Coherence and beam shaping in quantum-cascade lasers
Quantum-cascade lasers (QCLs) are semiconductor lasers operating in the mid-IR and terahertz range. Since their invention in 1994,1 their development has undergone tremendous progress, and they are now close to being exploited in real-world applications including trace-gas sensing, spectroscopy, and free-space communication. Their output power in continuous-wave mode is already in the watt range, with wall-plug efficiencies above 10% at room temperature.2 Complex aspects such as single-mode and surface emission have been addressed successfully.3,4 However, there are still a number of unresolved issues that are crucial for developing successful applications, including beam shaping. Because of the small dimensions of semiconductor light sources, the output light beam is typically broad and highly astigmatic. On the basis of adequate resonator designs, one can improve the beam shape and thus make bulky and expensive external optics obsolete.
A second outstanding issue is related to coherent superposition in laser ‘Y junctions,’ which is most commonly used to achieve high-power output in single-mode operation. However, it can also be used in monolithic active interferometers. In turn, this could be used as a sensing device, for example by applying some functional material that binds special polymers or biochemical molecules to one of the interferometer branches. Interaction with the leaky part of the waveguide mode causes a phase shift and hence a change in the emitted power due to destructive/constructive interference.
Divergence of the output beam can be dealt with using a ring-shaped resonator with a radial light-outcoupling grating on top (see Figure 1). The ring-type emitting area naturally forms a circularly shaped far field, and the large overall emission area narrows the beam. The grating is crucial for light extraction using the surface because QCLs are intersubband devices and therefore restricted to transverse-magnetic polarization. Light propagating normal to the epilayers—and hence normal to the surface—cannot be generated directly. One needs a diffractive element to produce surface emission. Second-order distributed-feedback (DFB) gratings are an ideal choice in this context as they also allow for mode selection, thus resulting in single-mode surface-emitting lasers.
Our group was the first to construct grating-coupled surface-emitting ring-multimode5 and single-mode6 QCLs. Multimode devices are equipped with shallow gratings that introduce only a small coupling, which is insufficient for single-mode operation. The far fields are ring shaped to point shaped, depending on the detuning extent of the lattice period. If the grating period matches the center of the emission spectrum, the far field forms a single spot with a full width at half maximum (FWHM) of approximately 10°. The FWHM is limited by the width of the emission spectrum since slight detuning causes angled emission. The lasing threshold is similar to those of Fabry-Perot-type cavities. This is expected because the waveguide loss is unchanged, and mirror losses for standard cavity lengths are comparable to the outcoupling loss introduced by the grating.
The far fields emitted by single-mode devices show a clearly reduced FWHM to below 3° (see Figure 1). This is due to the spectral narrowing induced by the DFB grating, which has a higher coupling coefficient. Along with the increased coupling leading to single-mode operation, the outcoupling is also increased. This affects the lasing threshold in a negative way but benefits the slope efficiency so that at typical operating currents the efficiency is identical to that of multimode devices.
For active interferometric QCLs, we first investigated the physics of light coupling in Y-branched QCLs.7, 8 Here, light starts as a single beam and is then monolithically split into two branches. The light is reflected at the ends of the two branches and again combined into the single stem. Through experimental analysis of cavities with different dimensions and wavelengths, we could determine phase-locking and modal-behavior dependence so that we could influence the coherence. As a result, we can now reliably fabricate Y-coupled resonators that provide a large degree of coherence. As a first application, we combined two Y junctions to form monolithic Mach-Zehnder interferometric (MZI) lasers,9 as shown in Figure 2. The performance of these devices is comparable to those of standard Fabry-Perot lasers with only small losses in the coupling regions.
We introduced a phase shift between the two branches by resistively heating one branch to change the refractive index. Next, we compared the output power with that of a uniformly heated device (see Figure 2). The additional continuous-wave current first increases the output power as the additional carriers overcompensate the negative effect of heating. As the heating is further increased, however, emission is reduced. Finally, constructive interference causes the MZI-laser signal to increase again.
We have addressed two important topics for mid-IR QCLs by using special resonator designs: a proof of principle for MZI based sensing devices as well as the very promising concept of ring-shaped DFB QCLs for improving the emitted far field. For the latter, we expect improvements due to an adopted waveguide as well as by moving to substrate emitting devices. For MZI QCLs, an altered waveguide could increase the overlap of the leaky part of a waveguide mode and hence, might be useful in a micro fluidic device.
We acknowledge support from the Austrian Science Fund's (FWF) ADLIS (Advanced Light Sources) program, the Austrian 'Nano Initiative' project PLATON (Processing Light - Advanced Technologies for Optical Nanostructures), and the Society for Micro- and Nanoelectronics (GMe).
