Bias in the analysis of geological materials for gold using current methods

Abstract

Abstract A consistent pattern has been observed in the results obtained for Au in three soil and two porphyry copper ore samples serving as control reference standards in geochemical analyses. The mean reported by wet chemical methods (regardless of the measurement technique) was less than the mean by fire assay-based methods which, in turn, was less than the mean reported by laboratories using direct instrumental neutron activation analysis (INAA). These data have been obtained from 16 laboratories, some employing more than one method. Compared to INAA, values obtained for Au (at the 30–300 ppb level) using aqua regia (AR) dissolution were low by 24–42%, while those reported by fire assay-based methods were low by 14–26%. Studies of these samples and 32 rocks of widely varying composition revealed that the amount of Au remaining in the residue after AR attack ranged from 4% to 59% of the total. Gold dissolved by AR was determined by graphite furnace atomic absorption spectrometry (GFAAS) after extraction into MIBK (methyl isobutyl ketone), whereas Au remaining in the residue was determined by INAA. The slope of the line obtained by plotting Au by INAA vs. the sum of Au by AR/GFAAS and Au in the residue was not significantly different from 1.0. Modifications to the AR procedure such as prior attack by HCl or HF did not improve the recovery of Au in the three reference soils. However, reduction of sample size from 10 g to 1 g while maintaining the volume of AR at about 30 ml did increase results for Au. Furthermore, addition of HF in the attack on 1-g samples yielded results virtually identical to those obtained by INAA. It is thought that the poor extraction efficiency by AR is due to non-wetting of the larger sized sample, a lack of intimate, prolonged contact between the grains of the sample and the acid mixture and the presence of insoluble gangue surrounding Au particles. Constant agitation of the sample during evaporation with AR is desirable. The mean values obtained for Au in the soils by fire assay methods were not significantly different from the results by INAA (low by 14–19%); this was not the case for the two copper ore samples (low by 26%). This probably reflects the difficulties encountered in fire assay by high concentrations of Cu which hinder effective collection and separation of Au into the Pb button. However, the accuracy of the INAA method has not been established and is dependent upon measurement procedures and the degree of certainty associated with the Au values assumed for the reference materials employed for calibration. While estimation of precision does not present a problem, accuracy is difficult to assess in the absence of certified reference materials for Au at concentrations in the 10–300 ppb range. However, it is concluded that methods based upon AR dissolution can lead to low results, the magnitude being dependent upon the sample matrix and the mineralogical association of the Au present.