In the initial phase of our experiment, we evaluated seven SUPRAS, formed from either hexanol or decanol as the key amphiphile components (Table 1) to simultaneously extract a diverse array of environmental contaminants.

SUPRAS 6 and 7 were synthesized using decanol, as different chain alcohols provide varying hydrophilic/lipophilic balance (HLB) and influence the SUPRAS composition and nanostructures, thus tuning the SUPRAS for different extraction properties. Both short- and long-chain alcohols can achieve good extraction efficiencies for polar and nonpolar compounds. Additionally, the ratio of alcohol to THF to water during SUPRAS formation plays a crucial role. Therefore, it is the complex equilibrium between these tuning factors that affects the extraction efficiency for compounds with different polarities [11, 16]. The formation of SUPRAS was successfully achieved, showing the inherent self-assembly properties of decanol-based amphiphiles. However, a significant challenge arose during the subsequent extraction process, when attempting to evaporate the decanol-based SUPRAS extracts using the N2 evaporation technique. Despite prolonged drying over several days, the evaporation of decanol-based SUPRAS extracts remained elusive under N2 flow. A vacuum evaporation system, such as a rotary evaporator, may offer a more effective approach; however, these decanol-based SUPRAS were not further tested in our method development.

SUPRAS 5, comprising Milli-Q, THF, and hexanol in a 30:60:10 ratio, respectively (Table 1), did not form the SUPRAS liquid phase. This is because at high concentrations of THF, the concentration of the coacervating agent (water) is too low to facilitate self-assembly and coacervation. Under these conditions, the concentration of water falls below the critical aggregation concentration (CAC) required for the system to form SUPRAS. As a result, the system does not form SUPRAS but rather a mixture of hexanol monomers or micelles diluted in THF:water [8, 11, 17]. Nevertheless, we proceeded to employ SUPRAS 5 for the extraction process. However, the average recovery (%) based on the recovery of surrogate standards representative for each group of contaminants was the lowest for many of the contaminant groups analysed (PAHs, O-PAHs, musks, PHTHs, and CUPs; Table 2, Table S5). We did not consider SUPRAS 5 further as the non-formation of SUPRAS was a large uncertainty and drawback.

The formation of the desired SUPRAS liquid phase in SUPRAS 1, 2, 3, and 4 was achieved and was therefore used for extraction. Among these, SUPRAS 2 displayed bigger variability between replicates and the broadest range of recoveries across different groups of contaminants. SUPRAS 1, 3, and 4 had the most consistent recoveries between replicates and among different contaminant groups (Fig. 2, Table S5). The best extraction performance was achieved using SUPRAS 1 and 4, where the average recoveries were above 50% across most of the contaminant groups, except for CUPs and PFAS, where recoveries failed to surpass 50% with any of the SUPRAS (Table 2). SUPRAS 1 and 4 yielded average recoveries of 53.3% and 56.2% for PAHs, 69.2% and 84.7% for nitro-PAHs, 117% and 82.6% for oxy-PAHs, 70.8% and 68.7% for musks, 72.8% and 62.7% for FRs, 83.8% and 74.2% for PCBs, 71.2% and 60.8% for OCPs, 81.3% and 68.2% for PHTHs, 39.6% and 40.9% for CUPs, and finally 45.5% and 45.4% for PFAS, respectively. Given that the bulk solution used to form SUPRAS for extraction of organic contaminants typically contains around 70% water [8, 10, 15], we selected SUPRAS 1 for further testing. We also estimated matrix effects from phase 1 analysis for SUPRAS 1 (Table S6).

Extraction efficiency of SUPRAS 1, 2, 3, and 4 in phase 1 tests. The green shaded area indicates acceptable average recoveries of 50 to 120% for each contaminant group

In the second phase, we broadened our assessment to include bisphenols, parabens, BADGE/BFDGE, HBCDs, OPFRs, dechloranes, and CPs in SUPRAS 1.

Among the bisphenols tested, we assessed the commonly used BPA, BPS, and BPF, with average recoveries of the surrogate standards of 29.8%, 45.9%, and 43.8%, respectively (Fig. 3, Table S7). Other bisphenols, such as bisphenol B, 2,4-bisphenol S, 2,2-bisphenol F, and bisphenol AF, exhibited average surrogate standard recoveries of 28.2%, 66.9%, 35.1%, and 23.0%, respectively. Recoveries of 31.9% and 46.9% were also achieved for BADGE and BFDGE, respectively. Four parabens—butyl, methyl, ethyl, and propyl—showed notably low recoveries, indicating less efficient extraction using SUPRAS 1 (Fig. 3, Table S7). Attempts to analyse HBCDs did not yield any data, marking the sole contaminant group where the SUPRAS method appeared ineffective.

