Comparative Assessment of Personal Exposures to Environmental Contaminants Using Silicone Wristbands and Fresh Air Wristbands Authors: Elizabeth Z. Lin, Nicholas J. Herkert, Duncan Hay, Ellen M. Cooper, Heather M. Stapleton, Krystal J. Godri Pollitt Published: 2026 ABSTRACT Wearable passive samplers are practical tools for personal exposure assessment, yet how sampler design and analysis methods shape the resulting measurements remains poorly characterized. We compared two widely used samplers, silicone and Fresh Air wristbands, worn simultaneously by 30 adults in Durham County, North Carolina, USA over 5 days with paired urine collection. All samples were analyzed using gas chromatography high-resolution mass spectrometry, covering a broad chemical space (m/z 53.4–1050), using five analytical approaches to separate the influence of sampler design from that of analytical workflow. Targeted analysis of 15 chemicals showed systematically different concentrations that were correlated between designs for several compounds: silicone wristbands had 33- to 189-fold higher levels of less volatile compounds, while Fresh Air wristbands captured 9- to 80-fold higher levels of more volatile compounds. The magnitude of these differences increased with the log K OA. Suspect screening annotated 1330 compounds in silicone wristbands versus 2261 in Fresh Air wristbands, with distinct molecular weight distributions: higher in molecular weight in silicone wristbands (80% at 100–300 MW) compared to Fresh Air wristbands (93% at 50–200 MW). Measurement of the flame-retardant TCPP from both wristbands correlated with its urinary metabolite. Sampler design and analytical methods each influenced measured exposures, underscoring the need for harmonization and well-documented methods to enable cross-study comparison. Introduction Exposure to environmental chemicals is associated with a wide range of adverse health outcomes, including respiratory and cardiovascular disease, cancer, and developmental disorders. − These chemicals originate from diverse sources, including consumer products, building materials, food products, pharmaceuticals, and furnishings, − and partition dynamically among gas-phase air, airborne particles, and settled dust. − Assessing personal exposure to environmental chemicals has traditionally been assessed using various sample matrices, each offering distinct advantages for characterizing different exposure pathways. ,− Settled dust analysis enables assessment of chemicals that accumulate on indoor surfaces over time, particularly less volatile organic compounds. Hand-wipe sampling directly measures chemicals transferred to skin through contact with surfaces and products. Active air sampling captures chemicals in both gas and particle phases, providing a measure of airborne contaminants from diverse sources. ,,,− Passive air sampling techniques, which collect airborne chemicals without pumps or power, have been used for environmental monitoring since the 1970s. More recently, these techniques have been adapted into wearable forms that enable continuous, unobtrusive monitoring of personal exposures across the varied environments in which individuals encounter in daily life. ,− The two most widely used wearable passive samplers for organic chemical exposure assessment are silicone wristbands and Fresh Air wristbands. ,, Both designs use polydimethylsiloxane (PDMS, also known as silicone rubber) as the sorbent material but differ in configuration and the routes through which they capture the chemicals. Silicone wristbands are flexible silicone loops worn directly on the wrist with contact with both air and skin. This design enables uptake through passive diffusion and particle deposition from air as well as through direct contact with skin, consumer products, and other surfaces. In contrast, Fresh Air wristbands house PDMS within a perforated case that prevents direct surface contact, restricting chemical uptake exclusively to airborne compounds. These differences in wristband design and air-only versus air-plus-surface sampling have important implications for interpreting measured exposures, as the configuration determines which environmental chemical reservoirs can be accessed. Chemical partitioning behavior, governed by the octanol-air partition coefficient (K OA), determines the distribution of different compounds between the air and surfaces. Compounds with a log K OA below approximately 9 are sufficiently volatile to remain predominantly in the gas phase and are therefore readily accessible to air samplers. ,, Conversely, those with a higher log K OA preferentially partition into the particle phase where they accumulate at concentrations that can exceed air concentrations by several orders of magnitude. ,, Consequently, wristbands that contact surfaces may therefore capture elevated levels of high-log K OA compounds through direct transfer, while air-only samplers detect primarily gas-phase chemicals. However, the behavior of these compounds (i.e., whether they partition to the gas or particle phase) can vary with dynamic changes in temperature, relative humidity, and other environmental conditions as people go about their daily routines. As a result, detections and concentrations in each phase can vary across individuals, time, and space, highlighting the need for wearable passive samplers to accurately capture personal exposures to these pollutants. Sampler selection must consider both the chemical properties of analytes of interest and the study’s exposure assessment objectives. Beyond the wristband design, analytical methods introduce another source of variability in exposure measurements. Chemical analysis typically employs gas or liquid chromatography coupled with mass spectrometry, with multiple options for sample preparation, extraction protocols, instrumentation configurations, and data-processing workflows. These methodological decisions affect the range of chemicals detected and the magnitude of their measured abundances. Interlaboratory comparison studies have demonstrated this variability. One study focused on semivolatile organic compounds (SVOCs) measured in air reported average relative standard deviations exceeding 