{"paper_id":"bef19c7a-5d5c-4537-b167-3e1c5f164de3","body_text":"The building and construction\nsector is a major industrial user\nof plastics, particularly of polyvinyl chloride (PVC). In Europe,\n71% of PVC is used in building and construction, contributing to 38%\nof all plastics used in the sector. 1 , 2  Flooring is\none major building and construction application (7–10% of PVC)\nemploying primarily flexible PVC, which is also used for flexible\nfilms, sheets, cables, and tubes (14% of PVC) and is generally extensively\nplasticized. 3 , 4  Many chemical substances are present\nin plastics, including residual monomers, additives, processing aids,\nand so-called “non-intentionally added substances” such\nas contaminants, by-products, and breakdown products. 5 − 7  Generally, PVC requires heat- and UV-stabilization (0.05–5\nwt %), flexible applications such as flooring require plasticization\n(5–65 wt %), and may contain large amounts of fillers (5–50\nwt %), in addition to other additives (such as colorants, antioxidants). 8 − 10  Among the additives, plasticizers and stabilizers are particularly\ninteresting, as they are used in comparatively large amounts and have\nbeen the subject of regulatory scrutiny in recent years. For example,\nPVC plastics have been notorious for their extensive use of multiple\nhazardous  ortho -phthalates as plasticizers and cadmium,\nlead, and tin as stabilizers. Consequently, the use patterns of plasticizers\nand stabilizers are changing in the PVC industry, shifting from well-known\nhazardous substances to alternative ones. 11 , 12\nMultiple  ortho -phthalate plasticizers have\nbeen\nassociated with various adverse health effects, including lower semen\nquality, altered anogenital distance, endometriosis, decreased testosterone,\nneurodevelopmental effects, attention-deficit hyperactivity disorder,\nautism, development of breast/uterine/testicular cancers, asthma,\nand type 2 diabetes, leading to increased regulatory scrutiny. 13 , 14  For example, bis(2-ethylhexyl) phthalate [DEHP, Chemical Abstracts\nService Registry Number (CASRN): 117-81-7] was added to the Authorization\nList of the European Union (EU)’s Chemicals Regulation, REACH,\nin 2011 with a sunset date in 2015, which means their use after this\ndate is prohibited on the EU market (unless authorization has been\nsought and granted, which is the case for DEHP in recycled PVC). 15 , 16  Similarly, benzyl butyl phthalate (BBP, CASRN: 85-68-7), di- n -butyl phthalate (DBP, CASRN: 84-74-2), and di- iso -butyl phthalate (DiBP, CASRN: 84-69-5) have also been\nadded to the EU REACH Authorization List as well as the Swiss Chemical\nRisk Reduction Ordinance (ORRChem). 122  While\ngroup-based assessment and regulation for other  ortho -phthalates is currently being discussed on the European level. 17  Phase-out of some  ortho -phthalates\nhas led to an increased demand for alternative plasticizers including\nterephthalates, trimellitates, cyclohexane dicarboxylic acid esters,\nphosphates, adipates, citrates, vegetable oil derivatives, and polymeric\nplasticizers (see Section S1 in the  Supporting Information  1). 11 , 18  Currently, these alternative\nplasticizers are generally less studied, and partially lack important\nphysicochemical and toxicological data. 18  While available hazard data indicate that many of them are likely\nsafer than  ortho -phthalates, some have shown some\ncause for concern, for example, tricresyl phosphate (CASRN: 1330-78-5,\nlikely toxic for reproduction), and tris(2-ethylhexyl) trimellitate\n(TEHTM, CASRN: 3319-31-1, likely persistent and endocrine disrupting). 18 − 20\nHeat- and UV-stabilizers have undergone a shift in recent\nyears.\nEarlier stabilizer systems were mainly based on cadmium and lead,\nknown for posing health and environmental risks. 21 , 22  The PVC industry in the EU voluntarily phased out cadmium- or lead-based\nstabilizer systems in 2001 and 2015, respectively, and replaced them\nwith (organo-)tin-, barium- and zinc–calcium-based systems. 11 , 23 , 24  Some of these replacements may\nlead to diverse adverse health effects. Organotins are known for their\nendocrine-disrupting potential, ecotoxicity, neurotoxicity, and liver\ntoxicity. 25 − 27  Barium exposure may lead to kidney diseases, neurological,\ncardiovascular, mental, and metabolic disorders. 28  Zinc possesses properties indicating hazards for human\nhealth and the environment. 29  Current regulation\nmainly covers legacy metal(loid) elements (chromium, cadmium, lead,\narsenic, mercury, nickel), which must be below 0.1 wt % in certain\nplastic products. 30 − 33\nBuilding and construction plastics contribute to long-term\nexposure\nto hazardous chemicals in two ways. On the one hand, due to the long\nlifetime of these plastics, legacy chemicals that have been phased\nout from new production and use may still be common in products that\nare in use. 34 − 36  On the other hand, the comparatively high recycling\nrate of building and construction PVC plastics (17% in the EU in 2012,\n16% in Switzerland in 2017) and the common practice of closed-loop\nrecycling can prolong the presence of hazardous chemicals through\ncontamination of new products. 37 , 38  This extends the consumer\nand occupational exposure to these substances. The EU waste framework\ndirective aims to increase the recycling rate further, while also\nproviding information on substances of very high concern (SVHCs) in\nproducts with the SCIP database established by the European Chemicals\nAgency (ECHA). 39 , 40\nPVC floorings have been\nidentified as a key source of indoor chemical\nexposure to hazardous chemicals, especially to multiple  ortho -phthalates and are an important source of recycled material. 41 − 48  Despite that, only limited information on the chemical compositions\nof PVC floorings is publicly available: (1) the SCIP database contains\n51 relevant entries (Sheet S11 in the  Supporting Information 2 ), and (2) the few conducted studies have typically\nhad a small sample size and tested a limited number of chemicals. 