Emissions of water-soluble polymers from household products to the environment: a prioritization study

Abstract

Water-soluble polymers (WSPs) are widely used in household products, including cleaning and personal care products. However, unlike insoluble plastic polymers, the environmental risks of WSPs are poorly understood. This study was performed to identify polymers in household use and characterize their emissions to the environment and key data gaps for prioritization. An inventory of polymers was developed and these were broadly grouped based on structure. Information from patents was combined with literature data to estimate down-the-drain emissions for each polymer. For the polymers with the highest emissions, predicted environmental concentrations for surface water and soil were estimated. A total of 339 individual polymers were identified and categorized into 26 groups. The polymers with the highest down-the-drain emissions were sodium laureth sulfate (1.6–3.4 g capita⁻¹day⁻¹), styrene/acrylates copolymer (0.1–0.8 g capita⁻¹day⁻¹), and monoethanolamine-laureth sulfate (0.4–0.8 g capita⁻¹day⁻¹). An analysis of available fate and ecotoxicity data for 30 key high-emission polymers indicated that several are lacking in data. In particular, no data were found for styrene/acrylates copolymer and copolymer of polyethylene glycol/vinyl acetate, and the environmental fate of polyquaterniums and polyol ethoxylate esters has been understudied, particularly in light of their hazard potential. However, a lack of reporting of key polymer properties hinders analysis. We recommend increased transparency in reporting of polymer identities moving forward as well as experimental work determining fate, removal, and hazard of the prioritized high-emission polymers that are lacking in data.

Introduction

The environmental impact of polymers is an area of increasing interest to scientific and regulatory communities, given the widespread use, emissions, and potential persistence of these substances (e.g., European Centre for Ecotoxicology and Toxicology of Chemicals (ECETOC), 2019; Huppertsberg et al., 2020; Koelmans et al., 2017). In particular, the presence and impact of macro, meso, micro, and nano plastics in the environment has been the focus of substantial research (e.g., Burns & Boxall, 2018; Derraik, 2002; Koelmans et al., 2017; Thompson et al., 2009). However, the environmental risks of nonplastic polymers, such as water-soluble polymers (WSPs), have been relatively overlooked (Arp & Knutsen, 2020; Huppertsberg et al., 2020).

Until now, polymers have been exempt from many regulatory schemes for low molecular weight (MW) chemicals. Polymers were excluded under Registration, Evaluation, Authorisation and Restriction of Chemicals (REACH; European Parliament and Council, 2006) due to the wide range of polymers on the market and the assumption that they are likely low concern due to their high MW (European Chemicals Agency (ECHA), 2023). However, a fundamental aim of REACH is to operate under the precautionary principle, with decision-making based on potential to cause significant harm despite scientific uncertainty (Hansen et al., 2007). Recent progress towards incorporation of “polymers requiring registration” into REACH remains limited by several outdated assumptions regarding polymer properties (Groh et al., 2023), with more recent scientific data indicating the need for further study and environmental risk assessment (ERA) based on both potential hazards and environmental exposure of polymers.

The assumption that polymers with MW > 1,000 g mol⁻¹ are generally biologically inert has been called into question (Groh et al., 2023). Uptake of polyethylene glycol (PEG) of up to 4,000 and 8,000 g mol⁻¹ has been observed in fish embryos and tadpoles, respectively (Nascimento et al., 2021; Pelka et al., 2017), and toxic effects have been observed for PEG and various cationic WSPs despite low or absent cellular uptake (Connors et al., 2023; Nascimento et al., 2021). This is despite the fact that PEG has historically been classified as nontoxic (Duis et al., 2021). Toxicity of PEG and polyvinyl alcohol (PVOH) to fish and frogs has also been observed (Zicarelli et al., 2024), although several other studies have again found PVOH to be nontoxic (Alonso-López et al., 2021; McDonough et al., 2024; Nigro et al., 2022). Cationic polymers have been extensively studied due to their toxicity towards algae, fish, and invertebrates (e.g., Costa et al., 2014; Cumming et al., 2008; U.S. Environmental Protection Agency (USEPA), 1997; Connors et al., 2023; Hansen et al., 2023; Rawlings et al., 2022).

There are several emission pathways of WSPs to the environment, given their widespread use in agriculture, wastewater treatment (WWT), and household products (Arp & Knutsen, 2020), with millions of tonnes of WSPs used in Europe annually (Huppertsberg et al., 2020). Water-soluble polymers in household products include ethoxylated compounds as surfactants in handwash (Cowan-Ellsberry et al., 2014), polycarboxylates as anti-redeposition agents in detergents (DeLeo et al., 2020; Soap and Detergent Association (SDA), 1996); and polyquaterniums as antistatic agents in hair products (Johnson et al., 2016), along with a range of other polymers with various functions. These polymers may be released down-the-drain to WWT and subsequently to surface waters or soil (Figure 1). Several WSPs have been measured in wastewater effluent, sewage sludge, and environmental waters up to the mg L⁻¹ and µg g⁻¹ range, including PEG, poly(vinylpyrrolidone), poly(N-vinylcaprolactam), poly(ethyleneimine), and cationic polyacrylamide (Antić et al., 2011; Jovancicevic & Schwarzbauer, 2023; Pauelsen et al., 2023; Vidović et al., 2022, 2023, 2024), showing considerable levels of WSPs have been released to the environment for over a decade, some of which are poorly biodegradable.

However, environmental concentration data for most WSPs are lacking (Duis et al., 2021; Huppertsberg et al., 2020). Characterization of emissions and environmental concentration is essential to determine exposure for ERA, and may also be useful in prioritizing WSPs for further data collection prior to full ERA (Groh et al., 2023). Pecquet et al. (2019) found that of 65 polymers identified in household cleaning products, 18 had insufficient data available to conduct an ERA. The authors recommended further prioritization based on usage volumes and concentration in products (among other criteria). The fate and effects of WSPs in cosmetic products (PEG, acrylic acid homo- and co-polymers, and polyquaterniums) were evaluated by Duis et al. (2021), and the authors highlighted a lack of exposure data limiting conclusive ERA, with insufficient analytical methods for polymer monitoring and a scarcity of usage volume data impeding determination of both measured and predicted environmental concentrations (MECs and PECs; Duis et al., 2021). Given the diversity and abundance of WSPs in current use, conducting even preliminary ERAs for all individual polymers is challenging, and it is essential to develop methods to estimate exposure when usage and emissions data are not directly available as input parameters for modeling.

In this study, PECs for surface water and soil of high emission polymers from household products were estimated based on a combination of publicly available product ingredients information, literature data, and emissions modeling, with application of various assumptions and chemical groupings to account for insufficient data. Biodegradability and ecotoxicity data were also assessed and used in combination with PECs to prioritize key polymers and their data gaps. Although the majority of these polymers are water-soluble and synthetic, insoluble and natural polymers were also included to allow all key polymers to be identified and prioritized. This was to highlight several key types of polymers that are likely to be released to the environment and provide preliminary concentration estimates and an assessment of research needs.

