Development of 1.3-μm GaAs-based vertical-cavity surface-emitting lasers

Abstract: Long-wavelength vertical-cavity surface-emitting lasers (VCSELs) are desirable as low-cost sources for optical metropolitan-area and access networks. In the development of 1.3-µm VCSELs, most attention today is given to monolithic GaAs-based solutions, although no established active material exists in this wavelength region. This thesis investigates the possibility of reaching the 1.3-µm telecom wavelength window using GaInNAs quantum wells (QWs) or 1.2-µm InGaAs QWs in conjunction with negative gain-cavity detuning in VCSELs. The work includes metal-organic vapor-phase epitaxy and characterization of InGaAs and GaInNAs QWs, realization of 1.3-µm InGaAs VCSELs as well as elements of optimization and analysis of such lasers. The evaluation of GaInNAs and InGaAs QWs has been performed using a number of characterization methods such as photoluminescence (PL), high-resolution x-ray diffraction, secondary-ion mass spectroscopy, and atomic-force microscopy as well as fabrication and evaluation of broad-area lasers (BALs). Both performance and growth reproducibility of GaInNAs QWs are considered and could be improved by using high V/III ratios. Nontrivial relations between PL and laser performance are pointed out and the technologically important but problematic combination of AlGaAs and GaInNAs in the same epitaxial structure is studied. Parallel to the work on GaInNAs, the possibility of extending the wavelength of InGaAs QWs towards 1.3 µm has been investigated. Generally better luminescence efficiency and laser performance are obtained for InGaAs than for GaInNAs QWs, but the gain-peak wavelength for InGaAs QWs is presently limited to about 1.24 µm due to strain-induced degradation. In this work the InGaAs QW growth is optimized for long wavelength and high luminescence. It is demonstrated that multiple QW structures can be grown with strain similar to that of single QWs, which is interesting for VCSEL applications. Record BALs with two to five InGaAs/GaAs QWs have low threshold current densities,  70 A/cm2 per QW at 1.24 µm. The main advantage of InGaAs QWs compared to GaInNAs QWs is that they represent a better-known material system with less complex and more stable growth. However, InGaAs QWs > 1.2 µm are on the verge of strain relaxation, and the possible consequences for laser production and reliability have to be considered. Using 1.2-µm InGaAs QWs, high-performance 1.3-µm VCSELs were achieved by negative gain-cavity detuning. Dynamic performance and surface reliefs to improve the single-mode operation have been investigated. The VCSELs have excellent high-temperature performance due to a smaller spectral distance between the gain-peak and the laser mode at elevated temperature. More specifically, a 1.27-µm single-mode device showed maximum output powers of 1.1 and 0.5 mW at 20 and 140ºC, which is state-of-the-art for GaAs-based long-wavelength VCSELs. In all, two methods for 1.3-µm GaAs-based VCSELs, GaInNAs and InGaAs QWs, have been investigated. GaInNAs is a difficult material but is still promising and several companies have predicted a near-future market introduction. However, the growth of GaInNAs is both complex and sensitive to growth fluctuations. On the other hand, gain-cavity detuned InGaAs-QW VCSELs show state-of-the-art performance at 1260-1290 nm with straightforward growth and processing. The devices exhibit good static and dynamic performance, and preliminary reliability tests indicate that there is no intrinsic problem. Both approaches are promising for application in real-world optical networks and deserve further attention.