100 g. A four-acid digestion uses 0.25 g. An aqua regia digestion uses about 0.5 g. Loss on ignition uses 1.0 g, heated at 1,000 °C for an hour. Fire assay for gold uses 30 g or 50 g. Put on one scale, the gap is large:

Preparation steps that change the answer

• ✔ Crush and split before pulverising. The split size decides how representative the pulp is.
• Pulverise to a stated fineness and ask the lab to report its sieve checks.
• Clean between samples with barren material, so a high-grade sample does not contaminate the next one.
• ⚠ Coarse gold: consider screen fire assay or a larger charge, because a 30 g aliquot can miss or over-count single grains.
• Keep the coarse reject and the pulp. They are your only route to re-assay without re-drilling.

Rock samples also answer questions soils cannot: what the host rock is, how altered it is, and whether an intrusion is the fertile kind. Those questions need whole-rock chemistry, covered two sections down, and they tie directly to field geological mapping, which tells you which rocks to sample in the first place.

XRF geochemical analysis: handheld screening versus the lab

X-ray fluorescence works by exciting atoms with X-rays and reading the energy each element gives back. It is fast, needs little preparation and does not destroy the sample. Those strengths make XRF geochemical analysis the default field tool on many projects. Its weakness is sensitivity, and the size of that gap surprises people who compare a handheld reading with a lab certificate.

The U.S. Environmental Protection Agency’s Method 6200 for field portable XRF is explicit. It calls the technique “a screening method to be used with confirmatory analysis” by AAS, ICP-AES or ICP-MS. Its example lower limits of detection, measured in clean quartz sand with long 100 to 600 second count times, are 50 mg/kg for copper, zinc and nickel, 40 mg/kg for arsenic and antimony, 20 mg/kg for lead and 10 mg/kg for molybdenum. Real soils with interfering elements and moisture do worse. The method adds that moisture between 5 and 20 percent causes minimal error, but saturated samples do not read well.

Compare that with the lower limits of a four-acid digestion finished by ICP in the lab schedule: 0.2 ppm copper, 0.2 ppm arsenic, 0.5 ppm lead, 2 ppm zinc, 0.2 ppm nickel, 0.05 ppm molybdenum and 0.05 ppm antimony.

Where handheld XRF earns its place

• ✔ Ore-grade decisions in the field: telling 0.5% copper from 2% copper in core boxes, where the XRF limit is irrelevant.
• Lithology and alteration: major-element ratios that separate rock types on a logging table.
• Choosing what to send: ranking samples so the lab budget goes to the right intervals.
• ⚠ Not for reporting low-level anomalies or light elements without lab confirmation.

Laboratory XRF on fused beads is a different animal. Fusing the sample into a glass disc removes most grain-size and mineral effects, which is why fused-bead XRF is a standard route to major oxides in whole-rock work. Treat “XRF” on a certificate as two possible methods and check which one was used.

Whole rock geochemical analysis: fusion versus four-acid digestion

Whole rock geochemical analysis asks what a rock is made of in total: the major oxides (SiO₂, Al₂O₃, Fe₂O₃, CaO, MgO, Na₂O, K₂O, TiO₂, P₂O₅, MnO) plus a suite of trace and rare earth elements. Geologists use it to name volcanic rocks that are too altered to identify by eye, to track alteration, and to judge whether an intrusion looks fertile for porphyry copper or other magmatic-hydrothermal deposits.

The standard route for major oxides is a lithium borate fusion. In the lab procedure, 0.100 g of pulp is mixed with lithium metaborate and tetraborate flux and fused at 1,000 °C, then dissolved and read by ICP-AES, with lower limits of 0.01% for each oxide. The trace-element version fuses at 1,025 °C and finishes by ICP-MS. For samples high in sulphides, the lab notes it may substitute a peroxide fusion.

What a four-acid digestion misses

Many companies run four-acid digestion on everything, because it is cheap and gives low detection limits. A University of Tasmania study published in Geochemistry: Exploration, Environment, Analysis tested what that costs. Zivkovic and co-authors analysed 410 samples both ways. Summed across all samples, four-acid results for zirconium and hafnium came to 76% and 82% of the fusion values. Individual samples lost up to 85% of their zirconium. Niobium, tantalum and titanium also under-reported, but by similar proportions, so their ratios held.

The practical sting is in the ratios. The authors found that a standard Zr/Ti against Nb/Y rock-classification plot gave wrong answers from four-acid data, pushing samples towards more mafic and alkaline names. Zr/Hf fertility indicators were biased too, producing false-positive anomalies. Their recommendation: use more than one decomposition method when an interpretation depends on large parts of the periodic table. If only four-acid data exist, stick to plots whose ratios survive, such as Ti/Nb against V/Sc.

“Four-acid digestion recovered 76% of the zirconium that fusion found, summed over 410 samples.”

Assays for gold and base metals: fire assay and ore-grade methods

Gold breaks the rules the other methods rely on. It is rare, heavy and often present as a few discrete grains, so a quarter of a gram of pulp is a lottery ticket. Fire assay solves this with a large sample and a chemical concentration step. In the lab procedure, the pulp is fused with lead oxide, sodium carbonate, borax, silica and a silver inquart. The lead button collects the precious metals, and cupellation leaves a silver bead that holds the gold. The bead is dissolved in aqua regia and read by atomic absorption.

