Comparison of reference libraries for the detection of tire-derived microplastics (TMPs)

The term microplastics (MPs) refers to plastic particles ≤5 mm in diameter that enter the aquatic environment (Andrady 2011). MPs can be classified as either primary or secondary. Primary MPs are manufactured with diameters ≤5 mm, whereas secondary MPs are derived from plastic debris (Takada 2018). Tire-derived MPs (TMPs) are estimated to account for 60% of all secondary MPs discharged to aquatic environments (Lassen et al. 2015), indicating that terrestrial emissions are important. In addition, it is estimated that 34% of coarse tire wear particles and 30% of coarse brake wear particles are estimated to be deposited in the oceans (Evangeliou et al. 2020). Studies conducted in the Netherlands have shown that approximately 1,800 tonnes of MPs from tire wear are discharged into surface waters each year (National Institute of Public Health and the Environment Netherlands 2016). MPs in the marine environment contain toxic chemicals resulting from the incorporation of antioxidants and flame-retardant additives (Teuten et al. 2009). Furthermore, the toxicity of tire wear particles may differ from the toxicity of tire rubber (Wik & Dave 2009). In fact, 6-PPD (N-(1,3-dimethylbutyl)-N′-phenyl-p-phenylenediamine) quinone derived from tire rubber has been reported to be toxic to coho salmon (Tian et al. 2021). Although some removal is achieved at wastewater treatment plants (Carr et al. 2016; Tanaka et al. 2019), MPs are transported to the ocean via stormwater runoff (Takada 2018). In some countries, stormwater treatment ponds have been installed for urban roads and highways (Liu et al. 2019; Rosso et al. 2022), but the treatment efficiency is lower for smaller particle sizes (Rasmussen et al. 2024). Furthermore, the removal of MPs in wastewater treatment plants is worse in the range of less than 100 μm (Tanaka et al. 2019). Therefore, it is imperative to determine the presence of TMPs on roads as a significant source of environmental pollution.

Micro-Fourier transform infrared (μFTIR) spectroscopy has been widely used to measure MPs due to its informative spectral database (Tagg et al. 2015). Several studies have used FTIR to measure MPs on roads (Okamoto et al. 2019; Roychand & Pramanik 2020; Yukioka et al. 2020; Nishimagi et al. 2023) or in water samples (Leads & Weinstein 2019; Ziajahromi et al. 2020). However, the lack of suitable reference standards led to the failure to detect tire-derived material among MPs (Liu et al. 2019), highlighting the need to develop standards for TMPs (Ziajahromi et al. 2020). In addition, there is a wide range in the number of TMPs reported between studies. Nishimagi et al. (2023) found 2–9 pieces/m² of TMP, while Oyama et al. (2023) reported a range of 40–180 pieces/m² and Yukioka et al. (2020) reported a much smaller range (0.4–4.5 pieces/m²). Although various factors (e.g. road structure) may have contributed to the differences, the use of different reference libraries could have resulted in different detection levels. Some recent studies have used scanning electron microscopy (SEM) (Kang & Kim 2023) or pyrolysis gas chromatography mass spectrometry (py-GCMS) (Barber et al. 2024; Oliveira et al. 2024) to detect TMPs, but they are not directly comparable with other studies due to methodological differences. In addition, these studies only quantified TMPs by weight, which lacks information on particle size and the potential impact of additive leaching. In order to discuss the leaching potential of chemicals into the ecosystem, it is necessary to establish and improve the μFTIR technique to detect TMPs on a particle basis.

Generally, a high-quality, dedicated database is essential for MP detection by μFTIR. Commercial polymer libraries only include standard reference material consisting of white or clear fragments or powders, which are different from plastic particles found in the environment (Frond et al. 2021). Current databases lack reference spectra relevant to modified or weathered polymers, which need to be included for further studies of surface modifications (Huppertsberg & Knepper 2018). The acquisition of standard weathered polymer reference samples or appropriate libraries remain a challenge (Cai et al. 2019). In addition, the use of conventional diamond crystals is difficult for the detection of carbon black-containing materials (Sarma et al. 2018). In addition, material degradation produces different spectra (Lima et al. 2023), which can interfere with the detection process. To overcome these challenges, we created new reference libraries in this study. Specifically, we developed four different reference libraries from different sources for TMP detection, which are (i) 16 rubber or rubber-like materials from the PKJ-STD library, (ii) a library of 7 purchased rubbers, (iii) a library of 4 scrap tires, and (iv) a library of 26 worn tires. Furthermore, the characteristics of these libraries are discussed based on the correlation coefficients with 659 validation particles from the rotary pavement tester.

