Development and validation of an analytical pyrolysis method for determination of microplastic polymers in atmospheric aerosols

Plastic pollution in the environment is a problem of growing concern. Plastic has been detected in all major ocean basins [1] and is increasingly reported in aquatic and terrestrial environments [2]. In recent years more attention has been drawn to the importance of environmental pollution of microplastic (MP, sizes of 1 µm - 5 mm) and nanoplastic (NP, < 1 µm). MP and NP (MnP) particles can be released directly to the environment (primary MnP) but can also form through fragmentation and degradation of macroplastic (forming secondary MnP) [3], [4]. MP pollution has been reported in several environments, including oceans [5], freshwater [6], sediment [6], soil [7], polar ice [8], wildlife [9], food meant for human consumption [10], [11], [12], and within the human body [13], [14]. MP ingestion has been shown to have detrimental effects on aquatic organisms [15] but the health effects on humans are poorly understood due to limited amount of research on the topic. Inhalation of PM_2.5 and exposure to MP have both been linked to adverse health effects [16], [17], [18]. In recent years MP pollution has been detected in the atmosphere [19], where it can be transported over long distances [20], [21] and potentially provides a source of human exposure. It has been detected in indoor air [19], [22], urban [19], [23], [24], continental, remote [25], and marine air [26], [27], [28] as well as higher up in the free troposphere [21], [29]. Airborne MP also has the potential to impact climate through radiative effects [30], influence on precipitation [31], and cloud formation [32]. Sources of airborne MP include brake and tire wear [33], resuspension of deposited MP [34], and the ocean which can act as both a source and a sink [26], [35]. MP concentrations in indoor air exceed that of outdoor air [22], while air concentrations in urban environments exceed those of sub-urban and remote areas [19], [25]. Globally the most produced plastic polymers are polyethylene (PE) 26.9 %, polypropylene (PP) 19.3 %, polyvinyl chloride (PVC) 12.9 %, polyethylene terephthalate (PET), polyurethane (PU), polyamide (PA) and polystyrene (PS) < 10 % each [36]. Therefore, we focus on these polymers in this study.

Pyrolysis gas chromatography coupled to mass spectrometry (py-GC-MS) is an effective analytical tool to identify types of polymers and determine mass concentration of MP [37], [38], [39], [40]. Analysis using py-GC-MS is mainly limited by polymer mass, while not being sensitive to uncertainty due to subjective visual evaluation nor limited by the size of the individual polymer particles [41], [42]. In contrast, some other analytical methods used for MP analysis such as Fourier transform infrared spectroscopy and Raman spectroscopy have a lower size limit of 10 and 2 µm, respectively [41], [42], while atmospheric particles of health relevance are typically in the range from a few nanometers to about 10 µm in diameter. This is relevant as there are indications that the number concentration of airborne MP particles increases with decreasing particle size [20], [21], [25], implying that NP could be of considerable importance to atmospheric processes and human health. Particles with diameters < 10 µm are respirable and particles < 100 nm can enter the bloodstream via the lung upon inhalation [18], [19], [20], [21], [25].

Pyrolysis involves heating a sample under oxygen-free conditions to a temperature that leads to the thermal decomposition of non-volatile analytes into characteristic volatile fragments that can be analyzed with a gas chromatograph [43]. Py-GC-MS shows great promise as a method for the analysis of MP within atmospheric samples with complex matrices and allows for the reporting of mass concentrations [40], [44]. Due to the occurrence of secondary reactions during pyrolysis, a complex matrix containing a multitude of organic compounds, as are typically present in atmospheric aerosol samples, can interfere with the analysis of polymers. Reactive matrix components can either react with the final polymer pyrolysis products or with intermediate pyrolysis products before they form the final product. Furthermore, certain polymer pyrolysis products can also react with each other, which can lead to inaccurate quantification [45], [46], [47].

Here we develop and validate a novel py-GC-MS analysis method, based on our previous work on PS [48] for the identification and quantification of seven polymers, PE, PP, PVC, PET, PU, PA, and PS on the nanogram scale, individually and in a mixture. The method was validated using two aerosol matrices: particles from candle burning with low (1 μg per sample) and high (2 μg per sample) aerosol mass and urban PM_2.5 aerosol particles of low (7 µg per sample) and high (53 µg per sample) aerosol mass loading. Methods for minimizing matrix effects on final polymer concentration estimates were developed, including a thermal desorption step prior to pyrolysis and the inclusion of deuterated-D8 polystyrene (DPS) as an internal standard (IS). The study provides insight into the behavior of common plastic polymers during co-pyrolysis as well as matrix effects from aerosol matrices containing elemental and organic carbon. The method allows for determination of common airborne polymers using direct py-GC-MS analysis without any sample pre-treatment step. Application of the method allows for estimating concentration levels and understanding the transport pathways of atmospheric MnP.

