A Critical Review of Environmentally Persistent Free Radical (EPFR) Solvent Extraction Methodology and Retrieval Efficiency

Abstract

Long-lived environmentally persistent free radical (EPFR) exposures have been shown in toxicology studies to lead to respiratory and cardiovascular effects, which were thought to be due to the persistence of EPFR and their ability to produce reactive oxygen species. To characterize EPFR exposure and resulting health impacts, it is necessary to identify and systematize analysis protocols. Both direct measurement and solvent extraction methods have been applied to analyze environmental samples containing EPFR. The use of different protocols and solvents in EPFR analyses makes it difficult to compare results among studies. In this work, we reviewed EPFR studies that involved solvent extraction and carefully reported the details of the extraction methodology and retrieval recovery. EPFR recovery depends on the structure of the radical species and the solvent. For the limited number of studies available for review, the polar solvents had superior recovery in more studies. Radicals appeared to be more oxygen-centered following extraction for fly ash and PM samples. Different solvent extraction methods to retrieve EPFR may produce molecular products during the extraction, thus potentially changing the sample toxicity. The number of studies reporting detailed methodologies is limited, and data in these studies were not consistently reported. Thus, inference about the solvent and protocol that leads to the highest EPFR extraction efficiency for certain types of radicals is not currently possible. Based on our review, we proposed reporting criteria to be included for future EPFR studies.

Graphical Abstract

1. Introduction

Environmentally persistent free radicals (EPFR) are long-lived, resonance-stabilized radicals formed in the cool zone of a combustion process¹. The concept of persistent free radicals was first introduced in the 1950s², and the signal of these radicals was detected by Pryor and co-workers in the 1970s^3,4. Dellinger et al.¹ proposed a detailed mechanism of EPFR formation in which the substituted aromatic compounds act as molecular precursors to react with transition metal oxides. The process starts with the physisorption of precursors onto the surface of transition metal oxides. Later, chemisorption takes place through elimination of water and/or hydrogen halide molecules, and electrons are then transferred to form surface-associated EPFR (Figure 1).

Figure 1.

Carbon-centered (left image) and oxygen-centered (center and right images) EPFR. The black dots in the figure represent the unpaired electrons and the thick black line indicates the particle surface.

Different precursors interact with transition metal oxides to form various EPFR. The structure of the absorbate, together with the properties of the absorption site under certain combustion conditions, determine the structure of the newly-formed EPFR and their different reactivities^1,5. Delocalization of the unpaired electron could lead to production of either a carbon-centered EPFR or an oxygen-centered EPFR.

Some recent studies investigated other mechanisms for EPFR formation on different molecular precursors and under varied reaction conditions. Studies provided evidence for EPFR formation in the presence of metal oxides other than those involving transition metals^6-8. Vejerano et al.⁹ discussed EPFR formation mechanisms on metal oxides and engineered nanomaterials (ENMs) during combustion and found that the molecular precursors and the type of metal oxide present affected concentration and lifetime of EPFR. D’Arienzo et al.¹⁰ revealed the formation of EPFR with unsubstituted benzene as the precursor. Borrowman et al.¹¹ explored EPFR formation through heterogeneous oxidation of ozone with polycyclic aromatic compounds. Zhu et al.¹² examined EPFR formation on microplastics under light irradiation. These studies provide insight into different EPFR formation pathways and their environmental impacts.

A growing body of research has examined persistence of EPFR. The lifetimes of EPFR range from hours to years^13-16 and theoretically may be infinite under a vacuum⁵. In contrast, traditional free radicals such as hydroxyl radicals (•OH) have half-lives on the order of 10⁻⁹ s¹⁷.

The toxicity of EPFR stems from their persistence in the environment coupled with their ability to generate •OH, which may lead to the downstream generation of other reactive oxygen species (ROS)¹⁸ including peroxyl (RO₂•) and alkoxyl (RO•) radicals. These ROS could induce oxidative stress in biological systems^19-22. In vitro and in vivo studies of inhalation exposure to EPFR containing (PM) identified cardiac^23-26 and pulmonary^27-31 dysfunction and effects on the central nervous system (CNS)^32-36. Saravia et al. (2013)³⁷ also studied adverse health effects of PM-associated EPFR on infants.

