From Spectra to Structure: AI-Powered 31P-NMR Interpretation

Phosphorus-31 Nuclear Magnetic Resonance (31P-NMR) spectroscopy is a powerful technique for characterizing phosphorus-containing compounds in diverse chemical environments. However, spectral interpretation remains a time-consuming and expertise-dependent task, relying on reference tables and empirical comparisons. In this study, we introduce a data-driven approach that automates 31P-NMR spectral analysis, providing rapid and accurate predictions of local phosphorus environments. By leveraging a curated dataset of experimental and synthetic spectra, our model achieves a Top–1 accuracy of 53.64% and a Top-5 accuracy 77.69% at predicting the local environment around a phosphorous atom. Furthermore, it demonstrates robustness across different solvent conditions and outperforms expert chemists by 25% in spectral assignment tasks. The models, datasets, and architecture are openly available, facilitating seamless adoption in chemical laboratories engaged in structure elucidation, with the goal of advancing 31P-NMR spectral analysis and interpretation.