Extraction efficiency of surrogate standards with SUPRAS 1 for A PFAS; B PHTHs and CPs; C bisphenols, BADGE, and parabens; D PBDEs, NFRs, and OPFRs; E CUPs; and F PAHs and substituted PAHs. The bars represent average of three replicates, the error bars show standard deviation, and the dashed lines indicate a recovery of 100%

Other flame retardants, including PBDEs, NFRs, and OPFRs, demonstrated good recoveries, ranging from 72.8 to 128% for PBDEs, from 44.2 to 102% for NFRs, and from 22.0 to 123% for OPFRs. Triethyl phosphate had the lowest recovery (22.0%) of the group, attributed to evaporation during extraction due to its higher volatility. A similar trend was observed for PAHs. Volatile PAHs like naphthalene evaporated entirely, while less volatile ones with higher molecular weights showed consistently better recoveries, averaging between 33.1 and 60.6% (Fig. 3, Table S7).

Four PHTHs—DEP, DiBP, DHxP, and DEHP—displayed recoveries of 126%, 40.0%, 48.5%, and 94.1%, respectively. Additionally, the chlorinated paraffins exhibited an average recovery of 61.0% (Fig. 3, Table S7). The results indicate that the extraction of these contaminants with SUPRAS is efficient.

For fifteen PFAS, average recoveries for fourteen ranged from 39.7 to 166%. However, FOSA displayed a very low recovery (7.86%), attributed to its volatile nature and evaporation during extraction, impacting detection and recovery (Fig. 3, Table S7).

Finally, CUPs had consistently low recoveries, ranging from 4.62 to 22.1% (Fig. 3, Table S7). The pesticides that have > 50% detection frequencies in European outdoor air [53] and were included in our study—atrazine, chlorpyrifos, metazachlor, metolachlor, tebuconazole, and terbuthylazine—all demonstrated very low recoveries, varying from 4.62% for chlorpyrifos to 21.7% for metazachlor (Fig. 3, Table S7). Previously, Peyrovi and Hadjmohammadi [54] found that the recoveries of selected CUPs, such as chlorpyrifos, were dependent on the chain length of the alkanol used (undecanol) and pH, with the most optimal being below or equal to pKa of their targeted pesticides. This may also explain the low recoveries in our study.

The SUPRAS-based extraction procedure developed in this study was found to be suitable for several groups of contaminants, notably flame retardants (PBDEs, NFRs, and OPFRs), certain bisphenols, PHTHs, PFAS, and chlorinated paraffins. However, it showed limitations with the extraction of HBCDs, and volatile compounds like triethyl phosphate, some PAHs, parabens, and FOSA, as well as very polar compounds, for example, selected CUPs. The low recoveries of volatile compounds are a limitation in the applicability; however, these compounds are typically of lesser importance in settled dust samples, which are dominated by less volatile compounds. Other sample handling strategies that could reduce losses during volume reduction steps may further expand the applicability of the proposed method to more volatile compounds as well. Low recoveries of very polar compounds such as CUPs could be improved by adding a salting out agent (e.g., using water with Na2SO4 instead of only water) to ensure the compounds are not extracted in the equilibrium solution. Alternatively, acidifying the water before SUPRAS formation below the pKa could prevent losses of acidic compounds. Furthermore, testing other SUPRAS, which have proven efficient in the simultaneous extraction of polar and nonpolar compounds, such as diol [3]- or acid [55]-based SUPRAS should be considered. Finally, increasing the ratio of SUPRAS to dust could enhance the recoveries.

In the final step, we compared results from SRM 2585 extracted with SUPRAS with the NIST SRM 2585 certified values where applicable, or with literature values when NIST certified values were not available (Fig. 4). Additionally, we also extracted SRM 2585 with the conventional methods, using hex:acet extraction for the non-polar contaminants and MeOH extraction for the polar contaminants. The results for comparison of NIST SRM 2585 certified values with those obtained by SUPRAS extraction and “conventional” hex:acet and MeOH extractions are presented in Table S8 and Figure S1.

Concentrations of selected contaminants extracted with SUPRAS compared to the NIST certified/literature concentrations in SRM 2585. The z-scores show deviation between the observed values and the certified or literature values, considering a 25% variation around those certified/literature values. The x-axis shows the vapour pressure (Pa) of the contaminants on a logarithmic scale

SUPRAS reproduced PAH certified values extremely well for 17 of 20 PAHs (PHEN, FLTH, PYR, B[a]A, CHRY, B[b]F, B[k]F, B[a]P, IND, D[ah]A, B[ghi]P, B[ghi]F, TRI, B[j]F, B[e]P, PER, D[a,c]A), in many cases better than the commonly used hex:acet extraction (Table S8). The most volatile PAH, naphthalene, was absent, due to evaporation during extraction. Anthracene and coronene concentrations were two times higher in SUPRAS compared to the NIST-certified values, however comparable to the hex:acet extraction values. The concentrations of synthetic musks, galaxolide (HHCB), and tonalide (AHTN) extracted with SUPRAS were 1200 ng/g and 1040 ng/g, respectively. These values were within the range of NIST-certified values, with concentrations of 1470 ng/g for galaxolide and 1700 ng/g for tonalide (Fig. 4).

