Combs in Quantum Cascade Lasers: Linear Chirp and Pulse Compression

Abstract

Quantum cascade lasers (QCLs) are bright, coherent sources, which operate at mid-infrared and terahertz wavelengths. The 2012 discovery that they also lock to form broadband combs has enabled applications such as dual comb spectroscopy, in which high signal to noise broadband spectra can be taken on millisecond timescales, impossible with the older globar technology. The locking is passive, requiring no external stimulus or separate absorber. Before the start of this thesis, it was already well understood that the comb forms by combination of spatial hole burning and four-wave mixing, and that the emission should tend to a constant intensity with time rather than a pulse train owing to the strong damping in the active region. Questions remained though on the exact nature of the state, experimental evidence lacking in the mid-infrared. This is a rich dynamical system, comprising parts with different characteristic timescales working together to form a stable comb emission. Understanding experimentally how the emission evolves with time serves as both a useful, granular comparison for simulation, and as a direct characteristic for applications which seek to take advantage of the higher peak brightness and lower electrical dissipation of a pulse driven laser. We use a technique which makes use of a boxcar integrator to track the spectral evolution from <10 ns to 1 us. We find significant broadening over these timescales (<1 us) compared to when driven with a constant bias, which could find application in similar space to quasi-CW driving of DFB lasers. Together, the spectral amplitude and phase determine whether the comb in the time domain is a pulse train, frequency modulated (FM, broadband continuous wave), or something else. For the QCL, models had predicted a so-called pseudo-random type of FM, whereby within each identical period, the laser would exhibit a complicated chirp pattern. For such an emission, standard pulse characterisation techniques are unsuitable. We use a coherent technique called SWIFTS (Shifted Wave Fourier Transform Spectroscopy), which works by spectrally resolving the the beats between neighbouring mode pairs, thereby directly measuring their phase differences. Applied to a broadband mid-IR comb, we found what is now understood to be quite a general behaviour in QCLs and other laser systems: a periodic linear chirp, laser sweeps its bandwidth from its longest wavelength to its shortest. Moreover, we confirmed 3 key properties of the comb: that the state was reproducible, stable in time, and tunes smoothly with current. Armed with knowledge of the phase and stability properties, the laser output can be converted to a stable pulse train by suitably delaying the different frequency components. Since the phase of the linear chirp takes on a simple form - broadly speaking a parabola - means for such lasers we do not need to use the more sophisticated/less commercially attractive spatial light modulator for example, which is difficult to come by affordably in the mid-IR. Instead, we opt for a grating based stretcher-compressor, a staple of the ultrafast community, which can be constructed relatively cheaply using off the shelf optical components. With this, we were able to modify the emission from ~CW with 134 ps period, to a train of pulses of width around 12 ps, and increase the peak intensity to average ratio from 4 to 40. For the future, this implies efforts can be concentrated on generating broader, flatter spectra, e.g. through bandstructure engineering or microwave injection, which in turn are expected to yield a more parabolic phase to better fit the compressor's transfer function. Ring QCLs are expected to behave very differently from Fabry-Perot ones, since spatial hole burning is absent when operating unidirectionally. Indeed, comb spectra in rings tend to be structurally different, and indeed much narrower, despite carrying comparable levels of optical power. We studied different ring QCLs, using a combination of dual comb spectroscopy and SWIFTS to measure their phase. In a first experiment, we find unidirectional lasing, and a quiet comb state with significant amplitude modulation. In a follow-up experiment, we find a laser emitting a sech2 spectrum, a signature of soliton emission, atop a CW background. Its measurement indeed confirms that this is a pulse, and with a grating filter, we separate it from the background. In this way, from a DC pumped QCL, are able to extract 3.7 ps pulses.