Hydroxyl radical oxidation of chemical contaminants on indoor surfaces and dust

Significance

In this study, we show that a common organic contaminant undergoes multiphase radical oxidation in the indoor environment to produce a myriad of transformation products that can be found in genuine indoor samples. These have relatively unknown toxicities, but humans are likely coexposed to all of them, which could result in greater adverse health effects than any single product alone. We hypothesize that most indoor contaminants will react via the same mechanisms we observe here, resulting in indoor human exposure to tens of thousands of unmonitored transformation products. This highlights the need for further study of the fate, toxicity, and human exposure to transformation products of indoor chemical contaminants.

Abstract

Humans are widely exposed to semivolatile organic contaminants in indoor environments. Many contaminants have long lifetimes following partitioning to the large surface reservoirs present indoors, which leads to long exposure times to gas-phase oxidants and multiphase chemistry. Studies have shown selective multiphase oxidation of organics on indoor surfaces, but the presence of hydroxyl radicals with nonselective reactivity and evidence of multiphase OH radical reactivity toward common indoor contaminants indicates that there may be additional unknown transformation chemistry indoors. We screened genuine indoor samples for 60 OH radical oxidation products of the common plasticizer and endocrine-disrupting contaminant bis(2-ethylhexyl) phthalate (DEHP) identified in laboratory experiments using nontargeted high-resolution mass spectrometry. At least 30 and 10 of these products are observed in indoor dust and DEHP films exposed to ambient indoor conditions, respectively, indicating that multiphase OH reactions occur indoors. Using the PROTEX model and a multimedia indoor chemical fate model, we demonstrate that these products have long indoor lifetimes and cause a higher potential for human exposure than DEHP. Some of these products are more active endocrine disruptors than DEHP itself, but most have unknown toxicities. Coexposure to all oxidation products will likely have an additive effect, leading to higher human health risks from indoor organic contaminants than previously thought. Due to the nonselective reactivity of OH radicals, it is likely that most indoor contaminants follow similar chemistry, and further study is needed to understand the prevalence and human health implications of such multiphase chemistry.

Organic contaminants sourced from building materials; appliances and furniture; infiltration of outdoor pollutants; personal care, cleaning, pest control, and fragrance products; and cooking, smoking, and biogenic emissions are widely found in indoor environments (1–4). Humans are exposed to these contaminants indoors through dermal uptake, mouthing-mediated ingestion, and inhalation (5, 6). Indoor environments, particularly residential spaces, are characterized by relatively low light intensity, low ventilation rates (especially for modern, energy-efficient buildings), and high surface area-to-volume ratios (7). This results in many chemical contaminants selectively partitioning to condensed-phase films, indoor dust, and other surface compartments indoors, even compounds which would be considered volatile in outdoor environments, resulting in long chemical lifetimes indoors due to low ventilation and gas-phase reaction loss rates.

These are ideal conditions for gas-surface multiphase chemistry. Multiphase ozonolysis of a variety of alkene- and aromatic-containing contaminants has been widely reported and is likely the dominant chemical fate for some types of semivolatile unsaturated compounds (8–11). Multiphase nitration mediated by nitrous acid (HONO) is observed for some contaminants, forming toxic products observed in indoor environments (12, 13). Multiphase chlorination is also observed following bleach cleaning (14). However, all these oxidants are relatively selective in their reactivity, interacting only with compounds containing specific functional groups (in the case of ozone or HONO) or are transient (in the case of bleach). Hydroxyl radicals are not selective, as they react with almost any organic molecule. OH radicals are formed through ozonolysis of alkenes and photolysis of HONO to produce low but consistent steady-state concentrations in the indoor environment, with spikes of OH concentrations observed during some indoor activities (15–20). There may also be potential for OH generation within surfaces through decomposition of organic peroxides or ozonolysis of alkenes. These OH radicals will react with indoor contaminants, as demonstrated by Alwarda et al., who showed OH reactivity with thin films of di-n-octyl phthalate, likely playing a significant role in the overall lifetime of this organic contaminant indoors (21). However, that work did not take advantage of modern analytical methods to characterize reaction products, and so it remains an open question whether multiphase OH radical chemistry could produce significant quantities of oxidation products indoors, what those products are, to what extent humans are exposed to them, and whether they exhibit any toxicological activity.