Materials and methods

The overall methodology of the study is summarized in Figure 2, with each stage of the workflow outlined in detail in each of the following sections.

Products, brands, and ingredients inventory

Household cleaning and personal care products released down-the-drain at point of use were identified from U.K.-based supermarket websites (See online supplementary material Data 1 ). The products included in the final dataset were laundry detergents, dishwashing detergents (for machine and washing by hand), toilet cleaners, and personal cleansers for skin (handwash, bodywash, soap bars, and bath liquids) and hair (shampoo and conditioner). Some product subtypes were analyzed together (e.g., liquid and powdered laundry detergent; toilet cleaner and bleach/disinfectant; and 3-in-1 personal cleansers and bodywash) under the assumption that usage patterns and polymer concentrations are likely similar.

Major brands for each product type were identified from websites of the top four U.K. supermarkets (Tesco, Sainsbury’s, Asda, and Morrisons; Coppola, 2021). For shampoo, conditioner, personal cleansers, and toilet cleaners, only brands listed by multiple websites were included in data collection, due to the large numbers of brands (> 45 in each case) initially identified. Supermarket own-brands were not included due to limited availability of ingredients data for some product types; formulations are also likely to be similar to other market brands.

Ingredients of all individual products from each brand were collated from publicly available information on brand and company websites, between April 2020 and May 2021. For some brands, information on ingredients was unavailable, so these brands were removed from the dataset. The number of brands included in the final study for each product type are shown in online supplementary material Data 2.

Polymer identification and grouping

Polymers were identified following the Organisation for Economic Co-operation and Development (OECD) definition of a polymer (OECD, 1991), summarized as criteria 1-4 in Table 1, with two additional criteria (in keeping with the OECD polymer definition) used to narrow the scope of the study. Although enzymes were excluded (criterion 5), some other identified proteins/polypeptides were included (gelatin, keratin, wheat gluten, and whey protein) as these may consist of multiple proteins across a range of MWs (Bragulla & Homberger, 2009; Farrugia et al., 1998; Vensel et al., 2014; Wang & Lucey, 2003).

Polymers were defined based on names listed in product ingredients, and where necessary, using information on chemicals provided by the European Chemicals Agency (ECHA, 2020), and databases such as PubChem, the Environmental Working Group (EWG) Skin Deep database, SpecialChem, The Good Scents Company (TGSC) Information System, ChemIDplus (EWG, 2021; Kim et al., 2023; SpecialChem, 2021; TGSC, 2021), and Sigma Aldrich/Merck (Merck, 2021). In cases where insufficient information was available to make a definitive assignment (e.g., no information on average number of monomer units or MW majority [criteria 2 and 4]), most were assigned as polymers to give more conservative emissions estimates. Ingredients potentially identifiable as polymers but not included in the final dataset are listed in online supplementary material Data 3. In addition, although most identified polymers are water-soluble, solubility was not an applied criterion and thus some identified polymers were insoluble (e.g., styrene/acrylates copolymer, silicones, etc.; see Results and discussion section), to allow all key polymers in the studied products to be identified and prioritized.

Identified polymers were broadly categorized into groups based on structure, monomers, and functional groups. Structural features for classification of key polymer groups are shown in online supplementary material Data 8 (see also Results and discussion section). As further information on polymer properties relevant to grouping (including MW ranges, partitioning behavior, and charge density; ECETOC 2019) were not reported, these could not be accounted for and thus, further refinements and subgroups will likely be necessary for higher tier exposure assessment.

Polymer concentration and market penetration

Concentrations of polymers (F_pol) in products were obtained from patents identified using Google Patents. Search terms included product types (e.g., “laundry detergent composition”), and either individual polymers (e.g., “styrene/acrylates copolymer”) or polymer groups (e.g., “polycarboxylate”). The most preferred concentration ranges (percentage by weight) for each polymer or group were recorded for a minimum of three patents (where possible), or from the first 3–5 pages of search results. For example, if a patent listed polymer concentration as “generally 0.5 to 15%, preferably 0.5 to 10%, more preferably 1 to 5 wt. %”, values of 1% to 5% were recorded. Concentrations deemed most representative of the acquired patents for each polymer group, while generally accounting for higher concentrations to provide more conservative estimates, were selected for use in emissions modeling (See online supplementary material Data 4). It was assumed that individual polymers within groups would perform similar technical functions in products, and thus products containing multiple members of a polymer group were assumed to have the same concentration ranges as products containing only one member of a group. For example, if a detergent product listed two alcohol ethoxylate polymers in its ingredients, it was assumed these would both contribute to total nonionic surfactant concentration, rather than each being used separately at the patented concentration of nonionic surfactants, to avoid unrealistically high estimations of polymer concentration. In some cases, polymer concentrations were difficult to estimate; for example, PVOH is most commonly used as a film surrounding detergent capsules or tablets, however, mass concentrations of such films were not found. Concentrations reflecting use of PVOH in a dissolved or dispersed form in the products were instead used.

Estimates of market penetration (F_prod) were calculated for each polymer group and product type, as the fraction of products containing one or multiple polymers belonging to each group (Equation 1, See online supplementary material Data 5).

where F_prod is the estimate of market penetration, N_prod is the number of products of a particular type containing one or multiple members of the polymer group, and T_prod is the total number of products of the selected type included in the dataset. This approach thus made use of widely available product ingredients data to estimate proportions of products containing each polymer type, allowing generation of usage estimates where market penetration and production and import volumes are not publicly available.

Down-the-drain mass emissions

Usage data (U_prod) in g capita⁻¹day⁻¹ for each product type were obtained from the literature (See online supplementary material Data 6). Masses of polymer groups emitted down-the-drain from each product type were estimated using Equation 2.

where M_DTT(prod) is the mass of polymer released down-the-drain from a particular product type (g capita⁻¹day⁻¹), F_prod is the fraction of products containing polymer type, F_pol is the fractional concentration of polymer in product (from % by weight), and U_prod is the product usage (g capita⁻¹day⁻¹). Ranges of F_pol values given by patents were used, giving a range of values for M_DTT(prod) for each polymer group.

Total M_DTT estimates for each polymer group were then obtained as the sum of estimates for each product type according to Equation 3.

To estimate M_DTT for individual polymers within groups, contribution to total group M_DTT(prod) from each polymer was calculated from its number of occurrences (as a fraction of total occurrences of all polymers in the group, for each product type), and thus applied to M_DTT(prod) and summed as above. This approach was taken to account for the fact that different product types contain these individual polymers to varying extents, but F_pol estimates were assumed to represent total contribution from each group (as described for derivation of F_pol above).

