Automatic detection of material phase transitions from spectroscopic data

Abstract

When using a temperature measurement method based on spectral radiance information for measuring the temperature of varying emissivity measurands, there is a need for a temperature reference at some point in time. In this work, such a reference is created from the spectral radiance data already used in the temperature measurement method. Knowledge of the measurand material's phase transitions and spectral radiance data is used to create a temperature reference. Through automatic identification of phase transitions from radiance spectra employing signal processing, the temperature is known at a certain instance in time, just like required by the temperature measurement method. Three methods for automatic identification of material phase transitions from spectroscopic data are examined and evaluated. The methods are based on derivatives, steady-state identification and cross correlation respectively. They are introduced and evaluated using experimental data collected from a solidifying copper sample. All methods proved to identify the phase transitions correctly. The addition of automatic phase transition identification supplements the existing temperature measurement method such that it becomes a stand alone, reference free method for measuring the true absolute temperature of a measurand with varying emissivity.