40 , 48 − 56  To our knowledge, only one recent study measured many PVC flooring\nsamples, which were from the United States market, using a nontargeted\nscreening approach. 55  Furthermore, various\nstudies focused on other PVC products or indoor dust, which may allow\nfor inferences on the possible chemical content of PVC floorings,\nbut with significant uncertainties. 57 − 70  Plasticizer handbooks suggest large variations across PVC flooring\nproducts from different times and regions, as well as across different\nPVC products. 3 , 9 − 12  Furthermore, dust samples may\nalso contain plasticizers from other products in the indoor environment.\nThus, a research gap remains regarding the chemicals present in a\nrepresentative selection of PVC floorings in markets other than the\nUnited States.\nIn this study, we aim to comprehensively understand\nthe chemicals\npresent in PVC floorings sold in Switzerland. Using a combination\nof targeted analysis and suspect screening, we present the occurrence\nand concentrations of legacy and novel substances in new PVC floorings\nfrom the Swiss market, focusing on plasticizers and metal heat/UV-stabilizers.\nIn addition, we analyzed the potential biological activities of selected\nflooring extracts by using several bioassays. We then contextualized\nour observations in terms of implications for human health, the environment,\nand the transition to a circular economy. Finally, we provide recommendations\nto researchers, policymakers, industry, and citizens.\n\nA total of 204 new flooring samples\nwere collected from various\ndo-it-yourself (DIY) stores and one flooring retailer for large-scale\nprojects in Switzerland during 2021 and 2022. The samples were first\nscreened using X-ray fluorescence (XRF) for their elemental compositions\nand using attenuated total reflection–Fourier transform infrared\nspectroscopy (ATR–FTIR) for their polymer compositions and  ortho -phthalate presence. Only PVC samples ( n  = 151), identified by XRF and ATR–FTIR screening, were further\nanalyzed. Targeted gas chromatography–mass spectrometry (GC–MS)\nwas used to quantify  ortho -phthalates. Alternative\nplasticizers were detected using suspect screening GC–MS. Furthermore,\na selection of samples underwent several bioassays to determine potential\nbiological activity. Detailed methods are described in the sections\nbelow.\nA breakdown by sample characteristics (i.e., color, hardness,\nnumber\nof layers, presence of a gray layer, and origin; these were manually\nassigned through simple visual inspection without specific testing)\ncan be found in  Figure  1 . For more details on the characteristics of each sample and the\nassignment of characteristics, see Sheet S1 in the  Supporting Information 2  and Section S2.1 in the  Supporting Information 1 , respectively. Information\nabout the presence of recycled PVC in individual samples could not\nbe obtained from the respective stores and retailers. Instead, the\npresence of a gray layer in a product was used as a nonconclusive\nproxy for recycled PVC, as color mixing and discoloration of insufficiently\nstabilized PVC at high temperature during recycling may cause a gray\nshade. 71 , 72\nSchematic overview of the characteristics of\nthe samples analyzed\nin this study and the analytical methods employed analytical methods.\nThe presence of a gray layer may be an indication of the use of recycled\nPVC in the product. 71 , 72  PVC = polyvinyl chloride, DIY\n= “do-it-yourself”, ATR-FTIR = attenuated total reflectance-Fourier\ntransform infrared spectroscopy, XRF = X-ray fluorescence, and GC–MS\n= gas chromatography–mass spectrometry.\nAn overview of the targeted\ncompounds and the suspect list compounds and further details (e.g.,\nCASRNs, suppliers) can be found in Table S4 in the  Supporting Information 1  and Sheet S2 in the  Supporting Information 2 . All reagents were of analytical\ngrade. The solutions were prepared and stored in amber glass vials.\nReference materials with certified levels of metals and metalloids\n(ERM-EC681m) and  ortho -phthalates (SPEX CRM-PVC001)\nwere used as quality controls for XRF and GC–MS respectively\n(Table S3 in the  Supporting Information 1 ).\nAll samples ( n  = 204) were screened with an ATR-FTIR\n(ThermoScientific\nNicolet iS5 with iD7 ATR module) to determine the polymer type and\nthe presence of  ortho -phthalates (further details\nin Section S2.3.1 in the  Supporting Information 1 , and all recorded spectra in the  Supporting Information 6  − Rawdata-ATR-FTIR). Non-PVC samples ( n  = 53) were not analyzed further.\nThe elemental composition\nof all samples ( n  =\n204) was determined using a hand-held XRF (Thermo Scientific Niton\nXL3 Gold Analyzer) with a plastic calibration (further details in\nSection S2.3.2 in the  Supporting Information 1 , and all XRF readings in Sheet S3 in the  Supporting Information 2 ). A certified reference material (ERM-EC681m—polyethylene\nhigh level) was used to check operation and equipment calibration\n(measured values had to be within 20% of the certified levels).\nOrtho -phthalate quantification\nwas performed on all PVC samples ( n  = 151) using\na validated method (protocol in the  Supporting Information 3 ) covered in the accreditation perimeter of a\nlaboratory complying with ISO 17025:2017. The sample preparation was\noptimized for  ortho -phthalate extraction and followed\nthe validated method (the  Supporting Information 3 ), including regular quality checks with the certified reference\nmaterial (measured concentrations had to be within 20% of the certified\nlevels). The polymer was dissolved in tetrahydrofuran (THF, CASRN:\n109-99-9), followed by matrix precipitation in acetonitrile (ACN,\nCASRN: 75-05-8) and filtration (0.45 μm nylon filters, BGB SF2503-2).