Group-III nitride materials, gallium nitride (GaN), aluminum nitride (AlN) and indium nitride (InN) have direct band gaps with band gap energies ranging from the infrared (InN) to the ultraviolet (GaN) and to the deep ultraviolet (AlN) wavelengths and covering the entire spectral range from 0.7 eV to 6.2 eV upon alloying. The invention of the GaN-based blue LEDs, for which the Nobel prize in Physics was awarded in 2014, has opened up avenues for exploration of IIINitride material and device technologies and has inspired generations of researchers in the semiconductor field. Group-III nitrides have also been demonstrated to be among the most promising semiconductors for next generation of efficient high-power, high-temperature and high-frequency electronic devices.

The need to build a sustainable and efficient energy system motivates the development of vertical GaN transistors and diodes for applications with power ratings of 50-150 kW, e.g., in electric vehicles and industrial inverters. The key is to grow GaN layers with low concentration of defects (impurities and dislocations), which enables an expansion in both voltage and current ratings and reduction of cost. Despite intense investigations and impressive advances in the field, defects are still a major problem hindering exploiting the full potential of GaN in power electronics. This Licentiate thesis focuses on the development of two different epitaxial approaches in MOCVD for reducing dislocation densities in GaN with controlled doping for power device applications: i) growth of planar GaN layers trough NWs reformation, which can be further exploited as templates for a subsequent growth of thick drift layers and ii) homoepitaxial GaN growth. Special attention is put on understanding homoepitaxial growth under different nucleation schemes and thermal stability of GaN. We have established conditions in homoepitaxy to deliver state-of-the-art GaN material with low impurity levels combined with a reasonable growth rate suitable for growth of thick drift layers.

The results are summarized in two papers: In Paper I we investigate GaN layers with different thicknesses on reformed GaN NW templates and highlight this approach as an alternative to the expensive GaN HVPE substrates. The sapphire used as a substrate limits to some extent the reduction of threading dislocations, however, the resulting GaN material presents smooth surfaces and thermal conductivity close to the bulk value, which suggests the potential of this approach to be integrated in GaN development as an active material for power devices on various substrates. In Paper II extensive study of homoepitaxial GaN growth by hot-wall MOCVD is presented together with results on the thermal stability of GaN under typical conditions used in our growth reactor. Understanding the evolution of GaN surface under different gas compositions and temperatures allows us to predict optimum homoepitaxial conditions. Analysis in the framework of Ga supersaturation of epilayers simultaneously grown on GaN templates and on GaN HVPE substrates reveals that residual strain and screw dislocation densities affect GaN nucleation and growth and lead to distinctively different morphologies on GaN templates and native substrates, respectively. The established comprehensive picture provides guidance for designing strategies for growth conditions optimization in homoepitaxy. We demonstrate homoepitaxial GaN-on-GaN grown under optimum growth conditions with state-of-the-art smooth surface with an rms value of 0.021 nm and an average TDD of 1.4·10⁶ cm_-2 which provide good basis for augmenting power device structures.Future work will be focused on GaN NWs reformation on different substrates, p- and n-type doping of homoepitaxial GaN with impurity control and the fabrication of pn power diode device structures for further processing and assessment by C₃NiT partners.

Group III-nitride materials, gallium nitride (GaN), aluminum nitride (AlN) and indium nitride (InN) have direct band gaps with band gap energies ranging from the infrared (InN) to the ultraviolet (GaN) and to the deep ultraviolet (AlN) wave-lengths, covering the entire spectral range from 0.7 eV to 6.2 eV upon alloying. The invention of the GaN-based blue LEDs, for which the Nobel prize in Physics was awarded in 2014, has opened up avenues for exploration of III-Nitride mate-rial and device technologies, and has inspired generations of researchers in the semiconductor field. Group III-nitrides have also been demonstrated to be among the most promising semiconductors for next generation of efficient high-power, high-temperature and high-frequency electronic devices. 

The need to build a sustainable and efficient energy system motivates the development of vertical GaN transistors and diodes for applications with power ratings of 50-150 kW, e.g., in electric vehicles and industrial inverters. The key is to grow GaN layers with low concentration of defects (impurities and dislocations), which enables an expansion in both voltage and current ratings and reduction of cost. Despite intense investigations and impressive advances in the field, defects are still a major problem which hinders exploiting the full potential of GaN in power electronics. 

