A new approach for mixture separation using NMR spectroscopy: Blind iterative source identification

Mixture separation can be a challenging but an engaging problem in a wide variety of fields. It is of particular interest for identification of compounds using nuclear magnetic resonance (NMR) spectroscopy. Currently, NMR data is most often acquired from isolated purified compounds as it can become very challenging to identify the components in a complex mixture system. NMR can identify individual samples with respect to their unique fingerprint spectra; however it has been traditionally poor at handling mixtures. Robustly designed techniques for separation of NMR spectra into subspectra from individual components would be of great interest in, for example, the pharmaceutical industry. Structural information about promising candidates from the drug design process must be extracted from samples that contain side products and impurities in addition to the target molecule. Diffusion ordered spectroscopy (DOSY) is one of the few NMR techniques that can be used for mixture separation [1]. It uses translational diffusion to effect a partial separation but requires that individual compounds have a substantial difference in their diffusion constants. In addition, the separation problem becomes more difficult when the component molecules have similar chemical functional group and overlap in their NMR spectra. Alternatively, blind source separation (BSS) is a technique that can be used to extract an individual NMR subspectrum from mixture spectra; further, if the molecules do not interact strongly - these individual subspectra are nearly identical to those of the corresponding pure compounds acquired under the same conditions [2,3]. BSS has been successfully demonstrated for separation using NMR [2,3]. However, application remains problematic for certain cases in which the data are of limited quality. Here, we present a new approach that can operate in the presence of phase or baseline imperfections in the NMR data which can affect the fidelity of the results when using the - onventional BSS algorithm. Coined as blind iterative source identification (BISI), the method will be benchmarked against both conventional BSS and DOSY (Fig. 1). Our initial results showed that the reliability and quality of the results were improved greatly compared to previous methods (Fig.1).