Intelligent Systems for Automated NMR Analysis of Complex Mixtures (autoNMR)

• Applicants: Professorin Dr. Yulia Monakhova; Professorin Dr. Margit Schulze
• Subject Area: Analytical Chemistry
• Term: since 2025
• Project identifier: Deutsche Forschungsgemeinschaft (DFG) - Project number 563667755

Nowadays, quantitative NMR (qNMR) is a well-established procedure to estimate the purity of chemical compounds and concentrations of multiple compounds in complex mixtures. Apart from experimental prerequisites for high precision qNMR (high signal-to-noise ratios (SNR), sufficient relaxation delays, good signal digitization), accurate pre-processing and integration are crucial. The smallest errors of qNMR are expected when the spectra are carefully processed by an experienced NMR spectroscopist. Existing automated routines for spectral processing are not able to verify the preprocessing and integration steps as well as the presence of impurities in the integral area while requiring reference databases for spectral profiling. As the size of molecules and the number of compounds in mixtures increases, spectral overlap in 1H NMR even impedes deconvolution techniques. In our project this problem will be overcome by switching to other NMR active nuclei such as 19F or to multi-dimensional NMR experiments such as heteronuclear single quantum coherence (HSQC) NMR. Even using these experimental techniques, sometimes multivariate/machine learning methods are required to mathematically resolve overlapping 1D and 2D NMR profiles. In the frame of this project intelligent systems will be developed for automated spectral processing of 1H, 19F and HSQC NMR experiments in case of baseline-separated and overlapping peaks. The routines will be easily implementable, fast and robust when operating under the different conditions usually encountered in the study of complex mixtures (low SNR, presence of overlapping peaks, and presence of artifacts). The project will be supported by the industrial partner Spectral Service AG, in particular high-field NMR experiments including theoretical and practical expertise in 1D and 2D qNMR. Industrial partner, as many other NMR laboratories, collected large NMR databases throughout the years. Researchers may not be aware that these data, when properly evaluated, can potentially uncover new valuable and more precise information about samples under analysis or can be used to automate analysis workflow. This possibility will be explored for retrospective analysis of a dataset (>300 samples) of 19F qNMR spectra used for routine purity determination of organic active pharmaceutical ingredients (APIs). In contrast to conventional high-field NMR spectrometers based on superconducting magnets, small permanent magnets were reinvented and further developed, which enabled the construction of NMR spectrometers that fit on a laboratory bench. In the current project, the 1H, 19F and 2D HSQC NMR low-field data will be modelled. The objects under study will be low- and high-molecular weight APIs as well as finished medicinal products. The results of this research will help to generally evaluate the applicability of Benchtop NMR instruments for challenging analytical tasks.

• DFG Programme: Research Grants (Transfer Project)
• Application Partner: Spectral Service AG