Advances in Raman hyperspectral compressive detection instrumentation for fast label free classification, quantitation and imaging

Bharat R Mankani, Purdue University


Multiple prototypes of hyperspectral compressive detection (CD) Raman spectrometers have previously been constructed in the Ben-Amotz lab and have proven to be useful for fast, label-free chemical identification, quantitation and imaging. The CD spectrometer consists of a volume holographic grating (VHG) that linearly disperses the Raman photons into its component wavelengths and all wavelengths are focused onto a digital micromirror devise (DMD). The DMD is an optical modulator that consists of an array of programmable 10μm mirrors that can reflect photons in either +12° or -12° to the incoming light. The DMD is tilted such that the +12° photons go back through the focusing lens and the VHG and is focused onto a single 150μm photon counting avalanche photodiode detector(APD).^ In chapter 1 of the thesis I describe the construction of a new CD Raman spectrometer capable of fast hyperspectral imaging that has better photon collection efficiency and fewer photon losses compared to its predecessors. The new spectrometer consists of a VHG and a DMD, however, the DMD is not tilted but is perpendicular to the incoming Raman photons. All the Raman photons modulated by the DMD are symmetrically detected in the +12° and -12° by two photon counting photomultiplier tube(PMT) detector modules. The new spectrometer avoids a double pass through the optics and hence has fewer losses associated due to reflection transmission of the optics. Full spectral measurements are made by consecutively scanning through columns of the DMD mirrors and measuring the intensity of photons associated with each wavelength. CD measurements are made by multiplexing wavelengths channels onto the detectors and can be done by applying optimal binary(OB) or Hadamard filters. The new optical design has a spectral window from 150cm-1 to 4000 cm-1 and the improvement in the photon collection efficiency allows classification and imaging speeds of 10μs per point with 13mW of laser power on the sample, and is significantly faster than measurements made with the previous prototype. ^ In chapter 2 of the thesis I describe the construction of a new instrument which is equipped with both a hyperspectral CD spectrometer as well as a traditional Czerny Turner spectrometer. A flip mirror after the Raman microscope directs the Raman scattered beam either towards the CD spectrometer (with the mirror down) or towards the Czerny Turner spectrometer. This instrument allows us to perform head to head comparisons of the two spectrometers using the same Raman scattered photons emitted by the sample. The CD spectrometer uses hardware optical filters to perform compressed chemometric measurements to classify chemicals. The traditional spectrometer uses the CCD to measure full spectral data and chemometric analysis is performed to extract lower dimensional chemical information post measurement. Chemical classification results obtained using two sets of chemicals with differing degrees of spectral overlap show that CD classification is comparable to full spectral classification in the high signal regime. However, for signals consisting of less than 1000 total photon counts, CD classification outperforms full spectral classification.^ In chapter 3 of the thesis, Raman spectroscopy is used to probe changes in vibrational spectra of nucleotide solutions and hanging droplets containing RNA crystals at different pH. Self-modeling curve resolution (SMCR) applied to full Raman is used to extract solute correlated (SC) Raman spectral components that contain solute spectra with minimal interference from the surrounding solvent. The goal of these studies is to show that Raman spectroscopy can be used to study biological molecules in aqueous environments, with minimal sample preparation and without the need of labels.^




Dor Ben-Amotz, Purdue University.

Subject Area

Analytical chemistry|Physical chemistry

Off-Campus Purdue Users:
To access this dissertation, please log in to our
proxy server