Empirical modeling of systematic spectrophotometric errors

Abstract

Spectrophotometers, as electro‐mechanical‐optical devices, perform at a finite level of accuracy. This accuracy is limited by such factors as monochromator design, detector linearity, and cost. Generally, both the diagnosis and correction of spectrophotometric errors require a number of calibrated standard reference materials and considerable effort and commitment on the part of the user. A technique using multiple linear regression has been developed, based on the measurement of several suitably chosen standard reference materials, to both diagnose and correct systematic spectrophotometric errors, including photometric zero errors photometric linear and nonlinear scale errors, wavelength linear and nonlinear scale errors, and bandwidth errors. The use of a single chromatic ceramic tile to correct systematic errors improved the colorimetric accuracy of a set of chromatic and neutral tiles by a factor of two for a typical industrial‐oriented spectrophotometer. Greater improvement can be achieved by increasing the number of tiles and performing a separate regression at each measured wavelength., These techniques have been extremely useful in improving inter‐instrument agreement for instruments with similar geometry. Copyright © 1988 Wiley Periodicals, Inc., A Wiley Company