Intercomparison of six ambient [CH2O] measurement techniques

Abstract

From May 29 to June 3, 1995 a blind intercomparison of six ambient formaldehyde measurement techniques took place at a field site near the National Center for Atmospheric Research in Boulder, Colorado. The continuous measurement methods intercompared were tunable diode laser absorption spectroscopy, (TDLAS); coil/2,4‐dinitrophenylhydrazine, (CDNPH); 1,3‐cyclohexanedione‐diffusion scrubber (CHDDS); and the coil enzyme method (CENZ). In addition, two different cartridge methods were compared: silica gel‐2,4‐dinitrophenylhydrazine (DPNH) systems and a C‐18‐DNPH system. The intercomparison was conducted with spiked zero air (part 1) and ambient air (part 2). The CH2O standards for part 1 were calibrated by several independent methods and delivered to participants via a common glass manifold with potential trace gas interférants common to ambient air (O3, SO2, NO2, isoprene, H2O). The TDLAS system was used to confirm the absolute accuracy of the standards and served as a mission reference for part 1. The ambient phase lasted 44 hours with all participants sampling from a common glass tower. Differences between the ambient [CH2O] observed by the TDLAS and the other continuous methods were significant in some cases. For matched ambient measurement times the average ratios (±1σ) [CH2O]measured/[CH2O]TDLAS were: 0.89±0.12 (CDNPH); 1.30±0.02 (CHDDS); 0.63±0.03 (CENZ). The methods showed similar variations but different absolute values and the divergences appeared to result largely from calibration differences (no gas phase standards were used by groups other than NCAR). When the regressions of the participant [CH2O] values versus the TDLAS values, (measured in part 1), were used to normalize all of the results to the common gas phase standards of the NCAR group, the average ratios (±1σ), [CH2O]corrected/[CH2O]TDLAS for the first measurement period were much closer to unity: 1.04±0.14 (CDNPH), 1.00±0.11 (CHDDS), and 0.82±0.08 (CENZ). With the continuous methods used here, no unequivocal interferences were seen when SO2, NO2, O3, and isoprene impurities were added to prepared mixtures or when these were present in ambient air. The measurements with the C‐18 DNPH (no O3 scrubber) and silica gel DNPH cartridges (with O3 scrubber) showed a reasonable correlation with the TDLAS measurements, although the results from the silica cartridges were about a factor of two below the standards in the spike experiments and about 35% below in the ambient measurements. Using the NCAR gas‐phase spike data to calibrate the response of the silica gel cartridges in the ambient studies, the results are the same within statistical uncertainty. When the same gas phase calibration was used with the C‐18 cartridges, the results showed a positive bias of about 35%, presumably reflecting a positive ozone interference in this case (no ozone scrubber used). The silica DNPH cartridge results from the second participant were highly scattered and showed no significant correlation with the TDLAS measurements.