Electron Paramagnetic Resonance (EPR) is a spectroscopic technique used to study samples containing radicals. The g-factor obtained from an EPR spectrum is characteristic of a specific molecular structure³⁸ and thus provides information about the radical. A higher g-factor indicates that the unpaired electron is closer to an oxygen atom. For carbon-centered radicals, such as phenyl radicals, g-factors are typically less than 2.0030 while for oxygen-centered radicals, such as semiquinone radicals, g-factors are typically greater than 2.0040^1,5,13,39,40. The g-factor of EPFR from an environmental sample is usually in the range of 2.0030–2.0040⁴¹.

There are two main approaches for using an EPR to measure the radical signal in a sample potentially containing EPFR. The first method is the direct measurement of solid samples^10,12,42,43 or of filter samples^13,44-48 within a quartz tube inserted into the EPR cavity. For filters, a process of folding or rolling to fit them within the EPR quartz tube may be required. Chen et al.⁴⁹ proposed a method to clamp a piece of a quartz sheet filter into a flat cell within the EPR cavity to avoid possible changes in sample uniformity caused by folding or rolling. The other approach for EPFR detection by EPR is to extract EPFR from the collection matrix using solvents and then measure radical signals in the extracts or residues^{15,16,40,49-55}.

EPFR have been detected in particulate matter (PM)^15,16,49,50, ENMs⁹, soil^40,52,56,57, soot^50,58, fly ash^53,59, and biomass^60-64. Despite many original studies^{15,16,40,49-55} and review articles^{9,37,41,65-69} focused on EPFR, there was no systematic method for reporting EPFR extraction methodologies and their impact on retrieval efficiency and radical characteristics. A critical review comparing EPFR solvent extraction methods and their retrieval efficiencies is urgently needed to promote reporting consistency in this emerging field.

In this review, we evaluated studies concerning the possible loss of EPFR signals and changes to the radical characteristics during the extraction processes. We present extraction methodology and retrieval data from reviewed studies. By comparing methods used and extraction recoveries among EPFR studies, we hope to provide some information to assist with selection of EPFR extraction methods. We aim to guide reporting in future EPFR studies with our review.

2. Methods

The literature search and screening processes were conducted based on the guidelines and the checklist of Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA)⁷⁰ (Figure 2). The broad literature search, modeled after the search conducted by the U.S. Environmental Protection Agency in the Integrated Science Assessment for Particulate Matter⁷¹, began with an advanced search using the “topic” tag:

Figure 2.

The stepwise scheme of the literature searching and screening process.

((TS="particulate") OR (TS="particulates") OR (TS="particle") OR (TS="particles") OR (TS="pm") OR (TS="pm10") OR (TS="pm(10) ") OR (TS="pm2.5") OR (TS="pm(2.5) ") OR (TS="pm1") OR (TS="pm(1) ") OR (TS="pm0.1") OR (TS="pm(0.1)") OR (TS="pmfine") OR (TS="pmcoarse") OR (TS="black smoke") OR (TS="tsp") OR (TS="black carbon") OR (TS="bc") OR (TS="elemental carbon") OR (TS="ec") OR (TS="organic carbon") OR (TS="oc") OR (TS="ultrafine") OR (TS="ufp") OR (TS="aerosol") OR (TS="aerosols") OR (TS="soot")) AND ((TS="environmentally persistent free radical") OR (TS="EPFR") OR (TS="environmentally persistent free radical*") OR (TS="EPFR*")).

An asterisk was used to capture both singular and plural forms of the term “environmentally persistent free radical” in the literature. Publications meeting the search string criteria over all years (1900 – 2021) were collected from all databases accessed by the Web of Science search engine. Databases included the Web of Science Core Collection, KCL-Korean Journal Database, the U.S. National Library of Medicine^® (NLM^®) Premier Life Sciences Database, Russian Science Citation Index, and SciELO Citation Index.