\nSubsequently, GC–MS analysis and quantification were carried\nout by using an internal standard calibration. Seventeen  ortho -phthalates were used as standards for the calibration curve and\nseven deuterated  ortho -phthalates were used as internal\nstandards, which were added to every sample (Table S8 in the  Supporting Information 1 ). The lowest calibration\npoint was reported as the method limit of quantification (LOQ) for\nthe target  ortho -phthalates. All analyses were carried\nout on an Agilent GC–MS system (GC: Agilent 7890A, MS: Agilent\n5975C) in single-ion mode with splitless injections. Compounds were\nseparated on a DB-5MS column using a temperature gradient from 80\nto 320 °C. Measurements were performed in batches, each containing\ncalibration solutions, samples, blank solutions (procedural blank\nand solvent blank), and reference solutions (a solution with known\nconcentration and a reference material extract). From the recorded\nchromatograms and mass spectra (available as Agilent files in the  Supporting Information 7  − Rawdata-GCMS-Phthalates),\ncompounds were automatically detected, identified, and quantified\nusing weighted (1/×) quadratic calibration curves (using the\nquantitative Agilent Masshunter workflow in the  Supporting Information 4 ). Further details on the  ortho -phthalate quantification workflow are given in Section\nS2.3.3 in the  Supporting Information 1 .\nAll PVC samples ( n  = 151) were\nscreened for other\nsubstances on our suspect list, mainly containing alternative plasticizers\nand antioxidants (Table S10 in the  Supporting Information 1 ). The suspect screening was conducted with low-resolution\nGC–MS using electron impact ionization (EI). The method parameters\nwere based on Löschner et al. (2011) but adapted and optimized\nfor detecting the substances on our suspect list (Section S2.3.4 in\nthe  Supporting Information 1 ). 73  Furthermore, for all substances on our suspect\nlist, the suitability of the extraction method was checked, a custom\nlibrary was created, and a dilution series for semiquantification\nand determination of the limits of detection (LOD) was conducted.\nThe analyses were conducted on an Agilent GC–MS system (GC:\nAgilent 7890A, MS: Agilent 5975C) in scan mode with splitless injections.\nGenerally, a nonpolar column (DB 5MS), a slow temperature gradient\n(8 °C/min), a high final temperature (40–300 °C),\na long runtime (55 min), and a wide scan range (30–800 amu)\nwere chosen to ensure elution, separation, and identification of all\ncontained compounds. Measurements were performed in one batch per\ndilution (40×/1600×), containing all samples, regular blanks,\nand regular suspect standard solutions.\nRecorded chromatograms\nand mass spectra (available as Agilent files in the  Supporting Information 8 -Rawdata-GCMS-Suspect) were analyzed\nfor (1) the presence and approximate concentration of the compounds\non the suspect list and (2) the presence of other unknown substances.\nFor this, the “Agilent MassHunter—Qualitative analysis”\nsoftware (workflow in the  Supporting Information 4 ), some manual processing, and Python-based data processing\n(Python scripts in the  Supporting Information 5 ) were employed. For the compound discovery, both “Find\nby integration” (considering all Lorentzian chromatogram peaks\nwith an area larger than 0.001% of the largest peak) and “Find\nby molecular feature” (limited to Lorentzian peaks with more\nthan 500 counts and the largest 200 compounds) were used. For the\ncompound identification, two libraries were used, (1) a manually created\ncustom suspect list library of the scanned suspect list standards\nand (2) the NIST 14 library (which contains fewer but more commonly\nused substances to limit overfitting the data). Results were considered\nacceptable if the mass error ranged from −0.3 to +0.7 Da and\nthe matching score was over 70. Further details on the suspect screening\nworkflow are in Section S2.3.4 in the  Supporting Information 1 .\nDue to time and resource constraints, only selected samples were\ntested using various bioassays. Randomly selected samples ( n  = 85) were tested for cytotoxicity using the MTT assay\nand induction of oxidative stress using the ROS assay. Several samples\nwere selected to cover maximal differences regarding their  ortho -phthalate contents and cytotoxicity, and were further\ntested for endocrine activity using the YES/YAS assays, mutagenic\nactivity using the AMES assay, and genotoxic activity using the planar-umuC\nassay (for sample selection, see Table S11 in  Supporting Information 1 ). The same extraction procedure as\nabove was used ( Section  2.2.3 ) except that the samples were concentrated after filtration\nfrom 6 mL to 300 μL using the Syncore system from Buchi (which\navoids losses of volatile substances). This was done because most\nbioassays have a low solvent tolerance (MTT/ROS: maximally 0.1 vol\n%). Due to the high volatility of THF, the sample volumes decreased\nduring the storage (−20 °C) and transport (max 20 °C\nfor 2–3 days) and were filled up to 300 μL with THF before\neach assay. The bioassays generally used procedural blanks, solvent\n(negative) controls, and positive controls unless otherwise required\nby the respective manufacturer protocol.\nThe highest possible test concentrations\n(1-μL concentrate\ncontaining the dissolved fraction of ∼1.1 mg PVC was used on\n1 mL cell culture medium) of the randomly selected extracts ( n  = 85) were screened for cytotoxicity using the MTT assay\nand for oxidative stress using the ROS assay. Both assays were conducted\non human liver cells ( Huh7 ), according to Christen\net al. 2014 (detailed conditions in Section S2.4 in  Supporting Information 1 ). 74  Samples\nwere categorized based on their cell viability in the MTT assay (<30%:\n“highly toxic”; 30–60%: “moderately toxic”;\n60–90%: “slightly toxic”; >90%: “not\ntoxic”,\nvariation of solvent control).\nEight selected extracts were screened for potential\nestrogenic, antiestrogenic, androgenic, and antiandrogenic activities\nusing the XenoScreen YES/YAS assays from Xenometrix (Allschwil, Switzerland).