Bayesian evaluation of broadband argon emission spectra for the determination of excited state population densities and temperatures

Emission spectroscopy takes advantage of the radiation emitted naturally by electronically excited gases or plasmas, which contains information about the population density of the emitting excited state. However, the reliability of emission spectroscopic measurements suffers from the fact that several approximations and uncertain parameters are required to invert the measurement model. For some of these parameters, uncertainties are relatively well quantified, e.g. for spontaneous emission coefficients; for others, auxiliary measurements have to be performed which introduce their own uncertainties. Using an atomic emission line measurement model, we present a Bayesian framework to infer plasma properties from emission spectroscopic measurements. The approach accounts for uncertainties in intensity calibration, spontaneous emission coefficients, lineshape properties, and other nuisance parameters. First results show the promise of this approach to propagate uncertainties from the raw spectral measurement to the final quantities of interest in a consistent fashion and to include prior knowledge in the data evaluation. The methodology supports the integration of experimental data and model predictions and increases the fidelity of emission spectroscopic measurements.