Systematic biases in measurement of urban nitrogen dioxide using passive diffusion samplers

Measurement of nitrogen dioxide using passive diffusion tube over 22 months in Cambridge, U.K. are analysed as a function of sampler exposure time, an d compared with NO2 concentrations obtained from a co-located chemiluminesc ence analyser. The average ratios of passive sampler to analyser NO2 at a c ity centre site (mean NO2 concentration 22 ppb) are 1.27 (n = 22), 1.16 (n = 34) and 1.11 (n = 7) for exposures of 1, 2 and 4-weeks, respectively. Mod elling the generation of extra NO2 arising from chemical reaction between c o-diffusing NO and O-3 in the tube gave a ratio (modelled/measured) of 1.31 for 1-week exposures. Such overestimation is greatest when NO2 constitutes , on average, about half of total NOx (= NO + NO2) at the monitoring locali ty. Although 4-week exposures gave concentrations which were not significan tly different from analyser NO2, there was no correlation between the datas ets. At both the city-centre site and another semi-rural site (mean NO2 con centration 11 ppb) the average of the aggregate of four consecutive 1-week sampler exposures or of two consecutive 2-week sampler exposures was system atically greater than for a single 4-week exposure. The results indicate tw o independent and opposing systematic biases in measurement of NO2 by passi ve diffusion sampler: an exposure-time independent chemical overestimation with magnitude determined by local relative concentrations of NO and O-3 to NO2, and an exposure-time dependent reduction in sampling efficiency. The impact of these and other potential sources of systematic bias on the appli cation of passive diffusion tubes for assessing ambient concentrations of N O2 in short (1-week) or long (4-week) exposures are discussed in detail.