\nFive selected extracts were screened for potential mutagenic activity\nusing the Ames MPF 98/100 assay from Xenometrix (Allschwil, Switzerland)\nwith  Salmonella typhimurium  strains\nTA98 (for detection of frameshift mutations) and TA100 (for detection\nof base substitution mutations), following the manufacturer’s\nprotocol which conforms with the OECD Test Guideline 471. Twelve selected\nextracts were screened for potential direct genotoxic activity using\nthe planar-umuC bioassay protocol of planar4 GmbH (Stäfa, Switzerland).\nFurther details are given in Section S2.4 in the  Supporting Information 1 .\n\nThe detection frequencies (DFs) and concentration\nranges of various\nelements in the 151 PVC samples are shown in  Figure  2 , and details are given in Sheet S1 in the  Supporting Information 2 . The most prevalent\nelements besides chlorine (which is part of the PVC matrix) are zinc\n(DF: 96%), iron (DF: 76%), barium (DF: 72%), titanium (DF: 68%), tin\n(DF: 58%), and vanadium (DF: 46%). Surprisingly, also several potentially\ntoxic metals and metalloids including chromium, lead, arsenic, and\nnickel are detected in 29 samples (DF: 19%; range 0.001–1.562\nwt %), with six samples surpassing a common regulatory reference level\nof 0.1 wt %, most of which contain lead ( Figure  4 ). None of the samples\ncontain cadmium or mercury.\n(A) Heatmap of the concentrations of selected\nelements and targeted  ortho -phthalates, the presence\nof alternative plasticizers,\nand the activities in bioassay tests, with every row representing\none sample. (B) Measured concentrations (top) and DF, (bottom) of\nselected elements and targeted  ortho -phthalates,\nalternative plasticizers, and bioassay tests. Note that the DF are\ncalculated in the following ways: elements had to be above the LOD\n(0.0004–0.0067 wt %) and  ortho -phthalates\nabove the LOQ (∼0.05 wt % for DiNP and DiDP, ∼0.005\nwt % for the others), whereas other substances were considered “detected”\nwhen they were (tentatively) identified in the suspect screening workflow.\nDF = detection frequency, LOD = limit of detection, LOQ = limit of\nquantification, and  p -umuC = planar-umuC.\nBiological activities of the tested samples ( n  =\n85). All markers show the results of cytotoxicity ( y -axis) against oxidative stress ( x -axis) in Huh7\ncells after 24 h exposure, with the mean values of three independent\nexperiments shown here. Special markers signify if samples were tested\nin additional bioassays, and the marker color signifies the result\nof the additional tests.\nPercentage of samples\nwith a reason for clear or potential concern.\nDEHP = di(2-ethylhexyl) phthalate (CASRN: 117-81-7); BBP = benzyl\nbutyl phthalate (CASRN: 85-68-7); DBP = di- n -butyl\nphthalate (CASRN: 84-74-2); DiBP = diisobutyl phthalate (CASRN: 84-69-5).\nCorrelations between some elements and the product\ncolor are observed.\nFor example, titanium concentrations strongly correlate with white\ncolor. Also, the concentrations of the toxic metals and metalloids\ncorrelate positively with the presence of a gray layer (which may\nindicate recycled PVC) and negatively with the number of layers (see\nSection S3.5 in the  Supporting Information 1 ). 71 , 72  Cadmium and lead had been the major heat\nstabilizers before they were voluntarily phased out by the PVC industry\nin the EU (cadmium in 2000 and lead in 2015). Today, heat stabilization\nfor PVC in the EU is achieved mainly using zinc–calcium, zinc–tin,\nand zinc–barium systems. 11 , 24  The observed elemental\ncompositions provide supporting evidence for this industrial shift.\nIn particular, no samples contain cadmium, indicating its phase-out\nin new products. Instead, most samples contain zinc, tin, and/or barium,\nsuggesting the wide use of novel heat-stabilization systems (noting\nthat calcium, another substance commonly used in novel heat-stabilizer\nsystems, was not measured in this study). Meanwhile, the presence\nof lead in several samples (DF: 9%) is most likely associated with\nrecycled PVC in products (see Section S3.5 in the  Supporting Information 1 ).\nThe  ortho -phthalate quantification with GC–MS\nshowed that 55 samples\n(DF: 36%) contain  ortho -phthalates, ranging from\n0.01–47 wt % ( Figure  2  and Table S13 in the  Supporting Information 1 ), most of which were also captured by ATR-FTIR screening\nwith a sensitivity of 78% and a specificity of 85% (see Section S2.3.1\nin  Supporting Information 1 ). The most\nprevalent  ortho -phthalates are DiNP (diisonoyl phthalate;\nCASRN: 68515-48-0; DF: 24%; 0.05–47 wt %), DEHP (DF: 19%; 0.003–20\nwt %), and DiDP (diisodecyl phthalate; CASRN: 68515-49-1; DF: 16%;\n0.05–28 wt %). This is despite DiDP and DiNP having a LOQ approximately\nten times higher (∼0.05 wt %) than the other  ortho -phthalates (∼0.005 wt %). They were mostly found in soft\nor medium-hard products (Section S3.5 in  Supporting Information 1 ). DBP, DiBP, BBP, and DEHP are regulated under\nthe EU REACH Authorization List and the Swiss ORRChem , which means\nthat their use is prohibited on the Swiss and the common EU market\nand new products shall not contain more than 0.1 wt % of these substances\n(unless an authorization has been sought and granted). 15 , 122  In the case of DEHP, specific authorization for recycled soft PVC\nwas granted in 2016 and is now expired. 16  Overall, 31 samples (DF: 21%) contain these restricted  ortho -phthalates (mainly DEHP), ranging from 0.003 to 21 wt %, with 24\nsamples surpassing the 0.1 wt % threshold.\nIn addition to  ortho -phthalates, the qualitative suspect screening shows\nthe presence of alternative plasticizers in 123 samples (DF: 81%,\nsee  Figure  2 , Sheet\nS1 in the  Supporting Information 2 ); the\nmost frequently detected ones are DEHT [Bis(2-ethylhexyl) terephthalate;\nCASRN: 6422-86-2; DF: 56%], DEHA [Bis(2-ethylhexyl) adipate, CASRN:\n103-23-1; DF: 19%], and Octicizer [2-ethylhexyl diphenyl phosphate;\nCASRN: 1241-94-7; DF: 13%]. Most alternative plasticizers were confirmed\nusing corresponding analytical standards and semiquantified (Table\nS14 in the  Supporting Information 1 ). However,\nsemiquantification results remain highly uncertain as no internal\nstandard was used and some signals were beyond the calibration range\n(even leading to implausible concentration estimates above 100 wt\n%; see Figure S9 in the  Supporting Information 1 ). With these uncertainties in mind, DINCH [1,2-cyclohexane\ndicarboxylic acid diisononyl ester; CASRN: 166412-78-8] and DEHT are\npresent in high concentrations. The overall estimated plasticizer\ncomposition per sample can be found in Figure S9 in the  Supporting Information 1 . Alternative plasticizers\nare more common in hard PVC samples with many layers (Section S3.5\nin the  Supporting Information 1 ).