(Invited) Achieving High-Power Single-Mode Operation in Vertical-Cavity Surface-Emitting Lasers Via Scalable, Higher-Order Mode Suppression Techniques

Abstract

Achieving high-power single-mode operation in Vertical-Cavity Surface-Emitting Lasers (VCSELs) has received renewed interest because of applications that span facial recognition and 3D imaging in mobile telephones, Light Detection and Ranging (LiDAR) for autonomous vehicles, and Pulse Amplitude Modulation (PAM-4) for high-speed (>100 Gbps) data transmission within data centers. A key challenge with oxide-confined VCSELs is that the strong refractive index contrast between the oxidized ~AlAs aperture and surrounding AlGaAs layers combined with the oxide aperture size create a photonic cavity that fundamentally supports multiple transverse modes. When the size of the aperture is reduced to create a single-mode cavity, the output power is limited by the electrically pumped gain region within the aperture. There is thus a fundamental trade-off between size and power which manifests itself in a limited capability to simultaneously achieve high power and single mode operation. Various methods have been employed by groups working on VCSELs to overcome this limitation. Neglecting external cavity methods which are fundamentally not scalable, the most promising results have been achieved by groups using etched surface relief where a pattern is etched in the top layer or layers to change the standing wave pattern in the cavity and reduce modal gain by lowering the magnitude of the electric field in regions overlapping with the quantum wells, which provide laser gain. A downside of this approach is that material is removed, impacting the uniformity of current injection across the area of the device. A key question of interest is therefore how to achieve mode suppression in large area devices in a way that does not impact current injection. In this presentation, novel methods for mode control that rely on spatial variation of modal gain across the area of the VCSEL aperture will be discussed, and results on high-power single-mode operations of VCSELs will be presented. Two distinct mode control methods will be reviewed. The first method of mode control uses deposited optical coatings on the top surface of the VCSEL to spatially modify the electric field standing wave in the quantum well heterostructure region. Through patterning the coating in either a pillar or anulus, specific modes can be suppressed by selectively changing the gain available for those modes. These filters have been made using both SiO2/TiO2 stacks and alternatively a single layer of silicon. The capability to pattern these layers in a wafer-scale process makes this method attractive for volume manufacturing. The second method of mode control utilizes impurity-induced layer disordering via patterned diffusion of Zn to selectively intermix Distributed Bragg Reflector (DBR) layers, thereby creating a top laser mirror whose reflectivity varies spatially across the aperture. Designing the diffused region to maximize overlap with high-order modes while minimizing overlap with the fundamental mode provides mode suppression and enhanced single-mode power. An added benefit of this approach is a reduction in resistance through the device top DBR layers brought about by the higher p-doping level in the Zn-diffused regions, minimizing Ohmic losses and heat generation. Through the use of strain control in the diffusion mask, diffusion front shape is also shown to be controllable in a way that can be used to minimize the impact of the diffused region on fundamental mode reflectivity. Tensile and compressive strain of diffusion masks are shown to have differing effects on vertical (into the device) versus lateral (across the area of the device under the mask) diffusion, allowing more precise spatial control of reflectivity. Using these methods, single-mode output powers in excess of 10 mW have been demonstrated, with side-mode suppression ratios in excess of 30 dB. These results will be discussed, as well as further work to enhance VCSEL performance. Figure 1