Automated Methods for Identification and Quantification of Structural Groups from Nuclear Magnetic Resonance Spectra...

Area of Science:

• Analytical Chemistry
• Spectroscopy
• Machine Learning

Background:

• Nuclear magnetic resonance (NMR) spectroscopy is vital for chemical structure elucidation and mixture analysis.
• Analyzing complex samples using NMR can be time-consuming and challenging.
• Existing methods often lack automation for structural group identification and quantification.

Purpose of the Study:

• To develop automated methods for identifying and quantifying structural groups in pure chemical components and mixtures using NMR data.
• To improve the efficiency and accuracy of NMR spectral analysis.
• To provide a foundation for predicting fluid properties of unknown substances.

Main Methods:

• Implementation of two automated methods: a group-identification method for qualitative analysis and a group-assignment method for quantitative analysis.
• Utilizing support vector classification for spectral data analysis.
• Training the methods on a dataset of nearly 1000 pure component NMR spectra (¹H and ¹³C).

Main Results:

• The developed methods achieve excellent prediction accuracy for both pure components and mixtures, including those not in the training set.
• The group-identification method provides qualitative insights into sample composition.
• The group-assignment method enables quantitative analysis by linking structural groups to ¹³C NMR signals.

Conclusions:

• Automated NMR spectral analysis using support vector classification is effective for identifying and quantifying structural groups.
• These methods offer a significant advancement over traditional, manual NMR analysis techniques.
• The obtained structural information can be integrated with thermodynamic models for predicting physical properties of chemical samples.