50% for 146 compounds across 15 participating laboratories, while another collaborative trial focused on nontargeted analysis found substantial differences in both the number and identity of chemical features measured across 16 laboratories using different analytical platforms. , However, few studies have systematically examined how specific analytical parameters contribute to measurement variability in wearable passive-sampler studies. The variance of protocols across samplers limits cross-study comparability and hinders the integration of wearable passive-sampler data across research efforts. A systematic evaluation of both wristband design and analytical methods is needed to inform best practices. Biomonitoring of urine and blood has traditionally been used to assess human chemical exposure, providing a measure of the internal dose that integrates absorption across exposure routes. Comparing these biological measures with estimates from wearable passive samplers is therefore useful for evaluating the exposure information that the samplers provide. Silicone wristband concentrations of organophosphate esters and other flame retardants have been shown to correlate with paired urinary metabolite and serum concentrations, supporting their use as predictors of the internal dose. ,, The objective of this study was to compare personal exposures characterized by silicone wristbands and Fresh Air wristbands and evaluate the influence of analytical protocols. We assessed how the two wristband systems differ in the diversity and magnitude of detected chemicals assessed using a combination of targeted and nontargeted analyses, examined the relationship between the log K OA and measured exposures, and investigated how analytical methodology, including temperature programming, injection systems, and internal standard normalization, influences measurement variability. Applying five analytical approaches across both wristbands allowed us to distinguish the contributions of sampler design from those of the analytical workflow. We also compared wristband-based personal exposure estimates with paired urinary metabolites to assess the extent to which measurements from each sampler corresponded to biomarkers of internal dose. Method Study Population and Design Study participants (n = 30, aged ≥ 18 years) were recruited from February to March 2022 in Durham County, NC, USA. Detailed population characteristics can be found in Table S1. Participants were asked to wear silicone and Fresh Air wristbands simultaneously on the same arm for 5 days and complete a series of questionnaires (see Supporting Information). Study protocols and related materials were approved by the Institutional Review Boards at Duke University (2017–0463) and Yale University (2000031698). All participants gave informed consent prior to providing any information or personal samples. Personal Exposure Assessment Silicone wristbands were prepared by the Duke team, while Fresh Air wristbands were prepared by the Yale team and shipped to Duke for deployment. Both wristband types were deployed and collected from participants at the same time each day (see Supporting Information). Participants were instructed to wear both wristbands on the same arm and follow identical care protocols throughout the five day assessment period. However, unlike silicone wristbands, which were worn continuously, participants were asked to remove the Fresh Air wristbands during water immersion activities (e.g., showering, bathing, and swimming), and they could place Fresh Air wristbands beside the bed during sleep if they preferred. Analytical Approaches Following collection, wristbands were analyzed under five analytical approaches (Approaches 1–5; Table ). Approaches 1 and 4 were established, laboratory-specific standard protocols for the silicone and Fresh Air wristbands, respectively; these were used for targeted quantification. The two wristbands differ as complete sampler-workflows, not only in sampler design and exposure route but also in extraction method, analytical protocol, targeted analyte panels, and wear instructions. Their contrast therefore reflects the combined difference between these two systems rather than sampler design in isolation. Approach 1 used solvent extraction followed by analysis on a Q Exactive GC Orbitrap with a programmable temperature vaporizer (PTV) inlet and a nonlinear temperature ramp, while Approach 4 used direct thermal desorption of the PDMS sorbent bar with a cooled injection system (CIS) and a linear temperature ramp. 1Overview of Five Analyses That were Conducted to Compare Environmental Chemicals Measured by Two Wearable Passive Samples Approach 1 Approach 2 Approach 3 Approach 4 Approach 5 Wristband SiliconeSiliconeSiliconeFresh AirFresh Air Extraction Method solventsolventsolventdirect thermaldirect thermal Sample silicone extract solutionsilicone extract solutionsilicone extract solutionPDMS sorbent barPDMS sorbent bar Injection System PTVCISPTV*CISCIS Instrumentation Q exactive GC orbitrap GC-MS/MS Laboratory DukeYaleYaleYaleYale Column Thermo Scientific TraceGOLD TG-5HTAgilent J&W HP-5 ms Ultra InertAgilent J&W HP-5 ms Ultra InertAgilent J&W HP-5 ms Ultra InertAgilent J&W HP-5 ms Ultra Inert Column specifications 30 m × 0.25 mm × 0.25 μm Stationary Phase 5% diphenyl/95% dimethyl-polysiloxane Max Column Temperature 400 °C350 °C350 °C350 °C350 °C Temperature Control nonlinearlinearnonlinearlinearnonlinear Modes EI, NCIEIEIEIEI m/z Range 70–105053.4–80070–105053.4–80070–1050aPTV: programmable temperature vaporizer; CIS: programmable temperature vaporizer; EI: electron ionization, NCI: negative chemical ionization. Overview of Five Analyses That were Conducted to Compare Environmental Chemicals Measured by Two Wearable Passive Samples PTV: programmable temperature vaporizer; CIS: programmable temperature vaporizer; EI: electron ionization, NCI: negative chemical ionization. The remaining three approaches were modifications of these standard methods, each reanalyzing the same silicone or Fresh Air samples to examine the effect of a specific analytical parameter on measured exposure. To examine the influence of the injection system and instrumentation, silicone