QUANTITATIVE ACCURACY OF THE SPECTRA
For most of its existence, this collection of gas-phase spectra has been the only one we knew of that was presented as having nearly correct absorbance values for a specified concentration-pathlength product. (The only quantitative collection.) Until recently we had no way of checking our numbers against any others. Basically, this collection has had to stand by itself. This is regrettable, because the inherent absorption strengths of lines and bands in an infrared spectrum are fundamental physical properties that deserve to be tabulated and published like all the other basic physical constants. Unfortunately, it is going to take a great deal of laboratory effort and soul-searching before these constants can be presented with even two significant figures (error of 1%, or less). At present we estimate our absorption coefficient accuracies to be within +/- 5%.
If compounds are chemically stable, relatively non-polar and have a room temperature vapor pressure of at least several Torr., they can be measured in an absorption cell with a fair degree of confidence, and the spectrum will yield correct absorption coefficients. The user must of course pay attention to errors that may be incurred because of temperature effects, pressure effects and the effects of incomplete spectral resolution. When a compound is thermally unstable, or when it reacts with wall materials, the measured absorption coefficients must be viewed only with a moderate degree of confidence. We advise users of our quantitative absorption spectra that if they have reason to suspect that our absorption coefficient values may be incorrect, they should carry out their own measurements. Then if they believe they have obtained a more reliable result than ours, they may use the Grams arithmetic functions to raise or lower our absorbances to agree with theirs. Our digitized, low-noise spectra will then still be usable in their quantitative analysis programs.
Some users choose to check QASoft against calibration gas mixtures. Frequently, carbon monoxide is the first compound to be checked out, and it is found that QASoft does not report the expected CO concentration. Then there is a search for an explanation. Is the calibration mixture at fault? Has the gas sample been mis-handled? Are the QASoft absorbances wrong? Does the disagreement arise because of instrumental factors that cause the sample and reference spectra to have different line shapes or line widths?
With a compound like CO that has individual lines in its spectrum, differences due to instrumental factors are likely to arise. For polyatomic molecules with broad absorption features, instrumental differences are less likely. If the user has confidence in his calibration mix and in his sample handling, then he can correct for the absorbance differences by adjusting the absorbances in the database spectrum. QASoft will then calculate and report out what he considers to be the correct concentrations.
Fortunately, the U. S. National Institute of Standards and Technology is now issuing standard reference spectra similar to ours. As their spectra come out, we are comparing ours with theirs, and in those few cases where the absorbance values are significantly different, we accept their values.
When QASoft reports a concentration that disagrees with the concentration that is supposed to be in the calibration mixture, that is a signal for the operator to examine his gas handling methods, the operating conditions in recording the spectrum and the quality of his calibration mixture.