Raman spectroscopy and multivariate analysis for quantifying uranium(VI) under uranyl nitrate crystallization-relevant conditions.

A fiber-optic Raman spectroscopy approach was developed for solution-phase, in situ quantification of U(VI) under uranyl nitrate crystallization-relevant conditions, where temperature fluctuations and chemically complex matrices can compromise calibration performance. A multivariate regression framework is presented that combines Raman spectroscopy, design of experiments, and chemometrics to predict U(VI) concentration in HNO3 in pre- and post-crystallization regimes. A D-optimal design was used to efficiently select calibration concentrations spanning the relevant composition space. Principal component analysis of Raman spectra acquired over a range of concentrations and temperatures revealed systematic trends, including a temperature-dependent redshift in the ν1 symmetric U(VI) stretch at high U(VI) concentrations near the solubility limit. Partial least squares regression (PLSR) and support vector regression (SVR) models were developed and evaluated from 20 to 60 °C in the presence of surrogate fission products (Cs, Sr, Ru, Nd, Re, Zr, and Mo) and varying HNO3 levels. SVR outperformed PLSR for U(VI) and HNO3 quantification by delivering strong accuracy despite temperature- and matrix-induced spectral variation. These results highlight the utility of Raman spectroscopy coupled with multivariate calibration for solution-phase quantification under uranyl nitrate crystallization-relevant conditions and provide a basis for future monitoring applications in used nuclear fuel recycling and other crystallization-focused process analytics settings.