A total of 120 references were identified through the initial stage of the search. References were discarded if they only contained abstracts, were written in other languages, contained irrelevant topics, were review papers, or consisted only of a model analysis without laboratory or field data. Because our objective was to compare the EPFR solvent extraction methods and their recoveries, studies were required to have an experimental process to be included in this review. A study using solvent extraction methods was recognized as one using solvents to extract EPFR from the collection matrix. Ninety-four references were screened to determine whether a solvent extraction method had been applied for the EPFR retrieval.

Overall, ten references were included in the detailed analysis for the comparison of extraction methods and their recoveries. Collected material from the studies included ambient PM^15,16,49,50, soil^40,50-52, coal⁵⁰, soot⁵⁰, fly ash⁵³, tar balls⁵⁴, and surrogate particles⁵⁵. Sample type, sampling strategy, filter material, extraction method, extraction recovery, g-factor, and EPFR concentration were abstracted from each remaining article, as available, and tabulated for comparison. The term “recovery” is a percentage comparing EPFR in the solvent extract with EPFR in the original sample. The recovery data are reported based on the change in sample instrument signal intensity or the change in EPFR concentration in these studies. Biases and limitations of each study were either drawn from the article itself or were inferred by us based on information provided in each study.

Following analysis of the ten papers, additional screening criteria were applied to categorize the entire reference pool of 94 articles for various research needs. Table S1 in the Supporting Information categorized the references first based on the sample origin, namely, environmental samples or laboratory-generated samples. “Environmental samples” included in the review were collected during field studies. In contrast, “laboratory-generated samples” refer to samples that were either synthesized in the laboratory or required certain treatments under controlled laboratory conditions to make the samples for the experiment. Examples of laboratory synthesized samples included surrogate particles made by metal oxides adsorbed to silica powders, biochar samples that were pyrolyzed in a laboratory under controlled conditions, and soil samples that were contaminated in a controlled laboratory setting. Within each category, references were then grouped based on the sample types. The full text of the ten references which provided the EPFR solvent extraction results were synthesized in the Results and Discussion section.

3. Results

3.1. EPFR Analysis

The number of studies regarding EPFR sampling has increased in the past decade. Among the 120 references identified since 2007, over half of the papers were published between 2016 and 2021. Among the 94 papers meeting one of our criteria of involving field or laboratory studies, 29 studied EPFR measured in environmental samples, and 65 examined surrogate samples synthesized or contaminated in a laboratory. The details of these studies were summarized in Table S1.

3.2. EPFR Solvent Extraction

Ten papers met our criteria for reporting the EPFR solvent extraction methodology and retrieval data. Various solvents and extraction methods including single solvent extraction and mixed solvent extraction techniques were used in different studies, and not every study reported the quantified recovery data. Solvents used in single-solvent extraction included methanol, isopropyl alcohol (IPA), dichloromethane (DCM), toluene, tert-butylbenzene (TBB), hydrochloric acid (HCl), sodium hydroxide (NaOH), acetonitrile (ACN), tetrachloromethane (CCl₄), and n-hexane. Among these, DCM was the most commonly used solvent, appearing in five studies^{15,16,49,51,55}. A v/v=1:1 acetone and DCM mixture was used in three studies^51-53, and a v/v=1:1 n-hexane and acetone mixture was used in one study⁵⁰. To extract the EPFR, researchers sonicated the samples in the solvent in four studies^51,53-55 and shook or vortexed the samples in three studies^16,40,52. A sequential pressure wash was applied in two studies^15,49, while a microwave accelerated solvent extraction system was used in one study for polycyclic aromatic hydrocarbon (PAH) analysis and EPFR extraction⁵⁰ (Table 1). These references were grouped by recovery method to facilitate analysis of the literature. The EPFR concentration data from these ten studies were provided in Table S2 (Supporting Information).

Table 1.

