TECHNICAL ARTICLE: NUGGET EFFECT IN THE LABORATORY
- Gold often occurs in metallic form
- It is practically impossible to pulverise
- Representative samples are hard to obtain
- Larger analysis portions help improve sampling
- Multiple replicates increase confidence in results
A limiting factor in analysis of geological samples for gold content is the difficulty of obtaining a statistically valid sample for analysis. Sampling techniques in the field or at the mine generally significantly outweigh sampling error in the laboratory, but are outside of this document’s scope.
Sample preparation is a relatively low-technology process involving drying, crushing, splitting and pulverising the sample however gold presents more challenges in this respect than most analytes. Unlike most commercial metals, which are usually found as oxides or other compounds, gold often occurs in its metallic form in nature, whether in large nuggets, small veins or discrete, microscopic particles, due to its relative chemical inertness. When combined with the extreme malleability, ductility and density of the metal, this leads to significant problems in dispersing particles evenly through a portion of rock, so that a representative sample can be taken.
Modern laboratory milling machines can reliably reduce most mineral material to less than 75µm particle size in a reasonable time however it is not easy to achieve the same result with metallic gold particles in the sample, particularly in soft matrices. Gold particles will usually be beaten into flakes, rolled into cylinders or smeared onto grinding surfaces during preparation2, but will rarely be reduced to particles as fine as the host rock. This leads to what is commonly known as “nugget effect” – the inability to take a representative subsample from a portion of pulverised, gold-bearing geological material. Subsamples will tend to bias low compared to the “true” overall gold grade, with occasional very high bias.
Some types of milling equipment are better at reduction of gold particle size than others and the ring and/or puck type of mill is preferred in most laboratories today. Particle size reduction can be maximised by ensuring that mills are not overloaded, and increasing pulverising time, though there are limits beyond which extra grind time is not beneficial as matrix material starts to re-aggregate.
This problem is evident in high-grade (ppm-range) samples and also low-level (ppb-range) samples. Bearing in mind that a 75µm particle of gold will contribute around 250ppb of observed Au value to a 25g subsample, it can be seen that this problem can be significant for geochemical exploration.
Many laboratories routinely check anomalously high gold values by means of a repeat assay; in the case of nuggetty samples, the repeat will usually come back low (as the larger gold particles are rare and poorly distributed in the sample), and a further analysis may then be requested which will probably also give a low result. There is a temptation to discard the high value as a “flier” however it should be recognised that this is a valid result and should be considered when evaluating the data set.
Careful sample preparation and fine grinding will minimise nugget effect, however the only way to totally overcome it is either to i) assay the sample to extinction and use the results to calculate the overall concentration, or ii) perform a screened assay in which coarse gold particles are screened out and assayed separately.
If you suspect coarse gold to be a problem in your samples, discuss the matter with the laboratory manager to identify the best approach.
1.“Strategies for Reducing Sampling Errors in Exploration and Resource Definition Drilling Programmes for Gold Deposits”. CR Stanley and BW Smee, Nov. 2007, Geochemistry: Exploration, environment, analysis, 7, 329-340
2.“Evaluation of Pulverizing Techniques for Samples Containing Particulate Gold”; Hollis G.A. and Whisson B.S., Proceedings of Perth International Gold Conference, October, 1988.
3.At LabWest, pulveriser bowls are routinely barren-washed between samples to avoid carryover.