wristband extracts were analyzed using a CIS inlet operated as a PTV under the same nonlinear temperature ramp as Approach 1 (Approach 3). To examine the effect of the GC temperature ramp, silicone extracts were analyzed using the linear ramp of the Fresh Air standard method (Approach 2), while Fresh Air wristbands were analyzed using the nonlinear ramp of the silicone standard method (Approach 5). Together, the linear (Approaches 2 and 4) and nonlinear (Approaches 1, 3, and 5) conditions allowed the effect of separation conditions to be evaluated across both samplers. Comparing all silicone approaches (Approaches 1–3) with all Fresh Air approaches (4–5), each spanning multiple analytical workflows, was further used to examine differences attributable to sampler design. Full extraction and instrument details for each approach are provided in the Supporting Information. Suspect Screening Analysis Data acquired from both wristbands were analyzed using all five approaches were processed using the same suspect screening workflow. − Compound Discoverer 3.2 (Thermo Fisher Scientific, Waltham, MA) was used to deconvolute spectra, annotate chemical features using the NIST 2017, Wiley 11th edition, and GC Orbitrap (Thermo Fisher Scientific) EI spectral libraries (over 740,000 unique chemical structures), and conduct peak alignment. All features were screened with a signal-to-noise threshold of 1.5, a total ion chromatogram (TIC) threshold of 100,000, an ion overlap window of 98%, and an accurate mass tolerance of 5 ppm. The retention time (RT) alignment window was set to 3 s. Relative abundances and possible identifications of compounds across all participants were then determined and exported. Using in-house scripts, annotations were designated as Level 2 (probable structure or close isomer), Level 3 (potential structure), or Level 5 (unknown, possibly with spectral matches) based on a composite score of retention index (RI) matching, the dot-product search index (SI) and reverse SI (RSI), the high-resolution mass filter (HRMF), and reverse HRMF (RHRMF). This scoring was based on community consensus-based scoring criteria that were previously presented. Blank feature filtering (BFF), duplicate removal, batch-wise median normalization, and TIC recalculation were then performed. In the BFF step, a feature was retained only if its maximum abundance across all samples exceeded twice a blank-derived threshold, defined as the mean blank abundance plus three times the standard deviation of the blanks (max­(samples) > 2 × [mean­(blanks) + 3 × SD­(blanks)]). Features that did not meet this criterion were removed in their entirety. Duplicate features were identified as those sharing the same chemical name or CAS Registry Number, which also removed coeluting isomers. Within each set of duplicates, the feature with the highest annotation confidence was retained (Level 2 > Level 3 > Level 5). When duplicates shared the same confidence level, the feature with the highest total score was retained, where the total score is the composite of the individual lines of spectral evidence computed in Thermo Compound Discoverer. Retention was therefore based on annotation confidence and total score, rather than on peak area. A manual review of spiked internal standards was performed to validate the method, in which each standard was evaluated across samples for mass accuracy (within 5 ppm), retention time drift (within 0.05 min), consistent detection (presence/absence), and reproducibility of peak area abundance, expressed as the relative standard deviation (RSD) across samples. Known target analytes were similarly confirmed. Biomonitoring in Urine Most participants (n = 28) started wearing the bands on a Sunday morning, typically around 9–10 am, and then removed the bands on Friday morning at around 9–10am. They were also asked to collect first morning-void urine samples on Monday, Wednesday, and Friday to measure metabolites of 6 organophosphate esters (OPEs), including bis­(2-chloro-isopropyl) hydroxy-isopropyl phosphate (BCIPHIPP), bis­(1-chloro-2-propyl) phosphate (BCPP), bis­(1,3-dichloro-2-propyl) phosphate (BDCPP), diphenyl phosphate (DPP), isopropylphenyl phenyl phosphate (ipPPP), and tert-butylphenyl phenyl phosphate (tbPPP). Urine samples were analyzed using previously published methods and urinary metabolite levels were adjusted for specific gravity. Detailed sampling procedures and sample analyses are provided in the Supporting Information. Statistical Analysis The comparisons used to attribute measured differences to sampler design, the injection system and instrumentation, and GC separation conditions are defined in the Analytical Approaches section (Table ). For targeted analyses, chemicals with more than a 50% detection frequency in samples were included in statistical analysis. The limit of detection (LOD) was defined as the average plus three times the standard deviation of field blank replicates. For measurements below the LOD, the values were replaced with LOD/2. Exposure concentrations were normalized by the days of wear, yielding daily mass-loading rates (ng/g PDMS/day). For suspect screening analyses of both wristband types, measurements were expressed as unitless peak areas and were not adjusted for wear duration. Meta-information was assembled for detected compounds, including chemical properties, product and function use, volatility, and chemical taxonomy. The US EPA CompTox Chemicals Dashboard was used to obtain chemical property information (vapor pressure, molar mass, predicted octanol-air partition coefficient or log K OA using the OPERA model) and product/use categories (e.g., pharmaceuticals, toys, cosmetics). The chemical taxonomy of compounds was classified by using ClassyFire. Wristband exposures were assigned a kingdom, superclass, class, and subclass. Spearman rank correlations were used to assess agreement between the two wristband designs for the overlapping analytes, the relationship between log K OA and the silicone/Fresh Air concentration ratio, and then association between paired wristband- and urine-based exposure estimates. Kruskal–Wallis and Wilcoxon rank sum tests were used to compare independent groups, including log