Preface

Air pollution sources emit a range of gases and particles. These pollutants encounter different chemical and physical processes in the ambient, resulting in their concentrations varying spatially and temporally over time scales ranging from seconds to days or longer and spatial scales of meters to kilometers (He and Dhaniyala 2012; Brantley et al. 2014; Van den Bossche et al. 2015; Seinfeld and Pandis 2016; Ranasinghe et al. 2016; Apte et al. 2017; Morawska et al. 2018; Li et al. 2018; Solomon et al. 2019). These air pollutants have been shown to impact human health and welfare (WHO 2006; Solomon et al. 2012; IPCC 2013). In the U.S., long-term measurements of EPA’s criteria pollutants (NO2, O3, CO, SO2, Pb, PM2.5, and PM10 [PM refers to particulate matter and here with respect to particles less than 2.5 µm or less than 10 µm aerodynamic diameter]) (https://www.epa.gov/criteria-air-pollutants) have been made at National Ambient Air Quality Standards (NAAQS) sites. Data obtained from these sites, particularly from urban areas, have been used to assess the long-term health impact of air pollutants (40 C.F.R. 58, 2018). However, these sites are spatially limited, with locations chosen based on population and prior incidences of NAAQS exceedance. Also, often, data from these sites have limited temporal resolution, with pollutants such as PM only reported as 24-h average, once every three days. Special field studies are often conducted to provide a wealth of information on pollutant concentrations and variability. However, these efforts are limited in their spatial and temporal range (see, e.g., Crawford and Pickering 2014; Solomon et al. 2008). Operation and maintenance of NAAQS sites require significant human resources, making further expansion of these sites very expensive. To overcome these obstacles, researchers have used recent advances in the development of new air pollution sensors that are small and inexpensive, to obtain data at high spatial resolution and in near real-time (Snyder et al. 2013). These sensors can: inform citizens about the pollutant concentrations that directly impact them, allowing them to better reduce their exposure to air pollution; aid researchers by providing high resolution data to support efforts linking health effects to sources; and help regulators develop improved emissions control reduction strategies and thereby reduce health and welfare risks from air pollution (Kumar et al. 2015).

In recent years, a large number of PM and gas sensors relevant for ambient monitoring have been developed and brought to market. Several studies have evaluated the performance of these sensors in a variety of settings, and some have been deployed in long-term studies at fine spatial scales (Brantley et al. 2014; Van den Bossche et al. 2015; Ranasinghe et al. 2016; Poladori et al. 2017; Papapostolou et al. 2017; Apte et al. 2017; Li et al. 2018; Morawska et al. 2018; AQ-SPEC 2019; EPA 2019a; Solomon et al. 2019; AE 2020; this issue of AS&T). Several publications, a few noted directly above, have detailed sensor development efforts, performance evaluation, and applications in the last decade. The PM sensors are almost exclusively based on light scattering, and several studies have shown that their response is influenced by temperature, relative humidity, particle mass concentration, and composition, the latter affecting particle refractive index and density. The Air Sensors International Conference (ASIC 2018), held in Oakland, CA, 12–14 September, 2018, was developed to “bring together stakeholders from academia, government, communities, and commercial interests to promote and advance air pollution sensors, improve the data quality from these sensors, expand the pollutants measured, and foster community involvement in monitoring air quality.” This dedicated issue of Aerosol Science and Technology (AS&T) and a companion issue in Atmospheric Environment (AE 2020) present selected papers from ASIC. The papers published in this issue include one related to sensor development and evaluation and 6 related to performance evaluation, the latter to better understand measurement uncertainty and account for variables affecting sensor output, such as T, RH, composition, particle size, and concentration.

Amanatidis et al. developed and evaluated a miniaturized radial differential mobility analyzer (DMA), referred to as the Spider DMA, to provide particle size classification in the 10–500 nm size range. The lightweight (∼350 g) sensor has an aerosol inlet geometry that ensures uniform particle distribution into the classifier region. With the novel design, the experimentally obtained transfer functions of the Spider DMA are seen to match theory to within 2%. The well characterized, miniature DMA is a critical building block for high resolution, low-cost measurements of ultrafine particles and nanoparticles.

A popular sensor used in several studies was the Plantower PMS5003 optical PM sensor. This sensor was evaluated by He et al., Magi et al., and Malings et al. using somewhat different approaches. He et al. conducted fundamental laboratory experiments of the standalone sensor using aerosols of different composition, size, and concentrations, and developed a transfer function to characterize sensor performance as a function of these variables. Magi et al. and Malings et al. conducted field experiments comparing the PurpleAir PA-II, which uses two Plantower units for PM sensing, to a Federal Equivalent Method (FEM). Both evaluated the response of the sensor to variations in T and RH with the goal to better match the data from the sensors with the FEM. Additional details of these three studies follow.

