Final Report Summary - EMCOS (Emission Comb Spectroscopy: Development of novel frequency comb spectroscopy methods based on stimulated Raman scattering)
Final Report for EMCOS (Proposal number 254478) over period 1-9-2009 to 1-9-2011
The idea of this project was to develop sensitive and precise detection techniques based on frequency comb lasers (including the laser system itself), for the detection, identification and investigation of molecules. Optical transitions in molecules are typically observed at mid-infrared wavelengths, and can be used to identify molecules and their structure. Initially two forms of Raman scattering were proposed for this purpose. One method was based on Raman-induced Stokes gain and anti-Stokes loss, providing absorption cross-section information. The other method was based on transient-index perturbation. By studying this with a dual frequency comb laser, using the technique of multi-heterodyning, the idea was that both techniques could be studied with a similar setup.
First a new dual frequency comb laser system was developed in a collaboration with IMRA, USA (dr. H. Ingmar and dr. M.E. Fermann) based on advanced Yb-doped fiber technology. It produces pulses at a repetition rate of 150 MHz, and with a pulse length of about 90 fs. At 150 MHz repetition rate, a data acquisition rate of more than 300 Hz is possible for Raman-shifts of up to 1000 cm-1. This still allows for video-rate data acquisition required for microscopy setups considered as a potential application of the proposed spectroscopy technique.
Apart from the two oscillator lasers, 3 amplifier sections were added. One amplifier produced narrow-bandwidth 'Raman' pump laser pulses. The highest pulse energy without any observable nonlinear pulse distortions was realized with a 15 μm core diameter step index fiber with an air-cladding. Operating at a repetition rate of 9.5 MHz, the Raman pump source delivered Fourier-limited pulse durations of 5.3 ps at pulse energies of more than 50 nJ corresponding to 0.5 W average power. The other amplifier sections (typical output 2-3 W) were spectrally broadened using 'photonic' fibers, resulting in a coherent spectrum ranging from 600 nm to more than 1600 nm with a very good coherence (complex-degree of first-order coherence g12 greater than 0.9).
In order to facilitate low-noise phase and timing control of the comb lasers, fast actuators were implemented in the laser cavity. For phase-locking of the repetition rate (the comb spacing), the saturable absorber mirror was mounted on a piezo. The carrier-envelope-offset frequency representing the second comb parameter was measured in a collinear f-2f-interferometer and controlled via the pump power.
In the context of the development of the fiber frequency combs, an additional study on improvements of the phase locking performance were carried out in a collaboration with the group of prof. dr. J. Ye (JILA, USA). This work was motivated by the recent success in eliminating broadband phase noise in fiber frequency combs by stabilizing the relative intensity noise of the pump laser to avoid phase noise, induced via a combination of intensity fluctuations and self-phase modulation. Therefore the pump power was stabilized, and the carrier-envelope phase was adjusted with an additional AOM. When closing the feedback loop on the pump power, the linewidth of the f:2f beat signal was reduced from 70 kHz to 10 kHz (resolution limited). Together with feedback on the AOM this scheme indeed showed superior locking performance resulting in 214 mrad residual phase error (integrated from 100 mHz to 1 MHz) at a locking bandwidths of 250 kHz.
The results of this work were published in Benko et al., Optics Letters Vol. 37, pp. 2196-2198 (2012). This extended phase-locking scheme was not implemented in the final laser system and the dual-comb-spectroscopy experiments, because it can be detrimental for heterodyning spectroscopy (due to contamination with zero-order light from the AOM, which influence cannot be filtered out easily in this case).
After the first year in which a tailored dual frequency comb system was constructed, we continued by building a Raman spectroscopy setup for the new laser system. As a first sample we used 1-propanol which is known to exhibit an exceptionally strong Raman transition at 820 cm-1. When pumped at 1050 nm, this corresponds to a transition wavelength of 1140 nm. Several schemes to measure the transition were extensively tested, but signal could not be found (even after consulting Raman spectroscopy experts).
The general consent was that it was all properly planned and designed, but after several months of testing still no signal could be detected. We therefore changed our scientific target towards a different but very related research area: mid-infrared frequency comb generation and spectroscopy. No approach has been shown to fulfill the requirements for a mid-infrared frequency comb spanning the entire molecular fingerprinting region, where molecular compounds exhibit their strongest absorption features.
