Determination of sulfur and sulfate half-ester content in cellulose nanocrystals: an interlaboratory comparison

Charged functional groups on the surface of cellulose nanocrystals (CNCs) are important for controlling the colloidal stability of suspensions and the interaction of the material with its environment. Quantification of surface groups is therefore an important metric for reproducible production of this nanomaterial. Here we report the results of an interlaboratory comparison (ILC) with twelve participants piloted by National Research Council Canada (NRC) that evaluated two methods for determining the sulfur content for an NRC reference material (CNCD-1) produced from CNCs extracted by sulfuric acid hydrolysis of wood pulp. Inductively coupled plasma-optical emission spectroscopy (ICP-OES) was used to measure the total sulfur content of CNCs after microwave assisted digestion in strong acid. Conductometric titration quantified the negatively charged sulfate half-ester groups on the CNC surface after dialysis and protonation using ion exchange resin. The data was analyzed with an ASTM method that employs h and k statistics to identify data sets for which the laboratory mean or standard deviation show excessive variation from the overall mean or standard deviation and the NIST consensus builder (NICOB) engine using the DerSimonian–Laird meta-analysis to determine a consensus estimate of the mean and its associated measurement uncertainty. The consensus estimates ± uncertainty for total sulfur and sulfate half ester content were (8771 ± 94) mg kg−1 and (252.0 ± 8.2) mmol kg−1, respectively. Neither approach identified any outliers in the data sets and the consensus estimates were in agreement with values obtained during certification of the reference material. The results confirmed that the difference between the sulfur content measured by conductometric titration and ICP-OES is larger than the combined 95% expanded uncertainties. The higher sulfur content measured by ICP-OES may reflect the presence of sulfur impurities that are not detected by titration. Overall the ILC provides validation for the protocols for the two methods and indicates that both are reasonably tolerant of small variations from the protocol. A more reliable method of assessing the titration endpoint and a quality control (QC) sample were identified as potential improvements to reduce variability for the conductometric titration.