Comparison of three analytical methods in the determination of gold in six Finnish gold ores, including a study on sample preparation and sampling

We used samples from six Finnish ore deposits to evaluate the efficiency of sample pretreatment procedures — crushing, splitting and grinding — and to compare three analytical methods based on the atomic absorption determination of gold following: (1) classical lead fire assay (FA); (2) the aqua regia leach (AR) followed by Hg coprecipitation of Au; and (3) the sodium cyanide (NaCN) leach. Sample size used for the method comparison is 20 g. The Au deposits and ore types were: Suurikuusikko and Osikonma¨ki, refractory ores in which Au is associated with arsenopyrite and pyrite; Pampalo and Kutemaja¨rvi ores with metallic Au and Au tellurides; and Jokisivu and Pahtavaara ores containing coarse-grained metallic Au. After crushing, the samples were split into three parts, one of which was put aside into storage. Two splits were further divided into two subsamples which were ground to two grades of fineness (<0.03 and <0.06 mm). The four subsamples thus obtained were analysed for Au using the three analytical methods. Each determination was performed five times on each of the four subsamples. According to t-tests on the FA results of the two splits, crushing and splitting produced samples of equal Au content in all six cases. Grinding to a finer grain size gave a significant difference in Au results only for the Pahtavaara ore sample. If the FA results are assumed to represent 100% recovery of Au, we obtained greater than 95% recoveries for all but the Suurikuusikko sample (87% recovery) by the AR leach method. We also obtained recoveries of over 95% by the NaCN leach method for the Pampalo, Kutemaja¨rvi and Pahtavaara samples, whereas recoveries for the other three samples varied between 73 to 92%. The AR leach was also performed on 1-g samples and the NaCN leach on 250-g samples. For three of the ore samples, decreasing sample size from 20 g to 1 g did not cause a significant difference in the variance of the Au results. Increasing the sample size from 20 g to 250 g significantly improves the representativity of only the Pahtavaara sample. For the Kutemaja¨rvi, Pahtavaara and Jokisivu ores, a sample larger than 250 g is needed in order to obtain a precision equivalent to that for reference samples.