Two independent methods of crystal-size distribution analysis were compared: the Bertaut-Warren-Averbach XRD technique (MudMaster computer program) and high-resolution transmission electron microscopy (HRTEM). These techniques were used to measure thickness distributions of illite crystals (fundamental particles) from sets of illite-smectites from shales and bentonites that had expandabilities ranging from 86%S to 6%S. The illite-smectites were treated with a polymer (polyvinylopyrolidone, PVP) to separate them into fundamental particles for XRD and HRTEM investigations.
A systematic difference between XRD and HRTEM results was observed: XRD (area-weighted distributions) detected a larger fraction of thick (>4 nm) and a smaller fraction of thin crystals as compared to HRTEM (number-weighted distributions). As a result, XRD-determined distributions have larger mean thickness values and larger distribution parameters (α and β2). The measurements performed by the two techniques were verified by modeling XRD patterns of the PVP-illites, using the measured distributions as inputs. The modeling indicated that the XRD-determined distributions are very accurate. Selecting broader thickness distributions in MudMaster further improved the modeling results. The HRTEM measurements underestimate the proportion of coarse particles, in particular in shale samples, and this inaccuracy is attributed to the effect of using number-weighted (rather than area-weighted) distributions and to inaccurate counting statistics for thick crystals.