K OA as well as squalene and cholesterol levels. For all analyses, significance was defined at α = 0.05 and p-values were adjusted for multiple comparisons using the Benjamini-Hochberg false discovery rate procedure. Statistical analyses were performed using R (version 4.2.1). Results The 30 study participants (80% female) ranged in age from 20 to 68 years (mean 36 years) (Table S1). Two participants reported exposure to cigarette smoke in their homes. Participants washed their hands between 1 and 10+ times per day, with half washing 7 to 9 times daily. Showering frequency ranged from 1 to 12 times per week, with most participants (n = 11, 37%) showering between 4 and 6 times weekly. Two-thirds of participants spent less than 12 h at home daily, and the same proportion spent 1 h or less in transit each day. Eighty percent of participants spent fewer than 12 h at school or work. Wristbands were worn on the left arm by 14 participants and on the right arm by 16 participants; dominant-hand information was not collected. Influence of the Wristband System on Detected ExposuresTargeted Analysis Silicone Wristbands Targeted analysis was performed on silicone wristband extracts for 111 compounds across four chemical classes (60 flame retardants and plasticizers, 10 phthalates/phthalate alternatives [PAEs], 25 polycyclic aromatic hydrocarbons [PAHs], and 16 pesticides; Approach 1) (Tables S2 and S3). A total of seventy-one chemicals were detected in at least one extract, 22 across all samples, and 43 were detected in at least 50% of participant samples (24 flame retardants/plasticizers, eight phthalates and PAEs, seven PAHs, and three pesticides) (Table S5). Concentrations spanned nearly 5 orders of magnitude across chemical classes, from PCB 11 (0.101 ng/g) to bis­(2-ethylhexyl) terephthalate (DEHT, 15,639 ng/g), with phthalates and phthalate alternatives measured at the highest levels overall. Compound-specific concentrations and detection frequencies for all four classes are detailed in the Supporting Information. Fresh Air Wristbands Targeted analysis screened 86 compounds (24 PAHs, 21 pesticides, 11 phthalates, 10 flame retardants, and 20 other compounds across various chemical classes; Approach 4) (Table S4). Ten analytes were not detected above the LODs after blank correction, and 37 chemicals were not detected in any participant samples. Thirty-nine chemicals were detected in at least one participant, 21 chemicals were found across all participants, and 33 in at least half of participant wristbands (11 phthalates, eight PAHs, three pesticides, two ultraviolet [UV] filters, two volatile organic compounds [VOCs], and seven other compounds) (Table S5). Concentrations spanned more than 4 orders of magnitude, from benzyl butyl phthalate (BBP, 0.404 ng/g) to homosalate (23,450 ng/g), with personal care product–related compounds measured at the highest levels overall. Additional information regarding concentrations and detection frequencies are available in the Supporting Information. Comparative Assessment of Silicone and Fresh Air Wristbands The targeted panels for both wristbands included 15 overlapping chemicals: six phthalates, two phosphate flame retardants (TPHP, TCPP), four low-molecular-weight (LMW) PAHs, and three high molecular weight (HMW) PAHs (Figure A,B and Table S6). Concentrations differed between the two designs for most of the 15 overlapping analytes, with silicone/Fresh Air ratios spanning from 0.01 (DMP) to 189 (BBP) (Table S6). Concentrations were most similar for pyrene, fluoranthene, and phenanthrene (ratios 0.8–1.5). Neither wristband consistently captured higher concentrations across all chemical classes. Comparison of 15 chemical exposures measured using targeted analysis of silicone and Fresh Air wristbands. Personal exposure concentrations of phthalates, polycyclic aromatic hydrocarbons (PAHs), and organophosphate esters (OPEs) for (A) silicone wristbands and Fresh Air wristbands, with mean concentrations indicated by diamond markers. (B) Smoothed trend illustrating the relationship between chemical partitioning properties (log K OA) and concentration ratios between wristband systems. The ratio of silicone to Fresh Air wristband measurements increases with increasing log K OA. Note DBP was quantified by the silicone wristband method while the sum of DBP+DiBP. Silicone wristbands had elevated mean levels for some phthalates, including BBP (189-fold higher) and bis­(2-ethylhexyl) phthalate (DEHP)+DEHT (33-fold higher) compared to Fresh Air wristbands. Conversely, Fresh Air wristbands captured higher mean concentrations of other phthalates, with dimethyl phthalate (DMP) being 80-fold higher and diethyl phthalate (DEP) being 27-fold higher than silicone wristbands. Phosphate flame retardants showed contrasting patterns between the two wristbands: TPHP concentrations were 114-fold higher in silicone wristbands, while TCPP levels were 9-fold higher in Fresh Air wristbands (Figure S1A,B). PAH measurements were generally higher in Fresh Air wristbands, with five PAHs showing elevated concentrations in this design: anthracene (LMW, 3.5-fold), acenaphthene (LMW, 3.1-fold), fluorene (LMW, 3.0-fold), chrysene (HMW, 2.6-fold), and pyrene (HMW, 1.3-fold). In contrast, only two PAHs were elevated in silicone wristbands: phenanthrene (LMW, 1.5-fold) and fluoranthene (HMW, 1.2-fold). The magnitude of concentration differences was strongly correlated with chemical partitioning properties, log K OA (rho = 0.81, p < 0.001) (Figure C). Agreement between the two designs varied by compound (Table S6). Correlations were significant for 7 of the 15 analytes after Benjamini-Hochberg correction, with the strongest agreement for DEP (ρ = 0.81), phenanthrene (0.74), and TCPP (0.73); agreement was weakest for compounds detected near the limit of detection in one design. Urine Comparisons We compared wristband exposure estimates with paired urinary metabolite concentrations (n = 30). TCPP measured by both wristbands was positively correlated with its urinary metabolite, BCIPHIPP, with comparable strengths for silicone (rho = 0.67; 95% CI, 0.42–0.83; p < 0.001) and Fresh Air wristbands (rho = 0.62; 95% CI, 0.32–0.81; p < 0.001). In silicone