To address this knowledge gap, we performed targeted laboratory experiments to identify 60 products of multiphase OH oxidation of bis(2-ethylhexyl) phthalate (DEHP) using nontargeted high-resolution mass spectrometric screening and have validated these experiments by showing that these products form at measurable levels in genuine indoor environments. DEHP is one of the most common contaminants found indoors due to its widespread use as a plasticizer with mean concentrations observed in indoor dust between 2000 and 2021 ranging from approximately 250 μg/g in North America to over 600 μg/g in Asia (22), and its aromatic ring, long alkyl chains, and ester functional groups exhibit reactivity that is likely to be shared by many other indoor contaminants. DEHP also exhibits significant toxicological effects following human exposure (23, 24), and many of the observed products are known to be active endocrine disruptors. Our results show that indoor contaminants in indoor films and dusts can be nonselectively oxidized by gas-phase OH radicals. This chemistry will occur for all semivolatile organic compounds reactive to OH radicals, producing tens of thousands of unknown oxidation products with simultaneous human exposures and unknown toxicities.

Results and Discussion

Experimental Oxidation of Bis(2-ethylhexyl) Phthalate.

We studied the decay of DEHP and the formation of oxidation products in a deposited 5 nm film of DEHP exposed to gas-phase OH radicals (Fig. 1) and found that DEHP was oxidized at rates relevant to indoor chemical lifetimes, consistent with previous work (21). Scaling our experimental conditions to those in the indoor environment (assuming an average OH radical concentration of 2 × 10⁵ molec cm⁻³), DEHP exposed to indoor air would have a half-life with respect to multiphase OH oxidation of about 2 wk. This value provides a rough upper estimate of the multiphase reaction rate of DEHP on indoor surfaces, although there is expected to be significant variability in OH radical concentrations between different indoor spaces and not all DEHP may be available to react with OH if it is buried within surfaces.

Fig. 1.

Degradation of Bis(2-ethylhexyl) phthalate in thin films and formation of total product signal upon exposure to OH radicals in the gas-phase. The upper x-axis shows an estimated indoor equivalent exposure time assuming an OH radical concentration of 2 × 10⁵ molecules cm⁻³. Error bars show SE with N = 3.

As DEHP degrades, it produces a variety of products with different levels of oxidation detected by nontarget and suspect screening using high-resolution mass spectrometry (HRMS). Overall, 34 unique molecular formulae were assigned with an additional 26 isomers separable with chromatography. The formation of these specific products was confirmed through additional oxidation experiments of deuterated standard of DEHP. Products were observed with combinations of between 0 and 4 oxygen additions, between 0 and 2 additional unsaturations, and bond cleavages resulting in loss of 1, 2, 3, 4, 5, 6, 7, 8, or 16 carbon atoms from the starting formula of DEHP (C₂₄H₃₈O₄). All products detected are summarized in SI Appendix, Table S1 with predicted structures and confidence levels for the highest-intensity products shown in SI Appendix, Fig. S1 and Table S2 and MS/MS spectra supporting these assignments in SI Appendix, Figs. S2–S13.

In Fig. 1, there is also evidence for DEHP degradation and product formation in the UVC control (experimental conditions with no H₂O₂ introduced into the flow tube). As discussed further in the SI Appendix, this was likely also multiphase OH radical oxidation due to the presence of residual H₂O₂ within the flow tube during control experiments. No similar degradation was observed for an H₂O₂ control experiment without UVC light.

Product Distribution and Mechanisms of the Multiphase OH Radical Oxidation of DEHP.

