Geology, assay & laboratory

Specificationmeasures the metal itself, by mass. The sample charge is defined by the assay ton: a weight chosen 'so each milligram of a noble metal in the sample is equivalent to one troy ounce in one avoirdupois ton (2000 lb) of ore' — 29.167 g in North America, 32.667 g where the long ton is used. USGS Bulletin 1445 tabulates flux charges for a 15 g sample - sodium carbonate, litharge (PbO), silica, borax glass, calcium fluoride, flour as reductant and potassium nitrate as an oxidant for sulphide ores - with the quantities varying by ore type; the table should be read in the source rather than reduced to a single set of ranges here. Fusion: furnace at 1000 degrees C, down to 900 degrees C for about 17 minutes, then back to 1000 degrees C for a further 15 minutes. Cupellation for gold with a silver collector runs from 820 degrees C to 880 degrees C. USGS takes the gravimetric route 'when bead weight exceeds 20 milligrams', with gold and silver then separated by parting in nitric acid.

Where it falls downOnly practical at high concentrations: USGS takes the gravimetric route only when the bead exceeds 20 mg, and below that the bead is too small to weigh accurately, which is why almost all exploration gold is finished instrumentally instead. Silver and palladium are partly lost into the cupel and into the slag; gold is the best behaved of the noble metals in a lead collection, which is exactly why lead fire assay is not used alone for the full platinum-group suite. Fire assay collects 'Ag, Au, Ir, Os, Pd, Pt, Rh, Ru' while 'eliminating virtually all the associated gangue minerals', but recovery of iridium, osmium and ruthenium into a lead button is poor. It says nothing about how the gold occurs, and a 15 g or 30 g charge is a very small sample of a coarse-gold rock.

Specificationmeasures gold (and silver, platinum, palladium) after physical pre-concentration, which is why its detection limits beat any direct digestion. A USGS field-study method (Open-File Report 2007-1126) using a 10 g charge with a dore bead and AA finish reports a gold detection limit of 0.002 ppm (2 ppb). A USGS fire-assay-plus-ICP-MS package for Au, Pt and Pd gives detection limits of 1 ppb for gold and palladium and 0.5 ppb for platinum. Typical laboratory charges are 30 g (approximately one assay ton) or 50 g.

Where it falls downEverything the fire assay front end misses, it still misses — and now the small instrument detection limit creates false confidence in a number whose real uncertainty is set by the 30 g charge, not by the instrument. High-grade samples exceed the calibrated range and must be re-run by a gravimetric or ore-grade method. For coarse-gold material a single 30 g charge is simply too small a subsample, which is the entire reason screen fire assay exists.

Specificationmeasures the whole coarse gold population instead of hoping to catch a representative share of it. A Spectroscopy Europe sampling column by Dominy, Purevgerel and Esbensen - a trade-magazine column rather than a peer-reviewed paper - describes the route as 'a total sample preparation and screen fire assay route', with pulp quality specified as '95 % to be P95 75 micrometres' - that is, 95% of pulps are required to achieve 95% passing 75 micrometres - and notes that for screen fire assay 'a high precision would be expected for undersize fraction' because all the coarse gold has been removed from it. The same work shows the field split being increased 'up from 2 kg to 4 kg' to address coarse gold, targeting 'a theoretical sample support of 3 kg m-1'. note: the common commercial configuration (approximately 1 kg pulp screened at 100 mesh / 150 micrometres, oversize assayed in total, undersize assayed in duplicate) is laboratory practice described only in commercial laboratory literature; it is widespread convention, not a published standard, and the specific mesh and mass are therefore not asserted here.

Where it falls downExpensive and slow, so it is applied selectively rather than to every sample, which introduces a selection decision that must itself be defensible. It fixes the subsampling error at the pulp stage but does nothing about the far larger error introduced earlier: sampling studies find 'a large component of the total error is introduced during sample collection, especially during primary sampling', with the field relative sampling variability in the two case studies reported reaching 42% and 34%. It also cannot fix a field sample that was too small to contain a representative number of gold grains in the first place.

