Technical note: Identifying a performance change in the Plantower PMS 5003 particulate matter sensor

Abstract

Communities, researchers, and citizen scientists have increasingly been deploying low-cost particulate matter (PM) sensors to understand pollution levels and personal pollution exposure. Currently, tens of thousands of these low-cost sensors have been deployed throughout the world. One of the most commonly used sensors is the Plantower PMS 5003. In 2021, PurpleAir (one of the largest sensor networks) noted a change in the performance of their PM sensors that could be identified by the ratio of PM counts in two size bins (≥0.3 μm and ≥0.5 μm). This study confirmed that in June 2021, an updated version of the Plantower PMS 5003 (PMS-U) was deployed between approximately June 2021 and January 2022. Our laboratory study revealed that the PMS-Us PM2.5 concentrations were biased low for raw concentrations (CF = 1) below 16 μg/m3. Our field study also suggests that this bias translates into an approximately 3 μg/m3 lower PM2.5 concentrations (raw) over the long term, and between March to May 2023, more than 10% of outdoor sensors in the PurpleAir network are PMS-Us. This level of bias would not affect some uses of low-cost sensor networks, such as during wildfires. However, this bias could be important for health or environmental justice studies. This study also developed and evaluated a correction method for PMS-U sensors to reduce the observed bias in PMS-U sensors.