For the Au-AA21 method the charge is 30 g and the reporting range is 0.002 to 1 ppm gold. Au-AA22 uses 50 g with the same range. Anything above 1 ppm goes to an ore-grade overlimit method (Au-AA25 or Au-AA26). The USGS manual on fire assaying remains a useful long-form reference for the chemistry.

Base metals follow a similar two-step logic. A trace package such as four-acid ICP reports copper up to 10,000 ppm (1%), and anything higher is re-run by an ore-grade method built for high concentrations. When you see a certificate with values stacked at exactly 10,000 ppm, it means “over the limit”, not “exactly 1%”.

Lower limits are from the documents cited above as checked in September 2026. Labs revise their schedules, so confirm current limits on your lab’s latest method sheet before you choose a package.

Reading a certificate without getting fooled

• “<0.2” is a result, not a zero. Decide how you will treat it (for example, half the detection limit) before you plot anything.
• ✔ Check the method code on every column. Two gold columns from different charges are not interchangeable.
• Look for the overlimit flag. Values pinned at the top of a range need the ore-grade re-run.
• ⚠ Watch units: ppm, ppb, g/t and percent all appear on the same certificate.

QA/QC: what makes geochemical data usable later

Good geochemical analysis is not just a good lab. It is a program of checks that proves, sample by sample, that the numbers are right. The checks are the same whatever the method: certified reference materials (standards) of known grade, blanks that should read near zero, and duplicates at several stages (field, coarse reject and pulp).

EPA Method 6200 gives one concrete example for field XRF: a calibration verification check must come within ±20 percent of the true value, and precision is judged acceptable when the relative standard deviation is no more than 20 percent. Exploration programs set their own tolerances, but the idea is the same. Decide the pass mark before the results arrive, not after.

These checks pay off years later. If a project reaches the point of estimating a resource, the person who signs the estimate has to be satisfied with the data. The JORC Code’s Table 1, and the CIM guidance behind NI 43-101 technical reports, both ask about sampling, sample preparation, assaying and QA/QC. A soil or rock program with no standards and no duplicates is a program someone will have to redo.

1. Write the QA/QC protocol first: insertion rates, standards to use, and what triggers a re-assay.
2. Use standards that match the grade range you expect, not just one high-grade standard.
3. Insert blanks after likely high-grade samples to catch carry-over.
4. Plot every batch on control charts as it arrives.
5. Send a share of pulps to a second lab for umpire checks.

Where satellite analysis fits before the first sample

None of the methods above can tell you where to put the first soil line on a 10,000-hectare licence. That is the step our satellite-based mineral detection is built for. It analyses multispectral and hyperspectral imagery, where each mineral and alteration zone reflects energy in its own spectral signature, and flags likely mineralised target zones, alteration halos and structures such as faults and fractures. Nobody sets foot on the ground during this stage.

• ✔ Input: coordinates, a KML/KMZ file or a polygon of your area, plus the target mineral.
• Delivery: 5 to 20 business days, depending on area and mineral complexity.
• Output: prospectivity heatmaps, high-potential zones with estimated location and depth ranges, geological interpretation, and PDF plus georeferenced GIS files that load into the same GIS as your geochemical analysis results.
• Cost and time: satellite screening can lower early-exploration cost by up to 80 to 85% and cut timelines from months to days, compared with starting on the ground.
• ⚠ Limitation: a satellite target is an exploration target, not an assay and not a mineral resource. Soil, rock and drill geochemistry confirm or reject it.

The sensible sequence is satellite screening, then soil and rock geochemistry on the best zones, then drilling. Our guide to remote sensing for mineral exploration covers the spectral side in more depth, and the mineral exploration guide shows where each step sits in a full program. For a view of the 3D deliverable, see the satellite-driven 3D mineral prospectivity mapping overview. We have scanned 100,000+ hectares for 20+ mineral types across 25+ countries.

Key terms

Lower limit of detection

The smallest concentration a method reports with confidence. Below it, the certificate shows “<” plus the limit.

Aqua regia

A mix of hydrochloric and nitric acids (3:1 in the standard version) that dissolves sulphides and loosely held elements but not most silicates.

Four-acid digestion

Hydrofluoric, nitric, perchloric and hydrochloric acids; dissolves most minerals, but resistant minerals such as zircon survive in part.

Lithium borate fusion

Melting the sample with a borate flux at about 1,000 °C so almost every mineral dissolves; the standard route to whole-rock oxides.

Fire assay

A lead-collection fusion on 30 g or 50 g of pulp that concentrates gold into a bead for measurement.

Overlimit

A value above a method’s upper limit, re-run by an ore-grade method.

Frequently asked questions

Reviewed September 2026 against US EPA SW-846 Method 6200, the ALS geochemical procedures lodged with the Northern Territory Geological Survey (report GR501), USGS Data Series 801, Zivkovic et al. (2023) in Geochemistry: Exploration, Environment, Analysis, the USGS fire assay manual (Bulletin 1445) and the JORC Code 2012 edition.

Detection limits and sample weights are those printed in the cited method documents and can change when labs revise their schedules; confirm with your laboratory before choosing a package. Satellite targets are exploration targets, not mineral resources or ore reserves, and must be confirmed by sampling and drilling. This article is general technical information, not investment advice.