116 Summary: Power Scalable Concepts
Table 3.3:Comparison compression techniques solid-core
PCF
Kagomé-type HC-PCF
KERR-effect bulk broaden-ing
cascaded-χ(2) bulk broaden-ing
peak power scalability
limited by criti-cal power
GW
demon-strated with Kagomé-type
HC-PCF and
capillaries54,316
requires multi-pass arrange-ment
GW
demon-strated355
average power scalability
250 W demon-strated272
questionable 500 W demon-strated336
90 W demon-strated
demonstrated compression factorsa
15 11 (1st stage),
2.4 (2nd stage)
6 (1st stage), 1.7/2.2 (2nd stage 10 fs and 7.7 fs, resp.)
6 in 3 stages
demonstrated compression efficienciesb
74 % 70 % (1st stage), 52 % (2nd stage, Fresnel losses
and mirror
bandwidths improvable)
≈ 50 %
(1st stage), 60 %/40 % (2nd stage 10 fs and 7.7 fs, resp.)
75 %
peak power enhancement factor
10 5.3 (1st stage),
0.66 (2nd stage)
2.1 (1st stage), 0.76/0.64 (2nd stage 10 fs and 7.7 fs, resp.)
3.3
phase shape at output (at ≈ 30fs)
mainly GDD in all-normal dis-persion regime
mainly GDD
for all-normal dispersion, higher-order terms for pres-sure gradient
higher order terms present, depending on broadening factor per plate
higher order terms present, depending on broadening fac-tor per crystal
and
phase-mismatch few-cycle
pulse genera-tion
demonstrated for low peak powers18,374,375
presented in sec-tion 3.1.2
presented in sec-tion 2.2.2
demonstrated
for longer
wavelengths, higher peak powers356,364,365
robustness fair fragile high high
a note that shorter input pulses lead to lower compression factors,
b refers to ratio of average power at out- and input.
is in strong contrast to the Kagomé-type HC-PCF experiments presented in section3.1.2.
Large apertures, easy replacement and cost-efficiency present additional benefits of bulk broadening. Those are especially important for “workhorse-type” laser systems. At the central wavelength of 1030 nm, few-cycle pulse generation with cascaded quadratic non-linerities is rather difficult due to the dispersion of n(cas)2 . The multi-pass arrangement simulated in section 3.2.1 appears to be the best solution to compress efficiently to the sub-10 fs regime. First experiments with KLM TD oscillators resulted in ≈ 30 fs pulses with a peak power enhancement from 60 MW to 230 MW62,376. This presents a record-high peak power for amplification free systems and is well suited for generating harmonics with tens of µW average powers and sub-50 nm wavelengths323.
The chapter has also introduced an alternative approach to carrier-envelope-offset fre-quency stabilization of high-power mode-locked (TD) oscillators. It takes advantage of the high sensitivity of the CEP to intracavity and pump power modulations, resp. It is based on a dual wavelength pumping scheme where the radiation of high-power laser diodes driven by a very stable power supply is combined with the radiation of low-power laser diodes driven by a power supply with large modulation bandwidth. A sub-400 mrad in-loop residual phase noise was achieved for the full oscillator power. Especially for frequency domain applications where the phase jitter can often be averaged out while the single pulse carrier-envelope-offset frequency is not decisive, this noise value (if also reached out-of-loop) is usable for applications. Nevertheless, the phase locking of the AOM approach was tighter than that of the pump power modulation approach due to the very high modulation bandwidth of the AOM (cf. Suppl.1). It is, however, expected that the residual phase noise can be further lowered by optimized PLF design or alternative dual wavelength schemes, for instance the one demonstrated in ref. 373 for low-power fiber oscillators. Pump power modulation, being more (average) power-scalable than in-tracavity loss modulation, may play an important role in future locking experiments of mode-locked oscillators with intracavity powers of several kW.