Out of the 60 product signals identified, 56 were semiquantified using the HRMS while 4 had intensities too low to effectively semiquantify. Fig. 2A shows the formation and further degradation of the major products formed (isomers are summed to produce a single signal intensity for each molecular formula). The most abundant products were consistently those produced by a single transformation event, including an O addition (as in C₂₄H₃₈O₅), a C─C or C─O bond cleavage (as in C₁₆H₂₂O₄) or an elimination resulting in an additional unsaturation (as in C₂₄H₃₆O₄) which can occur in tandem with an O addition (as in C₂₄H₃₆O₅). A mechanism for how these three types of reactions could occur is shown in Fig. 2B, although this mechanism does not include all possible pathways, nor does it show permutations where multiple reactions occur (i.e., the elimination reaction could result in an aldehyde on the leaving group and a radical on the remaining phthalate moiety instead of an aldehyde/ketone on the observed product). Detailed proposed mechanisms for the formation of C₂₄H₃₈O₅, C₂₄H₃₆O₅, C₂₄H₃₆O₄, and C₁₆H₂₂O₄ are given in SI Appendix, Fig. S16, including the Russell mechanism, an alternative pathway for RO₂–RO₂ radical termination not shown in Fig. 2B which is a likely formation mechanism for certain alcohols and ketones (25). These mechanisms are similar to the autooxidation pathways of alkanes, which produce many of the same functional groups and are also initiated by OH radicals (26–30), and autooxidation may have occurred in parallel with directly OH-initiated chemistry in our experiments. All the single-step transformation products showed primary behavior, forming quickly in early time points and then further reacting as demonstrated by the reduced signal intensities in later time points, generally to form products with more than one transformation event relative to DEHP. There was no consistently observed terminal product in high yield for this reaction, indicating that this reaction is likely to be highly dynamic indoors with a constant source of DEHP, with products of varying degrees of oxidation constantly being formed and then decaying away.

Fig. 2.

Selected products signal intensities normalized to the initial signal intensity of DEHP (A) where experimental oxidation time in hours is indicated by color, an example reaction mechanism (B), and product signal yields as a function of DEHP signal lost (C) from the OH radical oxidation of pure DEHP films. Error bars represent SE with N = 3.

One notable product is C₂₄H₃₆O₄, which is an alkene that was formed through an elimination reaction. Since alkenes are reactive to ozone, this product is likely to undergo further oxidation not observed in our experiments in the indoor environment and could be an additional source of OH radicals through ozonolysis. We confirmed the identity of this product by running an experiment where OH radicals were formed through photolysis of ozone in the presence of water (see the SI Appendix for details). In this experiment, while other products formed with a similar distribution to the H₂O₂-generated OH radicals, much less C₂₄H₃₆O₄ product was observed relative to other products, indicating that it is reacting with ozone. We also manually verified that the C₂₄H₃₆O₄ product signal could not be explained by in-source fragmentation and found that no precursor molecular signals were present at the same retention time.

The overall signal intensity yield for all observed products as a function of reaction extent (quantified as the fraction of DEHP degraded) for the experimental and control conditions is shown in Fig. 2C. The observed product yield was initially high, indicating that our methods successfully detected most of the first-generation products (i.e., products formed from a single oxidation step) formed by this reaction. However, as the reaction progressed, the observed product yield decreased. There are many possible reasons for this. First, the increasingly oxidized products observed later in the reaction tend to spread out the overall product signal over many molecular features with different m/z values and retention time. If some products have concentrations too low to be detected using this instrument, the overall product signal intensity observed would be reduced. Second, because no standards exist for most of these products, when comparing signal intensities between compounds, we assumed that their ionization efficiencies were equal. While the isocratic elution program used helps to reduce variability in ionization efficiency with a constant mobile phase composition, the increasingly oxidized products tend to elute earlier (SI Appendix, Table S1), where there is generally more background interference with ionization in the electrospray source, which could reduce overall product signal intensity (31). Third, oxidized products may not have similar ionization efficiencies as DEHP due to differences in polarity, the addition of functional groups, or in-source fragmentation (32). Fourth, some oligomerization could have occurred within the film due to RO˙-R˙ or R˙-R˙ chemistry, which would produce highly retained compounds not detectable using our chromatographic method.

