Independent component analysis applied to diffusion‐ordered spectroscopy: separating nuclear magnetic resonance spectra of analytes in mixtures

Nuclear magnetic resonance spectrum of a mixture contains the overall peaks of all the analytes. It is impossible to perform structural assignment on the mixture without the knowledge of individual spectra of the components. Spectral separation is thus an important means of teasing out pure components of a mixture before spectral assignment. We propose a strategy called diffusion‐ordered independent component analysis (DIFFICA) to achieve this task. This strategy applies independent component analysis algorithms to diffusion‐ordered spectroscopy (DOSY) to extract spectra of pure components in a mixture. DIFFICA was tested in a simulation and experimentally in two three‐component systems with and without water suppression, in 1D and 2D DOSY data. Pure spectra were achieved in both cases. The selection of diffusion parameters to guarantee pure spectra is guided by the distance correlation between separated spectra. Copyright © 2012 John Wiley & Sons, Ltd.