It is instructive to evaluate if the carrier-envelope-offset frequency stabilization schemes can be transferred from KLM to SESAM mode-locked TD oscillators which have not been stabilized at average power levels above 3 W so-far. In order to do so, the noise accumulation of the free beat note in Fig. 3.36 is compared to the transfer functions presented in ref. 236. The SESAM mode-locked Yb:YAG TD oscillator shows a strong resonance around 7 kHz as well and an abrupt phase behavior at this frequency which was attributed to the saturable absorber. The combination of high average and peak power makes it difficult to operate the SESAM in the well saturated regime. In particular, multi-photon absorption at the presence of several kW intracavity average power may lead to strong thermal lensing and rapid degradation. Consequently, SESAM mode-locking high-power TD oscillators generally appears to make CEP stabilization significantly more difficult than KLM where the saturation behavior is determined through the interplay of the Kerr-lens with hard and soft apertures. Moreover, the highest power SESAM mode-locked TD oscillator have been operated in a vacuum environment58,61 which was, for instance, not necessary for a 270 W KLM TD oscillator59. It can be inferred that high-power SESAM mode-locked TD oscillators are more sensitive to the distortions an intracavity AOM introduces, and thus should be preferentially stabilized by pump power modulations.
In summary, peak and average power scalable concepts for ultrashort pulse compression and CEP stabilization have been demonstrated. The presented schemes are important
118 Summary: Power Scalable Concepts
building blocks for upscaling the results of chapter2 to the latest and the coming gener-ation of mode-locked TD oscillators.
The previous chapters of this dissertation have demonstrated the progress in the devel-opment of KLM TD oscillators towards waveform control and few-cycle operation. So-far all research was concentrated on the optical octave from about 700 to 1400 nm.
In general, ultrashort pulse sources are most sophisticated in the near-IR and visible parts of the electromagnetic spectrum where active laser transitions of electronic nature are located. As mentioned before, the most powerful fs sources emit around 1 micron wavelength. Oscillators directly deliver more than 250 W of average power58,59, while amplifiers even reach kW levels28,32. For comparison, the 2 micron technology has reached 20 W from mode-locked oscillators151and about 150 W from amplifiers377. At 2.5 microns, lasers stay already well below 10 W378. Finally, broadband gain that allows femtosecond operation has only been demonstrated up to 2.9µm with approximately 300 mW average power379.
Consequently, the availability of mid-IR light sources stands in strong contrast to their tremendous multitude of applications, for example in fundamental physics, chemistry, environmental and life sciences. The spectral region contains the fundamentalvibrational transitions of all infrared-active small molecules, the most common functional groups as well as biomolecules like proteins, lipids, nucleic acids and carbohydrates. Importantly, mid-IR absorption results from the specific molecular structure which is exploited in various chemical analysis techniques380, to name only a few examples, in trace gas as well as human breath analysis381 and in early cancer detection382.
Particularly interesting techniques are frequency comb66, in the broadest sense frequency up-conversion383,384 and 2D infrared spectroscopy385. Frequency combs allow fast and highly precise data acquisition. Up-conversion spectroscopy avoids the need of mid-IR detectors and, in combination with coherent light sources and cross-correlation meth-ods, enables to extract temporal and phase information of molecular processes386,387. 2D infrared spectroscopy additionally resolves structural information of the samples under test. All these techniques need or at least strongly benefit from fs pulses. Compressing these to few-cycle durations has revealed tremendous strong-field effects, like the gas high-harmonic cut-off extension to the x-ray regime388 or the occurrence of Blochoscillations in semiconductors246,389. In these examples, high peak power is required for accessing the regime of extreme nonlinear optics. On the other hand, high average power and photon flux, respectively, results in light source brightness surpassing those available from syn-chrotron sources198, and consequently facilitates the use of (rather noisy) uncooled MIR detectors or detector arrays390. Moreover, high average power is important for experi-mental techniques with low yield such as stand-off gas detection391 and may avoid the need for multiple interactions with the sample in resonant cavities or multi-pass cells381. Here, the issue of the scarce availability of high-power fs sources in the mid-IR is addressed by taking advantage of the powerful KLM TD oscillator output and by down-converting it by means of three-wave mixing.
120 Optical Parametric Amplifiers for Frequency Down-Conversion