Sample Preparation Biases in Automated Quantitative Mineralogical Analysis of Mine Wastes

Mineralogical information is becoming increasingly important for the interpretation and prediction of the long-term leaching behavior of mine waste rock and tailings, yet the collection of quantitative mineralogical data for these materials is complicated by biases introduced during sample preparation. Here, we present experiments with synthetic reference materials, soluble mineral (gypsum) and pulverized weathered waste rock samples to investigate potential artifacts that can be introduced during the preparation of granular sample specimen for quantitative mineralogical analysis. Our results show that, during epoxy-molding, particle segregation due to size is more important than that due to density, both of which can be effectively circumvented by cutting molds perpendicular to the orientation of settling. We also determine that sacrificing sample polish to avoid phase alteration need not impede phase attribution as long as surface roughness and slope are calibrated with sample-internal contrast references. Finally, bootstrapping analysis shows that variability in geometric and mineralogical particle parameters due to unresolved sample heterogeneity is small compared with other biases, even at particle numbers <25,000 at sizes >150 & mu;m. Our results demonstrate the importance of quantifying potential sources of error during sample preparation in quantitative mineralogical studies on mine wastes.

Automated summary

Mineralogical information is crucial for predicting and understanding the leaching behavior of mine waste rock and tailings in the long term. However, biases introduced during sample preparation make the collection of quantitative mineralogical data challenging. This study investigates potential artifacts in granular sample specimen preparation for mineralogical analysis, using synthetic reference materials, soluble mineral (gypsum), and pulverized weathered waste rock samples. The results highlight the importance of quantifying potential sources of error in sample preparation for accurate mineralogical studies on mine wastes.