Performance of sample preparation for analysis of gold in samples of geological origin

We investigate the performance of sample preparation of gold ores using vibratory (bowl, ring and puck type) mills in common use in minerals analytical laboratories. The main criteria for effective grinding are using reduced grinding charge weights ≤ ca. 50% of nominal bowl capacity and using a grinding aid to prevent caking. We show that gold particles of millimetre scale can be comminuted to ≤ 100 μm by grinding in silica flour, bauxite, synthetic carborundum, or mixtures of silica and these materials using times of up to 5 minutes, and that 95% < 50 μm is achievable with extended grinding. This suggests that modified grinding technique can be used to make sample weights ≤ 5 g viable for routine analysis of gold in geological samples. We also demonstrate homogenisation of a gold- containing copper sulphide mineral flotation concentrate alone and in mixtures with silica by extended grinding at reduced charge weights. To support this work we develop a convenient new benchmark of gold ore sample preparation performance “G”, an apparent maximum gold particle size interpolated from replicate analytical variance in order to overcome the limitations of laborious sieve fraction analysis of gold particle size. We show useful agreement between G and sieve fraction analysis of gold particle size in samples and test the viability of G experimentally and by analysis of literature data.