Assessment of Uncertainties in Using Raman Thermometry Techniques to Determine the Local Thermal Conductivity of Uranium Dioxide (UO 2 )

Uranium dioxide (UO 2 ) is the standard fuel in light water, nuclear reactors, and its material properties, such as thermal conductivity, bear a great influence on the heat transfer efficiencies of the system. Laser flash analysis is the most common technique employed to measure the thermal properties of bulk UO 2 in post-irradiation examination, but there is interest in expanding measurement capabilities to situations where laser flash cannot be applied, such as spatial mapping of localized fuel properties. One potential method is Raman thermometry, where Raman spectra are related to the temperature of the material, but the uncertainties associated with relating temperature to UO 2 properties should be investigated. This work investigates the potential viability for measurements based on three existing thermometry techniques, termed Raman-based Thermophysical Property Models (RTPMs), by determining their potential uncertainty when applied to UO 2 . These are time domain transient techniques (which uses a cantilever sample), a two-laser, steady-state technique (which uses a thin film sample), and a frequency domain photothermal reflectance transient technique (which uses a semi-infinite sample). Using experimental data of the UO 2 Raman spectra as a function of temperature, a simulated temperature profile is generated. The thermal diffusivity ( α ) and/or thermal conductivity ( k ) are calculated by fitting the RTPMs to temperatures calculated from Raman spectra. The parameter sensitivity and uncertainty associated with each technique are calculated. For time domain methods, whose k and α errors range from 0.5 %–1.1 % to 1.7 %–209 %, respectively, it is concluded that time discretization of published spectra limits the accuracy of our proposed analysis method. For the two-laser methods, which exhibit a k error of 7.5 %, it is discovered that radiation cannot be neglected when thin films are sub-micrometer thick. For the frequency domain methods, which show k and α errors of 0.29 % and 6.8 %, respectively, the underlying fitting assumptions tend to produce large errors, except for a limited range of low frequencies that appear to achieve greater accuracy. In addition, a weak Raman peak is observed in UO 2 , which limits the effectiveness of transient RTPM-based measurements.