2. Experimental

2.3. Instrumentation

Analysis was performed using a pyrolysis unit (PYRO, Gerstel) located inside a thermal desorption unit (TDU, Gerstel) interfaced with a cooled injection system (CIS4, Gerstel) connected to a GC-MS (7890b and 5977a Agilent Technologies). The pyrolysis temperature was ramped at a slow rate from 300°C to 800°C at 5°C s⁻¹. The pyrolysis module was operated in splitless mode and the CIS4 in a 20:1 split ratio. The TDU temperature was ramped from 40°C to 300°C at 300°C min⁻¹ and held at 300°C for 2 min, while the CIS4 was cooled to −120°C followed by a temperature ramp to 325°C at 12°C s⁻¹ where it was held for 3 min. The pyrolyzer was also used for thermal desorption of aerosol matrix by ramping the temperature from 100°C to 300°C at 5° s⁻¹, see Table 2 for method details. A constant helium flow of 1 mL min⁻¹ carried pyrolysis products to the fused silica capillary column (HP-5MS Ultra Inert, 30 m × 0.25 mm inner diameter × 0.25 μm film thickness). The column was located inside a temperature-controlled oven with an initial temperature of 50°C that increased at a rate of 10°C min⁻¹ to 320°C where it was held for 5 min.

Data analysis was performed using Qualitative Analysis (10.0) Unknowns Analysis (10.1), and MS Quantitative Analysis (10.1) from Agilent MassHunter. The NIST17 library was used to match mass spectra. Data was further processed in MATLAB (23.2, The MathWorks Inc.) with the Curve Fitting Toolbox (23.2, The MathWorks Inc.).

3. Results and discussion

3.1. Method development

Fig. 1 shows the pyrogram for a polymer mix containing 2 μg of each polymer (PE, PP, PVC, PET, PU, PA, and PS) obtained from a slow ramp of the pyrolysis temperature (300–800°C at 5°C s⁻¹) and the MS operated in the scan mode (35–550 m/z). Peaks labelled by chemical name are from pyrolysis products that were also observed in the pyrograms from the individual polymers (Figures S1-S7) and were selected for the SIM mode. PA, PVC, and PET were also analyzed individually with flash pyrolysis (600°C) to compare with the slow ramp pyrolysis (see Figures S5-S7 for details). Secondary products were observed including benzonitrile, 2-chloroethylbenzoate and 1,4-benzenedicarboxylic acid, di-2-chloroethyl ester from reactions between pyrolysis products of PET and PVC as previously reported by Coralli et al. [45]. The pyrogram was divided into three sections for the SIM mode, see Figure S8 for a closer view of the sections. Table 3 lists the pyrolysis products chosen for SIM along with the name of the polymer that the product is an indicator of. At least one pyrolysis product specific to each polymer was selected for the SIM method, while several non-specific products were also selected. Generally, good peak separation was obtained, and most peaks chosen for SIM were well resolved.

Journal of Analytical and Applied Pyrolysis, 193, 2026, 107423: Fig. 1. Pyrogram obtained from a polymer mix with PE, PP, PVC, PET, PU, PA, and PS (2 μg of each). Analysis was performed with a slow ramp of the pyrolysis temperature (300–800°C at 5°C s−1) and an MS scan range of 35–550 m/z. Pyrolysis products that were selected for SIM are labeled with their chemical name. Secondary products are labeled by *.

Journal of Analytical and Applied Pyrolysis, 193, 2026, 107423: Table 3. Specific pyrolysis products included in the SIM method for the 7 polymers. Bold indicates a pyrolysis product (out of several) that is used for analysis and quantification of a specific polymer.

3.3. Investigation of polymer recovery and matrix effects

3.3.1. Polymer recovery

Fig. 3 shows the recovered polymer masses of 600 ng polymer mix standards spiked in matrices of candle burning and urban PM_2.5 aerosol samples. The red line indicates the polymer mass spiked onto the filter.

Journal of Analytical and Applied Pyrolysis, 193, 2026, 107423: Fig. 3. Polymer mass detected in spiked samples of candle combustion and urban PM2.5 aerosol samples. Error bars denote the associated uncertainty, and the red line indicates the polymer mass the sample was spiked with. Graph titles are the polymer detected while subtitles are the pyrolysis product used for quantification. Aerosol mass loadings of the samples were 1 μg (candle A), 2 μg (candle B), 7 μg (urban 3 d), and 53 μg (urban 7 d).