Surface water exposure

Estimates of surface water exposure (PEC_SW) for the top 10 polymer groups with the highest M_DTT were obtained across the studied products (Equation 4), based on the method given by the European Medicines Agency (EMA; 2018; adapted from ECHA 2016).

where PEC_SW represents the predicted environmental concentration in surface water (µg L⁻¹), M_DTT is the mass of polymer group released down-the-drain (µg capita⁻¹day⁻¹), F_WWT is the fraction removed from water in wastewater treatment, WW_INHAB represents the amount of wastewater per inhabitant per day (L capita⁻¹day⁻¹), and DF shows the dilution factor for entering surface water. Default values for WW_INHAB and DF of 200 L capita⁻¹day⁻¹ and 10, respectively, were used (EMA 2018; ECHA 2016). Wastewater treatment removal data (F_WWT) were obtained from the literature; Web of Science and Google Scholar were searched for specific polymers or groups (with the relevant polymer names being listed in online supplementary material Data 8) and “wastewater” or “wastewater treatment”. Where multiple values were available, both within and between different sources, the highest and lowest values were applied to the lowest and highest bounds of the M_DTT estimates to account for the most and least conservative scenarios. This also allowed for the fact that many groups contained a broad range of polymers, which may exhibit different properties and fate in wastewater treatment.

The PEC_SW estimates for individual polymers within groups were calculated from the contributions of each polymer as described above for derivation of M_DTT for individual polymers. This approach was taken due to the lack of specific structural information (e.g., molecular weight) for most polymers, as it allowed calculations to be made for polymer groups as a whole (reducing the requirement for structurally specific input data for F_WWT) whilst estimating relative exposure of individual polymers from their relative use in products.

Soil exposure

Concentrations of polymers in sludge following wastewater treatment were calculated using Equation 5 for the top 10 polymer groups with the highest M_DTT.

where C_SLUDGE represents the concentration of polymer group present in sludge (mg kg⁻¹, dry wt), F_SLUDGE is the fraction of polymer partitioned to sludge in WWT, and S_INHAB is the mass of sludge per inhabitant per day (kg capita⁻¹day⁻¹, dry wt). A value for S_INHAB of 0.074 kg capita⁻¹day⁻¹ was used (Guo et al., 2016). Values of F_SLUDGE were obtained from the literature and/or F_WWT (see Results and Discussion section).

Estimates of soil exposure (PEC_SOIL), assuming no degradation of polymers following emission, for sludge-amended soil after the first year of sludge application were determined for each polymer group according to Equation 6, based on guidance given by ECHA (ECHA 2016).

where PEC_SOIL is the predicted environmental concentration in sludge-amended soil (mg kg⁻¹), A_SLUDGE represents the dry sludge application rate to land (kg m⁻²yr⁻¹), D_SOIL is the soil mixing depth (m), and RHO_SOIL is the bulk density of soil (kg m⁻³; ECHA 2016; Guo et al., 2016). Default values for A_SLUDGE, D_SOIL, and RHO_SOIL of 0.5 kg m⁻²yr⁻¹, 0.2 m, and 1,700 kg m⁻³, respectively, were used (ECHA 2016; Guo et al., 2016).

Estimates of PEC_SOIL for individual polymers within groups were calculated using the same method as for estimates of individual M_DTT and PEC_SW as described above.

Biodegradation and ecotoxicity data

Biodegradation and environmental effects (hazard) data were gathered from the literature for the ten highest-emission groups to further assess data gaps and research needs. For three groups, data had been previously compiled (Human & Environmental Risk Assessment (HERA) project 2004, 2009, 2014a, 2014b), and thus these data were used in the present study. For the remaining seven groups, searches were conducted using the ECOTOX Knowledgebase (USEPA, 2000) and Google Scholar; search terms included generic group names (e.g., “polyquaternium”) and specific polymer names (e.g., “aziridine homopolymer”), and in Google Scholar, were combined with additional terms (“biodegradation”, “ecotoxicity”, “fish toxicity”, “algae toxicity”, or “daphnia toxicity”). Existing PECs and MECs were also collated from the literature to compare estimates of the present study and evaluate the accuracy of PECs.

Results and discussion

Identified polymers, market penetration, and polymer grouping

A total of 339 individual polymers were identified (See online supplementary material Data 8) across 1,353 products and 10 product types (laundry detergent, machine and hand dishwashing detergent, toilet cleaner, bodywash, handwash, soap bars, bath liquid, shampoo, and conditioner). For most polymers, certain key information was not reported in product ingredients, including chemical names, Chemical Abstracts Service (CAS) numbers, MW, and charge density. Chemical Abstracts Service numbers could not be identified in a meaningful way for most polymers, because many are associated with multiple CAS numbers, and specific polymer compositions used were not reported. Information on monomer ratios was unavailable, and often not all monomer types were specified for copolymers. All identified polymers are therefore reported in this study as listed in the product ingredients, with the stipulation that for many polymers, multiple naming conventions exist that may be ambiguous; for example, PEG-150 may refer to PEG of average MW 150 g mol⁻¹ or PEG with an average of 150 monomer units, and “sodium acrylates copolymer” and “acrylic copolymer” could contain the same or different monomers (e.g., polymers with CAS numbers 25035-69-2 and 25133-97-5 both have “acrylates copolymer” listed as an identifier by ECHA, despite having different monomers [ECHA 2020]). Despite these data limitations, key types of polymers likely to be released down-the-drain could be identified. As emissions data are severely lacking for most WSPs, and manufacture/import volumes are typically not publicly available (Duis et al., 2021), these data are a useful first step to addressing this data gap for exposure and risk assessment until further data become available.

The polymer identified in the most products was sodium laureth sulfate, an anionic ethoxylated fatty alcohol commonly used as a surfactant in home and personal care products (Robinson et al., 2010), present in almost half the products studied (Figure 3). Note that although number of monomer units (n) is often < 3 for alcohol ethoxysulphate compounds in household products (which technically does not fulfil the OECD polymer definition), longer chain lengths are also used (e.g., n = 8; HERA 2004, See online supplementary material Data 8). Therefore, sodium laureth sulfate (and other compounds with unspecified n) may include both polymeric and nonpolymeric material (based strictly on the OECD polymer definition). However, in reality there is no chemical cut-off between polymers with an average of 3 and 4 monomer units, and low MW “nonpolymers” will have similar properties to low MW “OECD polymers” and may contribute to similar environmental effects as a mixture. Other commonly occurring polymers (present in > 10% of products studied) included dimethicone, polyquaternium-7, styrene/acrylates copolymer, guar hydroxypropyltrimonium chloride, and polyquaternium-10 (Figure 3).

Although the vast majority of polymers assessed in this study are water-soluble, some polymers such as dimethicone and styrene/acrylates copolymer were not WSPs. However as nonplastic polymers, many have received little attention in the context of environmental risk. It was therefore considered appropriate to include all identified polymers in the dataset to provide an overview of key polymer types, many of which have been rarely studied (see Knowledge gaps and future applications section).

