Crystal growth of sapphire for substrates for high‐brightness, light emitting diodes

Abstract

Sapphire has emerged as the preferred substrate for high‐brightness, light emitting diodes (HB‐LEDs) made with a GaN epilayer. While the growth of most crystalline materials for electronic and laser applications normally converges onto a single, preferred method—Czochralski growth for silicon, lithium niobate and YAG (yttrium aluminum garnet), vertical gradient freeze (VGF) and Liquid Encapsulated Czochralski (LEC) for GaAs crystals (depending upon current density and device application), vapor transport for SiC, for example—a wide range of processes are employed for sapphire growth. These include Czochralski, Kyropoulos, EFG (Edge‐defined, Film‐fed Growth), Bagdasarov, classical Bridgman and several variants of Bridgman such as HEM (Heat Exchanger Method) and CHES (Controlled Heat Extraction System).This variety of approaches to grow crystals of what is basically the same material reflects a balance between acceptable defect levels, the costs of downstream fabrication steps such as wire sawing, and the wafer diameters demanded by the market. Each process has advantages and disadvantages. The Kyropoulos method, for example, probably yields crystals with the lowest defect levels, while EFG eliminates the wire sawing stage of wafer fabrication which is a major cost factor. Large sapphire crystals grown by HEM are frequently oxygen deficient, and a‐axis Czochralski crystals have higher dislocation densities than Kyropoulos boules.Facet formation on the c‐plane is an important criterion in selecting the growth methodology for sapphire. However, we will show that melt grown sapphire crystals have a curved interface that passes through other facet planes, and this can impact the growth process.In this paper we will summarize the various crystal growth methods currently employed to grow sapphire, discuss the strengths and weaknesses of each approach and also discuss the often overlooked thermodynamic aspects of this high‐temperature crystal growth process. We will attempt to show the long‐term trends in the evolution of the various growth processes and predict the ultimate winner of the contest to supply substrates.