Bias in carbon concentration and δ13C measurements of organic matter due to cleaning treatments with organic solvents

Interpreting the organic carbon content (TOC) and stable carbon isotopic composition (δ13C) of organic matter in the sedimentary rock record depends on our capability to accurately measure them, while excluding sources of contamination. This however becomes increasingly problematic as we analyze samples with ever-lower organic carbon content. Accordingly, organic solvents are sometimes used to remove contaminating traces of modern organic matter from ancient rock samples. However, especially for very low TOC samples, traces of solvents or their impurities remaining in the sample may contribute a significant organic contamination that can impact the bulk measurements of both TOC and δ13C values. This study, including three independent investigations performed in different laboratories, is the first detailed examination of the effect of cleaning treatments on the reliability of TOC and δ13C values in a range of natural rock samples and synthetic materials with low TOC content from below detection limit to 3330 ppm. We investigated the four most common organic solvents used to remove modern organic matter: dichloromethane (DCM), n-hexane, methanol and ethanol, and evaluated the effect of grain size and mineralogy. We find that (i) cleaning treatments with methanol, n-hexane and dichloromethane contaminate rock samples when used directly on sample powder, regardless of the grain size; (ii) this pollution buffers the natural variability and homogenizes the δ13C values of samples around the isotopic composition of the solvent, i.e. between −27 and −29‰; (iii) the extent of contamination depends on the solvent used, DCM being the most contaminating (up to 6000 ppm) and ethanol the only solvent that does not seem to contaminate rock samples above our detection limit; (iv) sample mineralogy also exerts an influence on the extent of contamination, clay minerals being more prone to adsorb contaminants. We conclude that the response of carbon concentrations and the stable carbon isotopic composition of organic matter in geological samples to cleaning treatments is neither negligible nor systematic when investigating samples with low carbon content.