Intercalibration of power meters used in the measurement of O(1D) quantum yields
(Mary K. Gilles and A. R. Ravishankara, NOAA Aeronomy Lab, Boulder CO 80305)
One of the key factors for accurately determining the quantum yields is the accurate measurement of the number of photons that were used to dissociate ozone. In all the experiments discussed here, the absolute photon fluence, i.e., the number of photons per cm2 per pulse, was not needed. However, to determine the relative dependence of the quantum yield as a function of wavelength, the relative fluence at various wavelengths were needed. The laser fluence was measured in all experiments using laser power meters. In almost all cases, they were the whole volume absorber type, where the laser energy is converted to heat and the evolved heat is measured. The following measurements were made to test the linearity of the power meters, i.e., the variation of the response with the laser fluence, and relative sensitivity of the power meters at various wavelengths used in this study. In the latter case, the laser meters from NOAA group (Talukdar et al., 1998) and the Japanese group (Takahashi et al., 1996a, 1998a) were inter-compared at four wavelengths.
The power meter used by the NOAA group was calibrated at 248 nm relative to the NIST standard. This calibration showed that the power meter was accurate to better than 3% at 9 mJ pulse-1 (close to the values used in NOAA studies) and to better than ~7% at 0.5 mJ pulse-1.
The Japanese power meter was compared to the NOAA power meter, where the energy per pulse were measured using each of the power meters. An excimer laser was used for generating 248, 308, and 351 nm, and a Nd:YAG pumped dye laser was used to generate 282 nm. The pulses from these lasers were passed through an aperture to generate a well-defined laser beam. This beam impinged on the power meter, which were mounted on positive position mounts. Each power meter was equipped with the same kind of mount and could be switched into an exactly reproducible position using the positive position mounts. The power meters were allowed to warm up for an hour, or so, and then they were “zeroed” to set the zero value. The fluence of the laser pulses was varied from ~0.5 to 25 mJ pulse-1. At each of the fluence setting, the power meters were positioned in front of the aperture and the energy was measured. The order in which the measurements were made between the two power meters was randomly altered. Also, the fluence values were altered randomly, rather than ramping it up or down monotonically. Such measurements were carried out at the four wavelengths noted above.
Figure A1 shows a plot of the fluence measured by the Japanese power meter as a function of that measured by the NOAA power meter in back-to-back measurements at 351 nm. The linear dependence of the Figure 1A, coupled with the linearity of the NOAA power meter in the NIST standard calibration, shows that the Japanese power meter also was linear with the laser fluence. If the two power meters agreed exactly, all the points should fall on the 1:1 line, which is shown in the Figure A1. The slope of the line in Figure A1 is 0.894 ± 0.030 and the intercept is essentially zero, 0.074 ± 0.36; the quoted errors are 2σ values from a linear least squares analysis of the data. Therefore, it is clear that the relative response of the Japanese and NOAA power meters differed by roughly 10% at this wavelength. Note that this difference in not the important factor, but the variation of this relative response as a function of the wavelength is the important issue.
Measurements described above for 351 nm were repeated at 248, 282, and 308 nm. The measured slope of plots such as Figure A1 in these cases are plotted as a function of the wavelength in Figure A2. Clearly, the differences between the two power meters were roughly the same at all wavelengths. The average value of the ratio was 0.93 ± 0.09, where the quoted error is the 2σ value. The difference of 5% (1σ) from the mean is shown as the hatched area. Further, there is no real trend in this ratio as a function of wavelength. If these points were fit to a line using linear least squares analysis the slope of the line was statistically indistinguishable from zero (4.6 ± 4.7 × 10-4). Therefore, it is safe to assume that there was no systematic difference in the measurement of the laser fluence as a function of wavelength in these two power meters over the range of wavelengths used for O(1D) quantum yields discussed here. It should be noted that although we have not measured the relative response at every wavelength used, they should not be different because of the principle of operation of these power meters.
We have not inter-compared all the power meters used by the various studies that are considered in this paper. However, they all should behave similarly with wavelength because the principles of the operation of all the power meters were the same. Further, the normalization of the quantum yield data to 308 nm removes the differences in response on an absolute scale. The relative response as a function of wavelength could still contribute to the uncertainty, but this contribution cannot be more than 15-20%, based on the experience with the two power meters discussed above. Such possible differences in the response are included in the estimation of the uncertainties in the evaluated values.