There are three important limitations to our experimental techniques. First, organoperoxides (including C₂₄H₃₈O₆, likely isomers E–G based on retention times) are likely artificially suppressed due to the use of UVC light to generate OH radicals. Similarly, the anhydride product (C₂₄H₃₆O₅ where a ketone is formed adjacent to the ester moiety) is likely to be hydrolyzed during the extraction process, forming C₁₆H₂₂O₄. Finally, relative to the indoor environment, our experimental conditions had high radical concentrations, and therefore, likely suppressed autoxidation mechanisms that could be occurring in the surface film by accelerating radical initiation and termination mechanisms, making the autooxidation propagation mechanism less important.

OH Radical Oxidation Products of DEHP are Observed in Genuine Indoor Samples.

Using the product retention times and MS–MS parameters identified in our laboratory experiments, we analyzed three indoor dust samples for the major OH oxidation products of DEHP using triple quadrupole tandem mass spectrometry (LRMS). In all three indoor dust samples tested, all major oxidation products were observed as shown in Fig. 3A normalized to DEHP concentrations present in each dust. When the NIST standard reference material (SRM) dust was exposed to the same laboratory OH oxidation conditions as the DEHP films were, all major product signals consistently gained intensity with increasing OH exposure time at the same retention times as in the unoxidized dust (Fig. 3B), indicating that the products observed in dust are identical to those formed during OH oxidation experiments. Similarly, when 5 nm films of DEHP were exposed to ambient indoor conditions in three different locations, the most intense primary products were uniformly observed forming (Fig. 3C). The highest product signal intensities were observed in the film placed near a window in a regularly occupied office, followed by those in a dark location in the same office, and then those in an erratically occupied common room (SI Appendix, Fig. S17). These differences are likely caused by differences in OH radical concentrations, as photolysis of nitrous acid near windows and ozonolysis of human skin lipids have both been reported to elevate OH radical concentrations indoors.(15, 17) We note that the signal intensities observed in indoor dust and ambient-exposed films are lower than would be expected based on the oxidation experiments we performed. While we cannot rule out that our experiments are overestimating the true environmental exposure of DEHP to OH (e.g. the OH mixing ratio in our indoor settings is lower than 2 × 10⁵ molec cm⁻³), there are several large uncertainties associated with product semiquantification specific to these environmental samples that may explain the differences in product signal intensities as discussed further in the SI Appendix.

Fig. 3.

Product signal intensities from fresh dusts (A), SRM dust exposed to gas-phase OH radicals (B), a film of DEHP exposed to ambient indoor conditions in an office (C), and chromatograms of product signals from a variety of representative samples as measured in LRMS (D). Error bars in (A–C) represent SE with N = 3. To facilitate comparisons between peak shapes and retention times across chromatograms in (D), the peaks of each sample retention time are corrected to the retention time of DEHP, and peak heights are normalized, so relative peak intensities do not represent relative compound concentrations.

