Results of the 2nd comparison exercise for EU National Air Quality Reference Laboratories (AQUILA) for TC, OC and EC measurements (2011)

The EC-JRC European Reference Laboratory for Air Pollution (ERLAP) has organized an inter-laboratory comparison for the measurement of total carbon (TC), elemental carbon (EC) and organic carbon (OC) in particulate matter collected on filters. To this comparison seventeen European Union National Reference Laboratories for air quality or delegated organizations participated, all using thermal optical analysis with the same analyzer (Sunset Lab off-line carbon analyzer). The aim of this comparison was to evaluate the performances of participants but also to study the effects of applying different thermal protocols, i.e. NIOSH and EUSAAR_2 protocols, currently in use in Europe for such analysis. In absence of a general consensus by the scientific community on the definition of a reference material for EC and, thus, of a standard analytical method, method performances [ISO5725-2] and laboratory performances [ISO 13528:2005(E)] were evaluated for TC and EC/TC ratio in the present comparison exercise. For TC, repeatability and reproducibility relative standard deviations ranged from 2% to 6% (sr = 0.017 × m + 0.227) and from 5% to 11% (sR = 0.038 × m + 0.389), respectively. For EC/TC ratio, repeatability and reproducibility relative standard deviations ranged from 2% to 10% and from 8% to 35%, respectively for the NIOSH-like protocol, and from 2% to 14% and from 4% to 19%, respectively for the EUSAAR protocol. (No satisfactory dependence was found upon EC/TC ratio). Furthermore, based on z-scores, three outliers were identified in the TC database when applying as standard deviation for proficiency assessment, σ*, that one calculated from data obtained in a round of a proficiency testing scheme. These outliers would also not comply with the DQO (i.e. expanded uncertainty, with a coverage factor of 2) of 25%, as in the EU Directive 2008/50/EC for PM at its limit value of 50 μg m-3. Laboratory performances were evaluated for EC/TC ratio, separately on the two data subsets from the NIOSH and EUSAAR_2 protocols using as σ* a common level of performance (i.e. 15%) that the inter-laboratory comparison coordinator would wish participants to achieve. Under this condition, four outliers were identified in the subset of data from the NIOSH-like protocol and one outlier in the subset of data from the EUSAAR_2 protocol.