A preliminary study on constructing a high-dimensional asynchronous spectrum to analyze bilinear data

Abstract A novel approach to constructing high-dimensional asynchronous spectra ( n D-Asyn) is proposed. Three theorems relevant to 1D slices of n D-Asyn are revealed. n D-Asyn is used to analyze bilinear data from mixtures containing multiple components obtained via hyphenated techniques. The spectral contribution of different components can be removed in a stepwise manner by increasing the dimensions of asynchronous spectra. As a result, the spectra of different components can be faithfully recovered even if the time-related profiles of different components severely overlap. Moreover, correct results can still be obtained via the n D-Asyn even if a considerable level of noise and baseline drift are present. The n D-Asyn approach is compared with MCR-ALS using different constraints in analyzing the data for a simulated and also for a real system. The n D-Asyn produced correct spectrum of every component. Only when complete constraints obtained from n D-Asyn method is utilized in the MCR-ALS calculation, correct spectra of all the components can be obtained. Thus, n D-Asyn can be used alone or in conjunction with MCR-ALS to analyze bilinear data containing contributions of multiple components.