Throughout all experiments and analyzed samples, the retention times and distributions of isomers were consistent as shown by the chromatograms in Fig. 3D and SI Appendix, Fig. S18, indicating that the same products formed in our laboratory experiments, in genuine indoor dust, and in films exposed to ambient conditions. In particular, the peak shapes for the C₂₄H₃₈O₅ and C₂₄H₃₆O₅ O-addition products were distinctive and consistent across samples (SI Appendix, Figs. S19 and S20). The broad C₂₄H₃₆O₄ signal in the oxidized film and dust samples is indicative of many isomers not separable by chromatography but the retention time of the signal was consistent between all samples where it was observed (SI Appendix, Fig. S21). Unsurprisingly, the distributions between product molecular formulae were somewhat different when comparing laboratory-oxidized DEHP films to ambient-exposed DEHP films to indoor dusts, as there was significant variability in oxidation conditions and DEHP concentrations between these different samples. One key difference is the large C₂₄H₃₈O₆ E–G isomer signal in the ambient-exposed DEHP films and vacuum and air purifier dusts relative to oxidized DEHP films or the SRM dust (SI Appendix, Fig. S22). These isomers are likely stable organic peroxides, supported by their suppression in experiments with UVC-generated OH radicals (which would photolyze these compounds) and their relatively long retention times corresponding to higher predicted octanol-water partitioning ratios predicted for organic peroxide structures (SI Appendix, Table S1). Organic peroxides are relatively unstable compared to other products which could explain the low initial concentration in the SRM dust and the subsequent growth of these isomers following OH exposure. As discussed previously, the use of UV light to generate OH radicals will photolyze organic peroxides generated in the oxidized pure DEHP films, resulting in the low relative organic peroxide signals observed there. Another notable difference is the lack of C₂₄H₃₆O₄ products in the ambient-exposed DEHP films visible in Fig. 3D, likely due to rapid ozonolysis of any alkenes produced relative to the rate of OH oxidation during those experiments. While it is possible that other oxidation products of DEHP are photolyzed by UV light, this is likely a minor reaction pathway as the product distributions in the DEHP oxidation experiments are otherwise similar to those found in indoor dust samples and DEHP films exposed to ambient conditions.

Implications for Human Exposure and Toxicity.

The evidence we have presented shows that the common indoor contaminant DEHP can react with OH radicals on environmentally relevant timescales to produce a variety of oxidation products that are observed across indoor environments. Applying the multiphase chemistry model presented in our previous work, under default conditions this process is likely to be a major sink for DEHP in the indoor environment, although significant uncertainties apply to this estimation (33). The presence of these products indoors raises the question of the extent to which humans are exposed to these transformation products. To address this question, a multiphase indoor chemical fate model and the indoor fate and human exposure modules of PROTEX model were applied to calculate indoor lifetimes, chemical fate, human exposure, and relative human whole-body lipid concentrations for the most abundant transformation products observed. Since experimental data are not available for most products, we used modeled partitioning ratios, biotransformation rates, and gas-phase OH radical reaction rates for all compounds (including DEHP) based on predicted structures given in SI Appendix, Fig. S1. These estimated parameters have large uncertainties (SI Appendix, Table S1) so the absolute values of results should be considered cautiously. Instead, we report values relative to DEHP and consider changes in exposure mechanisms and modeled in-body concentrations following OH oxidation indoors.

All OH radical transformation products of DEHP are predicted by the multiphase chemistry model to have long indoor lifetimes controlled primarily (>80%) by further multiphase OH radical reactions (SI Appendix, Fig. S23). The role of further multiphase chemistry in the lifetimes of these transformation products is supported by the presence of second and further-generation products in indoor dusts, including C₁₆H₂₂O₅, C₂₀H₃₀O₅, C₂₄H₃₆O₆, and C₂₄H₃₈O₆ (Dataset S1). All transformation products primarily partition into surface reservoirs, leading to very little loss to ventilation or gas-phase chemistry, but cleaning may also play a role in controlling the lifetimes of these products depending on cleaning frequency and efficiency.

Given these long chemical lifetimes and their presence in indoor samples, human exposure to these products is inevitable. Using the estimated physicochemical parameters and human biotransformation and elimination rates in SI Appendix, Table S1, the PROTEX human exposure module was applied to model average daily doses through different exposure pathways and resulting human body lipid-normalized concentrations of DEHP and its oxidation products assuming unit emissions for each.

