10th International Aerosol Conference September 2 - September 7, 2018 America's Center Convention Complex St. Louis, Missouri, USA
Abstract View
On the Inherent Measurement Uncertainty of Miniaturized PM Sensors
PAUL MAIERHOFER, Georg Röhrer, Alexander Bergmann, Graz University of Technology
Abstract Number: 453 Working Group: Aerosol Physics
Abstract Serious adverse health effects of PM on the human body raise a public desire for highly integrated, cost-effective and easy-to-use PM sensors. As the chase for smaller and smaller PM sensors is currently at an all-time high, it is not only fruitful but also necessary to have a closer look at limitations we encounter. Despite all the evident benefits miniaturized sensors are bringing, there is an inherent drawback in measurement accuracy, which is closely related to the discrete nature of particulates suspended in air. The miniaturization of those devices not only leads to smaller footprints of the devices themselves but also to smaller samples of air being measured. Even if a perfect measurement system is assumed which is able to correctly sense all particles in mass in any volume of air that is drawn through it, there is an inherent uncertainty in assigning a PM2.5 value representative for the ambient. This stems from the fact that particles are discrete and therefore stochastically distributed in the air. Consequently, there is an uncertainty according to counting statistics, as the number of investigated particles in small samples of air is also low. Additionally, the distribution of particle sizes and masses is not ideally captured by an arbitrarily small sample size especially since the size distribution of particles extends over orders of magnitudes. This distribution related uncertainty adds to the uncertainty resulting from counting statistics.
We investigated the expected measurement uncertainty for the described system by analytical means leading to the conclusion that both the exact size distribution of particles as well as the sample size have a major impact on the measurement uncertainty. By transformation of variables and use of the central limit theorem, this uncertainty was quantified and evaluated for various situations. Simulations using random numbers generated by a Monte Carlo algorithm were used to back the analytical results.
For an exemplary size distribution of urban aerosol at a concentration of PM2.5 = 25 µg/m3 and a sample size of 1 ml the 1 σ standard deviation is around 37% resulting solely from the fact that the number of particles in the sample is small, disregarding all other potential measurement errors.
As a conclusion the miniaturization of PM sensors and the connected miniaturization of microfluidic devices and therefore the sample size, leads to a significant uncertainty of the measurements. The only way to avoid these uncertainties is to avoid small samples e.g. by taking longer measurement times or guaranteeing relatively high flow rates, which seems contrary to miniaturization. Said uncertainty is not negligible, but is to be taken into account when developing future sensors.