wristbands, TDCPP and 4tBPDPP were also correlated with their corresponding urinary metabolites BDCPP (rho = 0.60; 95% CI, 0.24–0.84; p < 0.001) and tbPPP (rho = 0.60; 95% CI, 0.26–0.80; p < 0.001), respectively; these two OPEs were not included in the Fresh Air panel. More information on urinary metabolite levels and correlations is presented in Supporting Information. Influence of the Wristband System on Detected ExposuresSuspect Screening Analysis Silicone Wristbands Analysis of silicone wristbands detected 2,226 chemical features; of these, 1330 compounds were annotated using suspect screening analysis across all confidence levels (Table S7). The detected compounds were predominantly in the 200 to 300 molecular weight (MW) range (41%), followed by 100 to 200 MW (39%), and 300 to 400 MW (8%) (Table S8). Among the annotated compounds, 179 were classified as Level 2 confidence, with 78% (n = 138) of these found in the EPA CompTox Database, and 75% (n = 135) of these with predicated Log K OA values (Figure A). Average peak area was highest in log K OA ranges 8–9, 10–11, and 11–12, whereas counts of annotated compound were highest in ranges 7–8, 8–9, and 6–7. Comparative assessment of suspect screening results from silicone and Fresh Air wristbands. (A) Average peak area for chemicals (top panel) and percentage of annotated compounds by count (bottom panel) distributed by log K OA for both wristband systems. (B) Taxonomic classification of Level 2-confidence annotations by superclass, displaying the number and percentage of compounds detected in silicone wristbands (top panel) and Fresh Air wristbands (bottom panel). Asterisks (*) denote superclasses unique to a specific wristband. Analysis of chemical diversity using ClassyFire revealed nine distinct structural classes among the 138 compounds that had available InChIKey identifiers. Benzenoids were the most abundant class, followed by lipids and lipid-like molecules and organoheterocyclic compounds (Figure B and Table S9). Among the benzenoids, benzene and substituted derivatives (n = 46) were the most prevalent, along with naphthalene (n = 7) and phenol ethers (n = 7). The lipids and lipid-like compounds were dominated by fatty acyls (n = 19) and prenol lipids (n = 10). Benzothiazoles (n = 3) were the most commonly detected organoheterocyclic compounds. Fresh Air Wristbands Fresh Air wristband samples yielded 5370 chemical features; of these, 2261 compounds were annotated through suspect screening analysis. Most detected chemicals had molecular weights between 100 and 200 (67%), followed by 50–100 (26%) and 200–300 (7%). Level 2 confidence annotations totaled 661 compounds, with 67% (n = 441) of these identified in the EPA CompTox Database, and 65% (n = 432) of these with Log K OA values (Figure A). Average peak area was highest in log K OA ranges 8–10, whereas counts of annotated compound were highest in ranges 4–6 and 7–8. Structural classes were assigned to 441 compounds that had available InChIKey identifiers into 12 distinct chemical classes. Benzenoids, organoheterocyclic compounds, and organic oxygen compounds emerged as dominant classes. Benzene and substituted derivatives (n = 100) comprised the largest subset of benzenoids, followed by phenols (n = 21) and naphthalene derivatives (n = 19). The organoheterocyclic class was characterized by pyridines and derivatives (n = 17), indoles and derivatives (n = 8), and diazines (n = 7), while organic oxides (n = 57) represented the majority of the organic oxygen class. Comparative Assessment of Silicone and Fresh Air Wristbands Fresh Air wristbands captured a greater number of features (1.7-fold more annotated compounds) than silicone wristbands (Table S7). The difference was most pronounced for level 2 annotations, with Fresh Air wristbands showing 3.7-fold more level 2 annotations and 280-fold more level 3 annotations than silicone wristbands. Conversely, silicone wristbands had 1.5-fold more level 5 confidence annotations. The two wristband types captured chemicals with distinct distributions across log K OA. Silicone wristbands showed higher average peak areas for compounds with log K OA > 8 than for those with log K OA < 8 (p < 0.05), while Fresh Air wristbands showed no such differences. Differences were also observed in the counts of annotated compounds: 80% of annotated compounds detected in Fresh Air wristbands had log K OA values below 8. In contrast, in silicone wristbands, approximately half (53%) of the annotated compounds had log K OA values below 8 and half were above. Approximately 93% of annotated chemical features in Fresh Air wristbands had molecular weights between 50 and 200, whereas 80% of features in silicone wristbands ranged between 100 and 300 (Figure ). Chemical profiles across five analytical approaches comparing silicone and Fresh Air wristbands. Average peak area (left axis, bars) and count of detected chemicals (right axis, blue line) for annotated chemical features grouped by molecular weight range. Silicone wristbands were analyzed using Approaches 1–3, and Fresh Air wristbands were analyzed using Approaches 4–5. Approaches include wristband-specific standard analysis (Approaches 1 and 4) and alternative methods to assess hardware and analytical influences (Approaches 2, 3, and 5). Both wristbands detected chemicals from eight major structural classes: benzenoids, lipids and lipid-like compounds, organic acids and derivatives, organic oxygen compounds, organoheterocyclic compounds, phenylpropanoids and polyketides, hydrocarbons, and organohalogens. Fresh Air wristbands captured additional unique compound classes, including organic nitrogen compounds (n = 10), organosulfur compounds (n = 5), one alkaloid, and one organometallic compound. Silicone wristbands uniquely detected one homogeneous nonmetal compound. Influence of Instrument Hardware and Analytical Methods on Exposure Detection To evaluate how different analytical approaches impacted chemical detection and quantification, we assessed all compounds annotated through suspect screening regardless of confidence level across the five approaches using molecular weight distributions and peak area measurements. These approaches differ in several