EPFR extraction approaches and recovery

Study	Sample Type	Solvent	Treatment	Recovery	g-factor for Original and Extracted Samplesa
Truong et al., 201055	Surrogate particles (CuO/SiO2)	Methanol, IPA, DCM, to luene, and TBB.	Sonicated for 1 hr.	Polar, alcoholic solvents (IPA and methanol): > 90% (70% for catechol). Nonpolar hydrocarbon solvents (TBB and toluene): <10% (15-25% for catechol). DCM: ~20% to ~50%.	N/A
Kiruri et al., 201354	Tar balls	IPA and TBB.	Sonicated for 1 hr.	TBB extracts of TB12: weak signal but recovery data not available. No signal detected in other extracts.	Original: TB12: 2.0035. TB56, TB92, and TB450: ~2.0041. Extracts: TB12 in TBB: 2.0035. TB12 in IPA: ND. TB56, TB92, and TB450: ND.
Zhao et al., 201953	Fly ash collected from the fabric filter	Sequential extraction: methanol and then mixture of v/v=1:1 acetone and DCM.	Sonicated for 10 min and centrifuged for 5 min.	45.6 - 73.7%.	Original: 2.0032 – 2.0038. Extracts: 2.0035 – 2.0045.
Liu et al., 202052	Soil	Acetone and DCM mixture (v/v=1:1).	Vortexed.	N/A	Original: N/A Extracts: Bauxite soil: 2.00280 – 2.00297. Fluvo-aquic soil: 2.00294 – 2.00299. Chernozem soil: 2.00269 – 2.00291.
Zhao et al., 202051	Soil (montmorillo nite clay)	Single solvent: acetone, DCM, methanol, ACN, and CCl4. Mixed solvent: acetone and DCM (v/v=1:1).	Sonicated for 30-1200 s.	Acetone is the most effective solvent. Acetone and DCM mixture also extract EPFR with a slightly lower efficiency, while all others cannot effectively extract EPFR.	N/A
Yang et al., 201716	PM collected on quartz filter	DCM.	Slightly shaken for 3 min and soaked in dark for 6 hr.	N/A	Original: N/A Extracts: TSP: 2.00316 – 2.0034. PM>10: ND. PM2.5-10: N/A PM1-2.5: N/A PM1: 2.00323 – 2.00371.
Chen et. al., 201849	PM collected on quartz filter	Single solvent: DCM. Sequential wash: water, methanol, DCM and n-hexane.	Single solvent: submerged in DCM for 3 min and soaked in dark for 10 hr. Sequential wash: washed 3 times by each solvent.	6.8% for multiple solvents sequential wash. No comparable signal detected in single solvent extraction.	Originalb: mean of 2.0034. Single solvent extract: ND. Sequential washb: mean of 2.0038.
Chen et. al., 201815	PM collected on quartz filter	Sequential wash: water, methanol, DCM and n-hexane.	Washed 3 times by each solvent.	2.4 ± 0.7%.	Originalb: 2.0028 – 2.0033. Extracts’b: 2.0036 – 2.0040.
dela Cruz et al., 201140	Soil	Sequential extraction: HCl and then NaOH.	Shaken in HCl for 1 hr, and then precipitate was shaken in NaOH for 24 hr.	Around 5% is acid-base extractable.	Original: 2.0030-2.0039. Extracts: N/A
Wang et al., 201850	Soil, coal, soot, and total suspended particles (TSP) collected on glass fiber filter	Acetone and n-hexane mixture (v/v=1:1).	Microwave-assisted extraction.	TSP and soot: 20-30%. Soil and coal: non-extractable.	Original: Coal: 2.0046 ± 0.0000. Soot: 2.0044 ± 0.0000. TSP: 2.0041 ± 0.0001. Top soil: 2.0039 ± 0.0001. Extracts: N/A

^*IPA: isopropyl alcohol; DCM: dichloromethane; TBB: tert-butylbenzene; ACN: acetonitrile; CCl₄: tetrachloromethane; HCl: hydrochloric acid; NaOH: sodium hydroxide. TB₁₂: tar ball samples collected after 12 days of the Deepwater Horizon Gulf of Mexico oil spill. N/A: data are not available. ND: signal not detected.

^a:g-factor data provided in this table are from original samples and solvent extracts that were not subjected to any aging processes.

^b:Samples were measured using the quartz sheet-based method proposed by Chen et al.⁴⁹.

