Affiliated Professor University of Delaware Newark, Delaware, United States
In fifteen words or less, explain the significance of this contribution (Novel Aspect).: Resolution of more than two component spectra from a pair of source spectra
Abstract Text: Identification and resolution of spectral components of mixtures, such as pharmaceutical samples, is an important task for quality assurance and process control.Chemometric technique, like multivariate curve resolution, has been a well-established method of choice.Such a method works well if a reasonable set of initial estimates of the pure component spectra and a number of sample spectra encompassing a broad range of concentration variations are available.However, if the number of available spectra is somewhat limited, e.g., only the standard reference spectrum from the lab and that of an on-line sample collected during the production process, the task becomes more challenging.Two-trace two-dimensional correlation spectroscopy (2T2D-COS) is a variant form of 2D-COS, where only a pair of spectra are analyzed by spreading the dataset on a planar spectral map format.Among several interesting applications of 2T2D-COS, it has been realized that spectra of more than two components may actually be resolved from only a pair of source spectra.This seemingly surprising result is achieved not by the usual linear combination of source spectra but by the correlation-based classification of spectral signals into a set of distinct component spectra.The technique takes advantage of a simple fact that spectral features arising from the same component species cannot independently vary their intensities from each other.In other words, signals from the same species are synchronously correlated and can be discriminated as such.To achieve the effective resolution of mixture spectra, determination of the so-called characteristic spectral bands showing the least level of mutual correlation among themselves becomes a critical step.Recently, a systematic hierarchical procedure to identify characteristic bands based on 2D-COS has been developed.This method seems to work well with 2T2D-COS resolution of mixture component spectra.The performance of this newly introduced technique is demonstrated for the analysis of actual pharmaceutical samples.
.png)