Modeling, measuring, and minimizing the instrument response function of a tunable diode laser spectrometer

The instrument response function (IRF) of a spectrometer limits the accuracy of measured spectroscopic parameters by broadening recorded spectral lines/features. We describe methods to model the effects of the IRF on spectral data, to minimize the IRF widths, and to measure the resulting width of the spectrometer IRF. We have modeled the IRF of our Tunable Diode Laser Spectrometer as a Voigt function. A real-time method of eliminating the effects of low-frequency spectrometer drift has been implemented and has resulted in a substantial reduction in the width of the IRF, its residual Gaussian component reduced from about 0.00045cm-1 to about 0.0002cm-1. An accurate measurement of the IRF Gaussian width utilizes a computationally simple method making use of the spectral dependence of the RMS noise of each signal-averaged data point. Various noise sources affecting the spectrometer (preamp/detector noise, laser AM noise, and laser FM noise) are identified and separately quantified by use of the same method. The IRF Gaussian-width measurement can be automatically applied to each measured spectrum of an experimental data set. A related method is discussed which allows accurate determination of the spectral dependence of statistical noise appropriate for use in quantitative Chi-square fitting of absorption spectra. We explore simple, efficient numerical processes which can dramatically enhance the quality and usefulness of acquired spectral data, improving the ability to apply TDL spectroscopy to high-precision, quantitative measurements and the determination of detailed spectroscopic lineshape parameters. This paper provides a guide for interested readers to implement these developments in their own spectrometers.