Tire dust for validation purposes was collected using a rotating pavement tester (see Supplementary Fig. S1), which is designed to test different types of road surfaces under rotating tires. Tire dust is expected to contain tire particles from various sources because various types of tests have been performed on the tester. Tire dust was collected from the tester using a vacuum cleaner as in our previous work (Nishimagi et al. 2023; Oyama et al. 2023; Takagi et al. 2023). To avoid contamination, the nozzle, hose and filter of the vacuum cleaner were changed after each sample collection. Tire dust was collected in a disposable filter in the vacuum cleaner and then transferred to the laboratory in a sealed container. The contents of the filter were transferred and passed through a 4.75 mm stainless steel sieve (by the JIS Z 8801 test sieve standard) and then subjected to further analysis.

Samples from the rotary pavement tester were pre-treated for analysis as described below. Firstly, target particles were separated by density using a 35% potassium iodide solution. After separation, the particles were collected by overflow. A purification procedure was carried out to remove interfering substances by μFTIR analysis. Approximately 100 mL of a 30% hydrogen peroxide (H₂O₂) solution was added to the extracted sample. The organic matter was decomposed at 55 °C for ∼30 min, followed by a settling period of 24 h at room temperature to ensure complete decomposition. The sample was sequentially filtered by a stainless steel mesh filter with a mesh size of 105, 40 or 10 μm (DM-25-105, DM-25-040 or DM-25-010, respectively, Mossfil Co., Ltd, Japan). These filter sizes were decided in reference to previous studies (Kreider et al. 2010; Leads & Weinstein 2019). The number of particles obtained for validation purposes was 370, 199, and 90 for the 10–40 μm, 40–105 μm, and 105 μm–4.75 mm ranges, respectively. Spectra of these particles were compared with reference spectra described in Section 2.4 to calculate correlation coefficients and then assigned to the reference spectrum with the highest correlation coefficient in the reference library.

An infrared spectrometer equipped with an attenuated total reflection accessory and a germanium prism (Spotlight 200i, Perkin Elmer, USA) was used to identify MPs. Prior to analysis, background spectra were measured, and the prism was cleaned with methanol. Analysis was performed in the wave number range of 4,000–750 cm^–1. The use of a germanium prism allowed the observation of black particles. The acquired spectra were compared with libraries for identification purposes using Spectrum software (Perkin Elmer). TMPs were identified by calculating correlation coefficients between the sample and reference spectra by the Spectrum software (Perkin Elmer) and assigned to the spectrum with the highest correlation coefficient in a reference library.

Four reference spectra libraries were used: (i) 16 rubber or rubber-like materials from the PKJ-STD library (Table 1), (ii) a library of 7 purchased rubbers (Table 2), (iii) a library of 4 scrap tires, and (iv) a library of 26 worn tires (spectra for the 26 worn tires library are available in the Supplementary Data). The PKJ-STD library was the original library provided by the FTIR instrument manufacturer. Sixteen rubber or rubber-like materials (listed in Table 1) were selected from this library for identification and used for comparison in Section 3.1. In addition, three custom libraries were created. The 7 purchased rubber library consisted of a collection of spectra from seven purchased rubber samples, as shown in Table 2. The 4 scraped tire library was made from scraped tires provided by a car mechanic. The 26 tire wear library consisted of the pieces taken from the rotary pavement tester but with a diameter greater than 5 mm (as shown in Figure S2). After materials for each library were obtained, pieces of these materials were measured by germanium prism under the μFTIR instrument. The acquired spectra were processed into the instrument as a new library.

Of the 16 reference spectra in the PKJ-STD library, numerous validation particles were assigned to the YA0046 spectrum. The YA0046 spectrum is measured using the attenuated total reflection method with a germanium prism and ethylene propylene diene rubber (EPDM) as the reference material. The median correlation coefficients for the 10–40 μm, 40–105 μm, and 105 μm–4.75 mm particle size ranges were 0.501, 0.507, and 0.589, respectively. These similar values suggest that there are no significant differences between the particle size ranges. The AR0051 spectrum yielded fewer assigned particles than YA0046. The AR0051 spectrum was obtained by attenuated total reflection measurements for EPDM using a diamond prism. These results demonstrated the importance of the prism material type in achieving a good match. Several particles in Group A (Table 1) are pure rubber. Some validation particles were assigned to spectra in Group A, such as AR0050 (chloroprene rubber: CR), AR0052 (nitrile butadiene rubber), and YA0044 (silicone rubber), but the median correlation coefficients were generally <0.4. Only a few particles were assigned to Group B, which includes styrene-containing copolymers/resins. This finding suggests that these polymers are not the main plastic materials in tires. Overall, most of the particles showed a good match to the EPDM measurements made under observation conditions using a germanium prism (YA0046).