Specificationdissolves sulphides, sulphosalts, carbonates, many oxides, and metals adsorbed on clays and organic matter. does not dissolve the silicate lattice. Commercial ore-grade multi-element packages use a charge of a few tenths of a gram at a 3:1 HCl:HNO3 ratio and describe the attack as a partial digestion from which chromite, zircon and rutile are not dissolved; that description is commercial laboratory method text, no government or standards-body source giving the charge mass could be opened for this page, and neither is quoted here as though published. The closely related EPA Method 3050B nitric acid / hydrogen peroxide digestion, which is the source linked below and is not itself an aqua regia method, makes the same point in the clearest published language: 'this method is not a total digestion technique for most samples. It is a very strong acid digestion that will dissolve almost all elements that could become environmentally available', deliberately leaving silicate-bound elements undissolved; it uses 1-2 g wet or 1 g dry sample at 95 +/- 5 degrees C.

Where it falls downunder-reports, sometimes badly. Aluminium, potassium, titanium, zirconium, hafnium, chromium, barium, tin, tungsten, niobium, tantalum and the rare earths are all reported low, because the minerals that host them — feldspar, mica, zircon, rutile, chromite, barite, cassiterite, monazite — survive the acid. Volatile elements can be lost during digestion. Results are not comparable with four-acid or fusion results and must never be mixed in one dataset without a conversion study.

Specificationdissolves essentially all silicates, sulphides, carbonates and most oxides. A USGS 42-element package (Open-File Report 2006-1070) uses 0.25 g of sample in 'a mixture of hydrochloric, nitric, perchloric, and hydrofluoric acids at low temperature', reporting major elements (Al, Ca, Fe, K, Mg, Na, Ti) to a lower limit of 0.01%, arsenic and silver to 1 ppm; a companion USGS 40-element four-acid package (Open-File Report 2007-1126, 0.2 g charge) sets acceptance criteria of plus or minus 15 percent recovery and duplicate RSD no greater than 15 percent. The equivalent EPA microwave total-decomposition method uses about 0.5 g in '9 mL of concentrated nitric acid and usually 3 mL hydrofluoric acid', ramped to '180 +/- 5 degrees C in approximately 5.5 minutes' and held there for 9.5 minutes.

Where it falls down'near-total' is a load-bearing word. EPA's total-decomposition method states the constraint in published form: 'a few refractory sample matrix compounds, such as TiO2, alumina, and other oxides may not be totally dissolved', and those resistant grains 'might sequester target analyte elements'. In practice zircon, chromite, spinel, rutile, cassiterite, monazite, barite and tourmaline resist the attack, so zirconium, hafnium, chromium, tin, niobium, tantalum, barium, titanium and some rare earths come back low. Barium needs separate care, but not for the reason often given: perchloric acid supplies no sulphate and barium perchlorate is soluble. Where the sample carries sulphur, oxidative digestion converts it to sulphate and precipitates barium as insoluble BaSO4; where it does not, a low barium result simply reflects undissolved barite. Either way the control is mineralogy. If those elements matter — for lithogeochemistry, for tracing, for a critical-minerals assessment — the correct method is a fusion, not a stronger acid.

Specificationdissolves effectively everything, including the resistant accessory minerals that four-acid leaves behind. A USGS 55-element package uses 'a sodium peroxide sinter at 450 degrees C' followed by water leaching and nitric acid acidification, determining 55 elements including the full rare earth suite, with major elements reported in weight percent and traces in ppm. The low sinter temperature (450 degrees C, versus a full melt at higher temperature) is chosen to limit crucible attack and blank contribution.

Where it falls downThe flux is the problem. A large mass of sodium peroxide is added to a small mass of sample, so the dissolved solids load is high, the dilution is large, and detection limits are correspondingly worse than an acid digestion for the same element. Sodium and any element present in the flux or crucible become unmeasurable or unreliable. Zirconium and nickel crucibles each contaminate the solution with their own metal. Sulphur is oxidised to sulphate and lost to the method as a speciated measurement. It is a specialist, higher-cost route, not a routine replacement for four-acid.

