PM_2.5 refers to airborne particles with an aerodynamic diameter equal to or less than 2.5 micrometers. From a health perspective, such particles are of great concern due to their small size and toxic constituents. Importantly, when inhaled, PM_2.5 can bypass the collection mechanisms of the upper respiratory tract and reach the gas exchange region in the lower lung and potentially enter the bloodstream (Feng et al., 2016; Martins and Graça, 2018). Exposure to ambient PM_2.5, especially those containing heavy metals from fossil fuel combustion, is one of the leading risk factors for disease burden, including respiratory, cardiovascular, metabolic, and neurological disorders (Jo et al., 2017; Ning et al., 2019; Peng et al., 2008; Polezer et al., 2018; Power et al., 2015; Roberts et al., 2019; Wang et al., 2017; Xie et al., 2019; Yan et al., 2020, Maciejczyk, et al, 2021). These particles have been associated with 3 million premature deaths (Lelieveld et al., 2015) and over 2.7 million preterm births per year, globally (Ghosh et al., 2021; Hamra et al., 2014; Huang et al., 2017).

A recent study evaluated PM_2.5 concentration in 78 subway stations in the Northeast US, including Boston, New York, Philadelphia, Washington DC, and found the average PM_2.5 concentration on underground stations and on-trains was 315 μg/m³ and 194 μg/m³, respectively. On average, particles in the sampled underground stations were found to be 52% iron and 31% carbon by mass (Luglio et al., 2021). One study of 30 stations in New York City (NYC) found that the concentration of black carbon (BC) in the underground stations was 2 to 7 times higher than street level values, and the PM_2.5 concentration was 3.5 to 20 times higher than at street level (Vilcassim et al., 2014). Another study collected air samples from highways, aboveground and underground stations, inside train cars, urban street-sides, and parks. PM_2.5 mass concentration was the highest on the subway platforms among all sampled categories (Wang and Oliver Gao, 2011). Furthermore, an earlier study examined teenagers’ exposure to iron, manganese, and chromium with personal sampling techniques and identified the NYC subway system as their primary source of exposure to these chemical constituents (Chillrud et al., 2004). The NYC Subway, one of the largest subway systems in the world with 472 stations, accommodates approximately 5.5 million daily riders (MTA, 2020) and is reported to be heavily polluted with alarmingly high concentrations of PM_2.5 rich in iron, both on-platforms and in-train cars (Luglio et al., 2021; Vilcassim et al., 2014). However, to the best of our knowledge, studies that have been conducted so far rely on small samples, hence are unable to provide a complete characterization of the particulate matter exposure in the NYC subway systems. The study being reported here builds on our two previous studies (Luglio et al., 2021; Vilcassim et al., 2014), and extends the findings with a large number of samples of on-train and on-platform PM_2.5 measurements with two objectives. (1) investigating both on-platform (total = 333) and in-train car PM_2.5 concentrations for the entirety of nine subway lines, namely the #1, #3, #5, #6, B, C, F, M, and R lines. This large number of samples will provide a more reasonable understanding of PM_2.5 exposure of subway riders on a typical weekday. (2) identifying the chemical composition of PM_2.5 on-board and on-platform for these different subway lines.

Studies of subway particulate matter have also been carried out in a number of cities. The Los Angeles (LA) metro system was found to be less polluted than NYC’s; the average concentration on the underground stations in LA ranged from 9 to 130 μg/m³ (Kam et al., 2011). The Mexico City subway system follows a similar pattern of concentrations of PM_2.5 as the one in LA, ranging between 60 μg/m³ and 93 μg/m³ (Mugica-Álvarez et al., 2012). However, because of the variation in measurement techniques and instruments, the results from the different studies are not always directly comparable (Nieuwenhuijsen et al., 2007). In Asia, one study sampled selected stations from three subway lines in Shanghai and found the average concentration of PM_2.5 ranged between 82.5 and 178 μg/m³ (Guo et al., 2014). Similar concentration levels were found in the subway system of Tianjin (B. Q. Wang et al., 2016), Seoul (Park and Ha, 2008), Tehran (Kamani et al., 2014), and Taipei (Cheng et al., 2008). One study sampled four stations in Beijing and found the concentration of underground stations varied significantly (56 μg/m³ – 291μg/m³) on different days of the week (Pan et al., 2019). Furthermore, several studies have been carried out in European cities. For example, one study found that the deepest stations (>20 m under the ground) of the central subway line in London had significantly higher concentrations (~500 μg/m³) than the stations located from 0 to 10 meters below the ground. Time-series measurements of one station in London for several days also exhibited a high correlation between PM_2.5 concentration and train frequency (Smith et al., 2020). Thus, it is likely that many varying factors contribute to the range of PM_2.5 concentrations measured in different subway systems throughout the globe.

