Fluorescence Suppression in Raman Spectroscopy using Low Wavenumber Anti-Stokes Scattering

Raman spectroscopy is a highly effective and important technique, able to provide specific molecular information for sample identification in a variety of fields. One of the key limitations of the method, particularly in biological samples, is interference from fluorescence. Fluorescence emission can be in the same wavelength region as the Raman signals and can be much stronger. Therefore, it can drown out the Raman signals in photon shot noise and other fluorescence-induced spectral artefacts or cause issues due to CCD saturation. This can severely limit the effectiveness of Raman spectroscopy. Many methods, from simple fluorescence background fitting to complex techniques such as SERDS, have been developed to suppress fluorescence and improve Raman signal-to-noise ratio (SNR). These methods of fluorescence suppression all vary in their effectiveness and limitations. This thesis explores low wavenumber anti-Stokes Raman scattering (LWARS) as a new method for fluorescence suppression in Raman spectroscopy. Fluorescence emission is generally significantly weaker in the anti-Stokes region than in the Stokes region. While Raman signals are also weaker in the anti-Stokes region this effect is less significant at lower wavenumbers. Anti-Stokes Raman spectroscopy also has the advantage of being able to use longer excitation wavelengths without the Raman peaks of interest exceeding the spectral sensitivity limit of CCDs. Using longer excitation wavelengths typically results in less fluorescence being induced as the lower energy photons are less likely to have the required energy to do so. This means, that in the presence of strong fluorescence, LWARS may provide higher SNR when compared to other methods of fluorescence suppression. LWARS was first compared to Stokes Raman scattering of the same molecular vibration within the same wavelength region, by using different laser excitation wavelengths (830 nn for anti-Stokes and 785 nm for Stokes). This was done at 350 mW laser power and a 3 s acquisition time. In the presence of the fluorescent dye Patent Blue V, LWARS SNR was found to be 0.75 (± 0.07) times that of its Stokes equivalent. This inefficiency was believed to be due to the Patent Blue V absorption peak being at much shorter wavelengths than the near-infrared (NIR) excitation wavelengths. The experiment was repeated with HITCI, a fluorescent dye with high absorption in the NIR region. This was done at 100 mW laser power and a 6 s acquisition time. LWARS was found to have 1.52 (± 0.10) times higher SNR than its Stokes equivalent. The rapid photobleaching of HITCI in water also allowed for a comparison of LWARS and photobleaching. LWARS was found to provide fluorescence suppression equivalent to 12.5 minutes of 100 mW 785 nm laser illumination. This was achieved instantaneously and without the risk of sample degradation. The experiment was again repeated with HITCI dissolved in ethanol using a 75 mW laser power and an acquisition time of 3.75 s. The relative SNR was found to be 1.45 (± a0.13) which was a negligible change, within the experimental error of the HITCI in water relative SNR. Adding a gelatin phantom between the laser and the fluorescent sample and increasing the acquisition time to 10 s resulted in a relative SNR of 1.49 (± 0.18), i.e. functionally unchanged from the non-gelatin version of the experiment. LWARS was also compared to the spatially-offset Raman difference spectroscopy (SERDS) method with 100 mW laser power and a 4 s acquisition time. The two methods were found to give roughly identical SNR despite SERDS being more experimentally and computationally complex. LWARS also has the advantage of reducing the fluorescence intensity in the raw measured spectra mitigating CCD saturation issues. This experiment also explored the fluorescence suppression potential of anti-Stokes SERDS. However, it was found to provide worse SNR than LWARS and Stokes SERDS, while maintaining the increased complexity of Stokes SERDS.