Raman spectroscopy

Concept

Raman spectroscopy uses the phenomenon of Raman scattering, which is explained clearly by the following wikipedia image:

CC BY-SA 3.0, Link

The key idea being that:

• Most light absorbed by a molecule is elastically scattered - the incident wavelength is the same as the emitted wavelength. This is called Rayleigh’s scattering.
  ${\lambda }_i={\lambda }_e$
• A really small group of molecules are in a higher vibrational state. They absorb light and inelastically scatter it - the emitted wavelength is higher/redder than absorbed wavelength. This is called Stokes scattering.
  ${\lambda }_i<{\lambda }_e$
• A similarly (or even more) small group of molecules are in a lower vibrational state. They absorb light and inelastically scatter it - the emitted wavelength is lower/bluer than absorbed wavelength. This is called anti-Stokes scattering.
  ${\lambda }_i>{\lambda }_e$

Raman spectroscopy (or spontaneous Raman spectroscopy) uses this idea to identify the wavelengths emitted by a molecule. Because the wavelengths are dependent on the vibrational states, which are in turn dependent on the molecule structure, this method can be used to identify a molecule based on the inelastically scattered light.

The basic Raman spectrometer is simple:

1. Illuminate the sample with photons of exactly the same wavelength - i.e. a laser beam.
2. Filter Rayleigh’s scattering using an optical notch filter.
3. Use a spectrometer to detect the other wavelengths in the scattered light (i.e. diffract the filtered light, and put it on a line CCD sensor to measure intensity along wavelengths).

Stimulated Raman spectroscopy

The key constraint with the above idea is that the intensity of spontaneously inelastically scattered photons is very low (around one in a million). Stimulated Raman spectroscopy uses a modification of the above idea.

By Shay beer - I made it by myself with power point, CC BY-SA 4.0, Link

The idea of stimulated emission is the same as the one used to generate a LASER beam. This uses the non-linear effect, wherein presence of a photon helps stimulate another photon with the same wavelength. So instead of waiting for light to be spontaneously emitted, stimulated Raman spectroscopy has another beam, the stokes beam, that stimulates emission of a particular wavelength of light.

This idea can be used for anti-Stokes scattering based spectroscopy too. There, the molecule is sent to a higher rotational/vibrational state after the stimulated emission of a Stokes photon.

By YangWenlong - Own work, CC BY-SA 4.0, Link

The idea of using a second beam to stimulate specific photons, if there are rotational and vibration states, is very cool. The downside is that the stimulated photons add onto the laser beam, so it is hard to detect. I found a couple of ways this is solved:¹

1. Varying the Stokes beam amplitude sinusoidally at a constant frequency and using a lock-in amplifier to extract the extra photons as an addition to the baseline amplitude.
2. Using anti-Stokes scattering instead of Stokes scattering. This provides a high signal-to-noise ratio as both the pump and the Stokes photos can be removed using an optical notch filter. The disadvantage is that getting a, anti-Stokes photon requires two pump photons, so it would require much more (likely power or exponential) laser power.

Current approaches

These above ideas assume single laser wavelengths for both pump and Stokes lasers. Ideally, varying both pump and Stokes will allow us to capture more data along a range of bandwidths, increasing the richness of the spectral signature.

I found a couple of methods that use Femtosecond lasers to do this.

1. https://pmc.ncbi.nlm.nih.gov/articles/PMC3637845/
2. https://arxiv.org/pdf/1011.6576

From a quick glance at the optics, the idea seems to be to use the broad frequency spectrum of femtosecond lasers as the source of multiple wavelengths. By using chirping, the pulse is stretched so that the laser beam is longer in time and spread out in wavelength - higher wavelengths arrive first and lower wavelengths arrive later (this is the inverse of the solution to dispersion problems with femtosecond lasers or rather using dispersion creatively). This allows each laser pulse to be photons whose wavelengths vary with time.

Using this method, both pump and Stokes beams are time and wavelength varying pulses. In addition, the Stokes beam amplitude is varied with a constant frequency to use lock-in amplifiers.

The actual details are much more complicated. I will add more notes when I read more into this.

Footnotes

1. I will add more as I come across more methods. ↩