Integrating AF4 and Py-GC-MS for Combined Size-Resolved Polymer-Compositional Analysis of Nanoplastics with Application to Wastewater

Nanoplastics are a widespread pollutant, but their reliable characterization in environmental samples remains difficult. We introduce a new workflow combining asymmetrical flow field-flow fractionation with multiangle light scattering (AF4-MALS) and pyrolysis-gas chromatography–mass spectrometry (Py-GC-MS). AF4-MALS provides cleanup and size separation down to 1 nm, while Py-GC-MS identifies and quantifies polymers in each fraction.

Validated on standard particles and applied to wastewater, the method identified polystyrene and potentially PVC nanoplastics, while polyolefins and PET were below detection limits. Quantification limits ranged from 0.64 ng (PS) to 180 ng (polyolefins), with 8.8 ± 1.8 ng/mL PS measured in untreated wastewater. Despite challenges with recovery and matrix effects, this combined approach represents a promising step toward comprehensive nanoplastic analysis.

The original article

Integrating AF4 and Py-GC-MS for Combined Size-Resolved Polymer-Compositional Analysis of Nanoplastics with Application to Wastewater

Maria Hayder*, Cloé Veclin, Aislinn Ahern, Aleksandra Chojnacka, Erwin Roex, Florian Meier, Gert-Jan M. Gruter, Annemarie P. van Wezel, Alina Astefanei

Anal. Chem. 2025, 97, 28, 15216–15224

https://doi.org/10.1021/acs.analchem.5c01766

licensed under CC-BY 4.0

Selected sections from the article follow. Formats and hyperlinks were adapted from the original.

Plastic is an abundant environmental pollutant. Plastic waste undergoes degradation, forming micro- and nanoplastics (MNPs). (1,2) While there is still no final consensus on the nanoplastics (NPs) definition, here we use this term for plastic particles sized 1–1000 nm. (3) NPs differ from microplastics (1 μm-5 mm) in their transport properties and bioavailability. (4−6)

Despite their expected ubiquity, research showing the actual presence of NPs in the environment remains limited (9−16) with no routine method to characterize NPs in environmental samples. (17) NP size distribution is a key in understanding the extent of pollution and its environmental impact. (18) Asymmetrical-flow field-flow fractionation coupled to multiangle light scattering (AF4-MALS) has been applied to investigate NP presence in the environment. (9,19−22) AF4 provides mild, nondestructive size-based separation, potentially useful for pretreating environmental samples. (15)

Particle size distributions must be combined with polymer type for identification, which AF4 cannot provide. A common approach for chemical analysis and mass quantification of MNPs is pyrolysis-gas chromatography–mass spectrometry (Py-GC-MS), (8,23−26) providing polymer-related information by thermally degrading polymers into identifiable fragments. Combining the two techniques could result in comprehensive size and polymer information on NPs. Until now, scarce work has focused on practically combining AF4 and Py-GC-MS. (15,21,27) One reason for this may be technical challenges when connecting both instruments, such as sample dilution during AF4 separation and the incompatibility of common AF4 eluents (nonvolatile salts and surfactants) with Py-GC-MS.

In this work, we present a novel approach for NP analysis in environmental water samples. We explore the capabilities and limitations of an offline workflow combining AF4-MALS (size-based separation and size distribution measurements and sample cleanup) and Py-GC-MS (polymer identification and quantification). To address low concentrations of NPs in environmental samples, we developed a large-volume injection (LVI) method for AF4, allowing for injection of 10 mL of sample. We believe this work is a significant step toward improving analytical approaches urgently needed in the NP field.

Materials and Methods

AF4-MALS Measurements

AF4-UV-MALS measurements were performed using an AF4-UV-MALS system (AF2000 MultiFlow FFF system, Postnova Analytics, Landsberg am Lech, Germany) with an SPD-20A UV/vis absorbance detector operated at 280 and 254 nm (PN3212; Shimadzu, Kyoto, Japan) and a 21-angle MALS detector (PN3621). Data were acquired by the AF2000 control software version 2.1.0.1 (Postnova Analytics). More details and the scheme of the applied system can be found in Section S2 and Figure S1. A 10 kDa PES membrane and 350 μm spacer were employed. 0.25 mM ammonium carbonate was selected as a carrier liquid compatible with Py-GC-MS (volatility and clean thermal decomposition). A stability study (Section S3) showed that it does not enhance agglomeration in the AF4 (comparison with other salts in Table S1).

