There is increasing concern about the impact of ambient Particulate Matter (PM) exposure on human health. Existing ambient air quality monitoring networks, because of their sparseness, are not able to capture the spatial-temporal variation of ambient PM concentration. Recent studies have shown that although low-cost PM monitors, such as Optical Particles Counters (OPCs), can be utilised for both personal exposure and high-density sensor network applications, these sensors can be subject to several limitations such as relative humidity (RH) interferences, the inability to measure smaller (e.g. ultrafine) particles and the assumption made about the constant aerosol density value used in the conversion from number concentration to mass concentration.

Water can condense on aerosol particles resulting, if uncorrected for, in particle size growth and thus overestimation of the measured PM concentration in a manner which depends on both RH and the chemical composition of particles. In contrast, failing to account for particles below the sensors’ size detection limit leads to potentially large underestimations of PM concentrations. To account for changes in the aerosol density values assumed by OPCs, the PM readings calculated from the particle size and concentration measurements require the application of an additional, often empirical, factor. Not accounting for this factor, results in a poor agreement between PM measurements from an OPC and a reference gravimetric instrument.

To address these issues, this work describes the development, application and assessment of a three-stage correction process which allows low-cost sensors to provide quantitative PM measurements.

The first step is to account for the particle hygroscopic growth under high RH conditions. This led to the development of two correction methodologies: 1) a particle size distribution correction algorithm (based on κ-Köhler theory) which corrects for the particle hygroscopic growth after the measurements, and 2) a prototype particle drying system (consisting of a heated inlet) which dries the particles before measuring their size. Results from field studies across winter and summer are presented to show the feasibility and accuracy of the two methodologies.

The application of the correction algorithm resulted in a significant improvement in the determination of the correct particle size and concentration, with the overestimation of PM measurements (calculated in the size range of the low-cost sensor) reduced from a factor of 4.60±0.10 to 1.00±0.10. Similarly, the implementation of the particle drying system provided comparable results with the overestimation of PM measurements reduced from a factor of 1.90±0.20 to 0.90±0.10. However, the performance of the correction algorithm is dependent on the correct assumption of the particle chemical composition as opposed to the application of a heated tube on the sensor inlet which ensures drying of the particles.

The second step is to estimate and include the contribution of particles below the low-cost sensor size detection limit in the calculation of PM concentrations. An algorithm to determine the concentration of particles below 300 nm diameter based on the particle size distribution of reference measurements between the low-cost sensor and the reference instrument size detection limits was developed. The application of the estimation algorithm resulted in a reduction for the underestimation of the total PM concentrations from a factor of 0.40±0.10 to a factor of 1.10±0.10.

Finally, optical reference instruments try to exploit a semi-empirical correction factor to scale the calculated PM values to match the reference gravimetric methods measurements. This work investigated the possibility of applying a similar correction factor to low-cost sensor measurements. It was shown that, by exploiting the ratio between PM2.5 and PM1 measurements, an equivalent correction factor for low-cost sensors can be derived, thus enabling the Alphasense OPC-N2 PM measurement to be accurate within 25% relative to the certified reference equivalent data.

This work shows that, when appropriate corrections are applied, portable low-cost sensor can provide PM measurements which are comparable with readings from reference standard instruments. Finally, a conceptual design to exploit the dependence of PM on RH for chemical composition derivation is also included. Even though the sensor used in this work is the Alphasense OPC-N2, the presented methodologies are generally applicable to all OPC-based PM sensors.