For the 10–40 μm range, most particles were assigned to CR, followed by polybutadiene rubber and EPDM. The median correlation coefficients for the 10–40 μm, 40–105 μm, and 105 μm–4.75 mm particle size ranges were 0.503, 0.57, and 0.541 respectively. Little information is available on the composition of tires, but tires are predominantly composed of natural and synthetic rubbers, together with additives such as carbon black. The specimens identified were mainly CR, which is not natural rubber pieces, whereas tires are typically blends of different rubber materials. These results suggest that pure rubber spectra may not be sufficient for tire material identification.

The results showed that most of the validation particles originated from the tire tread rather than the sidewall. This was expected as the tire tread was in direct contact with the road surface. The letters N (no pre-treatment) or Y (pre-treatment) were used to indicate whether or not pre-treatment had been carried out. Unexpectedly, most of the validation particles were assigned to N (no pre-treatment) spectra, implying the need for care to avoid excessive degradation of the reference material. The median correlation coefficients for the tread samples in the four scrap tire libraries were 0.673, 0.708, and 0.681 for the particle size ranges of 10–40 μm, 40–105 μm, and 105 μm–4.75 mm, respectively. These coefficients with the 4 scraped tire library are higher than the number with the 7 purchased library, suggesting that the use of the tire itself is necessary to detect its constituents.

In this study, we examined the spectra of a total of 659 validation particles from the rotary pavement tester and compared their spectra with four reference libraries for the detection of TMPs by μFTIR. The correlation coefficients with these four reference libraries were compared to discuss the characteristics of each library. To the best of our knowledge, there have never seen this kind of attempt for TMP detection. Comparison with the 16 rubber or rubber-like materials from the PKJ-STD library suggested that an identical observation environment, specifically the use of a germanium prism, is necessary to maintain measurement accuracy. The use of a diamond prism is insufficient for the detection of black rubber (Sarma et al. 2018). Therefore, spectra need to be acquired under identical observation conditions with measurement.

The comparison between the 4 scraped tire library and the 26 tire wear library suggests the need for an appropriate scraping process. To obtain a sample from the tread or sidewall of a used tire for the 4 scraped tire library, a knife was used to cut out material. This type of cutting process does not occur on the road; instead, tires wear away to produce debris or dust as they accelerate, decelerate, or negotiate bends. The 26 tire wear library particles were collected on and around a rotary pavement tester, which rotates tires at a specific speed to produce debris or dust by simulating actual driving on a road. Tire wear particles produced by this method have been found to more accurately reflect the characteristics of real tire wear particles on the road. Of the 26 particles in the tire wear library, the most common particle was number 4,537, and the median correlation coefficients were 0.856, 0.881, and 0.902 for the 10–40 μm, 40–105 μm, and 105 μm–4.75 mm particle size ranges, respectively. These high coefficients demonstrated that naturally occurring tire wear particles are more suitable for the detection of TMPs on the road. Further research is awaited to investigate the applicability of this library to other samples.

Four reference libraries were evaluated for their ability in the detection of TMPs. The comparison of these four libraries showed that the 26 tire wear library is the most useful one due to its tire origin, natural wearing process, and the same setup used for material observation. Use of the 26 tire wear library is recommended for further TMP studies.

The authors are also grateful to Nippon Expressway Research Institute Company for the opportunity to sample the rotary pavement tester.

All authors contributed to the study's conception and design. Material preparation, data collection, and analysis were performed by Z.S. Manuscript was written by H.S.

This research was supported by the Environment Research and Technology Development Fund (JPMEERF20205R06) of the Environmental Restoration and Conservation Agency of Japan and Tokyo Metropolitan Government Advanced Research Grant Number (R4-2).

All relevant data are included in the paper or its Supplementary Information.

The authors declare there is no conflict.