The study conducted by Truong et al.⁵⁵ was the first to evaluate extraction of EPFR by solvents. The samples consisted of surrogate EPFR particles made from six organic precursors adsorbed onto 5% CuO/SiO₂ particles. Extraction was carried out by sonicating the particles in a solvent for 1 hr. Five solvents were tested individually. Another study of EPFR on tar balls⁵⁴ used a similar extraction method. Both studies compared the extraction efficiency of polar and nonpolar solvents. The two polar solvents (IPA and methanol) used in the Truong et al.⁵⁵ paper resulted in >90% recovery, while the nonpolar solvent (TBB and toluene) extracted <10% EPFR from the matrix. DCM exhibited a recovery of 20 to 50%. All recovery data were calculated based on reduction in EPFR concentration. A g-factor range of 2.0040 to 2.0065 was reported in the study without including a specific g-factor of radicals produced by each precursor. In the study of tar balls collected from the Gulf of Mexico following the Deepwater Horizon oil spill of 2010⁵⁴, the TBB extract from the sample collected 12 days after the spill (TB₁₂) had a g-factor of 2.0035, which was unchanged after extraction. Recovery of the TB₁₂ sample from TBB extracts was not reported in the study. No EPFR signal was detected in the IPA extract of TB₁₂ samples and the TBB and IPA extracts from samples collected after 56 (TB₅₆), 92 (TB₉₂), and 450 days (TB₄₅₀). The studies by Truong et al.⁵⁵ and Kiruri et al.⁵⁴ both used IPA and TBB as the polar and nonpolar extraction solvents and applied the same treatment process to the samples. The notable differences in recovery observed in these two studies could be due to variations in the chemical composition and properties of the surrogate particles and tar ball samples used in the two studies.

Three studies used the mixture of v/v=1:1 acetone and DCM as the extraction solvent^51-53. Zhao et al.⁵³ added a step of methanol extraction before fly ash samples were subjected to the extraction by the acetone-DCM mixture. Both steps involved a 10-min sonication followed by a 5-min centrifugation. Liu et al.⁵² vortexed the soil samples in the acetone-DCM mixture. Zhao et al.⁵¹ sonicated the surrogate soil samples in the acetone-DCM mixture while single solvent extraction was also investigated using acetone, DCM, ACN, and CCl₄ individually. Among these three studies, only Zhao et al.⁵³ reported a quantified EPFR recovery of 45.6 to 73.7% based on decreased EPFR concentration in the extracts. The fly ash samples exhibited a g-factor of 2.0035-2.0045 in the extracts compared to 2.0032-2.0038 in the original samples⁵³, indicating that the solvent extractable EPFR contain more oxygen-centered radicals. Liu et al.⁵² did not measure the g-factor and EPFR concentration in the original soil samples before extraction, thus no recovery values could be discerned. Zhao et al.⁵¹ presented qualitative data that acetone as the single extraction solvent has the highest extraction efficiency and the acetone-DCM mixture could extract EPFR from the matrix with a lower efficiency. The g-factors and EPFR concentrations before and after extraction were not reported in this study.

Another group of studies compared EPFR recovery using a single DCM solvent with a sequential solvent pressure wash^15,16,49. Yang et al.¹⁶ lightly shook the filter pieces in DCM for 3 min and then soaked the samples for 6 hr in the dark for the extraction. The authors did not measure the EPFR signal in the original samples before extraction, thus no EPFR recovery was reported for this study. The g-factor and EPFR concentration were reported under different collection air quality conditions, as shown in Table 1 and Table S2. Chen et al.⁴⁹ adapted this method and extended the soaking time to 10 hr for extraction. The authors compared EPFR recovery of this single solvent extraction to the results with multiple-solvent extraction by a sequential pressure wash. This method was applied in the newly proposed quartz sheet-based method, in which the filter slices could be directly clamped by a slide and coverslip and measured in the EPR cavity resonator, leaving the samples intact. The sequential solvent pressure wash method was discussed in another paper by Chen et al.¹⁵, where filter punches were washed three times with 10 mL of water (for extracting water-soluble matter), followed by methanol, DCM and n-hexane (for extracting water-insoluble matter) in that order. Chen et al.⁴⁹ reported that the sequential solvent wash resulted in an average EPFR recovery of 6.8% based on the concentration change, with the average g-factor slightly increased from 2.0034 to 2.0038. However, no signal was detected in single DCM extraction. Chen et al.¹⁵ compared EPFR extractions for different components of particles and observed that an average of 2.4 ± 0.7% of the EPFR were solvent-extractable. A g-factor range of 2.0036-2.0040 in the PM sequential washed solvent extracts compared to 2.0028-2.0033 was also reported¹⁵. Increased g-factor in the extracted EPFR indicated that the extract contained more oxygen-centered radicals, potentially from quinones. The authors also measured the non-extractable sample residual, which showed no appreciable difference of g-factor (changes within 0.0001) and EPFR concentration (changes from 0% to - 11%).

