UC Santa Barbara

III-N vertical-cavity surface-emitting lasers (VCSELs) show promise for numerous communications, lighting, display, and sensor applications due to their low threshold current, high beam quality, and arraying capabilities. Primarily, research has been focused on using c-plane based devices, but non-basal growth planes provide an interesting alternative due to a reduced quantum confined Stark effect; higher material gain; lower transparency current density; and inherent polarized emission. The anisotropic gain leads to VCSELs and VCSEL arrays where each laser is polarization locked along the a-direction. At UCSB, an m-plane VCSEL was first demonstrated in 2012 under pulsed injection and in 2018 under CW operation. Through that time, the device performance has improved and the polarization properties of the VCSELs has been experimentally verified. However, the wavelength of m-plane lasers is severely limited due to poor indium incorporation and high defect formation, inhibiting their adoption in many applications. This led to the question of how the benefits of using m-plane can be retained, such as the inherent polarization, while expanding the available wavelengths.

The answer that was developed in this thesis is the use of a semipolar growth plane with higher indium uptake. After developing an epitaxial growth recipe and optimizing processing parameters for semipolar planes, we achieved the first demonstration of semipolar (20(21) ̅) VCSELs which were experimentally shown to be polarization locked along the a-direction and emit in the blue region. The devices had a 5λ cavity length, an ion implanted aperture, and a dual dielectric DBR design and showed an improvement in the differential efficiency, threshold current density and total output power relative to m-plane VCSELs with the same design. However, there were issues. The devices were only able to lase under pulsed operation, up to a 70% duty cycle. Focused ion beam images in conjunction with COMSOL modeling was used to identify the key structural features that contributed to the high measured thermal impedance. Nearfield images suggest that the LP01 mode was lasing near the edge of the aperture. This commonly observed spatial misalignment introduced additional sources of loss beyond the expected material absorption loss, including mode overlap with the implanted region and the metal contacts. The effect of these absorbing layers on the device performance relative to simulation models was estimated and highlighted the need for proper mode control.

To improve the optical confinement, devices using a buried tunnel junction (BTJ) scheme to confine the current were fabricated and were found to lack the excess losses due to absorption seen on the initial semipolar samples. Significant filamentation was observed on these samples and several characterization methods, including optical and thermal nearfield images, were used to identify the source of the filamentation. Further comparison of multiple BTJ samples with different index guiding showed that the mode behavior was driven by the interplay of inhomogeneous current injection and index guiding. The cause of the inhomogeneous current injection is projected to be due to doping variations in the p-GaN but still requires further investigation for verification.