parameters simultaneously, including laboratory, analytical column, injection system, GC temperature ramp, and acquisition m/z range, and therefore, they do not constitute a fully controlled factorial design. Each comparison below reflected the combined effect of the parameters that differ between approaches rather than the isolated effect of a single parameter. Assessment of the types of chemicals detected and their measured abundances across methods. MW distributions and peak area measurements were used to quantify the methodological differences. Separation Conditions GC temperature ramp programming was associated with differences in both the detection and abundance of chemical features across MW ranges (Figure and Table S10). Linear temperature ramps with slow and steady ramp rates (Approaches 2 and 4) were associated with greater detection and higher peak areas for lower MW compounds (MW ≤ 200), while nonlinear ramps (Approaches 1, 3, and 5) were associated with increased detection of higher MW compounds (MW > 200). This pattern was also observed in the matched pairwise contrasts that differ primarily in the temperature ramp, Approach 4 versus 5 (Fresh Air) and Approach 2 versus 3 (silicone). The temperature ramp covaried with the acquisition m/z range, as the linear approaches acquired m/z 53.4–800 and the nonlinear approaches acquired m/z 70–1050 (Table ); the wider high-mass range may contribute to the greater detection of higher MW features under nonlinear conditions. No compounds above 500 MW were detected under linear ramp conditions in either approach, whereas nonlinear ramps additionally detected compounds above 400 MW (14, 2, and 12 compounds in Approaches 1, 3, and 5, respectively) and showed elevated signal intensity in the mid-to-high molecular weight range (Tables S8, S11, and S12). Instrumentation Injection systems, columns, and laboratory differences were also associated with differences in chemical detection and quantification. To examine these effects, silicone wristbands were analyzed using a nonlinear temperature ramp that was compared between Approach 1 (programmable temperature vaporizer, PTV; TG-5HT column) and Approach 3 (cooled injection system, CIS, programmed as PTV; HP-5 ms column). Although both used the same instrument model and nonlinear ramp, they differed in injection hardware, analytical column, and laboratory, so the observed differences reflect the combined contribution of these factors rather than injection hardware alone. Peak area variability ranged from 0.06- to 20-fold between the two approaches across seven MW groups, and chemical detections varied by 6- to 12-fold across MW groups (Table S11). Wristband Type Sampler design influenced the chemical classes detected, and this pattern was consistent across analytical workflows. Benzenoids were the most frequently detected class across both wristbands, but the second most frequently detected classes differed substantially. Silicone wristbands (Approaches 1, 2, and 3) consistently showed dominance of benzenoids and lipid-like molecules across all analytical methods. Benzenoids represented 32–43% of detected compounds, while lipids and lipid-like molecules accounted for 12–17%. Together, these two classes comprised 47–55% of all detections in silicone wristbands. In contrast, Fresh Air wristbands (Approaches 4 and 5) revealed more diverse chemical profiles, with benzenoids comprising a smaller proportion of total detections (25% in Approach 4, 18% in Approach 5) and greater representation of organic acids and organoheterocyclic compounds. Effectiveness of Internal Standards in Reducing Measurement Variability Internal calibration using spiked standards is designed to account for systematic differences caused by extraction methods, instrument settings, and analytical approaches. To evaluate this approach, we assessed the effectiveness of using three deuterated spiked internal standards, acenaphthene-d10 (MW = 164.14), phenanthrene-d10 (MW = 188.14), and pyrene-d10 (MW = 212.14), in reducing variations for four corresponding PAHs: acenaphthene (MW = 154.07), fluorene (MW = 166.07), phenanthrene (MW = 178.07), and pyrene (MW= 202.07) (Figure S2). All seven compounds were part of the targeted analyte panel for both wristbands and were manually validated against the analytical standards. To assess variability arising from analytical conditions while holding the wristband constant, we compared approaches 1, 2, and 3, which all analyzed identical silicone wristband extracts. Even under these controlled conditions, the internal standards themselves varied substantially in peak area across the three approaches, with coefficients of variation (CVs) of 120, 75, and 102% for acenaphthene-d10, phenanthrene-d10, and pyrene-d10, respectively (Table S13). PAH exposure measurements similarly showed high variability in peak areas across approaches before internal standard correction (CV 74–91%) (Table S14). Applying internal standard corrections did not consistently reduce this variability: CVs decreased for phenanthrene (37%) and pyrene (34%) but increased for acenaphthene (111%) and fluorene (158%) (Table S15). These findings indicate that internal standards across laboratories without proper optimization provide limited improvement in measurement consistency across analytical approaches. Discussion This study provides the first systematic comparison of wearable passive samplers and analytical methods for the personal exposure assessment. By evaluating silicone wristbands and Fresh Air wristbands deployed concurrently with 30 adults, we demonstrated that both designs and the analytical workflow affect the diversity and magnitude of measured exposures. These findings have implications, particularly for large-scale and longitudinal studies, where analytical consistency and cross-study comparability are essential. Wristband System Influences Exposure Profiles The silicone and Fresh Air wristbands differ as complete sampler-workflow systems, and the standard comparison cannot fully separate sampler design from the accompanying differences in analytical workflow. Comparisons across the experimental approaches, together with the relationship between the differences