The aim of this thesis is to perform an in-depth investigation of the growth of GaN and AlGaN under several nucleation mechanisms provided by different underlying substrates. In that regard, four different epitaxial approaches based on different nucleation schemes have been studied: (i) growth of planar GaN layers trough NWs reformation. We investigated GaN layers with different thicknesses on reformed GaN NW templates and highlight this approach as an alternative to the expensive HVPE GaN substrates. The sapphire used as a substrate limits to some extent the reduction of threading dislocations, however, the resulting GaN material presents smooth surfaces and thermal conductivity close to the value for bulk GaN. (ii) Homoepitaxial GaN growth. We developed a hot-wall MOCVD epitaxial approach that enables low surface roughness and appropriate impurity levels for advanced vertical power device architectures. A comprehensive picture of GaN homoepitaxy on different GaN surfaces, GaN templates on SiC and HVPE GaN substrates, is established on the basis of experimental results and thermodynamic considerations. (iii) GaN growth on GaN NWs templates by hot-wall MOCVD resulted in an atomically flat smooth surface with reduction of threading dislocations when the optimum annealing conditions have been employed. (iv) Heteroepitaxial growth of low Al composition n-Al_xGa_1-xN on SiC substrates revealed 700 nm crack-free epi-layers for an Al composition up to 12%. The highest mobility corresponds to an Al content of 6.5% where we also get a reduction in screw and edge dislocations. The results show the potential application of Al_xGa_1-xN(x= 0 - 0.12) as the active material for drift layers. 

Some of the epitaxial approaches developed in this thesis have been already implemented in the growth of power devices such as quasi-vertical GaN FinFETs on SiC substrates and fully-vertical GaN FinFETs on HVPE GaN substrates. 

Grupp III-nitrider är halvledare med direkta bandgap där bandgapsenergierna spänner från det infraröda till djupt ultravioletta banden. Tillräknade i den gruppen är galliumnitrid (GaN), aluminiumnitrid (AlN) samt indiumnitrid (InN) som tillsammans kan realisera alla bandgapsenergier från 0.7 eV (InN) till 6.2 eV (AlN) genom legering. Utvecklingen av GaN-baserade blå LED:er, som tilldelades 2014 års Nobelpris i fysik, har öppnat många nya dörrar inom III-nitridforskning och skapat många nya tillämpningar av halvledarmaterial. Till exempel har grupp III-nitrider påvisats mycket lovande som nästa generations högeffekts- och högfrekvenskomponenter inom elektroniken. Efterfrågan på hållbara och effektiva energisystem har drivit utvecklingen av vertikala GaN-transistorer och dioder för tillämpning inom 50-150 kW omfånget, så som elektriska fordon och industriella växelriktare. Nyckeln ligger i att växa lager av GaN med låg konsentration av defekter (orenheter och dislokations), som både kan öka spänningsfönstret och strömstyrkan och samtidigt reducera kostnaden. Defekter har däremot varit svåra att kontrollera och trots mänger av framsteg är det fortfarande den stora utmaningen för att fullt kunna utnyttja potentialen av GaN inom elektronik.

Målet i denna avhandling är att utföra fördjupade undersökningar av GaN- och AlGaN-tillväxt vid olika betingade tillväxtmekanismer som funktion av tillväxtsubstrat. Fyra olika epitaxiella tillvägagångsätt har studerats med tillhörande nukleationsmekanismer. (i) Tillväxt av plana GaN-lager genom nanotråd-reformation. Vi har undersökt GaN med olika tjocklekar på omformade GaN nanotråd-mallar och påvisar att metoden är ett alternativ till dyra HVPE GaN-substrat. Safiren som används som substrat begränsar till viss del en reducering av slingrande dislokationer men den resulterande GaN-ytan är jämn och har en termisk ledningsförmåga nära GaN i bulk. (ii) Homoepitaxiell GaN-tillväxt. Vi utvecklade en hetväggs MOCVD-epitaxi som möjliggör en låg ytojämnhet och en låg nivå av orenheter för avancerade vertikala högeffektsarkitekturer. En omfattande ter-modynamisk och experimentell bild har etablerats av GaN homoepitaxi på olika GaN-ytor, GaN-mallar på SiC samt på HVPE GaN-substrat. (iii) GaN tillväxt på GaN nanotrådmallar via hetväggs MOCVD resulterar i en atomärt jämn yta med en reducering av slingande dislokationer när optimerad glödgning har utförts. (iv) Heteroepitaxiell tillväxt av låg-nivå Al inblandning i n-Al_xGa_1-xN på SiC-substrat leder till 700 nm sprickfria epi-lager, för Al-inblandning upp till 12%. Den högsta uppmätta mobiliteten ficks vid 6.5% Al där också en reducering av skruv- och kant- dislokationer noterades. Resultaten visar på potentialen för n-Al_xGa_1-xN (x = 0 - 0.12) som aktiva material i driftlager.

En del av de epitaxiella tillvägagångssätten som utvecklats i denna avhandling har redan implementerats i tillväxt av högeffektskomponenter så som quasi-vertikala GaN FinFETs på SiC och vertikala GaN FinFETs på HVPE GaN-substrat.