Amplitude modulation and stabilization of quantum-cascade lasers

Quantum-cascade lasers (QCLs) are unipolar semiconductor sources that operate in underexplored portions of the electromagnetic spectrum. Applications such as high-sensitivity spectroscopy and frequency synthesis in the mid-IR and terahertz regions stand to benefit from QCLs possessing novel functionalities that enable amplitude and phase modulation with low injected power. Moreover, by implementing electrical feedback, these functionalities could also be used to stabilize laser frequency with metrological precision, thus realizing ideal local oscillators for extremely sensitive coherent detection.

To this end we have developed a three-terminal device with integrated functionalities for mid-IR wavelengths (5–10μm) that allows electrical control of the QCL's complex refractive index. We have also shown that metallic waveguides embedding QCLs are ideal for high-frequency modulation (up to 50GHz) and demonstrated a system that can actively stabilize the frequency and phase of a 2.7THz QCL to the round-trip frequency of a commercial mode-locked, femtosecond fiber laser.

The QCL is the only opto-electronic device based solely on electrons. This ‘unipolarity’ is typical of three-terminal devices such as transistors but has not yet been exploited. Recently, we conceived and realized a three-terminal quantum-cascade device that permits us to alter optical losses through electrical control. To date, modulating the intensity of QCLs was achieved by directly modifying the driving current. Our device achieves function by changing the voltage in the control region, which varies cavity losses.

The three-terminal device consists of a waveguide with two vertically stacked core sections, including an active laser region—gallium indium arsenide/aluminum indium arsenide emitting at 7.5μm—and a control region that can be independently controlled by two separate contacts with a common ground (see Figure 1). The control region is constructed of asymmetric quantum wells (AQWs) that permit electrical tuning of the absorption peak over a large spectral range.¹ By applying a bias to this region, we can add, or completely remove, extra losses to the optical cavity. The use of three terminals means we can drive the laser independently of the electrical power injected into the control region. This enables us to modulate more than 400mW of optical power by injecting only ~1mW of electrical power into the control region. Consequently, our approach employs a much lower voltage swing and drastically reduces the injected power required for direct amplitude modulation. This technique could be advantageously applied to distributed-feedback lasers to avoid the linewidth enhancement observed for low-frequency modulation that is typically induced by thermal effects.

The nonradiative recombination times of QCLs are on the order of picoseconds, which dramatically affects their dynamic properties. In particular, contrary to interband-diode lasers, we expect a flat intrinsic-modulation frequency response function. These short lifetimes, coupled to the very high photon density characteristic of QCLs, results in efficient, direct amplitude modulation of the gain medium up to very high frequencies. Recently, we showed that terahertz QCLs embedded in a downscaled version of a standard microwave microstrip waveguide perform extremely well in high-speed amplified modulation compared with classical lasers based on dielectric waveguides.² The advantage of this architecture is that the metallic waveguide similarly confines the fundamental, transverse electromagnetic mode at tera- and gigahertz frequencies, providing nearly 100% overlap factors between the modes and the laser active medium (see Figure 2). These excellent confinement properties are combined with relatively small amplitude attenuations at gigahertz frequencies (~2.5dB/mm at 40GHz) by virtue of the fact that in terahertz QCLs the doping level of the active region is fairly weak. These properties have allowed us to demonstrate amplitude modulation of terahertz QCLs beyond 30GHz, the highest ever reported for a QCL. Analysis of our data indicates velocity matching between the microwave and optical fields, yielding an ultrawide bandwidth of ~70GHz for a 1mm-long ridge.

We have also been investigating techniques to precisely control the emission frequency and phase of QCLs. Our results show emission-frequency phase locking of a 2.7THz QCL to the repetition rate of a mode-locked femtosecond laser with relative frequency stability below 1Hz (see Figure 3).³ Functions that currently are a prerogative of microwave frequencies, such as beam steering and coherent detection with a distributed clock, could be imported into the terahertz range (~1 to 30THz). Moreover, such a technique greatly improves the coherent detection of radiation emitted by terahertz QCLs, opening new avenues for the study of these lasers in a regime of mode locking and for their use in sensing applications.

In conclusion, we have shown that QCLs are unipolar opto-electronic devices that offer increased opportunities for amplitude and frequency modulation and stabilization over diode lasers. These capabilities have allowed us to develop new integrated functions and to show that QCLs are compatible with technologies such as microwave and femtosecond-frequency combs. As a next step, we plan to investigate how to merge these devices with very-well-established technologies such as microwaves and telecommunications. We believe that by ‘hybridizing’ QCLs with these mature technologies, we are spurring new ideas for system applications and, at the same time, helping to move forward research in the mid-IR and terahertz frequency ranges.

This work was partially supported by the Délégation Générale de l'Armement.