For the small volume injection (SVI, 1 μL), a PSC stock suspension (323 μg/mL final concentration) in carrier liquid was used. Injection was performed for 3 min using the autosampler (injection flow F_inj = 0.2 mL/min). For LVI (10 mL), the stock suspension was diluted in carrier liquid to a final concentration of 32.3 ng/mL so that after the injection, the analyte mass (323 ng) is the same as in SVI. The LVI injection was performed manually for the time deemed optimal after the initial experiments (see the Results and Discussion section). Recovery calculation and the flow rate program are described in Section S4. The fraction collection started 1 min later than the elution. Eight fractions were collected for 7 min each. Experiments were performed in triplicate.

Py-GC-MS Measurements

Py-GC-MS measurements were performed on a Shimadzu GCMS-QP2010 Plus system (Kyoto, Japan) with an Optic-4 programmed-temperature vaporization (PTV) injector as the pyrolysis chamber (ATAS GL, Veldhoven, The Netherlands) and a Focus XYZ autosampler (ATAS). More details regarding the PTV are listed in Section S5. Target polymers (polystyrene, PS, polyvinyl chloride, PVC, polyethylene terephthalate, PET, polyethylene, PE and polypropylene, PP) were chosen due to their large production volume and the resulting expected abundance in the environment. (30) 55 μL of the liquid sample was injected at the inlet temperature of 50 °C. Injection repeatability and absence of memory effects were confirmed. Pyrolysis was performed at 550 °C (Table S2). The GC oven temperature was ramped from 50 to 320 °C (Table S3). Target pyrolysis products and the corresponding m/z and t_r values were selected (Table S6). The MS was operated (Table S4) in the scan mode (m/z 60to 300). To establish the indicative m/z for targeted polymers, no quantification was required. To obtain clear, high signals, manually cut minimal amounts of each solid polymer were placed in glass inserts for injection, and pyrolysis was performed at a column flow of 1.2 mL/min and split flow of 200 mL/min at 550 °C, followed by the GC–MS measurement as described above.

For liquid injections, targeted polymers were dissolved (dissolution conditions are in Table S5). Stock solutions were diluted in THF. Calibration curves were measured for each polymer individually (Figure S4). Limit of detection (LOD) and limit of quantification (LOQ) values were determined based on 26 THF blanks (methodology in Table S6).²⁵For PET, there were no visible signals in blanks. LOD and LOQ were determined from the calibration curves as the lowest signal-yielding concentration and the lowest concentration in the linear response range, respectively. (26)

Sample Handling

The schematic representation of the proposed workflow is depicted in Figure 1. Directly before analysis, samples were sonicated (A) for 10 min three times in a RK510H bath (frequency 35 kHz, nominal power 160 W, Bandelin, Berlin, Germany), with 10 min breaks in between. Subsequently, samples were filtered (B) over 1 μm PES syringe filters into a glass container. After AF4 measurements (C), fractions were collected into glass tubes (D), which were then capped with Miracloth (rayon), frozen, and freeze-dried (E; Heto PowerDry LL1500 Freeze-Dryer, Thermo Fisher Scientific, Waltham, MA, USA). Filtration and freeze-drying validation procedures are described in Section S6.

Anal. Chem. 2025, 97, 28, 15216–15224: Figure 1. Schematic representation of the described workflow.

Next, samples were resuspended (F) in THF. One replicate was prepared by adding 500 μL of THF to each dry fraction and vortexing the tubes for 10 s. To increase the resuspension efficiency, the other two replicates were prepared by vortexing for 20 s and sonicating for 10 min (no great difference in signal intensities was observed in the results). THF was added slowly via the tube walls. Directly afterward, the suspension was transferred into glass vials. Such samples were injected into a Py-GC-MS system (G).

Results and Discussion

Py-GC-MS Method Development

Pyrolysis products typical for the targeted polymers and their m/z values were found in the literature (Table S7) and confirmed by pyrolysis of solid samples in our instrument (Table S6).

