RESUMO
An inter-comparison of techniques for long-term sampling of atmospheric ammonia (NH3) was conducted with a view to establishing a national network with > 50 sites. Key requirements were for: a low cost system, simplicity and durability to enable a postal exchange with local site operators, a precision of < +/- 20% for monthly sampling at expected NH3 concentrations of 1-2 micrograms m-3, a detection limit sufficient to resolve the small NH3 concentrations (< 0.2 microgram m-3) expected in remote parts of the UK, and a quantitative means to establish quality control. Five sampling methods were compared: A, a commercially available membrane diffusion tube (exposed in triplicate), with membranes removed immediately after sampling; B, the above method, with the membranes left in place until analysis; C, open-ended diffusion tubes (exposed with 4 replicates); D, a new active sampling diffusion denuder system; and E, an active sampling bubbler system. Method D consisted of two 0.1 m acid coated glass denuders in series with sampling at approximately 0.3 l min-1. These methods were deployed at 6 locations in the UK and the Netherlands and compared against reference estimates. Method D was the most precise and sensitive of the techniques compared, with a detection limit of < 0.1 microgram m-3. The bubbler provided a less precise estimate of NH3 concentration, and also suffered several practical drawbacks. The diffusion tubes were found to correlate with the reference at high concentrations (> 3 micrograms m-3), but were less precise and overestimated NH3 at smaller concentrations. Of the passive methods, A was the most precise and C the least precise. On the basis of the results, method D has been implemented in the national network, together with application of method A to explore spatial variability in regions with expected high NH3 concentrations.