A Computational Model for Holographic Sensing

Abstract

An existing analytical concept based on spectral decomposition has been developed more than hundred years ago, and is presently close to its limits in terms of performance and reliability, in particular, for complex samples. For molecules, a spectrum is a very complex pattern of sharp lines and continuous bands. So, in a classical spectrometer, detection is pruned to overlapping errors when two or more components of a sample have overlapping lines, and their separation is, generally, a non-unique problem. Indeed, a line can be assigned to, at least, two different transitions (in the same or different atom/molecules in a sample). Such an assignment based on line positions and transitions has limitations, and may not work at all for complex samples. As samples are getting more and more complex, the problem becomes increasingly intractable. In particular, algorithms and data processing to analyze complex spectra become very complex, require sophisticated peak analysis, etc. A mathematical "inversion" procedure for assignment and identification of components (species) also becomes unstable. That is, the current situation has all signs of a critical bottleneck, and requires an innovative approach. Meanwhile, selectivity is the first priority for many industries and applications. For instance, in the field of air toxics detection, the US EPA requires 189 components to be detected and regulated, and it is highly doubtful that any existing spectrometer is able to analyze reliably such a complex gaseous medium.