Exposure to DEHP sourced indoors and all its oxidation products is predicted to be largely (>80%) via mouthing-mediated ingestion (SI Appendix, Fig. S24A), reflecting the tendency for these compounds to partition into surface compartments and indoor dust. The whole-body lipid-normalized concentrations resulting from this exposure relative to those predicted for DEHP are shown in Fig. 4A (right column of each pair) scaled to the signal intensities observed in each indoor dust sample relative to DEHP (left column of each pair). Exposure predictions for experimentally oxidized DEHP films are also given in SI Appendix, Fig. S25. Oxygen addition products are predicted to preferentially accumulate in the human body over indoor dust relative to DEHP, while C─C bond cleavage products are less likely to accumulate in the human body (SI Appendix, Fig. S25). Overall, the oxidation products of DEHP are predicted to be enriched in the human body by approximately a factor of two over their concentrations in indoor dust relative to DEHP. This result is explained primarily by differences in the physicochemical properties of these products as depicted in Fig. 4B, where the human body lipid-normalized concentration of each compound is shown as a function of their octanol–air and air–water partitioning ratios (K_oa and K_aw) assuming products are neutral and biotransformation rates equal to that of DEHP. Oxygen addition products tend to have higher K_oa and lower K_aw values than DEHP due to the addition of polar functional groups reducing vapor pressures and enhancing water solubility. Some products are also acids and may dissociate depending on the pH of their environment, further reducing their vapor pressure and increasing water solubility beyond the values in Fig. 4B. In turn, low vapor pressures and high water solubilities increase the rate of mouthing-mediated ingestion due to increased partitioning into indoor surface compartments and indoor dust (SI Appendix, Fig. S24B) and may also directly enhance uptake in humans via increased dietary absorption efficiency. Children will be exposed to these products significantly more than adults, as they have much higher rates of mouthing-mediated ingestion through mouthing behaviors relative to their body mass (34). In contrast, shorter-chain products with lower K_oa values have reduced rates of mouthing-mediated ingestion. Many products also have longer modeled biotransformation half-lives than DEHP, which could also contribute to the enhanced predicted human body concentrations of these products relative to DEHP.

Fig. 4.

(A) Relative signal intensities of the oxidation products of DEHP in untreated indoor dusts (first columns of each pair) and resulting modeled whole-body lipid-normalized concentrations in humans (second columns) normalized to DEHP, assuming the signal intensities in the first columns represent the average concentration in the whole modeled indoor environment. Product distributions are assumed to be equal to the signal intensity distributions. Many of the most abundant oxidation products are predicted to have enhanced uptake in the human body relative to DEHP, resulting in relative human body lipid-normalized concentrations of these products approximately two times higher than in indoor dust. This enhanced uptake is driven by changes in the products’ physicochemical properties shown in (B), where the contour plot shows modeled whole-body lipid concentrations relative to DEHP as a function of the octanol–air and air–water partitioning ratios (K_oa and K_aw), assuming the same biotransformation and air degradation rates as DEHP. DEHP and each of its oxidation products (assumed to be in neutral states) are overlaid on top of this contour based on their modeled physicochemical properties, showing how the oxidation products’ increased K_oa and decreased K_aw enhance the uptake of these products relative to DEHP, while products with decreased K_oa values tend to have lower uptake. SE bars are shown only for DEHP for visual clarity.

The formation of and exposure to these oxidation products introduces significant uncertainty in the human health risk associated with DEHP contamination in indoor environments. Most of these products have no available toxicity data. Where studies have been performed (because human metabolites overlap with observed OH oxidation products), certain products exhibit activity toward nuclear receptors with the potential to cause adverse health effects. Mono(2-ethylhexyl phthalate) activates the peroxisome proliferator-activated receptors α and γ, while 4-hydroxy- bis(2-ethylhexyl) phthalate inhibits estrogen receptors α and β (35–38). Exposure to these and other metabolites of DEHP is associated with reductions in male sperm quality, alterations in thyroid function, obesity, kidney diseases, and ADHD development in children in epidemiological studies (39). While most individual products are formed at low concentrations relative to DEHP, the combination of enhanced exposure to and biotransformation and absorption of many oxidation products with similar structures and possibly toxicological effects could induce a significant additive human health impact. Further study is needed to investigate the impacts of DEHP oxidation products on human health.