For the PCBs and OCPs, the findings indicated overall consistency between SUPRAS, hex:acet, and NIST-certified values for all PCBs and most OCPs. Concentrations of p,p′-DDT were half of the NIST-certified value (111 ng/g) in SUPRAS (59.9 ng/g) but were also substantially underreported in the hex:acet extraction (35.3 ng/g). SUPRAS extraction was not effective for PeCB, deviating by 680% deviation from the NIST-certified value, attributed to a very low recovery of PeCB surrogate, due to the volatility of the compound, resulting in misleading concentration calculation.

Out of the 12 PFAS analysed, eight PFAS (PBFA, PFHpA, PFNA, PFDoA, PFHxS, PFTriA, PFOS, and PFUnA) were within the 25% accepted variation of the literature values. PFHxA (429 ng/g), PFDA (67.2 ng/g), and PFOA (928 ng/g) were all approximately double the concentration in SUPRAS, although maintaining the same order of magnitude with the NIST certified/literature values (260 ng/g PFHxA, 38.1 ng/g PFDA, and 567 ng/g PFOA). In contrast, PFBS concentration was lower in SUPRAS (18.0 ng/g) compared to the literature value (40.8 ng/g) but again, remained within the same order of magnitude (Fig. 4, Table S8). The recoveries of PFAS surrogate standards were often below 50%, however, very consistent, which allowed for accurate quantification of PFAS in SRM 2585.

Most PBDEs (congeners 28, 47, 85, 99, 100, 153, and 154) largely aligned with previous data, although BDE-183 was two times higher in SUPRAS (106 ng/g) compared to the NIST-certified value (43.0 ng/g). Further clarification is necessary for BDE-209, which was not detected in SUPRAS-extracted SRM 2585 while the NIST SRM 2585 certified value is 2510 ng/g. With the hex:acet extraction, we observed a concentration of PBDE 209 of 2940 ng/g, which suggests that hex:acet extraction is a more optimal method (Table S8). We suspect that the large molecular size and low solubility of PBDE 209 could contribute to inefficient extraction with SUPRAS. The highly brominated PBDE 209 structure might hinder its interaction with the surfactants involved in the SUPRAS extraction, as was noted for HBCD.

Five PHTHs (DMP, DEP, DnBP, BBzP, and DEHP) demonstrated good consistency with literature values (Fig. 4). DiBP was two times higher in SUPRAS (11.9 ng/g) compared to the literature value (6.5 ng/g), although within the same order of magnitude. DiNP, however, was three times lower in SUPRAS, with a concentration of 66.6 ng/g, compared to the concentration of 199 ng/g in the literature (Table S8). However, we note that SRM 2585 is not certified for PHTHs; therefore, some greater uncertainty is expected.

Four NFRs, HBB, PBEB, EH-TBB, and syn-DP, extracted with SUPRAS developed in this study exhibited concentrations within the 25% accepted variation of the literature values [45, 51]. BEH-TEBP and PBBZ were two times higher, while anti-DP was two times lower in SUPRAS compared with the literature values, however maintaining the same order of magnitude. Only TBP-AE showed a significant discrepancy between SUPRAS (0.308 ng/g) and the literature value (6.0 ng/g) (Fig. 4, Table S8). As with PHTHs, this may be, in part, due to a lack of certified values.

To our knowledge, there is no SRM 2585 data available for substituted PAHs and CUPs in the NIST certificate or existing literature. Consequently, we only conducted a comparative assessment for substituted PAHs and CUPs between the SUPRAS extracts and hex:acet (substituted PAHs) and MeOH (CUPs) extracts.

Of the 35 CUPs analysed, only 11 (azinphos-methyl, carbaryl, diazinon, dimetachlor, chlorpyrifos, chlorsulfuron, isoproturon, pendimethalin, prochloraz, propiconazole, and tebuconazole) were detected in MeOH extracts, while only five (carbaryl, diazinon, chlorpyrifos, pendimethalin, and tebuconazole) were detected in SUPRAS extracts. The concentration of carbaryl was 1760 ng/g and 413 ng/g, diazinon was 282 ng/g and 226 ng/g, chlorpyrifos was 538 ng/g and 504 ng/g, pendimethalin was 21.7 ng/g and 45.8 ng/g, and tebuconazole was 0.584 ng/g and 2.17 ng/g for SUPRAS and MeOH, respectively. Only two compounds (diazinon and chlorpyrifos) were in agreement. Due to large variability between the CUP data and generally low recoveries (< 30%) of the surrogate standards, SUPRAS-based extraction using the current method is not recommended for CUPs.