The styrene trimer was chosen as the identifier for PS as the styrene monomer may be a product of other substances, and the styrene dimer was measured with a lower intensity than the trimer.

For PET, acetophenone was selected as the indicative pyrolysis product as it does not require derivatization, unlike benzoic acid. Although a derivatization agent might improve the measurement quality for the polar substances, its reaction mechanism is not yet fully understood. We thus avoided its use to minimize the risk of it reacting with other components of the complex samples.

Distinguishing between PE and PP is challenging as these two polyolefins have a similar structure and produce a similar mass chromatogram in the form of a “comb” with each peak corresponding to a different oligomer length. Additionally, they yield pyrolysis products sharing the same m/z values. Although the literature suggests m/z 69 and 97 can be used to distinguish between PP and PE (Table S7), we have observed both values in the separate analyses of individual polymers. A chromatogram of the PE and PP mixture proved that these two cannot be certainly separated (see Figure S3 and further discussion). We followed the literature in reporting m/z 69 as belonging to PP and m/z 97 to PE, however, only the polyolefin origin can be ascertained (see Section S7).

Conventionally, Py-GC-MS methods use solid standards, bringing large uncertainties to quantification. (23) Here, we use liquid injection. Liquid injections of the same polymers were performed to confirm the presence of selected m/z and to observe the retention times (Table S6). Calibration curves were measured (Figure S4). We selected 55 μL of injection volume for Py-GC-MS, which is at the upper end of our instrumental possibilities without applying backflush. A detailed investigation on different mixtures and polymer ratios was out of the scope of this study; however, the presence of one polymer may influence the pyrolysis efficiency of another one. (33) We ascertained the accuracy of our measurements by running QCmix in each batch instead.

Wastewater Analysis

The analysis results of an untreated wastewater sample processed using our workflow are shown in Figure 3. Figure 3A shows the overlay of the AF4-MALS fractograms (three LVI injections, conditions identical with those for PSC standards), where signals are high but irregular, displaying a broad tailing peak approaching baseline in the second half of the elution window and a residual peak. This suggests a continuous particle size distribution in the sample, as expected. Although the peak shapes are repeatable, size measurements by MALS vary between injections. This variability is probably related to the sample’s nature. Despite all efforts to homogenize the influent sample differences between replicates may still occur, e.g., in terms of particulate matter. The measured radii differ from the PSC elution profile (Figure S2) ranging from ∼270 to 700 nm, and vary more than those in the PSC mixture, possibly due to particle agglomeration in the organic matrix (34) or different shapes of environmentally occurring NPs. In wastewater, NPs compose only a small fraction compared to other particulates and organic matter, which might influence MALS results, with larger particles masking the smaller ones. For radius calculations based on the MALS measurements, the errors generally increase with increasing analyte size. (35) The size distribution pattern measured by MALS is repeatable. The particle sizes are highest in the first part of the fractogram, they later decrease and subsequently increase again, which may indicate mixed normal and steric elution, where the elution order is reversed, (36) which complicates size separation and data evaluation. Particle agglomeration could contribute to such a behavior. Agglomeration depends on the matrix properties; thus, mixed-mode/steric elution is probably sample-dependent.

Anal. Chem. 2025, 97, 28, 15216–15224: Figure 3. (A) AF4-MALS fractograms of three wastewater replicates; (B) polymer composition and concentrations detected in the untreated wastewater by Py-GC-MS in the respective fractions. “–” indicates signal < LOD, “+”-signal < LOQ.

Conclusions

A novel approach for NP analysis is proposed, aiming at simultaneous preconcentration, size estimation, and polymer identification/quantification for nanometer plastic particles. Technical capabilities, limitations, and possible solutions for the challenging LVI-AF4-MALSPy-GC-MS coupling are widely discussed for the first time. Our offline workflow was tested on standard particles and on wastewater, offering original insight into the potential for NP simultaneous size and mass analysis in complex matrices. PS NPs were successfully quantified in the wastewater. Recovery remains the greatest area for improvement, with an urgent need to optimize resuspension. Possible directions in platform development are suggested. Since AF4-Py-GC-MS is sought-after in the NP research, we believe our work could be used to broaden the knowledge on NP occurrence and fate.