Time-correlated single photon Raman spectroscopy

Abstract

While time-resolved Raman spectroscopy offers high molecular specificity, it can be particularly challenging in the presence of noise. A stronger Raman signal can be obtained by exciting a sample within the visible spectral region (VIS), thus exploiting the proportionality of the Raman scattering to the wavelength of excitation (1/ λ4). However, acquiring the Raman spectrum while exciting in the VIS often comes with the price of introducing a strong fluorescence background, in many applications. This can be limiting, especially when it comes to resolving the Raman spectra of samples such as biological tissue, which pose particular challenges to commercial Raman systems due to their high fluorescence emission across a wide spectral excitation range.

To deal with this problem there are a number of widely used techniques that can be adopted to either avoid or reject the fluorescence, such as using near infrared lasers (where the background fluorescence is less intense), applying background subtraction algorithms and signal processing to the acquired data, exploiting time-gating techniques, the use of Kerr cells, and other different and combined forms of the above.

This study contributes to the Raman Spectroscopy field by demonstrating a suitable alternative to these commonly used methods, exploiting a time-resolved technique to enable the simultaneous acquisition, separation and subsequent analysis of both Raman and fluorescence from a single set of data, while exciting in the VIS.

For this purpose, we built a Raman spectrometer that incorporates a detector called “Ra-II”, recently developed in Robert Henderson’s lab. Ra-II is a new 16.5 giga-events CMOS line sensor, consisting of 512 pixels having 16 singlephoton-avalanche (SPAD) detectors per pixel capable of ultrafast in-pixel timecorrelated single photon counting (TCSPC) histogramming.

Compared to previous time-resolved work, the spectrometer is able to separate Raman from fluorescence without the use of time-gating, instead relying entirely on a unique feature of the sensor: in-pixel on-chip TCSPC timestamp histogramming. Additional features include zoomable time resolution from 50ps to 6.4ns, and adjustable on-chip time-delay with 63ps resolution, allowing acquisition of both Raman and time-resolved fluorescence decay data in a single exposure. These characteristics of the sensor allow the spectrometer to be operated with 532nm pulsed laser excitation sources, instead of conventional near-infrared lasers thus offering enhanced Raman scattering, since scattering emission is proportional to λ-4.

The system has been carefully calibrated using a Neon calibration lamp from OceanOptics and benchmarked using well-documented samples, such as crystalline samples of calcite, diamond, quartz, and also pure distilled-water, fluorescein, paracetamol, silicate glass and sesame oil. Comparison data has also been obtained from a continuous wave excitation commercial Renishaw InVia spectrometer. Lastly the system has been used in a challenging application in trying to separate fluorescence from Raman signals in a set of biological samples obtained by mixing goose fat with pig liver, with the aim of identifying significant variations in the Raman signal according to the percentages of the biological material in the mixture.

The study builds on our last publications on noise and laser selection in Raman, highlighting the importance of power and repetition rate as well as how different patterns of activation in the SPAD array (Activation Matrixes), in combination with the time-zooming capabilities of Ra-II, allow for significant and controlled improvement in the quality of the collected data. Thus, enabling the time-resolved separation of Raman scattering from background fluorescence of various samples when exciting in the visible range. The first preliminary results procured using biological material provide reason to believe that this method is far from being close to its limit, considering that all the following results have been collected using an already 7-years-old SPAD technology whose capabilities have since been pushed significantly forward.