Eighteen N-PAHs and nine O-PAHs were analysed, of which three N-PAHs and eight O-PAHs were detected in SUPRAS extracts and five N-PAHs and five O-PAHs were detected in hex:acet extracts. N-PAHs varied greatly between SUPRAS and hex:acet extracts, with only 7-nitrobenz[a]anthracene having comparable concentrations of 38.5 ng/g and 30.0 ng/g for SUPRAS and hex:acet, respectively. All O-PAHs except 1-naphthaldehyde were detected in SUPRAS extracts; however, they all had twofold higher concentrations compared to the hex:acet extracts. The concentration of 1,4-naphthoquinone in the SUPRAS extract was 122 µg/g, which seems highly implausible and can be attributed to a very low recovery (13%) of the surrogate standard.

The evaluation of SUPRAS for a wide range of compound groups, including FRs, PHTHs, PFAS, CPs, bisphenols, and others, has shown promising results. However, it is crucial to acknowledge and address specific limitations that have surfaced during this assessment.

While the SUPRAS selected in this study proved effective for many compound classes, challenges arise when dealing with volatile compounds, such as naphthalene, TEP, TBP-AE, FOSA, and PeCB. For instance, deuterated naphthalene and 13C-PeCB exhibited peak heights less than 10 times the noise level, leading to inaccuracies in the quantification of native naphthalene and PeCB. This observation confirms that the most volatile compounds are absent due to the evaporation step. In terms of additional method optimization, we recommend avoiding full drying of the extracts, but instead employing a solvent exchange method or to add a small volume of keeper, such as nonane, during the evaporation step. The evaporation of SUPRAS before LC–MS analysis may be unnecessary, as the extract is generally compatible with this technique. However, there are challenges related to implementing novel methods in routine instrumental analysis and concerns regarding the impacts of SUPRAS on instrumentation. While evaporation to dryness is typically unnecessary, if required for consistency with existing instrument practices, calibrants can be added to maintain accuracy and consistency. Calibrants can account for potential retention time shifts and variations in ionization, which can affect MS signals. When considering the loss of compounds due to evaporation, and if they are compatible with LC–MS, this approach should be noted as a potential solution. We employed sonication instead of the typically used vortex stirring during the extraction process. This alteration may have affected the extraction efficiency of the contaminants into the SUPRAS, potentially contributing to the lower recoveries observed. Additionally, the method faces difficulties with large, non-polar brominated compounds like HBCD and PBDE 209, as well as very polar compounds like CUPs. These findings highlight the need for further method optimization, or even a selection of different SUPRAS when these specific compounds/compound classes are of particular interest. The area of SUPRAS is developing rapidly, and there are multiple different SUPRAS, which have been shown to be efficient for the extraction of organic contaminants. These include for example cubosomic [56], vesicular [57], or magnetic [58] SUPRAS and should be tested in future studies.

When aiming for a wide-scope target screening method, some compromises need to be made as compared to typical targeted methods focusing on a single group of compounds. It is not possible to clean up the samples in a way that they meet the demands of the single specific targeted methods, because this will affect the detection of other groups of substances. The absence of a cleanup step in the extraction process can become an issue for the lifespan of GC components. This can lead to retention time shifts and suboptimal chromatography after a relatively small number of extracts. In addition, the lack of cleanup steps may contribute to challenges with matrix-related effects, requiring extra attention and caution during GC and LC–MS analyses. The matrix effects in this study showed high variability, with substantial matrix suppression for nitro-PAHs and oxy-PAHs and some NFRs, PCBs, and OCPs indicating impacts on analyte detection. Conversely, some compounds, such as many CUPs and PFAS, exhibited matrix enhancement (Table S6). This highlights the importance of considering matrix effects in analytical methods to ensure accurate quantification and reliable results and adjustments or calibrations might be necessary to account for these effects. For instance, due to the lack of a cleanup step in SUPRAS extracts in the quantification of DDT, the GC inlet and column became rapidly contaminated, resulting in a degradation of the 13C-p,p′-DDT, hindering accurate DDT quantification. While some matrix issues can be addressed through extract dilution or split injection, this approach may lead to higher limits of quantification and potential data loss due to lower concentrations of compounds of interest. Incorporation of cleanup steps or GC liner selection could enhance the instrument’s performance and mitigate these challenges.