OCR Text |
Show Te rf ^ 18 The value used for the bulk nonresonant susceptibility can influence greatly the inferred value of the temperature.27 To quantify this effect for these data, a series of test spectra was calculated for four temperatures using the expected value of the f/S factor. Temperature fits were performed on these spectra fixing the value for the f/S factor in the fitting function successively to four different fractions of the expected or true f/S factor. These results, shown in Fig. 11, indicate that substantial errors are possible at high temperatures if the true f/S factor is higher than that used in the fitting library. The largest errors are produced by unexpectedly low values of the nonresonant susceptibility or unexpectedly high values of the N2 mole fraction. These errors produce inferred temperatures that are too low. At high temperatures, moderate errors in the f/S factor (±50%) result in temperature errors of approximately 60 K. A similar study was made of the effect of a poorly known pump line-width since the K-T formalism is more dependent on this parameter than is the commonly used Yuratich expression. This radiation is a function of the particular Nd:YAG rod, its environment, as well as second harmonic conversion parameters. Test spectra were calculated for several pump widths for a temperature of 2200 K and fit using a library calculated with a pump width of 0.65 cm . This was done for the case of K-T spectra fit by a K-T library, Yuratich spectra fit by a Yuratich library and K-T spectra fit by a Yuratich library. These results are shown in Fig. 12. The K-T form has a maximum error slope of 56 K/cm , a value less than the maximum Yuratich slope of 92 K/cm . Assuming the experimental spectra are more closely modelled using the K-T form, K-T test spectra were fit using a Yuratich library producing the top curve in Fig. 12. This procedure always overestimates the true temperature, in this case by ^50 K. The data shown in Fig. 6 were also analyzed using the Yuratich formalism. The results are consistent with the top curve of Fig. 12. In particular, at temperatures near 2500 K, the Yuratich model produced inferred temperatures ^50 K higher than those shown; at 2300 K the difference was approximately +40 K; they were ^20 K higher at temperatures near 1900 K; and the difference decreased to approximately +15 K at 1700 K. |