An intercomparison of measurement systems for vapor and particulate phase concentrations of formic and acetic acids

Abstract

During June 1986, eight systems for measuring vapor phase and four for measuring particulate phase concentrations of formic acid (HCOOH) and acetic acid (CH_3COOH) were intercompared in central Virginia. HCOOH and CH_3COOH vapors were sampled by condensate, mist, Chromosorb 103 GC resin, NaOH-coated annular denuders, NaOH impregnated quartz filters, K_2CO_3 and Na_2CO_3 impregnated cellulose filters, and Nylasorb membranes. Atmospheric aerosol was collected on Teflon and Nuclepore filters using both hi-vol and lo-vol systems to measure particulate phase concentrations. Samples were collected during 31 discrete day and night intervals of 0.5–2 hour duration over a 4-day period. Performance of the mist chamber and K_2CO_3 impregnated filter techniques were also evaluated using zero air and ambient air spiked with HCOOH_g, CH_3COOH_g, and formaldehyde (CH_2O_g) from permeation sources. Results of this intercomparison show significant systematic and episodic artifacts among many currently deployed measurement systems for HCOOH_g and CH_3COOH_g. The spiking experiments revealed no significant interferences for the mist chamber technique and results generated by the mist chamber and denuder techniques were statistically indistinguishable. The condensate technique showed general agreement with the mist chamber and denuder methods, but episodic bias between these systems was inferred from large and significant differences observed during the first day of sampling. Nylasorb membranes are unacceptable for collecting carboxylic acid vapors as they did not retain HCOOH_g and CH_3COOH_g quantitatively. Strong base impregnated filter and GC resin sampling techniques are prone to large positive interferences apparently resulting, in part, from reactions involving CH_2O_g to generate HCOOH and CH_3COOH subsequent to collection. Significant bias presumably associated with differences in postcollection handling was observed for particulate phase measurements by participating groups. Analytical bias did not contribute significantly to differences in vapor and particulate phase measurements.

Files

Additional details