Specificationdissolves the whole sample, including zircon, rutile, monazite and the other acid-resistant accessories, and is a true total decomposition for silicate rocks. Chromite is the known exception: it dissolves incongruently in lithium tetraborate flux (Merkle, X-Ray Spectrometry 33, 2004), which is why chrome-bearing material goes to a peroxide fusion instead. The only open USGS fusion-XRF procedure available for this page is written for coal ash rather than whole rock, and its figures should be read with that scope in mind. It uses '600 mg of coal ash with 5.400 g of lithium tetraborate', a roughly 1:9 sample-to-flux ratio, fused for '20 minutes at 1,100-1,200 degrees C' or 'one hour at 1,100 degrees C' in a furnace, with a few drops of dilute hydrobromic acid as a non-wetting agent, cast into platinum-gold moulds as glass discs. Eleven major and minor elements (Na, Mg, Al, Si, P, S, K, Ca, Ti, Mn, Fe) are determined with detection limits 'as low as 1 microgram per gram for elements having atomic numbers greater than 26', rising to 'several hundred micrograms per gram for aluminum and silicon' and 'several thousand micrograms per gram for sodium'; precision is '2 to 5 percent for coal ash and 10 percent for determinations of elements in the whole coal'.

Where it falls downVolatile elements are lost in the melt — sulphur, chlorine, mercury, arsenic, antimony, selenium and cadmium cannot be determined on a fused bead, and the USGS procedure warns that the higher temperature produced by a Meeker burner in a Claisse fluxer may volatilise sodium from certain samples. The heavy flux dilution raises detection limits, making the method poor for low-level trace work. Sulphide-rich and metal-rich samples attack platinum crucibles and are often unsuitable without pre-roasting. USGS also identifies a practical constraint that has nothing to do with chemistry: 'the most important factor limiting the accuracy of analysis is the restricted availability of standard reference materials'.

Specificationmeasures emitted light intensity at element-specific wavelengths, in solution — so it measures only what the digestion put into solution, and nothing else. EPA's method covers 31 elements in aqueous solutions and waste digests. It is deliberate about detection limits: 'IDLs are necessarily instrument-specific' and must be established in each laboratory as 'the mean of the blank results plus three times the standard deviation of 10 replicate analyses'. In geological use, a USGS four-acid ICP package reports major elements at 0.01% and elements such as arsenic and silver at 1 ppm.

Where it falls downFour interference classes are recognised: spectral (line overlaps and background emission), physical (viscosity and surface tension effects from high dissolved solids), chemical (molecular compound formation, ionisation and solute vaporisation effects) and memory (carry-over from a previous sample). Dynamic range at the low end is limited — for many trace elements ICP-OES simply cannot reach the concentrations exploration geochemistry needs, which is why ICP-MS exists. High total dissolved solids from a fusion degrade both precision and detection limits. And the cardinal limitation is not the instrument at all: an ICP-OES result on an aqua regia digest reports what aqua regia dissolved, not what the rock contains.

Specificationmeasures ion counts at specific mass-to-charge ratios in solution. EPA's method applies to 'determination of sub-microgram/L concentrations of a large number of elements in water samples and in waste extracts or digests', validated through multi-laboratory testing for 23 listed elements, though the method notes that 'ICP-MS has been applied to the determination of over 60 elements in various matrices'; instrument detection limits 'will generally be < 0.1 microgram/L' in simple matrices, though 'for less sensitive elements (e.g. Se and As). IDLs may be >= 1.0 microgram/L'. In geological application a USGS fire-assay ICP-MS package reaches 1 ppb for gold and palladium and 0.5 ppb for platinum; a USGS sodium-peroxide-sinter ICP-MS package covers 55 elements including the full rare earth suite.

Where it falls downFour interference classes are specified: isobaric elemental (two elements producing ions at the same mass, requiring isotopic correction), molecular/polyatomic (multi-atom ions such as ArCl+ and MoO+ that land on analyte masses — the classic case is 40Ar35Cl+ on 75As, which is why arsenic in a chloride-rich digest needs care), ion current tailing from an intense signal at an adjacent mass, and physical (matrix effects on nebulisation and ion transmission, handled by internal standardisation or dilution); memory and carry-over between samples are treated separately again. High dissolved solids suppress signal and clog the interface, so fusion digests must be heavily diluted, which erodes the sensitivity advantage. Ore-grade samples saturate the detector and must be re-run diluted or by another method.

