Implementation of a Raman spectral decomposition technique that allows effective identification of complex mixtures. The computationally and memory efficient software enables new functionality to be added to portable hand-held devices.
Raman spectroscopy is an established method for identifying unknown materials across various sectors. Conventional analysis methods are based on comparing the measured spectrum with a reference spectral library of known chemicals to find the best match. While effective for identifying a single spectrum from a library, a sample composed of a mixture of different chemicals provides a greater challenge.
Edinburgh researchers have developed a Raman spectral decomposition technique based on a new fast sparse approximation method. Inputting a set of reference spectra and an unknown mixture yields the identity of mixture elements and their contribution percentages. It also has the capability of detecting cases where the mixture has a spectrum outside the reference library. The method is highly computationally and memory efficient, which means that it can run on a low power real-time platform. Implemented as a hardware independent C package, which can handle a given library and input spectrum, the technology enables use with hand-held devices. This provides a portable, non-invasive approach for identification of real-life mixtures of chemical substances.
A hardware independent C version of the mixture-matching algorithm has been prepared. Performance has been successfully demonstrated in the identification of real mixtures in different measurement scenarios, including where components are close to noise level.
Please note, the header image is purely illustrative. Source: Forance via Shutterstock