QCLs for Spectroscopy
Applications and Background: Spectroscopy
In the context of increasing demand for sensitive, selective, fast and portable detectors for trace components in gases and liquids, the near and mid infrared range are very good candidates for spectroscopy detection technics as a lot of molecules present strong vibrational absorptions in these domains. Some applications are the detection of atmospheric pollutants and improved medical screening capabilities for early detection of diseases and drug abuse. In the project “IrSens 2” funded by the Nano-Tera project framework we build together with other academic and industrial partners (EMPA, FHNW, UniNE, Alpes Laser) a laser absorption spectroscopy based compact portable platform that can be used to detect any molecules at those wavelengths both for liquids and gases. To give a proof of concept, the different partners chose to detect the nine most important environmental and greenhouse gases with trace-gas sensitivity.
Laser absorption spectroscopy aims at rotational and vibrational states of molecules, which are probed by means of energetic excitation with a laser (Figure 1). Since mid-infrared quantum cascade lasers (MIR-QCLs) emit in the MIR range of the electromagnetic spectrum (3-10 um) they can access the fundamental roto-vibrational transitions of many molecules.
FLASH COMIC (http://www.qoe.ethz.ch/research/t-bbmirqcl)
Figure 1: Cartoon displaying the principle of laser-based gas spectroscopy
MIR-QCLs are especially interesting, because of (i) the possibility to precisely tailor the active region for a target emission wavelength (i.e. measure any gas with absorption fingerprint in this spectral region), (ii) the possibility to exert precise mode control directly in the device or externally (needed for high resolution spectroscopy) and (iii) the possibility to fabricate low dissipation devices with a good wallplug efficiency (required for portable applications).
Active region optimization
The active region (or gain medium) of a MIR-QCL can be tailored to generate laser light at a certain frequency. Controlling elements for this purpose are the thickness of individual layers, material system, alloy composition of the material, doping and strain. In principle there are many active region designs lasing at the same frequency, but depending on the details in the design a given active region will lase more or less efficient. The optimization on one individual parameters is obviously impossible since the merit function will inevitably be a complicated function of all the parameters. Therefore we use a genetic approach, which gives the possibility to find a global optimum even in presence of many local maxima and allows a large exploration of the parameter space. Starting from an initial reference structure, we perform a sequence of random variations of the thicknesses and we evaluate the efficiency on the population. At the end, few individuals with the highest efficiency are retained to build the next generation (Darwin’s selection). Figure 2 compares the electrical and optical performance, as well as the wallplug efficiency for a reference active region design and its genetically optimized version.
Not all applications demand for efficiency as merit function for this optimization algorithm. Similarly, we optimized some active regions for a broad gain, which would enable lasing of a single device in broad spectral region or the fabrication of lasers, which selectively emit at distant spectral locations. Alternatively to a broad spectral gain, active regions with narrow gain in different spectral regions can be stacked to achieve the same effect. In this case, however, the total thickness of the active region has to be distributed among the individual stacks, which will in turn lower the gain
Typically, a single MIR-QCLs can be used to scan a portion of the MIR range, in which some absorption lines are situated. The span of the portion that can be scanned depends on the design of the active region, as well as the mode control strategy. For single-mode MIR-QCLs we work with two established strategies for mode selection: photonic crystals and external cavity setups. The difference between these two strategies is the locality of the mode control. Photonic crystals are typically implemented as one- or two-dimensional corrugations directly on the waveguide during the fabrication. These structures act as a reflector (and therefore amplifier) for a certain wavelength. In external cavity setups mode control is not exerted directly on the laser, but rather with external optical elements in a corresponding setup.
Mode control via external cavities
In external cavity lasers the mode control is not established directly during the semiconductor fabrication, but rather through external elements such as gratings and mirrors (Figure 3, inset). The advantage of such a setup is the possibility to use the whole gain profile of the active region for single mode lasing. While scanning the external optical elements the amplified frequency can be continuously tuned over the whole spectral gain. On the other hand these setups tune slower than electrically or thermally tuned lasers, due to the necessity for mechanically moving parts. Figure 4 shows a summary of External Cavity-QCL tuning achieved using different active region designs. The results are compared to a standard MIR-QCL, where mode control is established via a DFB grating, to illustrate the increased tuning range of External Cavity-QCLs. Different tuning curves are achieved either in pulsed or continuous wave operation employing different active region designs.
Mode control via one- and two-dimensional photonic crystals
In case of photonic crystals one- or two-dimensional corrugations are etched close to or into the active region. Most commonly used are so-called distributed feedback (DFB) gratings. With this technique the reflectivity of a certain wavelength is increased. Light at this wavelength in turn will be amplified and the laser will emit it with high coherence and efficiency. Although the tuning ability of such a laser is then strongly reduced to a few tens of nanometer wavelength compared to external cavity setups, the tuning can be very fast, decoupled from mechanical movements and is usually sufficient to cover some absorption lines of one or more gases.
The power of a MIR-QCL is proportional to the volume of the gain medium. Thus, it is limited by the dimensions of the laser active region for DFB based QCLs. The width of the active region has to be close to the desired wavelength, in order to suppress lateral modes and maintain spectral purity. We use two-dimensional photonic crystals as a promising alternative, which allows for wider active regions and maintaining single mode emission at the same time. Owing to the increased gain volume we can achieve a multifold of the power output compared to a DFB cavity.
In our work on multicolor DFBs we combine broad-gain or stacked active regions together with multiple DFB gratings. These devices are able to lase selectively in spectrally separated locations but produce only one optical output beam. They are promising for applications for multiple trace gas sensing, since they allow to access various gases while remaining with a low number of optical laser beams. Thus the optical complexity of such a system is strongly reduce as compared to one which utilizes several individual lasers. Figure 4 sketches some of the device concepts which are able to selectively emit multi-color light. A multi-section DFB device for example unifies two colors in a single laser ridge, which can be used independently. Artefacts arising do to this device geometry can mostly be remedied by the application of anti-reflection coatings (Figure 6).
Besides the precise control of the optical mode, portable spectroscopy applications rely on a well-controlled thermal budget. The thermal strategy can be heavily influenced by the laser fabrication process, since QCLs struggle with waste heat removal at high duty cycle operation. Oftentimes, to achieve the narrow spectral linewidth for high resolution, continuous wave operation is needed. In order to cope with the thereby dissipated thermal power expensive and sophisticated techniques like episide down mounting together with strong water-cooling lines are needed. Alternatively we develop advanced processing techniques, such as the inverted buried heterostructure fabrication strategy for good thermal conductance and narrow devices. With this process, combined with an active region designs genetically automatized for low dissipation, we achieve single-mode devices with an optical output power of 5 mW, a significantly reduced power consumption and an electrical dissipation at room temperature as low as 1 W (Figure 6).