Uni- and multivariate calibration of tempe­rature from the neodymium fluorescence spectra in nanocrystals of yttrium-gadolinium oxide and yttrium-gadolinium garnet

The use of neodymium-doped nanocrystalline powders of yttrium-gadolinium oxide and yttrium gadolinium garnet to increase the sensitivity of local fluorescent optical temperature sensors is considered. Based on the temperature dependences of the neodymium fluorescence spectra in this powders, univariate (using fluorescence intensity ratio from thermally coupled energy levels of the activator) and multivariate (using the partial least squares method) calibration models are developed. When using the spectral range 860 – 950 nm falling into the first biological transparency window (700 – 980 nm), both calibration models have a standard deviation of about 10 % and are comparable in accuracy. The spectral variables selection by searching combination moving window in the multivariate model made it possible to reduce the root mean square error for yttrium-gadolinium oxide by more than 12 times (from 9.8 to 0.8 °C), and for yttrium-gadolinium garnet by more than 2 times(from 8.7 to 4.0 °С). The result obtained indicatesthe proposed neodymium-doped nanocrystalline powders and multivariate methods of calibration can be used to localise areas with febrile temperatures for biological and medical purposes.