Deposition sampling

Atmospheric samples were collected for 24 hours following the U.S. Environmental Protection Agency (EPA) sixday monitoring schedule during two measurement periods: January and May 2007. Samples were collected using a denuder and filter sampler (DFS) consisting of two sampling trains, each with a sizeselective cyclone inlet, two annular denuders and a threestage filter pack. The cyclone inlets had a 50percent cutpoint diameters of 2.5 micrometer and 10 micrommeter, to facilitate calculation of fine (Dp less than 2.5 micrometer) and coarse (Dp 2.5–10 micrometer) particulate nitrogen concentrations. Air then flowed through the annular denuders in series: the first was coated with citric acid to collect ammonia gas (NH3) and the second was coated with sodium carbonate (Na2CO3) to collect nitric acid vapor (HNO3) (Possanzini et al., 1983; Allegrini et al., 1987). Air then flowed into the filter pack, which contained three filters: Teflon filter for particulate nitrate (NO 3 ) and ammonium (NH 4 ), nylon filter for volatilized particulate HNO3, and citricacidimpregnated Teflon filter for volatilized particulate NH3 (Allegrini et al., 1987). Collection of gas phase HNO3 and NH3 using denuders and afterfilters was designed to reduce bias due to volatilization of particulate NH4NO3 from Teflon filters (Hering and Cass, 1999). Particulate NH4NO3 volatilization increases with increasing temperature (Russell et al., 1983). Starting in May 2007, quartz fiber filters (prebaked at 600 degree C for 12 hours) were used in place of Teflon filters so that aerosol carbon could also be measured.

The designed flow through the DFS assemblies was 16.7 l min-1, controlled using a vacuum pump and a critical orifice. The flow rate for each DFS sampling train was measured before and after the start of the sampling period using a calibrated rotameter to be within 10% of the design flow rate. The flow rate for each sample was taken to be the average of the flows measured before and after each sampling period. The DFS was mounted in a weatherproof wooden box with access holes for the sizeselective cyclones and two cooling fans. An automatic timer was used to control sample duration and an hour counter was used to measure the actual sample duration to within 0.1 h..

Annular denuder and filter samples were extracted within 48 hours of collection. Each annular denuder was extracted with two sequential washings with 5 ml of nanopure water. Filters were extracted by sonication in 10 ml of nanopure water. Sample extracts were then analyzed for NO 3 and HNO3 (as nitrate) using automated colorimetry on a flow injection analyzer (QC 8000, Lachat Instruments, Loveland, CO) and for NH‚ 4 and NH3 (as ammonium) using the 118 indophenol method on an autoanalyzer (TrAAcs 800, BranLuebbe, Inc., Delavan, WI) (Allegrini et al., 1987; Chow, 1995). Quartz fiber filters were first cut in half: one half was extracted as above and the other half was analyzed for total carbon using flash combustion elemental analysis (PE 2400, PerkinElmer, Inc., Waltham, MA). The extracts from the citricacidcoated annular denuder and citricacidimpregnated filter were neutralized to pH 7 using concentrated NaOH. All samples were stored at 4 degree C between collection and analysis.

Method blanks were prepared for each sampling date during the study. Method blanks were collected by coating clean denuders with the same solution used for that sample date and loading new filters in a filter pack. The denuders and filters were then extracted using the same procedure as normal samples. Method blank extracts were handled, stored, and analyzed in the same manner as normal samples.

Atmospheric concentrations of HNO3, NH3, and particulate NH4NO3 were calculated from the filter and denuder extract concentrations. The mean of the concentrations measured for the fine and coarse size samples from the sodiumcarbonatecoated and citricacidcoated denuders were used to estimate the concentrations of HNO3 and NH3, respectively. Concentrations of particulate NH4 4 were calculated as the sum of the concentrations from Teflon (or quartz fiber) and citricacidimpregnated filters. Concentrations of particulate NO 3 were calculated as the sum of the concentrations from Teflon (or quartz fiber) and nylon filters. Method blanks were subtracted from all samples. Sample concentrations less than the atmospheric detection limit were reported as 0 microggram N m-3.