(74b) Real-Time Spectroscopy Via Multivariate Optical Computing

Quantitative detection of discrete chemical species is a valuable tool across a multitude of industries from inline process control to standoff threat detection. Historically, laboratory-based multivariate spectroscopy has been employed for analytical measurements of chemical or physical characteristics of solids, liquids, slurries, and gases. These analytical measurements are the result of processing optical spectra with pattern recognition algorithms (e.g. Partial Least Squares) on a processing unit (e.g computer or microcontroller). When migrating the spectroscopy tooling outside of the laboratory environment, additional challenges await like environmental stability and data bandwidth to/from the data processing unit in order to generate data in near real-time. Ultimately, the quantitative detection results are achieved with a spectroscopic system that possesses a large Size, Weight, Power, and Cost (SWAPc). The SWAPc problem is exaggerated when point detection spectroscopic tooling is replaced with hyperspectral imaging (HSI) systems as both the instrument complexity and raw data generation increases. Outside of the manufacturing environment, infrared HSI systems have been fielded for the detection of hazardous chemical and biological compounds, tag detection (friend versus foe detection), and other defense critical sensing missions over the last two decades. An ideal, real-time spectroscopic measurement tool for either point detection or imaging would possess the sensitivity and specificity of a laboratory spectrometer with minimal data bandwidth requirements to yield a direct analyte prediction with a low SWAPc.

Multivariate Optical Elements (MOE) are custom, wide bandpass, interference filters that exploit the capabilities of Multivariate Optical Computing. MOEs are encoded with one of many possible spectral patterns by using the optical transmission and reflection characteristics of the interference filter to detect/measure a complex chemical signature in the presence of a strongly interfering background. Simple instruments incorporating MOEs can realize the advantages of a multivariate calibration whereby the prediction of an analyte concentration or physical quality may be obtained without measuring the spectrum discretely through the optical equivalent of a dot product between a spectrum at discrete wavelength channels and a regression vector. Ultimately, an optical regression whereby an incident intensity of light is implicitly multiplied by the transmission or reflection properties of the interference filter replaces the complicated steps of a digital regression in a hardened apparatus where the chemometric advantages may be realized in a simple instrument that a non-expert can operate. Additional advantages of optically de-multiplexing spectroscopic signals include improved precision, optical throughput (i.e. increased stand-off range), improved reliability and ultimately the reduction of instrumentation cost. When coupled with a two-dimensional focal plane array or camera, MOEs may yield real-time chemical distributions within a scene.

This presentation summarizes the design and implementation of MOE instrumentation and its advantages as compared to traditional spectroscopic tooling with an emphasis on representative applications from inline process control to stand off threat detection.