and chemical partitioning, however, suggest that sampler design and the associated exposure route were a dominant driver of these differences. When each sampler was compared across multiple analytical workflows (Approaches 1–3 for silicone and 4–5 for Fresh Air), the differences that persisted were more likely attributable to sampler design than to any specific workflow. Sampler design determines the routes through which chemicals are collected. Fresh Air wristbands sample exclusively from air through diffusion and deposition, while silicone wristbands collect chemicals from both air and surfaces (including skin) through diffusion, deposition, and direct contact. These differences are driven by how compounds partition between air and surface reservoirs, as reflected by their log K OA. Both targeted and suspect screening analyses demonstrated this partitioning-driven sampling selectivity. Among the 15 chemicals quantified through targeted analysis, concentration differences between wristband types increased with log K OA. Suspect screening results confirmed this trend across 135 annotated compounds detected in silicone wristbands and 432 detected in Fresh Air wristbands. Fresh Air wristbands predominantly captured compounds with log K OA below 9, while silicone wristbands detected compounds across a broader range extending up to 12. These distributions are consistent with previous indoor measurements of SVOCs collected simultaneously in air and dust. , Dodson et al. measured 76 SVOCs in air and dust in 49 California homes and used a mixed-effects model to investigate the relationship of gas-phase air to dust concentrations. Compounds with a log K OA between 5 to 7 had higher relative concentrations in air than in dust, while those with log K OA values between 10 to 12 had higher relative concentrations in dust. The equilibrium model established by Weschler and Nazaroff for predicting the partitioning of SVOCs between the gas phase and settled dust provides support for our observed relationship between log K OA and wristband-specific chemical distributions. The partitioning-driven sampling behavior observed in our study was consistent across three chemical classes: PAEs, organophosphate esters, and PAHs. Among PAEs, silicone wristbands captured higher levels of DEHP+DEHT, diisononyl phthalate (DINP), and BBP (log K OA 9.8 to 13), while Fresh Air wristbands collected elevated concentrations of DMP, DEP, and DBP (log K OA 5.7 to 8.8). Past studies have aligned with these patterns. Blanchard et al. measured 57 SVOCs in 30 French homes and found that DEHP and DINP were the most abundant in settled dust, while DEHP, DBP, and DINP had the highest concentration in air. Measurements from homes across China, , USA, Canada, and Germany revealed similar partitioning trends, where DMP, DEP, DBP, and di-isobutyl phthalate (DiBP) maintained predominantly presence in air; while DEHP, DBP, BBP, and DINP were consistently enriched in dust. Organophosphate esters exhibited similar partitioning-dependent patterns in our study. Silicone wristbands were found to have higher TPhP concentrations (log K OA ≈11), while Fresh Air wristbands had elevated TCPP (log K OA ≈8.8). Published measurements across homes and offices in Canada, the Czech Republic, the United States, and China reported that TPHP was predominantly found in settled dust, and TCPP was prevalent in the gas phase. , Davis et al. further demonstrated that TPHP was most abundant in dermal patches, followed by settled dust, and was minimally detected in airborne gas phase. These findings reflect the stronger affinity of TPHP for surface reservoirs. PAHs in our study also followed expected partitioning-driven trends. Silicone wristbands had higher phenanthrene and fluoranthene concentrations (log K OA 7 to 9), while Fresh Air wristbands measured higher concentrations of fluorene, acenaphthene, and anthracene (log K OA 6 to 8). Studies that simultaneously assessed PAHs in air and settled dust in Kuwait and California homes, similarly reported that lower-molecular-weight PAHs (anthracene, acenaphthene, fluorene) remain predominantly in the gas phase, while higher-molecular-weight PAHs (phenanthrene, fluoranthene, pyrene, chrysene) partitioned between air and dust depending on temperature and indoor conditions. These patterns align with our observations, except for fluoranthene (log K OA ≈8.8) and phenanthrene (log K OA ≈7.6), whose elevated concentrations in silicone wristbands suggest additional uptake through surface contact. Beyond environmental contaminants, silicone wristbands uniquely detected skin-derived compounds. Suspect screening analysis identified more lipids and lipid-like molecules in silicone wristbands compared to Fresh Air wristbands, including cholesterol and squalene, which were absent in Fresh Air wristbands. These endogenous biomarkers further demonstrate that silicone wristbands capture surface-bound chemicals through physical contact, a sampling mechanism unavailable to air-only wristband samplers. Despite these differences in collection routes, wristband types were positively correlated across participants, demonstrating that both designs captured consistent interindividual variability. For TCPP, wristband-based exposure estimates from both designs showed comparable correlation with the corresponding urinary metabolite measurements, indicating correspondence with the internal dose for this compound. This agreement was shown for specific organophosphate esters and may not be extended to other chemical classes. Both wristband systems are suitable for relative exposure comparisons within studies provided consistent participant wear and deployment protocols are maintained. Instrumentation and Analytical Methods Impact Exposure Profiles Beyond the sampler design, analytical approaches emerged as critical determinants of exposure variability. Our evaluation revealed that analytical parameters and instrument settings, including injection inlets and GC temperature ramp programming, affected both the range and abundance of detected chemicals even when internal standards were used to control for systematic differences. The peak area variability and differences in chemical