Integrated Sensing and Processing Acoustic Resonance Spectrometry (ISP-ARS) for Sample Classification

Abstract

This research introduces a novel process analytical technique, integrated sensing and processing acoustic resonance spectrometry (ISP-ARS), and compares ISP-ARS with conventional full-spectrum ARS for the characterization of solid fuel premixes used in a new pill safe designed to protect narcotics. In ISP-ARS, the acoustic excitation waveform is the analog implementation of the chemometric weight function, encoded for this work on an MP3 player and used to distinguish between fuel samples, sparing post-collection multivariate computation. In effect, the detector directly outputs the sample identity. Repeated measurements of batches of similar fuel mixtures over time produce an automatic projection of similar spectra into corresponding three-dimensional probability density contours, thus, forming the analyte classification directly. For the characterization of ten different fuel mixtures, full spectral ARS resulted in a median intermixture distance in multidimensional standard deviations (MSDs) of 185.1, while ISP-ARS resulted in a median MSD distance of 81.3. The median cross-validation MSD was 1.41 for the ARS and 1.58 for the ISP-ARS. Distances in MSDs greater than three are considered separable, and MSDs less than three are indistinguishable. The classification procedure correctly identified all samples for both analytical techniques. ISP-ARS is an effective, simpler, and more rapid alternative to full spectrum ARS that can be implemented with a commercial MP3 player and used as an inexpensive spectrometric sensor for dynamic data-driven application simulations (DDDAS) and process analytical applications.