AGREEMENT BETWEEN FTS MEASUREMENT OF INFRARED ABSORPTION INTENSITIES OF NEAT LIQUIDS MEASURED IN SEVERAL LABORATORIES.

Abstract

Measurements of infrared absorption intensities of neat liquids at $25^{\circ}$C have been made at 1to $2cm^{-1}$ resolution in several laboratories as part of a project of the Commission on Molecular Structure and Spectroscopy of the International Union of Pure and Applied Chemsitry. The project was successful in establishing a recently published set of secondary standards of intensity measurements on liquids which are believed accurate to 2 to 3%. The purpose of involving several laboratories was to determine the extent of agreement between intensities measured on different FTS instruments under the normal conditions of use in a well run laboratory. People were asked to record and send to us transmission spectra of the cell full of liquid and of the empty cell, both referenced against the empty sample compartment. It was asked that the instrument should be well-aligned, but only at the manufacturer's usual standard. It was also asked that liquid transmission cells of good but normal quality should be used, and that good laboratory practice should be followed. All of the workers but one used a room-temperature detector. In our laboratory these transmission spectra were converted to imaginary refractive index, k, and molar absorption coefficient,$E_{m}$, spectra, after processing them to correct for reflection and other effects. The agreement found between the k or $E_{m}$ spectra from different laboratories was at the few percent level on peak heights and slightly better on integrated intensities(areas). This agreement will be illustrated by showing spectra and the comparison of integrated intensities. From our own experience, much of the disagreement is likely due to the disturbance of the infrared beam when a liquid cell is inserted into it, but to date the only instrumental modification that has been made to identify the sources of error is to ensure that the cell is held very firmly at $90^{\circ}$ to the beam. There was poor agreement in a few cases, but in these cases the band shapes and widths were also anomalous, suggesting poor alignment of the Michelson interferometer in the instrument.