While we have selected DEHP as a model compound for this study, the OH radical is a nonselective oxidant. Since OH oxidation products of DEHP, a relatively inert molecule, can be observed in the indoor environment, it is likely that OH oxidation products of many other indoor contaminants are also present. Some transformation products of other contaminants formed through different indoor multiphase chemistries have been identified, including nitration products of bisphenol A and nicotine (12, 13) and ozone oxidation products of oils, bisphenol A, nicotine, polycyclic aromatic hydrocarbons, and tetrahydrocannabinol (9, 11, 40–42). However, the nonselective nature of OH radical chemistry contrasts with these more selective nitration and ozonation reactions, and given the thousands of organic contaminants identified in indoor dust alone (3), there may be tens of thousands of additional unmonitored transformation products present in the indoor environment. These products have unknown toxicities and human health risks. Identifying and studying these compounds will be a difficult undertaking, as in many cases, OH radical chemistry produces many low-intensity products, especially when a compound has many reactive sites on an alkyl chain like DEHP. This is common in other phthalate esters, nonphthalate plasticizers, and organophosphate flame retardants. However, mixture toxicity following possible coexposure to tens of thousands of low-concentration transformation products in addition to the parent molecules indoors is a health risk that cannot be ignored. Future work Identifying, quantifying, and characterizing the exposure routes and toxicity of these transformation products, prevalence in samples from other indoor environments, and sources of OH radicals which promote their formation will improve a holistic understanding of the impacts of indoor contamination on human health.

Materials and Methods

Dust Sampling and Extraction.

Three indoor dust samples were used in this study. Two samples were collected from a residential apartment in Toronto, Ontario, Canada in November of 2022. The vacuum dust was collected from the floor of the whole apartment using a consumer vacuum cleaner, and the air purifier dust was collected by brushing off dust from the front of the HEPA filter in a consumer air purifier, both representing about 2 wk of dust accumulation. Both dusts were sieved through a 250 μm mesh to remove large debris. The third dust sample was the NIST Standard Reference Material (SRM) 2585, which was used without additional treatment. The SRM dust is a homogenized mixture of dusts from residences, hotels, and commercial cleaning service vacuums, and therefore represents a broad range of indoor environments. All dust samples were stored at 4 °C until use. Dust samples were extracted in 80:20 methanol:water and sonicated for five minutes, then spiked with DEHP-d4 (Toronto Research Chemicals) as an internal standard for quantification. Based on the reported value of DEHP within the SRM 2585 dust, this procedure had an extraction efficiency of 105% ± 11% (43). The SRM dust has a reported DEHP concentration of 570 μg/g, (43) and the measured DEHP concentrations in the apartment vacuum and air purifier dusts are 90 ± 22 μg/g and 230 ± 50 μg/g respectively. Samples were filtered through a 0.22-μm nylon syringe filter prior to LC-MS analysis. Throughout this study, glass was used whenever possible over plastic to minimize contamination.

DEHP Film Deposition and Extraction.

Thin films of DEHP and deuterated DEHP (DEHP-d4) were produced by depositing 100 μL of 31.9 μM DEHP solution in 90% HPLC-grade dichloromethane:10% HPLC-grade acetonitrile onto an 18 mm diameter glass coverslip. When dried, this would produce a theoretical film thickness of 5 nm if the DEHP were evenly distributed. Following oxidation, films were extracted in 80:20 methanol:water and sonicated for 5 min, then spiked with either DEHP-d4 or DnOP as internal standards for quantification for an average extraction efficiency for DEHP of 107 ± 5%. Films exposed to ambient conditions were filtered through a 0.22 nylon syringe filter prior to LC-MS analysis.

Ambient DEHP Film Exposure.

DEHP films were exposed to ambient conditions near an office window, on a shelf far from the window within the same office, and in a common room within the chemistry building on the University of Toronto campus for 42 d in December 2022 and January 2023. Within this building, the temperature is maintained between 21 and 24 degrees Celsius and in the winter the relative humidity is generally 30% or below. The indoor–outdoor ratio of ozone is approximately 0.7, leading to ozone concentrations varying between approximately 3 and 15 parts per billion based on publicly available data collected in downtown Toronto (44). Three coverslips were analyzed at 0, 9, 21, and 42 d, and three blank coverslips (containing no DEHP) at each location were extracted at 42 d.

