Nonlinear Microscopy

In coherent Raman scattering, vibration levels of certain molecules are excited by several pulses of appropriately selected wavelengths. These vibrational levels are specific to the respective molecule, which means that samples can be examined for their chemical composition in a marker-free and non-invasive manner. This makes Raman microscopy a popular method in a wide variety of fields such as biology, medicine and pharmacy. However, in order to excite the vibrational levels, the excitation pulses must fulfill certain requirements, such as synchronized repetition rates, and a broad and tunable wavelength range. This in turn places specific requirements on a light source that is to be used for Raman microscopy. Furthermore, in addition to the desired coherent Raman processes, parasitic processes often occur that lead to a background signal without chemical information. Mechanisms are therefore needed to suppress these background signals and thus make the resulting microscopy images more meaningful. Another challenge, especially for the detection of structures in the sample, is the low-noise detection of the signal, for which optimized detectors are used.

Background correction for stimulated Raman scattering using frequency modulation

© AG Fallnich

In stimulated Raman scattering (SRS), pulse trains with two different wavelengths are used to excite vibrational levels of certain molecules and thus to generate a Raman signal. This signal appears as a small intensity modulation on one of the pulse trains and is detected using a lock-in amplifier. However, in addition to this process, which carries information about the chemical composition of the sample, other processes can also take place without the involvement of a vibrational level, which result in an intensity modulation of the pulses as well and thus generate a parasitic background signal.

In order to suppress this parasitic signal, we have implemented a mechanism for frequency modulation of the pulses directly within the light source. One of the pulse trains is modulated in its optical frequency, i.e., in its wavelength, at half the pulse repetition rate, so that only every second pulse generates a Raman signal. The other half of the pulses cannot excite a vibrational level due to the not appropriately selected wavelength, but the wavelength-independent parasitic processes still take place. By using a lock-in amplifier for detection, the signals are directly subtracted from each other so that background correction takes place in real time. In our work, we were able to reduce the background by a factor of up to 8. This method also has the advantage that the resulting signal has 30% less noise than conventional methods in which the background correction is performed digitally afterwards. See the publication by Krstin, Thomas and Nick.

Frequency modulation to suppress the non-resonant background in coherent anti-Stokes Raman scattering

© AG Fallnich

Coherent anti-Stokes Raman scattering (CARS) describes a process of coherent Raman scattering in which the resulting signal has a shorter wavelength than the excitation beams. However, light at this signal wavelength can also be generated by another process that is independent of the vibrational levels of a molecule and, therefore, does not contain any chemical information: this is called a non-resonant signal. Since the non-resonant signal cannot be distinguished from the actual CARS signal due to its wavelength, this leads to a background signal during detection.

We have implemented a method for frequency modulation of the excitation beams directly in the light source, whereby an automatic background correction takes place during the detection of the signal. With frequency modulation, one of the excitation beams is modulated in its wavelength, i.e., two successive pulses have a slightly different wavelength. As a result, one pulse is used to generate the resonant CARS signal in combination with the non-resonant signal. The other pulse can no longer generate a resonant CARS signal due to its shifted wavelength, but the non-resonant signal is still generated. If the entire signal is detected with a lock-in amplifier, the signals of two consecutive pulses are effectively subtracted from each other, so that only the resonant CARS signal without background remains. This method therefore allows background correction to be carried out in real time, whereby we have achieved an increase in contrast by a factor of 18 and an improvement in sensitivity by a factor of 40. See the publication by Thomas and Kristin.

Stimulated Raman scattering using a fiber-based light source with fast wavelength tunability

© AG Fallnich

Stimulated Raman scattering can be used in microscopy to display not only magnified images of samples, but also their chemical compositions. For this purpose, specific vibrational levels of molecules are excited using two pulses whose wavelength difference corresponds to the energy of the vibrational level. In order to fully examine a sample, light sources are required that can be tuned in their wavelength in order to address as many vibrational levels as possible.

We have used a fiber-based light source whose wavelength can be changed within a wide range in just 5 milliseconds, whichs allows to obtain a lot of information about the sample in a short amount of time. This enables hyperspectral imaging of a wide range of samples. In addition, we utilized the wavelength tunability to rapidly record Raman spectra of different substances. Since stimulated Raman scattering requires low intensity modulations to be measured on high intensity pulses, the noise behavior of a light source is also of great importance for Raman microscopy. We have characterized the intensity noise of the fiber-based light source, which at -153.7 dBc/Hz is only a few decibels above the fundamental limit of shot noise. See the publication by Kristin and Thomas.