detection demonstrate that without standardization, chemical intensity measurements cannot be directly compared across laboratories and analytical methods. GC temperature ramp programming had the most pronounced impact on the chemical detections. In this study, GC temperature ramp comparisons focused on two approaches: a linear method with a slow ramp rate and a nonlinear ramp method with stepwise, faster ramp rates. Linear temperature ramps enhanced detection of low-molecular-weight compounds, while nonlinear ramps enhanced higher-molecular-weight compounds. The observed patterns are attributable to differences in method optimization, as temperature programming can be tailored to enhance sensitivity for compounds across any molecular weight range. Additionally, hardware differences between laboratories, including injection systems and instrumentation configuration, likely introduce substantial variations in measurements, even when identical sample solutions were analyzed under controlled conditions. Most interlaboratory studies have acknowledged the impact of hardware on measurement variability and have attributed unexplained measurement variations to these factors. ,, However, few studies have investigated the specific hardware parameters (e.g., voltage settings for tuning particular elements) that may drive such variations. To address variability introduced by different analytical approaches, targeted methods with internal standards are often applied. In our data, internal standards showed limited effectiveness in mitigating measurement differences when analytical methods were adopted without a specific optimization. Even when the same silicone wristband extracts were analyzed across approaches, the internal standards themselves exhibited variability (75–120% CVs), and internal standard correction reduced variability for some PAHs while increasing CVs for others, with the lowest resulting CV remaining above 30%. These findings challenge the assumption that internal standards alone, without method optimization, can adequately control for differences across analytical workflows and laboratories. Study Limitations Several limitations should be considered in interpreting these findings. First, the study was conducted in a relatively small cohort (n = 30) recruited from a single geographic location (Durham County, North Carolina) during a single season (late winter to early spring), which may limit the generalizability of the findings to other populations, regions, and times of year. Second, wear instructions differed between the two wristband designs: silicone wristbands were worn continuously, whereas Fresh Air wristbands were removed during water immersion activities and could be set aside during sleep. These differences in wear conditions, together with the absence of information on participants’ dominant hand, may have contributed to differences in measured exposures. Third, the two designs were paired with different analytical workflows and targeted analyte panels. The customized targeted panels limited direct comparison to 15 chemicals between wristband types, highlighting the need for unified analytical approaches in future studies. The analytical approaches also differed in multiple parameters simultaneously, so we could not isolate which specific methodological components contribute most to measurement variability, and further research is needed to disentangle these factors. Fourth, we did not evaluate potential differences in uptake rates between the two wristband designs, which may contribute to the observed differences in the measured concentrations. The absence of codeployed active air, dust, and hand-wipe comparators further limited attribution of measured exposures to specific environmental reservoirs. Finally, to evaluate the influence of hardware and analytical workflows, the silicone wristband extract solution was analyzed at Yale University nearly one year after extraction at Duke University. Analyte loss may therefore have occurred during storage or transport and may have contributed to the observed differences among measurements obtained by using different analytical approaches. Future Directions This study demonstrates that both wristband design and analytical methodology influence measured exposures, underscoring the need for harmonized protocols throughout the entire exposure assessment workflow, both within and across laboratories, to ensure reproducible and comparable measurements. The observed variability has several implications. First, the two samplers capture complementary aspects of personal exposure, and the sampler selection should reflect this. The silicone wristband uniquely captures a dermal signature, including skin-derived compounds such as cholesterol and squalene and chemicals from direct contact with skin, surfaces, and dermally applied consumer products, along with less volatile, higher log K OA compounds that partition to surfaces and dust. The Fresh Air wristband samples airborne compounds and captured a broad breadth of chemicals overall, including volatilized consumer products and more volatile, lower log K OA compounds present in the air. Rather than one sampler being uniformly more informative, each provides a distinct window into personal exposure. Second, comprehensive method documentation, including instrumentation configuration, analytical procedures, and data-processing steps, is necessary to enable cross-study comparability and the investigation of exposure patterns across diverse populations, geographies, and time periods. Our findings suggest several directions for improving the comparability across wearable passive-sampler studies. The development of unified minimum targeted chemical panels would enable broader comparisons across sampler designs and studies. Harmonized data analysis workflows, including common spectral libraries and annotation criteria, together with consistent reporting of method documentation, quality control, and performance metrics, would reduce measurement variability and improve reproducibility across laboratories. Publicly accessible data repositories with comprehensive metadata would further support the cross-study data integration. Supplementary Material