Biodegradation and Ecotoxicity of Branched Alcohol Ethoxylates: Application of the Target Lipid Model and Implications for Environmental Classification

Abstract

With the advent of global regulations for safer detergent and an emphasis on a shift toward more environmentally friendly formulations, the environmental profile of surfactant chemistries have moved to the forefront of product formulation and design. The two cornerstones of surfactant environmental profiles are the ability to biodegrade in the natural environment and the ecological hazard profile. The objectives of this article are to describe biodegradation and aquatic toxicity data for a series of branched oxo‐alcohol ethoxylate (AEO) surfactants; to apply the target lipid model (TLM) for deriving model‐based threshold hazard concentrations (HC5) of AEO; and, finally, to accurately determine aquatic classifications for AEO surfactants for use in regulatory classification frameworks. Biodegradation results indicate a high level of biodegradability of branched AEO, with C8–C13‐rich oxo‐alcohols with 1–20 mol of ethoxylate meeting the readily biodegradable criteria. Results from acute and chronic toxicity tests indicated comparable or lesser aquatic toxicity versus linear AEO structures previously reported in the literature. The TLM model, applied a priori, resulted in good agreement with acute toxicity data (RMSE = 0.49) and is comparable to the root mean square errors (RMSE) previously determined for other narcotic chemicals (RMSE = 0.46–0.57). Model errors for invertebrates and fish were smaller than those for algae, with the TLM systematically overpredicting acute and chronic classification of two of seven branched AEO. Furthermore, TLM‐predicted HC5 values were determined to be sufficiently conservative, with 100% of observed chronic data (N = 79) falling above the HC5 threshold values, providing a useful tool for the risk assessment of AEO.

Predictive Toxicity Models for Chemically Related Substances: A Case Study with Nonionic Alcohol Ethoxylate Surfactant

“…Third, although limited chronic data prevented the estimation of chronic HC5s, comparisons of acute HC5s from ICE‐enhanced SSDs for 14 substances with their estimated chronic HC5s (Belanger et al 2006) produced an average acute‐to‐chronic ratio of 5, facilitating acute‐to‐chronic HC5 estimations. The latter results are consistent with a related study (Bragin et al 2019) documenting similarities between acute and chronic endpoints across several taxonomic groups (fish, invertebrates, and algae), leading to a similarly nonspecific acute‐to‐chronic ratio of 3.4. However, caution should be used because not all the test durations were standardized and the chronic studies were generally truncated designs.…”

“…Most AE–ICE predictions were within a 3‐fold difference of observed values, with models not grossly over‐ or underpredicting toxicity. Despite relatively limited species diversity and data availability, these models also produced reasonable prediction even for other nonionic surfactants outside the applicability domain (Brooke 1993; Dorn et al 1993; Bragin et al 2019). Analyses also showed that AE–ICE predictions were closer to the observed values than predictions from related ICE models with shared species pairs (Barron et al 2015; Raimondo et al 2015), further supporting the development of these types of models for groups of substances with similar chemical structures.…”

“…Similarly, some of the greater departures from observed values were based on AE-ICE predictions for the alga R. subcapitata. However, this may have been expected due to the nature of (Brooke 1993;Dorn et al 1993;Bragin et al 2019) endpoints and the multigenerational assessment compared with simple mortality or immobilization estimates for fish and invertebrates. Despite departures from observed values, most AE-ICE predictions were within the range of ≤5-fold of measured values.…”

“…The external validation dataset was further expanded to include sources of information for AE substances within and outside the applicability domain (i.e., chemical space boundaries defining the model). Alcohol ethoxylate substances within the applicability domain included AEs with similar alkyl chain lengths, alcohol ethoxylation, and linearity (data summarized in Human & Environmental Risk Assessment 2009), and data for related nonionic surfactants with different chemical characteristics (i.e., outside the applicability domain; Brooke 1993; Dorn et al 1993; Bragin et al 2019). Nonionic surfactants outside the applicability domain included: 1) branched and semilinear AEs covering alkyl chain lengths in the C12 to C15 range and alcohol ethoxylation in the EO5 to EO12 range (acute toxicity data only; Bragin et al 2019); 2) branched AEs covering alkyl chain lengths in the C12 to C15 range and alcohol ethoxylation of EO7 (Dorn et al 1993); and 3) branched nonylphenols including short carbon chain lengths of C4 (Brooke 1993) and ethoxylated (C9–EO9; Dorn et al 1993) substances.…”

“…FIGURE 2: Comparison of observed and interspecies correlation estimation predictions by predicted species (each panel), using external datasets within (Human & Environmental Risk Assessment 2000) and outside(Brooke 1993;Dorn et al 1993;Bragin et al 2019) the applicability domain, with sources represented by unique colors. Dots represent the mean predicted value, and bars indicate the 95% confidence interval predictions.…”