\nThe observed plasticizer profiles in  Figure  2  visualize the ongoing industrial shift from\nlegacy  ortho -phthalates such as DBP, DiBP, BBP, and\nDEHP to an increased use of other  ortho -phthalates\n(mainly DiNP and DiDP) and alternative plasticizers (mainly DEHT,\nDEHA, and Octicizer). 18  Interestingly,\nsamples typically contain one major plasticizer, either DiNP/DiDP\nor an alternative plasticizer (Figure S9 in the  Supporting Information 1 ). DEHP was generally present along\nwith other major plasticizers and at concentrations below the usual\nplasticizer range for flexible PVC (5–65 wt %). 9  This suggests that the presence of DEHP comes mainly from\nrecycling rather than intentional use.\nIn total, nearly 400 substances are tentatively\nidentified using chromatogram integration and NIST library matching,\nmostly without further confirmation (Sheet S9 in the  Supporting Information 2 ). Some more frequently detected substances\nidentified through library matches include oleamide (DF: 21%, CASRN:\n301-02-0), 5-hexen-1-ol (DF: 11%, CASRN: 821-41-0), dodecane (DF:\n11%, CASRN: 112-40-3), hexanamide (DF: 10%, CASRN: 628-02-4), and\nisobutyric anhydride (DF: 10%, CASRN: 97-72-3). Some of the tentatively\nidentified substances are or may be hazardous. For example, endocrine-disrupting\nbisphenol A (CASRN: 80-05-7) was present in two samples (confirmed\nusing an analytical standard, DF: 1%), possibly persistent, bioaccumulative,\nand toxic UV-326 (bumetrizole, CASRN: 3896-11-5) was likely present\nin the six samples (library match, DF: 4%) and short-chain chlorinated\nparaffins (SCCPs, matched by CASRN: 111-85-3 and CASRN: 73772-39-1)\nwere likely present in the two samples (DF: 1%).\nFrom the 85 tested samples,\n26 show some biological activities (DF: 17%;  Figures  2  and  3 ): (1) Seven of the 85\ntested samples show\nmoderate cytotoxicity and clear induction of oxidative stress (i.e.,\nthe ratio of oxidative stress/viability >1.5), whereas another\n11\nsamples display slight cytotoxicity. There is a clear correlation\nbetween cell viability and ROS ( Figure  3 ). (2) Endocrine\nactivities are observed\nin five of the eight tested samples and show no correlation with the\ncytotoxicity ( Figure  3 ). (3) For one of the\nfive tested samples,\nmutagenic potential cannot be ruled out (Figure S13 in the  Supporting Information 1 ). (4) Genotoxic activity in the planar-umuC\nassay is observed in 11 of the 12 tested samples (Figure S14 in the  Supporting Information 1 ), with one showing activity\nin the 1:1000 dilution, five showing activity in the 1:100 dilution,\nand another five showing activity in the 1:10 dilution.\nSeven of the 85\ntested samples show\nmoderate cytotoxicity and clear induction of oxidative stress (i.e.,\nthe ratio of oxidative stress/viability >1.5), whereas another\n11\nsamples display slight cytotoxicity. There is a clear correlation\nbetween cell viability and ROS ( Figure  3 ).\nEndocrine\nactivities are observed\nin five of the eight tested samples and show no correlation with the\ncytotoxicity ( Figure  3 ).\nFor one of the\nfive tested samples,\nmutagenic potential cannot be ruled out (Figure S13 in the  Supporting Information 1 ).\nGenotoxic activity in the planar-umuC\nassay is observed in 11 of the 12 tested samples (Figure S14 in the  Supporting Information 1 ), with one showing activity\nin the 1:1000 dilution, five showing activity in the 1:100 dilution,\nand another five showing activity in the 1:10 dilution.\nGenerally, these observed biological activities do not\ncorrelate with the product characteristics such as color, hardness,\nor presence of a gray layer, nor the content of specific chemicals\ndetected in this study (except for endocrine activity, which only\noccurred in samples containing  ortho -phthalates,\nsee  Figure  2 ; however,\nthis result needs to be read with caution, as only eight samples were\ntested). Furthermore, although cytotoxicity and oxidative stress correlated\nstrongly, they could not be used to predict other biological activities\n( Figure  3 ).\n\nOver time, the dominant use of DEHP has been replaced\nby other  ortho -phthalates (e.g., DiNP, DiDP) and\nalternative plasticizers\n(e.g., DEHT, DINCH, DEHA, and Octicizer), as demonstrated by the changing\nsubstances reported in this study and previous studies on PVC flooring\n(see Table S24 in the  Supporting Information 1 ). 48 − 54  Meanwhile, some regional differences in this industrial transition,\nespecially in alternative plasticizer use, can be observed. For example,\nthe major emerging plasticizers identified in Switzerland in this\nstudy are DiDP, DiNP, DEHT, and Octicizer. However, a study in Norway\nthat focused on phosphor plasticizers and flame retardants mainly\nfound TBEP [tris(2-butoxyethyl) phosphate, CASRN: 78-51-3] and TPhP\n(triphenyl phosphate, CASRN: 115-86-6). 52  A nontargeted study in the United States found DEHA, DEP, DBP, BBP,\nTXIB (2,2,4-trimethyl-1,3-pentandiol diisobutyrate, CASRN: 6846-50-0),\nand ATBC (acetyltributyl citrate, CASRN: 77-90-7). 55  Note that these differences could also be due to different\ntarget substances, instrumentation, and/or extraction procedures.\nOther products made from PVC, mainly medical devices and toys, have\nfrequently been studied for their plasticizer content (Table S26 in\nthe  Supporting Information 1 ). For PVC\nmedical devices, DEHP has been present in high concentrations (up\nto 40 wt %) and has only partially been replaced with alternatives\n(e.g., DiNP, DEHT, DINCH, TEHTM, and ATBC) in recent years. 