This paper is an investigation of the effects of errors in the analog-to-digital converter (ADC) of a Fourier transform infrared (FT-IR) spectrometer on the photometric accuracy of that spectrometer. The effect of ADC errors on the spectrum after Fourier transformation is calculated analytically for monochromatic, two-line, and wide square band emission spectra. Numerical modeling is used to extend the analysis to absorption spectra, and to include the effects of noise on the amplitude of absorbance bands. These analyses showed that ADC errors can generate artifacts throughout the spectrum, although the largest effects occur at sharp spectral features. Errors as large as 8% in the amplitude of absorbance bands can be produced by ADC errors of one least significant bit (LSB), in a 15-bit ADC. These results were confirmed qualitatively by measuring the net height of the oxygen vibrational band in silicon at 1107 cm−1 using four different ADC circuit boards in the same FT-IR spectrometer. At the highest signal levels, the ADCs disagreed by as much as 4%, even though their static transfer characteristics (which were measured in a separate experiment) exhibited errors of less than ± 2 LSBs.
Aslan Baghdadi, Warren K. Gladden, and Donald R. Flach, "Nonlinear Effects of Digitizer Errors in FT-IR Spectroscopy," Appl. Spectrosc. 40, 617-628 (1986)