We used the second year of the project to demonstrate such a laser system using difference-frequency mixing, in a collaboration with prof. dr. M. Marangoni from the Politecnico di Milano (Italy). The process is based on a χ(2) interaction in a nonlinear crystal, where two electric fields, pump and signal, are mixed, resulting in an idler field whose wavelength is determined by the energy difference between the two. By a proper choice of these wavelengths, the idler field can be easily shifted into the mid-infrared spectral region. Moreover, if pump and signal field are pairs of phase-coherent pulses emitted from the same source, the generated idler field is carrier-envelope-offset phase-free and requires only stabilization of the comb spacing.
It is based on the two Yb:fiber frequency combs described before. For each system the output was split in two branches, providing tunable seed and fundamental wavelength pump pulses for the subsequent DFG process. As the seed we used the longest wavelength Raman-solitons from supercontinuum generation in a 25 cm long highly nonlinear suspended-core fiber. Average powers from 1.4 to 1.9 W and about 20 mW for pump and seed pulses, respectively, were available for the subsequent difference frequency generation. For this, pump and signal beam are re-combined with a dichroic mirror and focused in a 500 μm long GaSe crystal for a Type-I (e - o = o) difference frequency interaction. Behind the GaSe crystal, a Ge long-pass filter separates the residual pump light from the idler output. Tuning of the idler output is possible by changing the Raman-soliton wavelength from 1.15 to 1.65 μm and the launched pump power in the highly nonlinear fiber, respectively, with a corresponding adjustment of the temporal overlap and phase-matching angle. They span the extremely large 3-10 μm range with a spectral width varying between 210 and 710 nm. This constitutes a significant improvement of the tuning range compared to previously reported mid-infrared sources based on difference frequency generation.
The maximum power measured with a calibrated pyro-electric detector to 1.5 mW at 4.7 μm corresponds to a peak-power spectral density of about 77 μW / nm and 0.9 μW / comb mode, respectively. The power levels remain in excess of 1 mW in the range from 4 to 5.5 μm, while decreasing to few hundreds of μW at both ends of the tuning range. The results were published in A. Ruehl et al., Optics Letters 37, pp. 2232-2234 (2012).
Significantly higher conversion efficiencies were possible by using a 40 mm long PPLN crystal but with the drawback of a reduced tuning range.
In conclusion, we have developed a dual Yb-fiber frequency comb system and difference-frequency mixing to cover the whole fingerprinting spectral region in the mid-infrared for the first time. This system is ideal for heterodyning frequency comb spectroscopy, a method which combines extreme frequency accuracy with a wide spectral coverage and fast acquisition speeds.
The idea of this project was to develop sensitive and precise detection techniques based on frequency comb lasers (including the laser system itself), for the detection, identification and investigation of molecules. Optical transitions in molecules are typically observed at mid-infrared wavelengths, and can be used to identify molecules and their structure. Initially two forms of Raman scattering were proposed for this purpose. One method was based on Raman-induced Stokes gain and anti-Stokes loss, providing absorption cross-section information. The other method was based on transient-index perturbation. By studying this with a dual frequency comb laser, using the technique of multi-heterodyning, the idea was that both techniques could be studied with a similar setup.
First a new dual frequency comb laser system was developed in a collaboration with IMRA, USA (dr. H. Ingmar and dr. M.E. Fermann) based on advanced Yb-doped fiber technology. It produces pulses at a repetition rate of 150 MHz, and with a pulse length of about 90 fs. At 150 MHz repetition rate, a data acquisition rate of more than 300 Hz is possible for Raman-shifts of up to 1000 cm-1. This still allows for video-rate data acquisition required for microscopy setups considered as a potential application of the proposed spectroscopy technique.
Apart from the two oscillator lasers, 3 amplifier sections were added. One amplifier produced narrow-bandwidth 'Raman' pump laser pulses. The highest pulse energy without any observable nonlinear pulse distortions was realized with a 15 μm core diameter step index fiber with an air-cladding. Operating at a repetition rate of 9.5 MHz, the Raman pump source delivered Fourier-limited pulse durations of 5.3 ps at pulse energies of more than 50 nJ corresponding to 0.5 W average power. The other amplifier sections (typical output 2-3 W) were spectrally broadened using 'photonic' fibers, resulting in a coherent spectrum ranging from 600 nm to more than 1600 nm with a very good coherence (complex-degree of first-order coherence g12 greater than 0.9).
In order to facilitate low-noise phase and timing control of the comb lasers, fast actuators were implemented in the laser cavity. For phase-locking of the repetition rate (the comb spacing), the saturable absorber mirror was mounted on a piezo. The carrier-envelope-offset frequency representing the second comb parameter was measured in a collinear f-2f-interferometer and controlled via the pump power.