In their laboratory experiments, He et al. challenged the standalone Plantower units with dry ammonium sulfate and sodium chloride particles having single-charged equivalent diameters between 0.1 and 0.7 µm. The sensor transfer functions were then validated by comparing predictions of sensor response to polydisperse ammonium sulfate particles with experimentally determined values. The tests with near mono disperse particles showed that the transfer function shapes of the different sensor channels matched that of the aerosol light scattering response, as might be expected with a nephelometer-like sensor. More importantly, the complex sensor response of the different size channels makes it difficult to convert the reported cumulative number concentration measurements to a size-selective value. Additionally, the sensor response is shown to have a complex dependence on particle composition. The physical-property based transfer function may allow researchers to relate and compare measurements made in different ambient conditions where some knowledge of size distribution and compositions may be available.

The sensor field evaluation by Magi et al. occurred in North Carolina over 16 months under conditions of high relative humidity (∼30–90%). The reference sampler was a Met One 1020 Beta Attenuation Monitor (BAM), an FEM that was located at a NAAQS site. A multiple linear regression model, considering sensor dependence on ambient T and RH, was developed to adjust the Purple Air PA-II measurements to approximate the mass concentrations reported by the BAM. After corrections, the authors report that the hourly corrected-PM2.5 data from the PA-II was accurate to within 3–4 µg/m³.

Malings et al. compared the PA-II as well as a Met-One Neighborhood Particulate Monitor (NPM) to a Met One 1020 BAM FEM. They evaluated up to 25 NPMs and 9 PA-IIs in Pittsburgh, PA, at urban background, industrial, traffic, and rural sites over extended periods of time (between 2 months and 294 days depending on the location). Precision of sensors of the same type were mostly below 2.5 µg/m³ with a correlation coefficient (r) usually greater than 0.9. They employed two correction approaches. The first was a physics-based approach that accounted for hygroscopic growth based on particle composition using the range of plausible compositions from EPA’s Chemical Speciation Network (CSN) (EPA 2019b) and mass measurements obtained using a 1020 BAM. Secondly, they developed a fully empirical correction that uses linear or quadratic functions of T and RH based on the same collocation dataset. The physics-based and empirical approaches achieved similar performance after correction, with a mean absolute error of 3–4 µg/m³ in the hourly average with respect to the BAM FEM.

The performance of the SHARP GP2Y1010AU0F PM sensor, which was used as part of a personal air filter test (PAFT) system, was evaluated by Hapidin et al. The SHARP sensor was used to determine the efficiency of the personal air filters tested in this study. The output response of the SHARP sensor was evaluated as a function of flow rate, particle diameter, and PM sources. The SHARP sensor was tested with four sizes of PSL particles (163, 216, 234, and 303 nm) and with burning incense and mosquito repellant, the latter sources with mass median diameter ∼100 nm. Results showed a linear relationship to particle number concentration for both PSL and other aerosol, although slopes and intercepts varied by size and source type.

Kuula et al. evaluated the Pegasor AQ Urban (Pegasor Ltd., Finland), which measures lung deposited surface area (LDSA). LDSA is directly relevant for health studies and this parameter is primarily associated with ‘ultrafine’ particles (<300 nm in aerodynamic diameter), as most particle surface areas are in this size range. Unlike optical sensors that have poor detection efficiencies for ultrafine particles, the AQ Urban is a diffusion charging-based sensor that is sensitive to these small particles and reports a signal value that is related to LDSA. Kuula et al. studied the accuracy and stability of the AQ Urban sensor with field deployment of the unit in three distinctly different urban locations in the Helsinki metropolitan area. The sensor measurements were compared with those made using a differential mobility analyzer (DMA). Sensors were operated for 12 months at the three sites and good agreement was observed between the AQ Urban LDSA and that calculated from the DMA measurements. Results indicated the AQ Urban can be used to assess local combustion emissions, which are difficult to measure with optical sensors, and provide associated LDSA concentrations that are important from a health perspective.

Salimifard et al. used controlled chamber experiments to evaluate four low-cost OPCs (Dylos, Speck, IC Sentinel, and OPC N2) to varying concentrations of biological (e.g., dust mite, pollen, and pet fur) and non-biological (e.g., silica) aerosols found indoors. An AeroTrak OPC was used as a reference method and all units reported particle number concentrations. Sensor responses were evaluated to varying particle type, size, and concentration of the different biological and non-biological aerosols. Of the particle characteristics studied, particle concentration had the largest effect on linearity. A more varied sensor response was observed for non-biological particles than biological particles to particle type and size.

The publications in this special issue demonstrate advances in design, development, and application of air pollution sensors in recent years. The exponential growth in the use of these sensors over the last decade has been driven by their low cost, small size, low power consumption, and portability. The integration of these sensors with advanced communication technologies has enabled both researchers and the general public to monitor and access air quality data in near-real time at spatial and temporal scales not previously realized. This information is beginning to allow individuals to better manage their specific exposures to air pollution and thus improve their quality of life, and is expected to become integral to decision-making by regulators and communities responsible for managing population exposures to air pollutants.

Paul A. Solomon
Independent Consultant in Air Quality, Henderson, Nevada, USA
[email protected]
Suresh Dhaniyala
Bayard D. Clarkson Distinguished Professor and Co-Director of CARES, Mechanical and Aeronautical Engineering, Clarkson University, Potsdam, New York 13699, USA
[email protected]