Specificationmeasures absorption of element-specific light by free atoms. EPA's flame method covers 25 elements from aluminium to zinc in 'ground water, aqueous samples, extracts, industrial wastes, soils, sludges, sediments, and similar wastes', all solids requiring digestion first; example lower limits of quantitation in reagent water are magnesium 0.001 mg/L, sodium and lithium 0.002 mg/L, beryllium, cadmium and zinc 0.005 mg/L, calcium 0.01 mg/L and copper 0.02 mg/L, with EPA stressing these are 'guidance purposes only'. For mercury, USGS cold-vapour AAS on a 0.1 g charge gives a detection limit of 0.006 ppm; a USGS regional package using nitric/hydrochloric digestion and cold-vapour AAS reports mercury to 0.02 ppm. Tellurium by flame AA on a 2.0 g charge is reported to 0.1 ppm.

Where it falls downOne element at a time, so multi-element work is slow and expensive. Chemical interference occurs where atoms stay bound in molecules because the flame is too cool — phosphate suppresses magnesium, calcium and barium unless lanthanum is added. Spectral interference from other elements' absorption lines can be reduced by narrowing the slit. High dissolved solids cause light scattering (background interference), and hot flames cause ionisation, suppressed by adding 1,000 mg/L of an easily ionised element such as potassium, sodium, lithium or caesium. Flame AAS has a narrow working range and poor sensitivity compared with ICP-MS.

Specificationmeasures elements in the solid, non-destructively — nothing is dissolved, so there is no digestion bias, but also no pre-concentration. EPA's field-portable method covers 26 elements in soil and sediment and gives example lower limits of detection in clean quartz sand with count times of 100-600 seconds: lead 20 mg/kg, barium 20 mg/kg, chromium 150 mg/kg, cadmium 100 mg/kg — with the explicit caveat that these vary greatly with matrix, instrument and conditions. Intrusive (prepared-sample) analysis requires homogenisation, drying 2-4 hours at under 150 degrees C, grinding and sieving to 60 mesh; 'microwave drying is not a recommended procedure'. The laboratory fused-bead route, for which the published USGS example is written for coal ash, uses a roughly 1:9 sample-to-flux lithium tetraborate fusion at 1,100-1,200 degrees C and achieves 2-5% precision on major elements.

Where it falls downPoor to blind on light elements: EPA Method 6200 states that 'generally elements of atomic number 16 or greater can be detected and quantitated by FPXRF', which leaves out aluminium (13), silicon (14) and phosphorus (15). That method dates from 1998 and detectors have since improved, so treat the cut-off as method-specific rather than a law of the instrument. Detection limits are far too high for gold and most precious-metal exploration. Matrix effects are severe — large vanadium concentrations interfere with chromium measurement and iron interferes with cobalt quantitation, because the interfering lines sit within a few tens to a couple of hundred electron volts of the analyte lines, and where lead exceeds arsenic by a factor of ten the arsenic result is unreliable and EPA advises that 'all critical soil samples be sent off site for confirmatory analysis'. Moisture between 5 and 20% causes minimal error but saturated samples do not work. The single biggest control on data quality is sample preparation, not the instrument: EPA states 'homogenisation is the most critical factor when comparing the results'. In situ readings sample only the few millimetres of material facing the window.

Specificationmeasures total carbon and total sulphur, converted to gas. USGS total carbon: 'approximately 0.20 g' of sample (0.100 g above 10% C), combusted at 1350 degrees C, CO2 'measured by a solid state infrared detector', working range 0.05% to about 50%, limit of detection 0.03% and lower limit of determination 0.05%, with duplicate RSD of about 1% over 0.23 to 49.0% C. USGS total sulphur: 'approximately 0.25 g sample. mixed with 1 g vanadium pentoxide flux', 1350 degrees C, SO2 by infrared detector, working range 0.05% to about 35%, lower limit of detection 0.0004% and lower limit of determination 0.0007%, mean duplicate RSD 3% over 0.05 to 32%. ASTM D4239 covers sulphur in coal and coke 'by high-temperature tube furnace combustion'.

Where it falls downIt measures total carbon and total sulphur, and nothing else — it cannot tell carbonate carbon from organic carbon, or sulphide sulphur from sulphate sulphur. Those splits require separate determinations (acid leach then re-combustion, or coulometric titration of carbonate carbon, reported by USGS with limits of 0.01-50%). Interferences are specific and real: 'high concentrations of fluorine and molybdenum will interfere with the detection of CO2 by coating the cell walls and the detector', with samples above 0.2% fluorine or 1% molybdenum requiring reduced sample weights. Organic-rich material can combust too fast for the detector unless a retardant is added.