Evaluating the PurpleAir monitor as an aerosol light scattering instrument

Abstract

Abstract. The Plantower PMS5003 sensors (PMS) used in the PurpleAir monitor PA-II-SD configuration (PA-PMS) are equivalent to cell-reciprocal nephelometers using a 657 nm perpendicularly polarized light source that integrates light scattering from 18 to 166∘. Yearlong field data at the National Oceanic and Atmospheric Administration's (NOAA) Mauna Loa Observatory (MLO) and Boulder Table Mountain (BOS) sites show that the 1 h average of the PA-PMS first size channel, labeled “> 0.3 µm” (“CH1”), is highly correlated with submicrometer aerosol scattering coefficients at the 550 and 700 nm wavelengths measured by the TSI 3563 integrating nephelometer, from 0.4 to 500 Mm−1. This corresponds to an hourly average submicrometer aerosol mass concentration of approximately 0.2 to 200 µg m−3. A physical–optical model of the PMS is developed to estimate light intensity on the photodiode, accounting for angular truncation of the volume scattering function as a function of particle size. The model predicts that the PMS response to particles > 0.3 µm decreases relative to an ideal nephelometer by about 75 % for particle diameters ≥ 1.0 µm. This is a result of using a laser that is polarized, the angular truncation of the scattered light, and particle losses (e.g., due to aspiration) before reaching the laser. It is shown that CH1 is linearly proportional to the model-predicted intensity of the light scattered by particles in the PMS laser to its photodiode over 4 orders of magnitude. This is consistent with CH1 being a measure of the scattering coefficient and not the particle number concentration or particulate matter concentration. The model predictions are consistent with data from published laboratory studies which evaluated the PMS against a variety of aerosols. Predictions are then compared with yearlong fine aerosol size distribution and scattering coefficient field data at the BOS site. Field data at BOS confirm the model prediction that the ratio of CH1 to the scattering coefficient would be highest for aerosols with median scattering diameters < 0.3 µm. The PMS detects aerosols smaller than 0.3 µm diameter in proportion to their contribution to the scattering coefficient. The results of this study indicate that the PMS is not an optical particle counter and that its six size fractions are not a meaningful representation of particle size distribution. The relationship between the PMS 1 h average CH1 and bsp1, the scattering coefficient in Mm−1 due to particles below 1 µm aerodynamic diameter, at wavelength 550 nm, is found to be bsp1 = 0.015 ± 2.07 × 10−5 × CH1, for relative humidity below 40 %. The coefficient of determination r2 is 0.97. This suggests that the low-cost and widely used PA monitors can be used to measure and predict the submicron aerosol light scattering coefficient in the mid-visible nearly as well as integrating nephelometers. The effectiveness of the PA-PMS to serve as a PM2.5 mass concentration monitor is due to both the sensor behaving like an imperfect integrating nephelometer and the mass scattering efficiency of ambient PM2.5 aerosols being roughly constant.

An evaluation of the U.S. EPA’s correction equation for Purple Air Sensor data in smoke, dust and wintertime urban pollution events

“…Conversely, oil mist had the largest MMD at 2.9 μm; its PMS5003 PM2.5 averaged only 0.23 times the filter-derived PM2.5. These lab results are consistent with the physical-optical model developed 55 for the PMS5003 by Ouimette et al (2022). The model predicted that the PMS5003 response decreases relative to an ideal nephelometer by about 70-90% for particle diameters ≥1.0 μm.…”

“…We show above that the PAS data, using the standard correction, is substantially under-reporting PM2.5 concentrations during dust events. The next question is whether the PAS data can give some information about dust events (i.e., the presence/absence of dust), despite significant issues with the reported size distribution (Ouimette et al, 2022). To address this question, we calculated the slopes of the PM1 to PM10 mass ratios and the 0.3 µm to 5 220 μm particle counts ratio, both using the PAS data for each event.…”

“…A number of field and laboratory studies have found that the PMS5003 size distributions are not correct. Several studies have reported that the PMS5003 tends to 3 California (Ouimette et al, 2022). So, at present, there remains some ambiguity over how the PAS reported PM2.5 concentrations respond to different aerosol types.…”

“…However the requirement that the A and B channel 0.3 and 5 5 µm counts agree within 30% reduces the number of data points by more than half. This largely reflects the low aerosol numbers in the 5 µm channel and therefore 285 relatively large variability, due to the factors identified in Ouimette et al (2022). It is possible that there is a better dust correction algorithm that could integrate more of the PAS measured parameters.…”

Performance evaluation of the Alphasense OPC-N3 and Plantower PMS5003 sensor in measuring dust events in the Salt Lake Valley, Utah

“…Our results suggest that it may be possible to obtain better resolved spatial estimates of PM10 concentration using a combination of PMS sensors (often publicly available in communities) and measurements of PM2.5 and PM10, such as those provided by FEMs, research-grade instrumentation, or the OPC-N3. low-cost PM sensors, which have flow rates on the order of 0.1 LPM (Sayahi et al, 2019;Ouimette et al, 2022;Alphasense Ltd, 2022). The OPC-N3 allows particle counting in 24-size bins for sizes ranging from 0.35-40 µm.…”