The 339 identified polymers were categorized into 26 groups (See online supplementary material Data 8), based on monomer type, polymer structure and functional groups, and expected functions in products, with the exception of one group (“other”; containing 15 remaining unrelated polymers). These 15 polymers were analyzed separately to obtain individual emissions estimates before being combined into a group. The most common polymer groups by market penetration included alcohol ethoxylate salts and alcohol alkoxylates (used as anionic and nonionic surfactants, respectively; e.g., Cowan-Ellsberry et al., 2014), and polyquaterniums (used as antistatic and film-forming agents; e.g., Johnson et al., 2016). Other key groups included polycarboxylates, silicones, polyethers and copolymers, and polyol ethoxylate esters (Figure 4, See online supplementary material Data 5). The most prevalent groups by market penetration differ by product type; for example, cationic silicones are in the top five polymer groups by market penetration for conditioner and soap bars, with cellulose polymers being prevalent in machine dishwashing detergents and toilet cleaners. Some product categories contained certain polymer groups in close to 100% of the products studied (e.g., laundry detergent, machine dishwashing detergent, shampoo), whereas other product types had no polymer groups present in more than about half of the products (conditioner, toilet cleaner). Soap bars had the lowest market penetration of all polymers, with all groups present in < 4% of the products studied.

Down-the-drain emissions

Total M_DTT estimates for polymer groups (i.e., cumulative emissions for each entire group) were in the range of 4.9E-05 g capita⁻¹day⁻¹ (amine/formaldehyde polymers) to 4.8 g capita⁻¹day⁻¹ (alcohol ethoxylate salts; Figure 5; see online supplementary material Data 7).

Laundry detergents were the major contributor to total M_DTT estimates for many polymer groups, including several of the 10 groups with the highest M_DTT. For example, laundry detergents contributed 72% to the M_DTT of cellulose and derivatives, 71% to polycarboxylates, and 59% to each of silicones and polyvinyl alcohol (Figure 6). Handwash and bodywash also were major contributors, collectively, for several groups, including polyol ethoxylate esters (86%) and polyquaterniums (70%). The high contributions from these three product types reflect high usage rates of 11.3, 10.3, and 8.3 g capita⁻¹day⁻¹ for laundry detergent, handwash, and bodywash, respectively (Eriksson et al., 2002; Garcia-Hidalgo et al., 2017; International Association for Soaps, Detergents and Maintenance Products (A.I.S.E.), 2019), which were notably higher than values for other product types (See online supplementary material Data 6). However, for a small number of groups, other product types contributed more significantly to M_DTT (Figure 6), reflecting higher concentrations and greater market penetration.

Estimates of M_DTT for individual polymers within groups ranged from 2.7E-06 g capita⁻¹day⁻¹ (wheat gluten, proteins/polypeptides group) to 3.4 g capita⁻¹day⁻¹ (sodium laureth sulfate, alcohol ethoxylate salts group). For some groups, M_DTT was largely made up of only a few specific polymers; for example, 70% of the total M_DTT of alcohol ethoxylate salts was from sodium laureth sulfate alone (1.6–3.4 g capita⁻¹day⁻¹; see online supplementary material Data 8). Other groups showed a wider distribution, with e.g., alcohol alkoxylates having the highest emitted polymer (laureth-4) contributing only 14% (0.2–0.5 g capita⁻¹day⁻¹) to the total group M_DTT (See online supplementary material Data 8). The polymers with the highest M_DTT overall were sodium laureth sulfate, styrene/acrylates copolymer, and monoethanolamine (MEA)-laureth sulfate, with estimated M_DTT of 1.6–3.4, 0.1–0.8, and 0.4–0.8 g capita⁻¹day⁻¹, respectively (See online supplementary material Data 8). However, for highly homogenous groups, for example, alcohol ethoxylate salts, study of exposure as a total mixture may be worthwhile, because these are likely to have similar fate behavior and ecotoxicological effects. Increased transparency in reporting of polymer structure and MW for other groups (e.g., polycarboxylates) will help clarify which subgroups may be necessary and which group members should potentially be assessed as a mixture.

Wastewater treatment and PEC calculation

The top ten polymer groups with the highest M_DTT (Figure 5) were prioritized for calculation of PECs, incorporating removal in WWT (F_WWT and F_SLUDGE) based on available literature data (Table 2). Grouped polymers were analyzed together for WWT removal to facilitate data collection and reduce data requirements, and due to the lack of available data for most polymers.

Some data on F_WWT were available for most groups (Table 2). For polyethers and copolymers, polycarboxylates, and cellulose and derivatives, data were available from OECD simulation experiments. For alcohol ethoxylate salts and alcohol alkoxylates, data from degradation experiments were combined with data from monitoring studies to estimate removal. Monitoring data were also available for silicones to estimate F_WWT, and data for PVOH were based on an extensive literature review with data from various degradation studies (Rolsky & Kelkar, 2021). However, for polyquaterniums, only modeling data were available, and for polyol ethoxylate esters and starch, data were not found. For polyol ethoxylate esters, values of F_WWT = 0 and F_SLUDGE = 1 were used (i.e., assuming all of this polymer type is either released in effluent or partitioned to sludge, respectively) meaning both PECs for this polymer group represent conservative worst-case scenarios for surface water and soil. For starch, significant biodegradation is likely; an estimate of 50% degradation was thus used from the data for cellulose, which will degrade more slowly than starch, to provide a conservative estimate while accounting for likely biodegradation. For these polymers, information on relative removal via partitioning and degradation was unavailable, and thus F_WWT = F_SLUDGE was applied for starch and cellulose, assuming the nondegraded fraction is entirely partitioned to sludge for PEC_SOIL calculations.

Moreover, F_SLUDGE = F_WWT was also applied to groups expected to have limited biodegradability (polycarboxylates, silicones, and polyquaterniums; i.e., no degradation assumed). For the remaining four groups (alcohol ethoxylate salts, alcohol alkoxylates, polyethers and copolymers, and polyvinyl alcohol), derivation of the fraction present in sludge (F_SLUDGE) was possible (i.e., accounting for degradation).

For five groups (alcohol ethoxylate salts, alcohol alkoxylates, polyethers and copolymers, silicones, and polyvinyl alcohol), removal estimates were relatively high, ranging from approximately 70% to close to 100% (Table 2), suggesting relatively small proportions of these polymers are released in treated effluent. For polycarboxylates, although the upper estimate of WWT removal was also high (98%), the lower estimate (9%) indicates high variation depending on polymer structure and MW (HERA 2014a, 2014b). However, due to the lack of MW data for polycarboxylate polymers identified in this study (with only generic names such as “polyacrylic acid” being reported), and lack of specific structural data for some polymers, it was not possible to further subdivide this group to refine removal estimates and subsequent PEC. In addition, styrene/acrylates polymer is insoluble and thus likely to be efficiently removed from water in WWT by flocculation; however, in the absence of specific WWT removal data for this polymer, the full range of 9%–98% removal applied to polycarboxylates was used (assuming 98% removal is realistic for an insoluble material, but 9% removal provides a conservative estimate where specific data are lacking), with the stipulation that PEC_SW are likely towards the lower end of the estimated range for this polymer, and vice versa for PEC_SOIL (see subsequent sections). For the remaining two polymer groups (polyquaterniums, and cellulose and derivatives), WWT removal was estimated at ≤ 50% (Table 2), suggesting relatively low removal rates for these polymers and high potential for release in WWT effluent.

