ShimNetV2-RM Neural Network for Post-Acquisition Correction of Spectral Lineshapes in NMR Reaction Monitoring

Area of Science:

• Chemistry
• Spectroscopy
• Computational Chemistry

Background:

• Nuclear Magnetic Resonance (NMR) is valuable for in situ reaction monitoring.
• Traditional NMR requires frequent magnetic field homogeneity correction (shimming), which is time-consuming during reactions.
• This leads to spectral distortions that can hinder accurate analysis.

Purpose of the Study:

• To develop an automated post-acquisition method for correcting spectral distortions in NMR reaction monitoring.
• To adapt a neural network (ShimNetV2-RM) for improving spectra acquired during dynamic chemical reactions.

Main Methods:

• A neural network, ShimNetV2-RM, was developed based on a prior model (ShimNet).
• The network was trained using well-shimmed pre- and post-reaction spectra to constrain parameters for reaction monitoring.
• The method was applied to copper-catalyzed azide-alkyne cycloaddition (CuAAC) reactions.

Main Results:

• ShimNetV2-RM effectively corrects spectral lineshape distortions that accumulate during NMR reaction monitoring.
• The neural network provides an automated alternative to manual post-acquisition correction methods.
• Successful application demonstrated on challenging CuAAC reactions.

Conclusions:

• ShimNetV2-RM offers a robust solution for improving NMR spectral quality during in situ reaction monitoring.
• This method enhances the reliability and efficiency of analyzing dynamic chemical processes using NMR.
• The approach is particularly useful under demanding experimental conditions where repeated manual shimming is impractical.