Search

Low-cost airborne particle sensors are gaining attention for monitoring human exposure to indoor particulate matter. This study aimed to establish the concentrations at which these commercially available sensors can be expected to report accurate concentrations. We exposed five types of commercial integrated devices and three types of "bare" low-cost particle sensors to a range of concentrations generated by three different sources. We propose definitions of upper and lower bounds of functional range based on the relationship between a given sensor's output and that of a reference instrument during a laboratory experiment. Experiments show that the lower bound can range from approximately 3 to 15 μg/m 3 . At greater concentrations, sensor output deviates from linearity at approximately 300-3000 μg/m 3 . We also conducted a simulation campaign to analyze the effect of this limitation on functional range on the accuracy of exposure readings given by these devices. We estimate that the upper bound results in minimal inaccuracy in exposure quantification, and the lower bound can result in as much as a 50% error in approximately 10% of US homes., (© 2019 John Wiley & Sons A/S. Published by John Wiley & Sons Ltd.)

Many consumer-grade low-cost particle sensors have recently entered the market, potentially providing a means of inexpensively monitoring indoor air quality and eventually being used to control building systems. In this study, we evaluate several such products: three bare sensors (BS) and five integrated devices (ID) containing one or more of these sensors and additional features. We collocated the sensors with reference instruments in a real indoor environment for two and half months. During this time, we conducted 27 particle emission events of several different types (incense, candle, toast, spray, cooking, humidifier and open window). All ID and BS exhibited an ability to respond to different PM sources, although the magnitude of their responses varied. The ratio of the time-integrated signals and peaks of the sensor signals to the reference instruments during each event varied from close to zero to around approximately three. Similarly, correlations between one-minute low-cost sensor data and similar reference instrument data varied widely among the sensors and devices tested, with R2 values between near zero and near one observed. [ABSTRACT FROM AUTHOR]