Predicted environmental concentration in surface water

Total PEC_SW estimates for whole polymer groups (See online supplementary material Data 9a) ranged from 0.8 µg L⁻¹ (alcohol alkoxylates) to 915 µg L⁻¹ (polycarboxylates). Estimates of PEC_SW for individual polymers ranged from 7E-05 µg L⁻¹ (coceth-7, alcohol alkoxylates group) to 512 µg L⁻¹ (sodium laureth sulfate, alcohol ethoxylate salts group; see online supplementary material Data 9c). The three polymers with the highest PEC_SW were sodium laureth sulfate (0.8–512 µg L⁻¹), styrene/acrylates copolymer (0.8–349 µg L⁻¹), and sodium polyacrylate (0.4–160 µg L⁻¹). Although the 10 polymers with the highest M_DTT all belonged to the alcohol ethoxylate salts, polycarboxylates, alcohol alkoxylates, and starch groups, the top 10 polymers in terms of PEC_SW also included polyquaternium and polyol ethoxylate ester polymers, with PEG-7 glyceryl cocoate, polyquaternium-7, and PEG-200 hydrogenated glyceryl palmate having PEC_SW of 20–99, 2.7–84, and 15–74 µg L⁻¹, respectively.

For polycarboxylates, PEC_SW from this study were in good agreement with literature values (See online supplementary material Data 10a). Total PEC_SW from polyacrylic acid (PAA) within this group estimated in our study ranged from 0.5 to 222 µg L⁻¹, and literature studies report PEC_SW of 70–570 µg L⁻¹ (DeLeo et al., 2020) and 43–110 µg L⁻¹ (HERA 2014a) for this polymer. Similarly, total PEC_SW from polyacrylic acid-maleic acid copolymers (PAA-MA) in the present study ranged from 0.4 to 166 µg L⁻¹, and literature studies report PEC_SW of 20–130 µg L⁻¹ (DeLeo et al., 2020) and 35–49 µg L⁻¹ (HERA 2014b). However, these literature data cover only PAA and PAA-MA, leaving other group members unstudied.

Literature data for silicones and polyquaterniums were also generally in good agreement with data of the present study (See online supplementary material Data 10a). Fendinger et al. (1997) reported MECs of dimethicone (polydimethylsiloxane) in receiving water of < 5–7 µg L−1 (however, note that all but one sample were below the limit of detection at 5 µg L⁻¹). Total PECSW for dimethicone and simethicone (which is a mixture of dimethicone and SiO2) was estimated at 0.8–7 µg L⁻¹ in our study. For polyquaterniums, the literature value of PECSW = 0.72 µg L⁻¹ for polyquaternium-68 (Australian National Industrial Chemicals Notification and Assessment Scheme [NICNAS] 2009) also shows close agreement with our study, with the PECSW of polyquaternium-68 being calculated as 0.02–0.6 µg L⁻¹. However, Cumming (2008) reported PECSW for polyquaterniums in Australia at 0.039–0.46 µg L⁻¹, which is significantly lower than the range of 5–142 µg L⁻¹ for the entire polyquaterniums group in our study. The author noted insufficient data for estimation of the mixture of polyquaterniums present, including range of charge densities and MW, and thus a general approach was applied to estimate PECSW using a “theoretical” polyquaternium with properties typical of other identified polymers (Cumming, 2008), similar to the broad WWT removal approach used in our study. However, the author also noted incomplete manufacture and import estimates (Cumming, 2008), whereas the methods of this our study did not rely on production or import volumes.

Comparison of estimates from our study with literature data for other polymer groups is less straightforward. For alcohol ethoxylate salts, surface water MECs range from 0.01 to 10.3 µg L⁻¹ (Popenoe et al., 1994; Sanderson et al., 2006; See online supplementary material Data 10a), and literature PEC_SW range from 0.42 to 54.87 µg L⁻¹ (Environment and Climate Change Canada (ECCC) 2019, See online supplementary material Data 10a). Although most literature data report values for specific ranges of carbon (C) and ethoxy (EO) chain lengths, most members of this group did not have reported EO chain lengths or MW in our study and thus only comparison to the entire group is possible. The PEC_SW of our study ranges from 1 to 731 µg L⁻¹ for this group, and although there is significant overlap with literature data, the higher upper estimate suggests greater transparency in reporting of polymer chain lengths and MW distributions may allow more direct comparisons and thus better evaluation of PEC_SW estimates.

Similarly, for alcohol alkoxylates, although most group members have reported C and EO chain lengths, each individual polymer still contains a distribution of chains across a range of MWs, and actual mixtures and distributions cannot be estimated from available data. The summed PEC_SW for alcohol ethoxylates (i.e., only ethoxy ethers, not propoxy ethers) with C = 9–18 and EO ≤ 21 (deemed to be generally representative of ranges in literature studies) is estimated at 0.4–198 µg L⁻¹ (See online supplementary material Data 10a). Ranges of literature MEC are reported as <0.011–50.9 µg L⁻¹ (Lara-Martin et al., 2011; McAvoy et al., 1998; See online Supplemental Data), and PEC_SW are reported as 0.06–16.76 µg L⁻¹ (ECCC) 2019; See online Supplemental Data), again showing overlap, but with the values in our study ranging to one order of magnitude higher. However, as described above, the lack of information on specific polymer composition in this study limits the usefulness of these comparisons. For example, polymers such as steareth-20 included in the summed PEC_SW of our study will contain polymer chains of 20 and lower (within the ranges reported in literature studies), as well as higher chain lengths, which are not reported in literature studies.

A similar challenge exists for polyethers, with the only directly comparable literature data being those of Pauelsen et al. (2023), who reported PEG MECs (quantified independently of MW) up to 11 µg L⁻¹ in surface water in Germany. This shows good agreement with total PEC_SW for all PEG polymers in our study, which is estimated as 0.6–30 µg L⁻¹ (See online supplementary material Data 10a). For other reported literature MECs, general estimates can be made from our study for comparison with key polymer chain lengths (See online supplementary material Data 10a), but these are again limited by the lack of data on MW distributions, and thus there is likely significant overlap between polymers included and excluded in summed PEC_SW with those reported in the literature.