The composition of subway particles is quite different from the PM composition in the urban outdoor ambient air, where particles created by the combustion of fossil fuels comprise the majority by mass (Davidson et al., 2007; Squizzato et al., 2018). Similar to the findings in NYC, some of the above subway studies included the analysis of the chemical composition of the particles in the subway air, concluding that iron (Fe) particles were the dominant element, accounting for over 40% by mass, where other transition metals include Cu, Ba, Cr, Si, Mn, and Zn (Guo et al., 2014; Loxham et al., 2013; Martins et al., 2016b; Moreno et al., 2015; Querol et al., 2012; Salma et al., 2007; Smith et al., 2020). The metal-rich particles in the subway systems are mostly generated from wear and friction processes at the intersection of rail–wheel–brake (Jung et al., 2012). Additionally, ambient air, ballast, electrification, diesel engines, and construction work in the tunnels could also be potential sources for the mineral particles (Aarnio et al., 2005). Other than the composition, contributing factors for the high particulate matter concentrations in subways include train frequency and station depth, ventilation system, ambient air quality, year of construction, etc. (Martins et al., 2016a; Wen et al., 2020).

While the adverse health impacts of exposure to ambient particles generated by fossil fuel combustion are well recognized (Maciejczyk et al., 2021), there is more uncertainty regarding exposure to iron-rich particulate matter, such as those experienced in subways. An available guidance level for exposure to iron-rich particulate matter is set by the Occupational Safety and Health Administration (OSHA), where the occupational exposure standard for welding fume (enriched with iron oxide) is limited to 5000 μg/m³ for an eight-hour work shift (NIOSH, 2011). Because subway PM concentrations would be below the occupational exposure limit set by OSHA, one might dismiss the health risks among subway workers and commuters. However, metals abundant in underground air have significant potential to contribute to oxidative stress by generating reactive oxygen species (ROS) (Figueroa-Lara et al., 2019). The co-presence of combustion-related sulfate with iron is of concern, as it is known to increase the solubility and oxidative stress from fine particle metal exposures (Fang et al., 2017). One study found that underground Mn, Zn, Ba, and especially Cu have more oxidative potential than Fe (Moreno et al., 2017a). The particles from a unit of magnetite (Fe₃O₄) rich air from the Stockholm subway is expected to be in the range of 40–80 times more genotoxic and 20–40 times more potent to induce oxidative stress as compared to air from a busy urban street (Karlsson et al., 2004). Another study found that PM sampled in the Paris Metro subway stations induced oxidative stress in macrophages in vitro, and lung homogenates in vivo (Bachoual et al., 2007). Additional research has found that employees who are highly exposed to airborne particles in the Stockholm underground system have higher concentrations of risk markers for cardiovascular diseases than employees with lower exposure (Bigert et al., 2008). Thus, there is burgeoning evidence that subway particles may have adverse effects on commuters and transit workers. Nonetheless, current studies rely on a limited number of samples that are inadequate for population-level exposure analysis. This study addresses the limitation of the existing body of knowledge by collecting real-time and gravimetric data on PM_2.5 concentrations and composition at a scale that has not been conducted before for New York City. The study provides a statistically informed place-based understanding of population exposure to air pollutants on public transit for a typical weekday and enables a number of areas for further research, particularly addressing the at-risk population.

