Determination of the isotopic composition of enriched materials using laser ablation molecular isotopic spectrometry: partial least squares and multivariate curve resolution for the determination of 15N content in enriched urea

A quantitative analytical method based on laser ablation molecular isotopic spectrometry (LAMIS) and multivariate analysis was developed and evaluated for the determination of the isotopic composition of enriched materials. The method consists preparing a concentrated solution of the enriched material, using small quantities of a sample (125 mg), and ensuring the economic efficiency of the analysis. Standard solutions of known isotopic contents are prepared by employing mixtures of urea highly enriched in 15 N and urea of natural isotopic ratio and analyzed by mass spectrometry. A small volume (30 μL) of these solutions is delivered to a filter paper disc (3 cm diameter). After drying, the disc, offering a homogeneously distributed analyte, is presented to a LAMIS equipment to acquire the vibronic emission spectra containing information about the isotopologues of interest. To illustrate the proposed method, the content of 15 N is determined in enriched samples of urea. In this case, each spectrum is normalized by the intensity of emission of the CN isotopologues for the electronic (Δν = 0) emission band at 387.1 nm, ensuring better accuracy. Selected regions and single wavelengths of the vibronic emission spectrum (Δν = + 1 or − 1) related to CN species were employed to construct multivariate partial least squares (PLS) and univariate regression models to predict the isotopic content of new samples. Besides, the LAMIS data set was evaluated by multivariate curve resolution (MCR) algorithm. The best MCR and PLS models presented similar results regarding the accuracy to determine 15 N content in enriched urea. MCR is capable of identifying spectral interferences and minimizing its effect. The results show that the proposed method based on LAMIS and PLS or MCR multivariate analysis can determine the 15 N content in the range 5–50% with a root mean square error of prediction (RMSEP) respectively equal to 0.5 or 0.7% (m/m) in comparison with reference results obtained by mass spectrometry. Graphical abstract