In recent years, the technology of the quantum cascade laser (QCL) has drastically improved, as proven by room temperature continuous wave (CW) operation, high optical power, low dissipation devices, and accessible spectral range [1]. As a result, QCLs have drawn a great deal of interest for their various applications in trace-gas sensing, defense countermeasures, spectroscopy and free-space communications [2]. The performance of QCLs is primarily dependent on the design of the active region, usually gown by MBE (molecular beam epitaxy). Nevertheless, the laser performance are also influenced by the thermal properties and optical losses introduced by the fabrication technology, in particular in continuous-wave operation. In that respect, the buried heterostructure (BH) approach, which is a standard and reliable telecommunications laser manufacturing technique, offers unique possibilities to achieve effective thermal dissipation and lateral single modes. In standard BH designs, the process starts by the definition and etching of the ridge mesa using a dielectric mask. The latter (SiO2 or SiN) is also used as the masking material for the selective regrowth by metal-organic vapor phase epitaxy (MOVPE) of a thick insulating InP (Fe) layer over the entire laser structures. Unfortunately, the selective regrowth of InP:Fe is not a trivial process. Indeed, it is extremely difficult to achieve an ideal planarized regrowth, i.e, completely burying the laser ridge, while maintaining a planar surface. In addition to growth conditions, the characteristics of the regrown InP layer strongly depend on the sidewall profile of the ridges, as well as on the physical parameters of the dielectric mask (thickness, width…) [3]. The complexity of the etching and regrowth process further increases for thick layer structures, such as for long wave QCLs, whose ridges can be >10 µm deep. In effect, deep ridge etching generates big undercuts in the structures and consequently, highly non uniform active region widths. Moreover, the thick selective regrowth step, needed for planarization, tends to favor the appearance of crystal growth artifacts (crystallites, voids, “rabbit ears”…) in the vicinity of the ridges, which deteriorate the performance of the regrown structures. In some cases, additional photolithographic and back-etching steps might even be required to improve the planarity of the structure. Another method that allows overcoming the aforementioned limitations is to use a modified buried heterostructure concept, with two MOVPE regrowth steps [4]. In this case, the waveguide pattering and etching is constituted only by the active layers (<3 μm). Hence, the active region is laterally insulated and planarized using selective growth of a much thinner InP:Fe layer. The dielectric layer is subsequently removed and the optical cladding is grown over all the structure. An additional advantage of a thinner regrowth is the reduced amount of defects due to the selective growth of InP:Fe. However, the undercut formed by the overhanging dielectric mask (needed for the selective growth in first MOVPE step), and the tapered profile of the active region can result in a reduced material thickness close to the upper edges of the active region. In this case, the material thickness can be too thin and induce carrier tunneling through the interface between the active region and the lateral regrowth (insufficient electrical isolation). In this perspective, we report an optimized BH-QCL approach obtained by adding an InP layer and a sacrificial InGaAs cap layer on top of the MBE grown active region. As shown in the attached figures (FIB/SEM cross-section profiles), the lateral thickness of the insulating and planar regrowth is increased and reduces potential leakage paths. Using this approach, BH QCL regrowths were successfully carried out with reproducible and good epitaxial growth quality (smooth and uniform regrown layers, planar surfaces, no “rabbit-ears” and defect free interfaces) around the etched ridges, permitting the fabrication of BH-QCLs with good beam quality and hundreds of mW of optical power at room temperature. The growth and optimization of the BH process, as well as the fabrication and testing of the laser devices will be presented and compared to conventional double trench devices.
Read full abstract