On the heterogeneity of biochar and consequences for its representative sampling

Representativeness of sampling has often been neglected for heterogeneous discrete materials such as biomass or products there from, potentially leading to inaccurate and non-reproducible/imprecise results. Here, we investigate how elemental and PAH concentrations in specimen non-probabilistically selected from a 750 kg biochar lot by grab sampling after initial mixing differ from those obtained from presumably probabilistically gathered samples by iterative mass reduction with fractional shoveling and repeated mixing. To assess reproducibility/precision, the concentration dependent relative standard deviations (RSD) are compared with the empirical Horwitz function, describing the within-laboratory RSD for multiple analyses, and by an F-test. While the precision was generally better for the samples, it was still satisfactory (RSD within 10%) for elemental concentrations in the specimen. However, replicates of specimen varied by 37% for PAHs. This number could be reduced to 4% in presumably representative samples. Accuracy as assessed by t-tests of specimen and samples revealed significant, but overall minor differences for some elements.