Laboratory DEHP Film OH Oxidation Experiments.

DEHP films were exposed to controlled elevated OH radical concentrations using a flow reactor apparatus (21). For each experiment, a set of coverslips containing 5 nm DEHP films produced by wet deposition of a solution in 90:10 dichloromethane:acetonitrile were placed on a Teflon bar and inserted into the quartz flow reactor irradiated by a single 254 nm low-pressure mercury UV lamp running the length of the flow tube. The H₂O₂ concentration within the reactor was approximately 160 ppm and the gas residence time was 6 min. The UVC/H₂O₂ OH source was chosen following a series of experiments with different types and numbers of UV lamps to minimize overall UV exposure and experiment time while providing a stable and elevated concentration of OH radicals. At each time point, three cover slips were extracted and analyzed for DEHP concentration and product formation. The flow tube average OH radical concentration in the flow reactor was 5.5 × 10⁷ molec cm⁻³ with a precision of 20% measured using the indirect approach described in Alwarda et al. (21); systematic errors may be comparable. Control experiments with H₂O₂ and no UV light were also performed and no DEHP degradation was observed. Additional details of the flow reactor apparatus and the OH measurement experiments are provided in the SI Appendix.

LC-MS/MS Analysis and Product Identification.

Quantification of DEHP decay and identification, MS/MS analysis, and semiquantification of its oxidation products in pure films and in the SRM dust were performed using liquid chromatography (LC) coupled to a high-resolution Q-Exactive orbitrap mass spectrometer (HRMS) with positive-mode electrospray ionization (ESI). Product semiquantifications in ambient-exposed films and unaged indoor dusts at high concentrations were performed using LC-ESI connected to a TSQ Endura triple quadrupole mass spectrometer (LRMS). Selected thin film and SRM oxidation samples were also analyzed on the LRMS to compare chromatograms. Separations were accomplished with 80% LC-MS grade methanol in 20% MilliQ water in an isocratic elution program through a HALO Phenyl-Hexyl column (100 × 2.1 mm, 2.7 μm pore size). Tentative product mass spectral features were identified in HRMS using a combination of suspect screening for expected products of the OH reaction with DEHP, bottom–up screening using the phthalate anhydride fragment ion, and direct MS1 nontarget analysis. Tentative features were then assigned molecular formulae where possible based on their exact masses and were filtered manually to remove features without feasible molecular formulae, features present in blanks, controls, and unreacted material, features without expected adducts and isotopologues, and features not present in an oxidation experiment with deuterated DEHP. This resulted in the final identified product list. LC-MS instrumental parameters, details of DEHP and product quantification using both instruments, additional details and rationale for product identification, and kinetic analysis are provided in SI Appendix.

Modeling Approaches.

Human exposure to DEHP and its transformation products and modeled using the PROTEX (PROduction-To-EXposure) model (45). Supplied with chemical partitioning and reactivity properties, PROTEX simulates the distribution of chemicals within an indoor environment, parameterized to represent an archetypal American home, comprising indoor air, hard surfaces, carpet, flooring, and walls and ceiling. It additionally simulates human exposure to chemicals through inhalation of indoor air, mouthing-mediated ingestion, and dermal absorption of chemicals through human skin. Fate and multiphase lifetime of DEHP transformation products were estimated using the multiphase fugacity model developed by Fahy et al. with multiphase OH reaction rates for transformation products scaled from that of DEHP based on their model-predicted gas-phase reaction rate constants (33). Unless otherwise stated, physicochemical properties for transformation products were calculated using an ensemble approach based on the structures in SI Appendix, Fig. S1 by taking the geometric mean of the values predicted by the EPIsuite, OPERA, IFSQSAR, and QSARINS estimation approaches where applicable, and are summarized in SI Appendix, Table S2 (46–50). Additional details of both models and their inputs can be found in the SI Appendix.

Supplementary Material

Appendix 01 (PDF)

Dataset S01 (XLSX)

Data, Materials, and Software Availability

All study data are included in the article and/or supporting information.