UC Santa Barbara

The III-Nitride materials system provides a fascinating platform for developing optoelectronic devices, such as solar cells and LEDs, which have the power to dramatically improve the efficiency of our power consumption and reduce our environmental footprint. Finding ways to make these devices more efficient is key to driving their widespread adoption. This dissertation focuses on the intersection of challenges in physics and metalorganic chemical vapor deposition (MOCVD) growth at the nanoscale when designing for device efficiency.

In order to create the best possible InGaN solar cell, a multiple quantum well (MQW) active region design had to be employed to prevent strain relaxation related degradation. There were two competing challenges for MQW active region design and growth. First, it was observed current collection efficiency improved with thinner quantum barriers, which promoted efficient tunneling transport instead of inefficiency thermally activated escape. Second, GaN barriers could planarize surface defects in the MQW region under the right conditions and when grown thick enough. A two-step growth method for thinner quantum barriers was developed that simultaneously allowed for tunneling transport and planarized V-defects. Barriers as thin as 4 nm were employed in MQW active regions with up to 30 periods without structural or electrical degradation, leading to record performance. Application of dielectric optical coatings greatly reduced surface reflections and allowed a second pass of light through the device. This both demonstrated the feasibility of multijunction solar integration and boosted conversion efficiency to record levels for an InGaN solar cell.

III-N LEDs have achieved state-of-the-art performance for decades, but still suffer from the phenomena of efficiency droop, where device efficiency drops dramatically at high power operation. Droop is exacerbated by the polarization-induced electric fields in InGaN quantum wells, which originate from a lack of inversion symmetry in GaN’s wurtzite crystal structure. These fields can be screened by using highly doped layers, but the extreme dopant densities predicted by simulation for complete screening may require using Ge as an alternative n-type dopant to Si. GaN:Ge layers with excellent electrical characteristics were grown by MOCVD with doping densities exceeding 1020 cm-3. However, their surface morphologies were very poor and they proved a poor screening dopant in LED structures. Using Si as the n-type screening dopant, LEDs with single QW active regions were grown, packaged, and tested. Biased photoluminescence showed strong evidence of complete polarization screening. The LEDs had low droop, but also low peak efficiencies. Possible explanations for trends in efficiency with varying QW width and field screening will be discussed.