57 , 58 , 63 − 70  The use of DEHP in medical devices in the EU had still been specifically\nauthorized until recently, which may explain these findings. 70 , 75  For PVC toys, due to increased regulatory scrutiny in the sector,\nDEHP and other commonly restricted  ortho -phthalates\nhave been replaced with alternatives comparatively early on [mainly\nwith ATBC, DEHT, TXIB, DINCH, ESBO (epoxidized soybean oil, CASRN:\n8013-07-8)]. 76 − 82  However, DEHP and other commonly restricted  ortho -phthalates are still widely found in many PVC toys across the globe\n(present in 11 of 118 toys in Switzerland, 89 of 700 in the EU, 17\nof 49 in New Zealand, and 1 of 1 in Jordan). 79 − 82  The wide presence of such well-known\nhazardous substances across a wide range of PVC products points to\nissues in monitoring and enforcement of existing regulations and may\npose a risk of contamination to any PVC product, including flooring,\nshould open-loop recycling occur.\nTo the best of our knowledge,\nmetals have not explicitly been studied\nin PVC floorings but only in other PVC products (Table S25 in the  Supporting Information 1 ). Similarly to this\nstudy, barium and tin have often been found to be the main heat-stabilizers. 83 − 85  Some studies have also reported the presence of lead and cadmium,\nwhich are not detected (cadmium) or only detected in a few samples\n(lead) in this study. 85 − 87  This may indicate an ongoing industrial transition\nto alternative heat-stabilizers.\nA wide range\nof chemicals was detected in this study, with many quantified as well.\nMany of them are hazardous or potentially hazardous substances. Using\na simple common regulatory threshold of 0.1 wt % (e.g., used as a\nthreshold under the EU Restriction of Hazardous Substances in Electrical\nand Electronic Equipment (RoHS) Directive, and as a reporting threshold\nfor SVHCs in articles under the EU REACH), 16% of the samples show\na clear cause for concern. These contain restricted  ortho -phthalates (16%), and some additionally contain lead (4%) and/or\nchromium (0.5%) ( Figure  4 ).\nMeanwhile, there\ncould be more samples of potential concern in addition to those samples\nshowing a clear cause for concern. First, an additional 11% of the\nsamples showed activity in one of the bioassays, indicating the potential\nto cause biological effects. Second, additional 16% of the samples\ncontaining some other hazardous plasticizers (several  ortho -phthalates) and stabilizers (barium and tin) above the common threshold\nof 0.1 wt % and thus may be of potential concern due to the toxicity\nof these chemicals. 13 , 14 , 25 − 28  Third, some samples contain the already restricted substances below\nthe common threshold of 0.1 wt % and are thus not counted as of clear\nconcern. However, several of these chemicals are endocrine disrupting\nor genotoxic and thus may have a safety threshold below 0.1 wt % (e.g.,\nlead has no safety threshold). Therefore, these samples may still\nbe of potential concern, accounting for an additional 16% of the samples.\nIn total, 16% of samples show a clear reason for concern and 35% of\nsamples are of potential concern ( Figure  4 ).\nFurthermore, this does not imply\nthat the other samples are guaranteed to be entirely safe. For example,\nwhile currently available evidence suggests that the detected alternative\nplasticizers may be safer than restricted  ortho -phthalates,\nmany are present in high concentrations in the samples (especially\nDINCH and DEHT) and their continued release may cause significant\nexposure and render them ubiquitous in the environment. 18 , 88  Furthermore, research on the environmental and human health effects\nof alternative plasticizers and stabilizers is ongoing, which may\nwarrant further assessment in the future. 18 , 25 − 28 , 89  In addition, some other hazardous\nsubstances could have been present in the samples but are not detected/quantified\nin this study, chemically and/or through bioassays.\nWhile many\nsubstances are detected in PVC floorings, one may question\nwhether they can be released from the products and result in actual\nexposure.\nFor metals and metalloids, it is not an easy question\nto answer, as release depends on their metal(loid) form (e.g., chemical\nspecies, matrix, particle size) and several other environmental variables\n(e.g., environmental pH, exposure route). 90  Previous studies have demonstrated the release of lead from PVC\nduring use and associated toxic effects on human health. 91 , 92  Thus, the continued presence of toxic metals in PVC products may\nto a certain extent pose a risk to humans and the environment.\nFor plasticizers, literature on exposure from PVC flooring and\nother sources has been abundant (see Section S4.3 in the  Supporting Information 1 ), with the following\nlearnings that are relevant to our results. For  ortho -phthalates, PVC floorings are a major contributor to indoor air\nand dust concentrations and are responsible for a large portion of\ntotal indoor exposure (low μg kg bw –1  d –1  range; ingestion or inhalation of dust, inhalation\nof airborne particles, and direct skin contact being the major exposure\npathways). 41 − 48  Together with dietary intake (which is the main exposure pathway,\nhigher μg kg bw –1  d –1 ), occupational exposure, and, for some individuals, medical exposure\n(low mg kg bw –1  d –1 ),\nrelevant health limit values can be approached or even exceeded, especially\nfor susceptible populations (e.g., toddlers). 41 − 44 , 70  Ongoing exposure is a particular concern, as recent meta-reviews\nsuggest that “safe levels” for typical health concerns\nposed by some  ortho -phthalates (e.g., endocrine disruption,\ndevelopmental toxicity) might be lower than the current regulatory\nhealth limit values, especially when also considering additive or\nsynergistic mixture effects. 13 , 14 , 88 , 93 , 94  This suggests that many existing PVC floorings will continue to\ncontribute to  ortho -phthalate exposure and potential\nnegative human-health outcomes and that recycling of such PVC floorings\nmay lead to further prolonging these.