dela Cruz et al.⁴⁰ extracted EPFR from contaminated soil by shaking the samples first in HCl for 1 hr followed by shaking the insoluble part in NaOH for 24 hr. This approach produced a 5% recovery of EPFR in the acid-base soluble component. The detailed g-factors and EPFR concentrations for the extracts were not explicitly reported in this study.

Wang et al.⁵⁰ sought to extract PAH from soil, coal, soot and total suspended particles (TSP). The authors reported a 20-30% EPFR signal decrease after extraction for soot and TSP samples, with no changes in coal and soil samples. The authors did not point out explicitly whether the extracts were measured. Based on the data and Wang et al.’s⁵⁰ explanation, we deduced that the EPFR signal in the residue was measured to calculate the extraction recovery. In this case, the g-factor and EPFR concentration in the extracts were not reported.

4. Discussion

4.1. Synthesis of Findings

The g-factor change after solvent extraction provides information about the influence of extraction on the radical structure for different types of samples. Among the ten studies^{15,16,40,49-55}, four^15,49,53,54 of them reported the g-factor of samples before and after the extraction process. For those, extractable tar ball samples showed no g-factor change while fly ash samples and PM samples exhibited increased g-factors after extraction. The fly ash and PM samples had more similar compositions. This result indicates that the structure and the composition of the sample matrix play an important role in EPFR formation and extraction. The remaining six studies lack g-factor data in the original samples, extracts, or both.

Overall, EPFR retrieval efficiency is closely related to the chemical characteristics of the samples and solvents^49,51,54,55. It is important to consider the possible influences of solvent extraction on the EPFR characteristics before applying the solvent extraction method. With a relatively low solvent retrieval efficiency, Chen et al.⁴⁹ compared the solvent extraction method for EPFR analysis with a direct EPFR measurement method and concluded that complete EPFR retrieval via extraction could not be guaranteed. Chen et al.⁷² investigated the organic carbon, elemental carbon and metal compositions of the solvent extractable samples and the non-extractable components of PM_2.5 samples collected in the study. The results showed that the EPFR were mainly non-extractable and that extracted EPFR exhibited a more complicated decay pattern with a rapid increase in concentration before the start of decay⁷². Zhao et al.⁵¹ investigated the solvent extractability of a benzo[a]pyrene/anthracene-EPFR under different solvent types and ultrasound parameters to optimize the solvent extraction process for this type of PAH-EPFR. Differences were observed in stability of the extracted EPFR compared with the surface-bound EPFR, which had longer lifetimes⁵¹. Zhao et al.⁵³ also compared solvent-extractable and non-solvent extractable components of fly ash. They normalized EPFR spin densities by total organic carbon concentrations and then tested correlations of individual metals with solvent-extracted EPFR and the non-extractable EPFR in the sample. Zhao et al.⁵³ observed a relationship of the metal conjugate with the amount of ROS generation for the solvent-extracted EPFR but not for non-extractable EPFR, which may suggest that the solvent extraction process could influence the toxicity of the samples. Truong et al.⁵⁵ observed differences in extractability by polar and nonpolar solvents for EPFR generated by conjugation of aromatic compounds to CuO-silica. Vejerano et al.⁷³ suggested that the solvent extraction process has the potential to destroy EPFR, which could modify the toxicity of PM samples.