Concerning determinants for on-train PM_2.5 concentrations, past studies have found that particles penetrate the train cabin from the platform when the door opens, making the on-train concentration rise (J. Wang et al., 2016). While this explains how concentrations may change as trains enter a station, it does not explain how they change when riding through tunnels. Understanding how air circulates within the subway car is essential to analyze these dynamics. It is reported that at any given time, 75% of the air inside the NYC subway cabin is recycled, and the remaining 25% is pulled from outside (Gröndahl et al., 2020). The inside cabin air is constantly pulled, cooled, and filtered during the recycling process, while being simultaneously mixed with filtered outside air, before being pushed back into the cabin. With this process, the inside train car air theoretically gets entirely replaced by outside subway air every three to four minutes. Often, subway cars use MERV-7 category filters to filter out the particles (Gröndahl et al., 2020). Experiments identifying PM_2.5 filter efficiency found that these MERV-7 category filters only remove between 2% and 21% of particles, whereas higher category filters such as MERV-16 and HEPA can remove upwards of 96% – 100% (Azimi et al., 2014; Zhao et al., 2015). Therefore, the filtrated air that is dispensed into the cabin is likely to possess a significant concentration of tunnel particles. Focusing on real-time measurements can provide insight into how the particle concentration changes in the subway train car while traveling through a tunnel.

Regarding on-platform concentrations, possible strategies to reduce PM_2.5 concentrations on subway platforms include the installation of platform screen doors (PSD), filtration devices, improved ventilation, and increased tunnel cleaning, as demonstrated in other studies (Chang et al., 2021). Since the particles are likely to be generated mostly by the friction between rail, wheel, and brakes, replacing the metal component of the friction surface, such as introducing a rubber-tired rail system, could reduce the production of heavy metal particles (Cartenì and Cascetta, 2018). In NYC, most subway stations and tunnels lie deep underground, making underground air and ambient exchange difficult. In such cases, forced mechanical tunnel ventilation could effectively reduce PM_2.5 concentrations (Moreno et al., 2014). Again, when a train approaches the station, piston wind is generated, which pushes air and particles from the tunnels to the platform (J. Wang et al., 2016). This piston wind is considered the main driver of high PM concentrations on platforms (He et al., 2018). Separating rail tracks from the platform with barriers such as PSDs could be an effective measure for controlling particle concentration on the platforms as they can block the tunnel wind from entering the platforms (Han et al., 2015; Martins et al., 2016b, 2015; Moreno et al., 2017b). Importantly, although PSDs can reduce on-platform pollution, a study suggested that they may increase on-train particle concentrations (Son et al., 2014). To mitigate this potential effect, the concentration inside the train cabin can be reduced with filters or subway cabin purifiers, as discussed above. Ultimately, however, reductions in the on-platform concentrations will be required to minimize in-cabin concentrations.

The alarmingly high concentrations of PM_2.5 observed in this study raise concerns about the potential impact on a subway rider or worker’s cardiopulmonary health. A few studies, conducted around the world, have examining human health endpoints after subway exposures, and found mixed results (Bigert et al., 2008; Klepczyńska Nyström et al., 2010; Liu et al., 2015; Sauvain et al., 2022). The subway system in New York City, however, is among the most polluted in the world (Xu and Hao, 2017), and generally exceeds those observed in these other studies. As such, it is important to investigate the potential health effects of commuting or working in this subway environment with extreme PM_2.5 exposure.

Although the study presented here offers a representation of the particulate matter scenario in the NYC subway system for a typical weekday, it is important to note that the study has certain limitations. While on-train and on-platform data were collected within the same season, the overall impact of seasonal changes on subway air quality is still unknown. Additionally, the concentration of particulate matter can vary within a day, with train frequency being a major determinant of subway PM concentration. Although we were not able to capture the temporal variability of concentration within a day, we did perform on-train measurements during peak hours to minimize bias and included the sampling time for all station measurements (Tables S1–S9) as a part of comprehensive data reporting. The results of this study may be limited by the shorter sampling time used to sample each platform, and capturing a larger number of stations with longer sampling times would require a more extensive campaign that should be considered for future work.