Analysis of non-spiked punch-outs from the candle particle and PM_2.5 filters showed no detection of PET and DPS (D-styrene), while the D-trimer signal yielded DPS masses from <LOQ to 6 ± 2 ng. Background contamination of PS (10 ± 18 ng) and PP (20 ± 77 ng) was found for all filters (see Table S2). PVC and PE were present in all samples with masses ranging from 30 ± 22 ng to 55 ± 22 ng for PVC and from <LOQ to 43 ± 300 ng for PE. PA was only detected in the candle A filter (<LOQ). High PU masses were found in the candle particle filters: 251 ± 67 ng (candle A) and 275 ± 67 ng (candle B) compared to the PM_2.5 samples where PU was not detected, suggesting that the candle particle filters were contaminated with PU in the sampling set-up. Background contamination was not accounted for in recovery calculations as they were below 15 % of the recovered mass for PVC, below 8 % for PS, PP, and PE, and not detected for PA and PET.

The candle particle samples mainly contain EC with only a minor fraction of OC. The PM_2.5 filters have higher aerosol mass loadings than the candle filters. The chemical composition of the urban aerosol matrix is not characterized for the specific samples, but in general ambient aerosol contains a larger fraction of organic compounds and inorganic salts (in particular ammonium sulfate and ammonium nitrate) than candle particles.

In the candle particle samples, good recoveries (70–120 %, see Table S3) are obtained for DPS (D-styrene and D-trimer) and PS for candle A, and for DPS (D-styrene), PP, and PA for candle B. Lower recoveries (<70 %) are found for PU, PVC, and PET in both candle filters, as well as for PS, PE, and DPS (D-trimer) in candle B. Poor recoveries of PE are found for both samples due to the large uncertainty (relative standard deviation>30 %). Generally, lower recoveries are found for candle B compared to candle A, suggesting that the content of EC and inorganic salts in the matrix influences the analysis of these polymers, except for PVC and DPS (D-styrene). The PA and PP masses are overestimated in candle A (>120 %), suggesting that caprolactam forms from secondary reactions with species in the matrix, as PA was below the LOQ of 3 ± 1 ng in the non-spiked sample (Table S2). For PP, the overestimation could be partly caused by the background level of 21 ± 77 ng in the non-spiked sample (see Table S2).

In the urban PM_2.5 samples the recovery is good (>70 %) for PP and DPS (D-trimer) for both the 3-day (low aerosol loading) and 7-day (high aerosol loadings) filters. PS and PA are well recovered in the 3-day sample (>80 %), but not in the 7-day sample (<40 %), while PVC is well recovered in the 7-day sample (88 %) but not the 3-day sample (40 %). Poor recoveries (<42 %) are found for PU, and PET in both PM_2.5 samples, and for DPS (D-styrene) recovery is also low (<70 %) in both samples. The recoveries of PE (>70 %) are associated with large uncertainties (RSD>30 %) for both PM_2.5 samples, hence PE is poorly recovered, as seen for the candle particle samples.

4. Conclusion

This study presents a method for direct analysis of mixtures of synthetic polymers in atmospheric aerosol samples using py-GC-MS. Seven common polymers were investigated: PE, PP, PVC, PET, PU, PS, and PA6, along with deuterated PS (D8-PS) as an internal standard. Calibration curves were determined in the 10–1000 ng range showing good linear responses for PE, PP, PA, PU, PET, and DPS, and a quadratic response for PS and PVC. The lowest LOD and LOQ were obtained for PVC, PS, and PA (LOD<1 ng, LOQ=3 ± 1 ng), and the highest for PU (LOD=25 ± 14, ng, LOQ=84 ± 46 ng).

The method performance was tested for aerosol matrices of aerosol particles from candle burning and urban PM_2.5. Generally, higher recoveries were obtained for the candle particle filters compared to the PM_2.5 filters, possibly due to the higher mass loading on the PM_2.5 filters as well as differences in reactivity between polymer pyrolysis products and matrix. Larger matrix effects were observed for high aerosol mass loadings for both the candle and PM_2.5 matrix samples for most polymers, except PVC (supposedly due to the formation pathway of naphthalene and interactions with matrix species). We found that satisfactory quantification of 600 ng of PS, PP, and PA could be obtained for low aerosol mass loadings of urban PM_2.5, while PE, PU, PET, and PVC were underestimated. The method is semi-quantitative for PE, PU, PET, and PVC as it can be used to determine a lower concentration limit of the polymers in aerosol matrices. We suggest that for direct pyrolysis of ambient aerosol particle samples, no more than 10 μg of matrix should be introduced to the pyrolysis unit if the expected MnP concentration is on the nanogram scale.

The use of DPS as IS showed that the D-trimer signal reflects the behavior of the polymers better than the D-styrene signal. The best linear correlation was found for DPS and PS, followed by PP. Further explorations should be carried out to find ISs that better represent other polymers, especially PE, PVC, and PA.

In summary, we present a method for the quantification (PP, PA, and PS) and semi-quantification (PE, PVC, PET, and PU) of common synthetic polymers on the nanogram scale in atmospheric samples with direct py-GC-MS. The method can be used as a tool to estimate concentration levels of atmospheric MnP, which is key in understanding emission patterns, transport pathways, and exposure levels of this emerging contaminant.