\nFor alternative plasticizers,\nfewer exposure assessments have been\nconducted. 18 , 95 − 98  Alternatives are found in similar\nconcentrations in indoor media, albeit slightly lower than  ortho -phthalates, and thus result in slightly lower exposures. 95 − 98  Currently, health limit values for alternative plasticizers (e.g.,\ntolerable daily intake, reference dose) are either yet to be set,\nor orders of magnitude higher than those for  ortho -phthalates. 18  As research on alternative\nplasticizers is still ongoing, this space has yet to be monitored.\nExposure to plasticized PVC, whether from indoor PVC floorings\nor during production and recycling, has a strong link to plasticizer\nconcentrations in biological tissues. Several biomarkers (e.g., urine\nlevels of metabolites, typically in the ng/mL range) were found to\ncorrelate with exposure to PVC floorings or with occupational exposure\nto PVC. 43 , 99 − 101  Several other studies\npoint to an association of asthma and allergies with residential PVC\nfloorings, and to the development of liver cancer in an occupational\ncontext (which is likely caused by vinyl chloride exposure, rather\nthan additives). 102 − 105\nSustainable circular economy\npractices should take the chemical level into account. 35 , 106  About 16% of the samples measured in this study contain legacy,\nregulated hazardous substances, such as DEHP and lead, at significant\nlevels ( Figure  4 ).\nInterestingly, these substances are mostly present at levels lower\nthan typically necessary for fulfilling their functions (DEHP for\nplasticization: 5–65 wt %, lead for heat-stabilization: 0.05–5\nwt %), suggesting their origins being ongoing uncontrolled recycling\n(i.e., recycling of contaminated waste materials containing these\nsubstances into new products) rather than intentional use. 9 , 36 , 107  While DEHP had been explicitly\nauthorized in recycled PVC materials, this practice was controversial. 16 , 108  Recycling contaminated materials into long-lived products, such\nas floorings, prolongs exposure to and hampers an effective phase-out\nof hazardous chemicals. In fact, recycling can result in legacy hazardous\nchemicals remaining in products and materials for many decades after\ntheir initial use. For instance, the new PVC flooring sampled in this\nstudy would stay relevant for waste managers until mid-2030 or later,\nsince floorings have a long lifetime of at least 10–15 years\n(this may even be increased by lifetime prolongation measures or reuse). 34  In other words, reuse, sorting, and recycling\nsystems decades from now will still have to deal with significant\namounts of hazardous substances in end-of-life products, requiring\nefficient identification tools, safe disposal options, and possibly\nnew virgin material to replace the disposed fractions.\nWhile\nidentifying products that contain legacy or other hazardous substances\nis an important tool for realizing a safe and sustainable circular\neconomy, it remains challenging. In this study, no single product\ncharacteristics (e.g., the presence of a gray layer, which hints at\nrecycled content; color; hardness), nor analytical technique, could\nserve as a simple proxy for identifying all samples of concern (Section\nS3.6 in the  Supporting Information 1 ).\nFor example, sorting out samples with a gray layer (66% of the samples)\nwould remove only about 67% of the samples with legacy hazardous substances\n(“ Sensitivity ”), while losing a significant\nportion, 65%, of (comparatively) clean materials (“100-specificity”).\nSeveral different screening tools are compared in Section S3.6 in\nthe  Supporting Information 1  and Figure\nS15 in the  Supporting Information 1 . In\nthis study, a combination of ATR-FTIR and XRF screening is the most\neffective for identifying the majority of concerning samples; however,\nsuch a combination can yet not measure many other hazardous chemicals\nor detect mixture effects. While bioassays may provide evidence for\nunknown hazardous substances and possible mixture effects, the bioassays\nemployed in this study are very time- and resource-intensive. While\nYES/YAS assays identified  ortho -phthalates well,\ntime and resource constraints associated with sample extraction, preparation,\nand subsequent testing make them not suitable to realistically serve\nas a screening tool. Meanwhile, high-throughput screening for cytotoxicity\nand oxidative stress cannot be used as an indicator to replace other\nbioassays, such as endocrine disruption, genotoxicity, and mutagenicity.\nOverall, to make bioassays efficient and helpful screening tools for\nproblematic plastics, their further development, including alternative\nsample preparation techniques (e.g., direct sample probing or leaching\nto water instead of organic solvent extraction and concentration)\nand higher sensitivities are needed.\nSome limitations and uncertainties remain, mainly stemming\nfrom sampling, solvent selection, and the selection of analytical\nand data analysis parameters.\nPVC flooring samples were collected\nfrom four largest DIY stores and one large retailer near Zurich, possibly\nleaving out supply chains for small- or medium-sized building projects.\nDespite our inquiries, we were not able to obtain sales or tonnage\ndata for individual PVC flooring products; thus, their relative importance\nremains unclear. Product characteristics such as color, hardness,\nor “containing a gray layer” were manually assigned\nand, thus, depended on individual perception. The recycling content\nof products was not openly communicated, thus gray layers were used\nas an initial proxy, but with uncertainties as the color of recycled\nmaterial may vary depending on the pretreatment (e.g., color separation)\nand posttreatment (e.g., coloration). 71 , 72\nOur\nstudy focuses on stabilizers, plasticizers, and several biological\neffects and covers neither all substances present in PVC floorings\nnor all biological effects that may be caused. To gain a complete\npicture, additional extraction procedures, solvents, analytical techniques,\nand bioassays including other cell lines and end points would be needed. 