Abstract
The thickness of epitaxially grown layers is commonly measured by infrared multiple reflection. The phase shift at the layer–substrate interface is very often neglected. It has been computed as a function of the wavelength, and for accurate measurement it is said to be advisable to apply this correction. By direct calculation from measured spectra, a phase shift independent of wavelength is found. The layer–substrate reflection coefficient, generally assumed to be constant, has been verified to vary exponentially with the wavenumber, yielding another constant phase shift. The measured constant phase shift cannot be explained from this dependence. An alternative procedure for measuring the thickness of epitaxial layers with a precision to within 0.5% is suggested.