Weighing up the options for gallium oxide crystal growth
What are the pros and cons of producing β-Ga2O3 by the two most common techniques, the Czochralski method and edge-defined film-fed growth?
BY JANI JESENOVEC AND JOHN MCCLOY FROM WASHINGTON STATE UNIVERSITY.
There’s much excitement surrounding ultra-wide bandgap materials, such as Ga2O3, AlN and diamond. They are renowned for a very high intrinsic voltage breakdown field, which makes them exceptional candidates for handling high powers.
Within this family of semiconductors, interest in Ga2O3 is on an exponential trajectory. As well as its great potential for making incredibly small and efficient power-switching devices that trim thermal losses, it is attracting attention for the fabrication of deep-UV detectors that are transparent to visible wavelengths.
For developers of all forms of Ga2O3 devices, a key decision is the choice of polytype. There are many to decide among, but arguably the most important of all is β-Ga2O3, the room-temperature stable phase. One of its most promising features is that it can be alloyed with Al2O3 to increase the bandgap and push the transmission window deeper into the UV.
Efforts at understanding the nature of β-Ga2O3 rely on the growth of high-quality material. This can be in the form of a bulk single crystal, or a thin film that is grown by one of many methods. Today, high quality single crystal films of β-Ga2O3 may be deposited by MOCVD, HVPE and MBE.
One of the merits of bulk β-Ga2O3 is that, unlike the other ultra-wide bandgap materials, it can be formed from the melt, using techniques such as the Czochralski crystal growth technique, which has been employed for the manufacture of silicon wafers for many years. Crystals of this oxide can also be produced by edge-defined film-fed growth, a technique employed for commercial production of β-Ga2O3. In addition, research is underway into the growth of β-Ga2O3 by other methods, such as the vertical Bridgman and optical float zone techniques.
In the remainder of this feature we shall review the progress made with the more common methodologies and detail insights provided by characterisation. Within this survey, we include a brief account of the contribution to this field by our team from the Institute of Materials Research at Washington State University.
Of the techniques we have mentioned for crystal growth, the two primary approaches – the Czochralski method and edge-defined film-fed growth – have much in common. They both involve pulling solid single crystals out of molten Ga2O3, held at temperatures above 1800 °C and housed in iridium metal crucibles. Ga2O3 produced by the Czochralski method tends to be cylindrical, while that formed by edge-defined film-fed growth involves the use of iridium capillaries, which can draw the material into the desired shape.
Pioneering β-Ga2O3 produced by the Czochralski method is a team at the Leibniz Institute for Crystal Growth, Germany. Commonly known as IKZ, this institute has demonstrated marked success, detailed in many publications and several patents.
Famous for producing β-Ga2O3 via edge-defined film-fed growth is Novel Crystal Technologies (NCT). This Japanese outfit, which manufactures and sells wafers of doped or un-doped β-Ga2O3, touts that crystals produced by this technique are superior to those made by the Czochralski method, due to the potential to produce very large wafers. However, edge-defined film-fed growth usually forms thin crystals, so the overall pulled volume is lower than that associated with the Czochralski method, which has the potential to scale and yield large cylindrical single crystals.
Delving into doping
For all semiconductors purity is critical. When small concentrations of other atoms are present, or defects added, this strongly affects the bulk properties. In the case of β-Ga2O3, there is the opportunity to intentionally replace the gallium that sits on its atomic site with other elements, via doping, to induce optical or electrical phenomena; and there is the threat that undoped β-Ga2O3 can be plagued with impurities. Common impurities in β-Ga2O3 produced by any method include silicon, iron, and chromium; iridium may also be introduced when crystals of this oxide are formed by the Czochralski method or edge-defined film-fed growth. Metallic impurities tend to originate from the precursor powders used in synthesis, or from the crucible that can contaminate the melt at high temperatures. Due to this, undoped β-Ga2O3 is referred to as ‘unintentionally doped’ material. Typically, this contains substantial background impurities, acting as dopants and enabling electrical conduction.
Intentional doping of β-Ga2O3, a topic undergoing much research, offers control over the electronic behaviour of this oxide. Material that is n-type, and thus dominated by electron conduction, may be formed by doping with silicon, germanium, tin, zirconium or hafnium. If electrically insulating behaviour is desired, appropriate dopants include iron, magnesium, zinc, nickel and copper. Realising p-type behaviour has proved far trickier. Some annealing experiments suggest marginal p-type behaviour may be realised due to hydrogen interaction, but this behaviour has not been observed in as-grown bulk crystals. Maybe this is not surprising, given that this difficulty has been observed in ZnO, another transparent semiconducting oxide. In the case of ZnO, this weakness has eventually limited the application of the material.
A property that may pique the interest of many is photodarkening, which has been demonstrated in copper-doped β-Ga2O3 by the Institute of Materials Research at Washington State University. Uncommon in semiconductors, this novel effect is long-lived in copper-doped β-Ga2O3, with UV excitation causing a sample held at room-temperature to darken and remain in that state for weeks (see Figure 2). Note that heating accelerates the reversal of this darkening.