109  For example, not all substances are soluble\nin THF or ACN, and mainly (semi)volatile compounds can be detected\nwith GC–MS. 110 , 111  Furthermore, some uncertainties\nare related to the substance identification. The low-resolution GC–MS\napproach employed in this study provides only approximate masses and\nthus leaves many uncertainties in library matching including possible\nmisidentification. To limit the number of matches, we relied on the\nsmaller NIST 14 library; however, with this procedure, we may have\nmisidentified substances. High-resolution mass spectrometry, newer\nlibraries (e.g., NIST 20 library), and other suspect lists (e.g.,\nNORMAN Suspect List Exchange) may help overcome this issue in future\nresearch. Another issue for substance identification was the identification\nof UVCBs, mixtures, and chemical products with different compositions\non the market (such as DiDP and DINCH). For example, differentiating\nindividual substances from a mixture [e.g., di(2-propylheptyl) phthalate\n(DPHP, CASRN: 53306-54-0) from DiDP] cannot be guaranteed with the\nstandards we employed. 112\nHazardous\nchemicals in long-lived or recycled products pose a challenge\nto the society as a whole. Our case study on PVC flooring shows that\n(1) hazardous substances are present in long-lived materials, (2)\nuncontrolled recycling is taking place, and (3) monitoring or screening\nof products containing hazardous chemicals is challenging, expensive,\nand time-consuming. Based on our experience, we recommend the following\nactionable points.\nImplementing current regulations, including\nthe phase-out of hazardous chemicals, does not sufficiently cover\nrisks associated with chemicals in long-lived or recycled products.\nThe presence of substances in products needs to be tracked and monitored\nthroughout their life cycle. Initiatives such as the SCIP database\nin the EU, and chemical audits by market surveillance bodies in different\ncountries, are valuable steps in this this direction. However, they\nshould ideally not only rely on self-reporting, extend to products\nalready in use, and (for the case of databases) make a clearer link\nto concrete products/waste streams in the real world. 40 , 113  Importantly, more stringent regulation based on the precautionary\nprinciple would be necessary to avoid burdening future recycled materials,\nto avoid undermining the idea and social acceptance of a circular\neconomy, and to ensure that only clean, safe, and recyclable materials\nare put on the market. 114  This may include\n(1) swift restriction of hazardous substances that show sufficient\nbut not necessarily conclusive evidence for health or environmental\nconcerns, (2) incentivizing simplification and harmonization of material\noptions, including chemical compositions, toward standard formulations,\nand (3) enacting extended producer responsibility toward true recyclability. 115\nIndustry action has in some cases preceded\nregulation of hazardous\nchemicals, including the early phase-out of lead and cadmium stabilizers\nby the EU PVC industry. 23 , 24  Learning from these\nexamples and utilizing existing industry-wide organizations (e.g.,\nVinyl Plus), manufacturers may pioneer and push for a swift phase-out\nof other hazardous substances and transition to safer and more sustainable\nalternatives. Furthermore, learning from the PET water bottles, manufacturers\ncould come together, along with other actors throughout the value\nchains, and establish positive lists that greatly simplify and harmonize\nmaterial options and chemical composition. 106  The recycling industry may enhance sorting by employing available\nfast screening techniques for hazardous substances at scale (e.g.,\nusing XRF for toxic metals and bromine, ATR-FTIR for  ortho -phthalates) and thereby avoiding at least some contamination of\nrecycled materials. However, such efforts are expensive and may not\nbe available to all recyclers (especially in low-income countries)\nand shift the burden of hazardous materials from the manufacturers\nto the waste management sector. Nevertheless, enhancing the traceability\nof chemicals throughout the life cycle of products is urgently needed,\nfor example, by labeling. This would make it possible to proactively\nreact to current or future findings, and make informed decisions on\nwhether and how to recycle materials or whether disposal is the most\nsensible option. In practice, digital product passports that are currently\nbeing discussed could include information on the chemical composition. 116\nFor consumers and designers, it is difficult\nto judge the safety\nof products based on visible characteristics; neither color, presence\nof layers, nor softness reliably predict the presence of hazardous\nsubstances. An independent, reliable, and easily interpretable label\nfor building and construction products (similar to the “Blue\nAngel” in Germany) may simplify consumers’ decision-making.\nFurthermore, citizens can and should demand more transparency, appropriate\nregulation, and industrial responsibility for hazardous chemicals\nin products. 117\nResearchers should\ndevelop or improve simple, fast, and ideally\ncomprehensive methods for identifying and removing hazardous chemicals\nin plastics. Ideally, this includes screening tools that can work\nwith present and future sorting infrastructure, tolerate contamination\nwell, and analyze the plastic directly with minimal preprocessing.\nImportantly, novel processes of removing hazardous chemicals will\nneed to ensure high-quality output materials and lower environmental\nburden compared to incineration and other final disposal options. 118 , 119  Furthermore, researchers should fill knowledge gaps regarding hazards\nof commonly detected emerging substances (e.g., DEHT, DINCH, tin or\nbarium stabilizers) and their mixture toxicity in realistic exposure\nscenarios, taking into account everyday exposure from other sources\nas well. 120","source_license":"CC-BY-4.0","license_restricted":false}