Predicted environmental concentration in soil

Total PEC_SOIL estimates for whole polymer groups (See online supplementary material Data 9b) ranged from 0.02 mg kg⁻¹ (polyquaterniums) to 39 mg kg⁻¹ (polycarboxylates). Estimates of PEC_SOIL for individual polymers ranged from 9E-06 mg kg⁻¹ (modified guar hydroxypropyltrimonium chloride, polyquaterniums group) to 15 mg kg ⁻¹ (styrene/acrylates copolymer, polycarboxylates group; see online supplementary material Data 9c). The three polymers with the highest estimated PEC_SOIL were styrene/acrylates copolymer (0.1–15 mg kg⁻¹), sodium polyacrylate (0.06–7 mg kg⁻¹), and sodium acrylic acid/MA copolymer (0.06–6 mg kg⁻¹), suggesting polycarboxylates are a key group in terms of environmental soil exposure to polymers. Members of the silicones and polyvinyl alcohol groups were also present in the top 10 polymers with the highest PEC_SOIL, with dimethicone and PVOH having estimated PEC_SOIL of 0.9–4 and 0.5–3 mg kg⁻¹, respectively.

Literature data for PEC_SOIL are scarce and even more limited than data for surface waters (See online supplementary material Data 10b). Measured environmental concentration data were available only for silicones, with concentrations of polydimethylsiloxane measured in sludge-amended agricultural soil ranging from < 0.41 to 10.4 mg kg⁻¹ (Fendinger et al., 1997). These values are in good agreement with the sum of PEC_SOIL for dimethicone and simethicone determined in our study (1–4 mg kg⁻¹). Literature PEC_SOIL for PAA (polycarboxylates group), reported to range from 0.47 to 4.37 mg kg⁻¹ (HERA 2014a), is also in close agreement with estimates of our study (0.09–9 mg kg⁻¹ for all PAA polymers; see online supplementary material Data 10b). Literature PEC_SOIL for PAA-MA (26.8–35.2 mg kg⁻¹; HERA 2014b) is, however, higher than in our study (0.07–7 mg kg⁻¹).

Literature PEC_SOIL for polyquaternium-68 has been reported at 0.0055–0.055 mg kg⁻¹ (NICNAS 2009), showing overlap with the range for polyquaternium-68 in our study (0.00009–0.009 mg kg⁻¹; see online supplementary material Data 10b), although ranging one order of magnitude higher. Comparison of data for alcohol alkoxylates and alcohol ethoxylate salts with literature data again remains challenging, as discussed previously, with broad comparisons suggesting close agreement for alcohol ethoxylates but not alcohol ethoxylate salts (See online supplementary material Data 10b) but further data on polymer mixtures and MW distributions being required for more in-depth comparisons.

Biodegradation data

Several of the identified polymers are likely to further biodegrade in the environment (Table 3). Alcohol ethoxylates and alcohol ethoxylate salts of typical chain lengths are readily biodegradable (HERA 2004, 2009), and natural polymers such as starch are expected to rapidly biodegrade. However, modified natural polymers may be less susceptible to biodegradation; for example, carboxymethylcellulose (cellulose gum) is not readily biodegradable (Menzies et al., 2023), and hydroxyethylcellulose has been classed as nonbiodegradable (Bading et al., 2024), with these two polymers being the highest emitted in the cellulose group (See online supplementary material Data 8). Polypropylene glycol (PPG) and PEG (making up the majority of the polyethers group) are typically readily or inherently biodegradable (Beran et al., 2013; McDonough et al., 2023; Menzies et al., 2023; West et al., 2007), with low MW PEG and PPG degrading rapidly in river water (Zgoła-Grześkowiak et al., 2006). However, high MW polyethers (≥ 14.6 kDa) take significantly longer to degrade in ready tests (e.g., 86% biodegradation after 160 days for PEG of 500 kDa; Bernhard et al., 2008; Menzies et al., 2023), and biodegradation in marine water is significantly slower than in freshwater (Bernhard et al., 2008; West et al., 2007). Polyvinyl alcohol is also readily biodegradable (McDonough et al., 2023; Menzies et al., 2023) but with negligible biodegradation in marine water (Alonso-López et al., 2021).

However, polycarboxylate polymers such as PAA and PAA-MA are not readily biodegradable (HERA 2014a, 2014b) and are slow to degrade in environmental matrices (Table 3), with biodegradation decreasing at higher MW. Similarly, polyquaterniums are likely to persist in the environment (although few data are available for these substances; Duis et al., 2021), and silicones are nonbiodegradable (e.g., Darracq et al., 2010; de Albuquerque Vita et al., 2023), although they may be removed by abiotic processes (reviewed by (Graiver et al., 2003). Polyol ethoxylate esters are generally lacking in data; a REACH registration dossier for polysorbate 20 (CAS 9005-64-5) states ready biodegradability (ECHA 2020), and bacterial strains isolated from environmental soil and sediment have been found to degrade polysorbates (Nguyen, 2018; Nguyen et al., 2021; Yeh & Pavlostathis, 2005), suggesting potential for biodegradation in the environment. However, environmentally relevant data are needed for other key members of the group to confirm this, particularly higher MW polymers (e.g., PEG-200 hydrogenated glyceryl palmate).

For readily biodegradable polymers, the PECs calculated in our study are likely to be reduced further from the point of release to the environment, and thus higher-tier modeling incorporating biodegradation may be useful to further refine PECs. However, it is worth noting that several of these biodegradable polymers have been detected in the environment (e.g., PEG has been detected in surface water at levels similar to PEC_SW of our study; Pauelsen et al., 2023). Furthermore, considerable environmental exposure to even biodegradable WSPs close to the point of release, before significant degradation has occurred, is likely and thus may be relevant in ERA.

Ecotoxicity data

Literature ecotoxicity data were collated for the top 10 polymer groups with the highest emissions to identify data gaps and highlight potential polymers of concern (i.e., those polymers for which PEC_SW exceed concentrations which cause ecological effects).

For polycarboxylates, alcohol ethoxylate salts, and alcohol alkoxylates, extensive data were available and previously summarized in the HERA reports (HERA 2004, 2009, 2014a, 2014b; Table 4), with acute and chronic data for all standard species groups (fish, algae, and crustaceans). Chronic toxicity for alcohol ethoxylate salts and alcohol alkoxylates was reported to range to < 0.1 mg L⁻¹, with upper estimates of PEC_SW for these groups of 0.7 and 0.4 mg L⁻¹ (See online supplementary material Data 9a), respectively, although as noted previously, most group members are readily biodegradable, which is likely to reduce PEC_SW. All recorded effect concentrations for polycarboxylates were > 1 mg L⁻¹ (Table 4), with PEC_SW ranging to 0.9 mg L⁻¹ for the entire group, suggesting a low potential for ecological hazard. However, some key polymers within the group lack ecotoxicity data, including styrene/acrylates copolymer.