In the context of the development of the fiber frequency combs, an additional study on improvements of the phase locking performance were carried out in a collaboration with the group of prof. dr. J. Ye (JILA, USA). This work was motivated by the recent success in eliminating broadband phase noise in fiber frequency combs by stabilizing the relative intensity noise of the pump laser to avoid phase noise, induced via a combination of intensity fluctuations and self-phase modulation. Therefore the pump power was stabilized, and the carrier-envelope phase was adjusted with an additional AOM. When closing the feedback loop on the pump power, the linewidth of the f:2f beat signal was reduced from 70 kHz to 10 kHz (resolution limited). Together with feedback on the AOM this scheme indeed showed superior locking performance resulting in 214 mrad residual phase error (integrated from 100 mHz to 1 MHz) at a locking bandwidths of 250 kHz.
The results of this work were published in Benko et al., Optics Letters Vol. 37, pp. 2196-2198 (2012). This extended phase-locking scheme was not implemented in the final laser system and the dual-comb-spectroscopy experiments, because it can be detrimental for heterodyning spectroscopy (due to contamination with zero-order light from the AOM, which influence cannot be filtered out easily in this case).
After the first year in which a tailored dual frequency comb system was constructed, we continued by building a Raman spectroscopy setup for the new laser system. As a first sample we used 1-propanol which is known to exhibit an exceptionally strong Raman transition at 820 cm-1. When pumped at 1050 nm, this corresponds to a transition wavelength of 1140 nm. Several schemes to measure the transition were extensively tested, but signal could not be found (even after consulting Raman spectroscopy experts).
The general consent was that it was all properly planned and designed, but after several months of testing still no signal could be detected. We therefore changed our scientific target towards a different but very related research area: mid-infrared frequency comb generation and spectroscopy. No approach has been shown to fulfill the requirements for a mid-infrared frequency comb spanning the entire molecular fingerprinting region, where molecular compounds exhibit their strongest absorption features.
We used the second year of the project to demonstrate such a laser system using difference-frequency mixing, in a collaboration with prof. dr. M. Marangoni from the Politecnico di Milano (Italy). The process is based on a χ(2) interaction in a nonlinear crystal, where two electric fields, pump and signal, are mixed, resulting in an idler field whose wavelength is determined by the energy difference between the two. By a proper choice of these wavelengths, the idler field can be easily shifted into the mid-infrared spectral region. Moreover, if pump and signal field are pairs of phase-coherent pulses emitted from the same source, the generated idler field is carrier-envelope-offset phase-free and requires only stabilization of the comb spacing.
It is based on the two Yb:fiber frequency combs described before. For each system the output was split in two branches, providing tunable seed and fundamental wavelength pump pulses for the subsequent DFG process. As the seed we used the longest wavelength Raman-solitons from supercontinuum generation in a 25 cm long highly nonlinear suspended-core fiber. Average powers from 1.4 to 1.9 W and about 20 mW for pump and seed pulses, respectively, were available for the subsequent difference frequency generation. For this, pump and signal beam are re-combined with a dichroic mirror and focused in a 500 μm long GaSe crystal for a Type-I (e - o = o) difference frequency interaction. Behind the GaSe crystal, a Ge long-pass filter separates the residual pump light from the idler output. Tuning of the idler output is possible by changing the Raman-soliton wavelength from 1.15 to 1.65 μm and the launched pump power in the highly nonlinear fiber, respectively, with a corresponding adjustment of the temporal overlap and phase-matching angle. They span the extremely large 3-10 μm range with a spectral width varying between 210 and 710 nm. This constitutes a significant improvement of the tuning range compared to previously reported mid-infrared sources based on difference frequency generation.
The maximum power measured with a calibrated pyro-electric detector to 1.5 mW at 4.7 μm corresponds to a peak-power spectral density of about 77 μW / nm and 0.9 μW / comb mode, respectively. The power levels remain in excess of 1 mW in the range from 4 to 5.5 μm, while decreasing to few hundreds of μW at both ends of the tuning range. The results were published in A. Ruehl et al., Optics Letters 37, pp. 2232-2234 (2012).
Significantly higher conversion efficiencies were possible by using a 40 mm long PPLN crystal but with the drawback of a reduced tuning range.
In conclusion, we have developed a dual Yb-fiber frequency comb system and difference-frequency mixing to cover the whole fingerprinting spectral region in the mid-infrared for the first time. This system is ideal for heterodyning frequency comb spectroscopy, a method which combines extreme frequency accuracy with a wide spectral coverage and fast acquisition speeds.