4.2. Limitations and Risk of Bias in the EPFR Solvent Extraction Literature Reviewed

Inference from this review is limited, because the objectives of the papers reviewed in this study were different. Some of them did not focus on optimizing the EPFR extraction method but instead applied a method to retrieve EPFR for the analysis. In such cases, two studies did not report the recovery data^16,52,. The reported recovery in the study conducted by Chen et al.¹⁵ was acquired after the sequential wash. The EPFR signal was measured after sequential extraction by methanol, DCM, and n-hexane. Thus, the extraction efficiency for each single solvent could not be specified. The recovery range reported in the study by Zhao et al.⁵³ is relatively wide, indicating that more replicates may be needed. Furthermore, there is no quantitative result on the EPFR extraction efficiency by different solvents or under various ultrasonic conditions in the study by Zhao et al.⁵¹.

The g-factor and concentration are important parameters for any EPFR study because quantification of changes in g-factor and EPFR concentration before and after the retrieval improve our understanding of possible reactions during the extraction process and the influence of different solvents on the radicals. However, several studies^40,51,54,55 did not provide systematic EPFR g-factor and concentration data. Discrepancies between the body of the manuscript and the supporting information were noticed in both the study by Zhao et al.⁵³ and the study by Yang et al.¹⁶. Zhao et al.⁵³ reported the data range for g-factor and concentration in all samples except for non-extractable samples. The g-factors of two non-extractable data points exceeding 2.003 were presented in the supporting information. There is a risk of mistakenly interpretating the non-extractable EPFR to be carbon-centered radicals, because the g-factor range of 2.003–2.004 was characterized for carbon-centered radicals with an adjacent oxygen atom from previous studies^1,5. With the very limited number of studies that reported the g-factor in the original samples and the solvent extract, it is hard to generalize the influence of the extraction process to the sample g-factor, which indicates the radical structures.

Some studies did not report important details of the extraction and measurement methods. For example, the extraction time was not presented in the study by Liu et al.⁵². In the study by Wang et al.⁵⁰, more details about soil sample collection would help to reduce the uncertainties. Moreover, the author did not clearly state which component of the sample after solvent extraction was measured for EPFR. In the study by Zhao et al.⁵³, no details were presented about how the solvent-extracted EPFR were measured. In this review, we identified a gap in the literature where the lack of a systematic reporting procedure makes it difficult to compare the recovery of EPFR extraction among studies.

4.3. Strengths and Limitations of Our Review

The studies reviewed in this paper included different sample matrices and EPFR structures. The results from these studies^49,51,55 have led to the conclusion that the structure of the radical species and the solvent were important factors for EPFR recovery. Thus, the recovery reported in the small number of studies identified in our review may not be generalizable for future studies if the EPFR species have changed.

5. Conclusions and Perspectives

EPFR are an emerging contaminant that has received increased research attention in recent years. EPFR can be generated during combustion processes, and they are believed to be related to certain adverse respiratory and cardiovascular effects due to their ability to generate reactive oxygen species.

The solvent extractability of EPFR depends on the chemical composition of the sample and the solvent type. We have discovered two main findings in this review. First, the polar solvent seems to have a superior EPFR solvent extraction recovery. Second, a higher g-factor in the solvent extracted samples were noticed compared to the original samples in fly ash and PM samples, indicating more oxygen-centered radicals were present in extracted samples, while the g-factor for extractable tar ball samples remained the same. However, with the very limited number of papers available for review in this manuscript, these two findings cannot be generalized.

We identified an important gap in the literature where studies involving EPFR are often inconsistent about reporting experimental details, such as retrieval and g-factor. We recommend that EPFR studies should report detailed information regarding sampling procedure, sample treatment, measurement protocol, and data. In the sampling procedure section, authors should include sample type, and detailed sampling information such as the type of sampler, setup height above the ground, sampling site location and description, distance from major emission sources, and relevant meteorological data. A detailed sample treatment method should be reported, and if a solvent extraction method is employed, important information including solvent type, quantity of samples and solvent, extraction time, and instrument type and parameters are also recommended for inclusion. Likewise, the analytical chemistry method should present information about which component(s) of the sample are measured, which instruments are employed, and operational parameters. The extraction recovery and EPFR g-factor and concentration before and after the extraction should also be reported. Regular reporting of this information will improve the state of the science through development of best practices on EPFR sampling and analysis.

Highlights.

• Important gaps exist in the EPFR literature regarding retrieval and analysis.

• We summarized the state of the science for EPFR retrieval via solvent extraction.

• Reporting criteria for EPFR studies were proposed.