Cellulose (and derivatives) and polyvinyl alcohol were the least toxic groups; although some data indicated moderate toxicity of carboxymethylcellulose (cellulose gum) and PVOH to crustaceans (acute half-maximal effect concentration [EC50] = 87.26 and chronic no observed effect concentration [NOEC] = 2.18 mg L⁻¹, respectively; Arfsten et al., 2004; Warne & Schifko, 1999), these concentrations were orders of magnitude above maximum PEC_SW for the groups (0.06 and 0.019 mg L⁻¹, respectively; see online supplementary material Data 9a). Silicones are also nontoxic; although some high toxicity has been reported with median lethal toxicity (LC50) values for fish as low as 3.16 mg L⁻¹ (Birge et al., 1978), aquatic ecotoxicity tests are generally considered inapplicable to this group of substances due to their insolubility in water (Stevens, 1998), with more relevant sediment and soil tests indicating no toxicity even at high polymer concentrations (Craig & Caunter, 1990; Henry et al., 2001; Stevens et al., 2001; Tolle et al., 1995). Similarly, although few data were found for starch polymers, toxicity is highly unlikely for the natural polymers of this group.

Polyethers such as PEG are generally considered nontoxic, with literature ecotoxicity data typically indicating no or low toxicity (Table 4). Although moderate toxicity to algae has been observed for PEG 400, with acute EC50 values of 18.51 mg L⁻¹ and 67.79 mg L⁻¹ observed for C. tenuissimus and P. tricornutum, respectively (Pastorino et al., 2022), these values are orders of magnitude above PEC_SW for the entire polymer group (up to 0.09 mg L⁻¹). However, Kutt and Martin (1974) reported 36% and 59% mortality after 2-day exposure of algae to 0.0125 and 0.05 mg L⁻¹, respectively, of a PEG/PPG copolymer of MW 2,700 ± 300 g mol⁻¹, suggesting high toxicity, which may be due to surfactant properties; maximum PEC_SW values for all PEG/PPG copolymers were estimated at 0.003 mg L⁻¹ in our study, approximately an order of magnitude below these observed effects. Similarly, although acute data for polyol ethoxylate esters all indicated moderate or low toxicity (Table 4), chronic NOECs for algae and crustaceans for polysorbate 20 (3.16 and 10 mg L⁻¹, respectively; Straub et al., 2014) indicated high toxicity, although again, these were higher than maximum PEC_SW for this polymer (0.058 mg L⁻¹; see online supplementary material Data 9c). Key members of this group, including PEG-7 glyceryl cocoate and PEG-200 hydrogenated glyceryl palmate, had no ecotoxicity data available.

The polymer group with the highest ecotoxicity potential is undoubtedly polyquaterniums, with acute effects observed at concentrations < 1 mg L⁻¹ for fish, algae, and crustaceans, indicating very high toxicity. However, none of the recorded effect concentrations for specific polymers were exceeded by PEC_SW of our study. For polyquaterniums -6, -16, -28, and -55, the lowest effect concentrations were 0.03, 0.12, 1.6, and 0.5 mg L⁻¹, respectively, and corresponded to acute EC50 values for algae (polyquaternium-6 and polyquaternium-16) and fish (polyquaternium-28 and polyquaternium-55; Cumming, 2008; Cumming et al., 2008; Hansen et al., 2023). Maximum PEC_SW values for these polymers determined in our study were 0.001, 0.0001, 0.0002, and 0.0002 mg L⁻¹, respectively (See online supplementary material Data 9c). For polyquaternium-10, the lowest effect concentration (acute algae EC50 of 0.04 mg L⁻¹; Cumming, 2008) was closer to but still greater than maximum PEC_SW (0.01 mg L⁻¹). No ecotoxicity data were found for polyquaternium-7 or guar hydroxypropyltrimonium chloride, the two highest emitted polymers in this group (maximum PEC_SW = 0.08 and 0.02 mg L⁻¹, respectively; see online supplementary material Data 9c).

However, comparisons of PEC_SW for these polymers (and polymers in other groups) with ecotoxicity data are overall limited by a lack of information on MW and charge density in our study, which will affect ecological effects (e.g., Hansen et al., 2023; Rawlings et al., 2022). It should also be noted that predicted no-effect concentrations (PNECs) and environmental quality standards are typically 1–3 orders of magnitude lower than directly observed EC50s and NOECs due to application of assessment factors to account for uncertainty and ensure protection of the majority of species (European Commission 2011). Where PEC exceeds PNEC, this indicates unacceptable environmental risk. Thus, although no individual polymers were predicted to have PEC_SW above effect concentrations, this does not necessarily preclude risk. Determination of PNECs in our study was limited by most individual polymers lacking ecotoxicity data for all three standard species groups (European Commission 2011). Predicted no-effect concentrations are available for PAA, PAA-MA, alcohol ethoxysulfates and alcohol ethoxylates, polysorbate 20, polysorbate 80, and polyquaternium-67 (HERA 2004, 2009, 2014a, 2014b; Simões et al., 2021; 2022; Straub et al., 2014). The latter two of these polymers were not identified in our study, and comparison of PNEC values to PEC_SW for the remaining polymers was again limited by a lack of structural information such as MW. Direct comparison was only possible for PAA and PAA-MA (because literature PNECs accounted for a range of MW) and polysorbate 20 (because the structure and MW of this polymer is well defined). The relevant PNEC and PEC_SW values for these polymers are shown in Table 5. Although PEC_SW values for both polycarboxylate polymers are significantly lower than their PNECs, for polysorbate 20, maximum PEC_SW (58 µg L⁻¹) is similar to the aquatic PNEC (63 µg L⁻¹; Straub et al., 2014), suggesting polyol ethoxylate esters may be a priority for further risk assessment; information on removal in WWT for this group may thus be useful to refine PEC_SW. In addition, PNECs for various polyquaternium-67 polymers with differing charge densities and hydrophobic modifications range to as low as 0.17 µg L⁻¹ (Simões et al., 2021; 2022). Because this polymer is less toxic than other polyquaternium polymers discussed above (Cumming, 2008; Cumming et al., 2008; Hansen et al., 2023), there is significant potential for the identified polyquaterniums to exceed safe concentrations in the environment (e.g., maximum PEC_SW for polyquaternium-6, which is more toxic than polyquaternium-67 based on available data, is 1.3 µg L⁻¹), although polyquaternium toxicity may also be mitigated by the presence of humic acid and suspended solids in the environment (e.g., Hansen et al., 2023; Rawlings et al., 2022). Predicted no-effect concentrations for individual high-emission polymers across all groups, requiring extensive ecotoxicity data across multiple species, are needed to characterize potential risk. In addition, the potential for some polymers to contribute to ecological effects as a mixture (e.g., polyquaterniums, for which ecotoxicity is directly related to cationic charge; Connors et al., 2023) may be significant for ERA.

Knowledge gaps and future applications

The availability of key data for the top three polymers (highest M_DTT) in each of the top 10 highest-emitted groups is summarized in Table 6. These data have been compiled and discussed in the above sections. This provides a list of 30 key polymers, based on down-the-drain emissions, which can be prioritized for further study based on associated data gaps (Table 6). Several polymers have associated REACH registration dossiers, although many have multiple CAS numbers with different levels of information available (ECHA 2020).

