Abstract

Understanding the structure of the OH⁻ absorption peaks at 1.25 and 1.39-µm is necessary for effective long-wavelength lightguide system design and production. This paper reports the use of a nonlinear regression algorithmf, based on the Marquardt method, to estimate the amplitudes, center wavelengths, and halfwidths of a twelve-parameter four-component Gaussian OH⁻ peak model at 1.39 µm and a six-parameter two-component Gaussian OH⁻ peak model at 1.25 µm. These models, combined with the Marquardt regression, provide accurate estimation of OH⁻ absorption, and the analysis is much faster than previously reported techniques involving measurement of infrared and Raman spectra. Numerical values for the Gaussian center wavelengths and halfwidths are tabulated, and correlation analysis and linear regression are used to determine the interrelationship of the six-component amplitudes. Simplified OH models incorporating the estimated parameter values are proposed. The OH^- peak equations are expressed in terms of the measured maximum loss of the 1.39-µm peak. A comparison of the closeness of fit is made between the twelve- and six-parameter models, the simplified models, and the measured maximum-loss model. The simplified models are shown to provide an accurate representation of the OH⁻ absorption profiles for even a limited set of measurement data.

© 1985 Optical Society of America