▎ 摘　　要

Purpose Graphene oxide (GO) nanomaterial has found wide potential industrial applications, but its life cycle environmental impact is not fully understood mainly because of lack of characterization factors (CFs) for the life cycle impact assessment. In this paper, we report the derivation of CF for freshwater ecotoxicity of GO based on the USEtox method. Methods The CF is derived based on the toxic effect factor, fate factor, and exposure factor of GO in the aquatic environment. The toxic effect factor is extracted from mechanistic toxicity studies available in the literature. The fate factor is derived with the colloidal method, and the exposure factor is determined through Langmuir adsorption isotherm for interactions between GO and dissolve organic carbon. Additionally, both fate factor and exposure factor are recalculated through the default mass-balanced model in USEtox. The apparent octanol-water partition coefficient (Kow) required in the mass balanced model is determined via experiment. Other parameters are calculated according to the apparent Kow. Results and discussion The study derives a CF of 777.5 potentially affected species (PAF) daym(3) kg(-1) for GO with a fate factor of 27.2 days and an exposure factor of 0.93. Sensitivity analysis suggests that variability from the effect factor is the dominant source leading changes in CF. The uncertainty of CF value can vary between similar to 1 and 10(3) PAF day m(3) kg(-1). Comparison between the colloidal and the mass-balanced models indicates that heteroaggregation may be underestimated by using the apparent partition coefficient, and thus, a much higher estimate of fate factor is obtained from the mass-balanced model. Additionally, empirical formulae in the USEtox to correlate other coefficients with Kow are not proper to calculate bioaccumulation and adsorption with dissolved organic carbon since a virtually a unit exposure factor is obtained. Conclusion The derived CFs can be readily incorporated into future toxicity assessment on GO. The fate factor is calculated in the colloidal model while adsorption of dissolved organic carbon onto GO surface should be derived from the Langmuir isotherm. Compared to the colloidal-based method, the conventional mass-balanced method may not be well applicable to GO due to the significant uncertainties in fate and exposure factors from applying the apparent partition coefficients. As three orders of magnitude variations in CF are caused by effect factor due to limited toxicity tests available for GO, more toxicological studies of GO on various species are needed in the future.