From Table 6, several polymers and polymer types can be identified as lacking in key data. Polyol ethoxylate esters as a group are severely lacking both environmental fate and effects data, and more research is thus warranted to determine similarities and differences between these and other polymeric nonionic surfactants, such as alcohol ethoxylates, particularly given the closeness of PEC_SW (this study) to PNEC (Straub et al., 2014) for polysorbate 20. In addition, the most highly emitted polyquaterniums are lacking in data, which is highly significant given the potential ecotoxicity and persistence of cationic polymers. Styrene/acrylates copolymer is also likely to require further research moving forwards, because it differs significantly to other polycarboxylates, is the highest-emitted polymer in this group (as well as one of the highest-emitted polymers overall), and does not have any corresponding fate or effects data. Most of these criteria also apply to copolymers of PEG/vinyl acetate (polyethers group). Several polymers identified in our study were highlighted by Pecquet et al. (2019) as having insufficient data available for conducting an ERA, including polyquaternium-10, polyquaternium-7, and styrene/acrylates copolymer; we have here shown high emission rates and therefore high potential for environmental exposure to these polymers, further suggesting they should be prioritized for further study. In addition, Pecquet et al. (2019) excluded polymers identified only by trade names and those lacking in CAS numbers from their dataset due to inadequate characterization. In our study, polymers were identified based on names listed in product ingredients, which are often more informative than CAS numbers (although, as discussed previously, information on polymer structure and MW was lacking in most cases). Therefore, although data of the present study may incorporate some materials that do not strictly fit the OECD polymer definition, there is also potential for inclusion of other polymers that do not have sufficient data for ERA but were excluded from analyses by Pecquet et al. (2019). Data are also somewhat lacking for cellulose and starch polymers; as these are naturally occurring, they may be of less cause for concern, although physical effects from release of large quantities of natural polymers should not be overlooked, and the assumption that these polymers cause minimal ecological effects should be confirmed for chemically modified variants.

Several assumptions were needed to account for a lack of data for most identified polymers when obtaining estimates of emissions and PEC in the present study. Although most polymer groups had some level of WWT removal data available to calculate PEC, this was often not specific to the highest contributing group members, and thus, further data are needed to refine estimates and the need for subgroups and obtain more specific PEC estimates for each polymer. For most polymers identified, including 23 of these top 30 polymers, MW data were not available, which further limits the ability to fully characterize polymers and their fate. Polymer naming conventions are also oftentimes ambiguous and for many polymers, do not reveal sufficient information on polymer structure. Increased transparency in reporting of key polymer properties for polymers in current use, particularly relating to MW, would significantly enhance further data collection and ERA. Although reporting of full MW distribution and mixture composition data may be unrealistic currently, even reporting of average MW would greatly facilitate risk assessment efforts and allow more specific analyses of polymers based on their individual properties, including ecotoxicity, WWT removal, and environmental biodegradation, which are all MW-dependent. Although the polymer groups established in this study are also a useful first step to indicate key polymer functionalities that are likely released to the environment, particularly in light of the severe lack of emissions data for most WSPs, many of these groups contain a broad range of polymers with (potentially) different MWs, additional monomer units (in the case of identified copolymers), and physicochemical properties (e.g., charge density), with insufficient information to determine these properties, in most cases, from available product ingredients data. For higher-tier exposure and effects assessment, it may be useful to test the extent to which these differences in polymer properties affect behavior, ecotoxicity, and subsequent environmental risk. This is likely to lead to the need for further refinement of groups and subgroups as more data become available. However, for some groups, determination of grouped PEC may be more relevant; for example, in the alcohol alkoxylates group, polymers such as “C11-15 pareth-7”, “C12-14 pareth-7”, and “C12-14 pareth-n” are listed under separate names but will contain many of the same components, and chain lengths outside this range will also exist as a distribution, indicating significant overlap between group members.

Despite the uncertainty in emissions and PEC data due to the assumptions applied, the data in this study are useful in providing both preliminary exposure estimates for a group of substances for which data are severely lacking and in prioritizing polymers for further research to fill these data gaps with more robust methods. Environmental concentrations have been modeled for initially unidentified polymers without the need for substance-specific usage or emissions data, such as manufacture and import volumes. The approach used allows identification of specific polymers without prior knowledge of polymer identities, meaning the full range of polymers used in the incorporated products can be accounted for. The down-the-drain emissions estimates may be useful for future exposure assessments and modeling, particularly in combination with addressing the knowledge gaps highlighted for key high-emission polymers in Table 6, with incorporation of exposure-based indicators into prioritization approaches having been recommended previously (Groh et al., 2023). As more data become available on identities and environmental fate behavior of these polymers and identified knowledge gaps are addressed, more complex models could be developed, combining emissions estimates of our study with fate and biodegradability data to refine PECs. Incorporation of other product types which may be released down-the-drain from household use, including toothpaste, moisturizer, fabric conditioner, deodorant, and others, may also be significant for future polymer identification and emissions estimates.

Conclusion

Results from the emissions modeling approach developed in this study suggest that a wide variety of WSPs found in household products are likely to be present in the environment. Several high-emission polymers are currently lacking in environmental data, including polymers from the polycarboxylates, polyethers, polyol ethoxylate esters, and polyquaternium groups, and further research is recommended to fill these data gaps by characterizing environmental fate and effects as well as the suitability of read-across approaches. Several polymers identified in this study have been detected in the environment already, and development of analytical methods to characterize other polymer types is critical. However, analyses in this study were hindered by a lack of reporting of key polymer properties and ambiguity in polymer naming conventions. Increased transparency in the identities of polymers used in industry as well as better characterization methods will greatly facilitate future research.

Supplementary material

Supplementary material is available online at Environmental Toxicology and Chemistry.

Data availability

All data except raw ingredients and brand data are included in the supplemental data. The aforementioned additional data are available on request from the authors.

Author contributions

Hattie Brunning (Conceptualization, Data curation, Formal analysis, Funding acquisition, Investigation, Methodology, Project administration, Visualization, Writing—original draft, Writing—review & editing), Brett Sallach (Project administration, Supervision, Writing—review & editing), Alistair Boxall (Conceptualization, Funding acquisition, Methodology, Project administration, Supervision, Writing—review & editing)

Funding

The present study was funded by the Natural Environment Research Council (NERC) as part of the Adapting to the Challenges of a Changing Environment Doctoral Training Partnership (ACCE DTP; grant number NE/S00713X/1) and by Reckitt as part of a collaborative partnership (Collaborative Awards in Science and Engineering (CASE) partner).

Conflicts of interest

None declared.

Disclaimer

The peer review for this article was managed by the Editorial Board without the involvement of Alistair Boxall.

Acknowledgments

The authors would like to thank O. Price, T. Hutchinson, V. Zanchi, and other colleagues at Reckitt for their feedback on the research and manuscript.

References