【World Express】Championing a true heavyweight material: Releasing the hope of Ga₂O₃

      With its incredible band-gap width, shallow donor energy level, and ability to grow by melt, Ga₂O₃  is the most promising material ever for power electronics.

      In addition to the record number of delegates and packed showrooms, those attending this year's CS International will remember the conference for the phenomenal level of interest in wide-gap power electronics. The field is incredibly exciting, as sales of SiC and GaN devices and modules are set to soar over the next decade, climbing from a combined annual revenue of less than $2 billion today to around $18 billion by 2030. This is good news not only for chipmakers, but also for those companies that make related materials, inspection instruments and device design software.

      The adoption of these two new categories of power electronics will benefit humanity. The conversion efficiency of electricity between alternating current and direct current will be improved. This has led to a reduction in the carbon footprint of many electrical systems. What's more, electric cars can go farther on a single charge, which helps allay concerns about range anxiety and makes a more compelling case for abandoning internal combustion engines.

      But will the revolution in power electronics end with the advent of SiC and GaN? Or is there more to come?

      Three speakers at CS International, Martin Kuball (pictured above), head of the Centre for Thermal Imaging and Reliability of Devices at the University of Bristol; Akito Kuramata, chief executive of Novel Crystal Technology(NCT), a manufacturer of Ga₂O₃ substrates; And Heather Splawn, CEO of HVPE specialist Kyma Technologies, both see Ga₂O₃  as a heavyweight material capable of handling incredibly high voltages and delivering switching devices with much higher efficiency compared to SiC and GaN. These properties make this oxide a very promising candidate for the ultra-high voltage market, where it could be deployed to support power grids, process electricity generated by wind turbines, and electric train applications.

      In terms of its physical properties, Ga₂O₃ is known for its wide band gap of 4.9 eV and a breakdown electric field of up to 8MV cm^-1. But how does it fare when benchmarking against the Baliga coefficient, a common standard for measuring the potential of power electronic devices? At first, although oxide seemed well ahead of SiC and GaN, it lagged behind two other heavyweights - Diamond and AlN. However, when one considers the doping level depth, which is a key factor in determining the capability of a power device, β-Ga₂O₃ has the upper hand due to the relatively shallow doping level (see Figure 1).

Figure 1: Once doping activity levels are taken into account, Ga₂O₃ emerges as the most promising UWB semiconductor according to the Baliga coefficient. This chart is taken from Y. Zhang et al. Semicond.Sci.Technol.35 125018 (2020).

      Another major advantage of Ga₂O₃ is the relative ease of crystal growth. Like Silicon, GaAs, and InP, it can be grown from melt, ensuring that substrates with low dislocation density can be produced with relative ease. But the same is not true of SiC and GaN, the former tends to be produced by vapor phase transmission method; The latter does not yet have the bulk material growth technology suitable for mass production of homogeneous power electronics.

Hurdles to overcome

      Because of all these advantages, devices made from Ga₂O₃  have achieved impressive results. "The performance has surpassed that of Silicon Carbide," Kuball said in a speech to delegates at CS International 2023. That doesn't mean commercial success is guaranteed, however, in part because the devices still have defects.

      Rather than dismiss these concerns, Kuball discusses them head-on. One such weakness is the high density of defects in the material, and the deep-level transient spectrum reveals many different defect states in the β-Ga₂O₃. More work needs to be done because it is still largely unclear what the fatal defect in this oxide might be.

Akito Kuramata is CEO of Novel Crystal Technology, a supplier of Ga₂O₃ substrates and epitaxial wafers. The company plans to expand into the production of diodes and transistors based on this oxide.

      Another concern is Ga₂O₃'s low thermal conductivity, which leads many to claim that the oxide will never be a viable material for power electronics due to overheating of the chips. But this problem can be solved with engineering, and Kuball argues that introducing diamond next to the active area could effectively take away heat. His team has already used this approach to improve the thermal management of GaN devices.

      The addition of diamond can actually provide simple and superior thermal extraction results. Kuball team's results show that electrical control of the device can be achieved by filling the N-type Ga₂O₃  with P-type diamond grooves formed by Superjunction Schottky Barrier Diodes. This is encouraging because Ga₂O₃  does not have P-type doping, which is a major problem, and integrating other P-type semiconductors is a promising solution.

      When any device is still in its infancy, people always worry about its reliability. Kuball and his colleagues have been studying the Ga₂O₃ trench FETs provided by Cornell University, and their experiments have found that the Al₂O₃ dielectric layer is faulty. Efforts now need to be made to improve the interface between Ga₂O₃ and Al₂O₃.

      The Bristol team also produced the trench Schottky Barrier Diode. Benchmarking these devices without field plates showed performance comparable to other leading teams in the field.

      In May 2022, Kuball's team enabled Europe's first commercial Ga₂O₃ MOCVD reaction grower-the Agnitron Agilis tool. Since then, they have enjoyed the opportunity to grow their own material, producing Ga₂O₃ and Al_x(Ga₂O₃)_1-x epitaxial layers for a variety of structures, including vertical devices.

      To improve the thermal conductivity of the device, they investigated a two-step growth technique on diamond, Kuball explained. The transmission electron microscope and scanning electron microscope images show the perfect surface combination of the 245 nm thick β-Ga₂O₃ layer, as well as different competing crystal orientations. The physical properties of the film are not much different from those of β-Ga₂O₃ grown on a sapphire substrate, and Kuball believes the experimental results so far have been very promising.

Figure 2: NCT plans to increase the diameter of the substrates it produces and expand its business to the production of Ga₂O₃ diodes and transistors, based on the Baliga coefficient. Figure taken from Y.Zhang et al. Semcond.Sci.Technol.35 125018 (2020)

Method of growth of crystal ingot

      NCT's Kuramata gave more insight into the growth of Ga₂O₃ bulk materials in his talk, where he discussed two growth techniques NCT uses to produce commercial materials: the EFG method and the Vertical Bridgman method.

      While the Sasayama-based Japanese company is better known for the production of substrates and epitaxial sheets, it also plans to produce chips, devices and their packages. Kuramata presented the company's roadmap, which includes the launch of 150mm and 200mm wafers in the next five years, as well as diodes and transistors in 2023 and 2025 respectively (see Figure 2 for details). The company's diodes are designed to provide a better price/performance ratio than Silicon Carbide Schottky Barrier Diodes, and transistor performance is expected to surpass Silicon IGBT by the same standards.

Figure 3: For high-speed growth of large diameter Ga₂O₃ materials, EFG growth provides unparalleled results.

      For the production of larger size materials, the EFG growth is leading the way, and NCT has reported the development of materials with a size of 6 inches. "This is currently the only way to produce large-size N-type substrates," Kuramata said.

      Another advantage of EFG growth is that it is the fastest growing method of all Ga₂O₃ growth techniques - the growth rate is 15mm/ hour, which is three times that of zone melting, the Cz method is only 2mm/ hour, and the Vertical Bridgeman method is only 1 mm/ hour.

      In EFG growth, engineers use capillary action to pull the molten Ga₂O₃ through the slit and grow on the seed crystal (see Figure 3). This produces a crystalline material with a plate-like geometry that typically has a defect density of about 10³ cm^-2.

Using the Vertical Bridgman technique, NCT has produced a 2 inch (010) substrate.

      Since the Ga₂O₃ industry is still in its infancy, it is not surprising that there are no orders for the 6-inch material today. However, Kuramata is confident that once such an order is received, NCT will quickly establish production of materials of this size.

      The Japanese company is also developing the production of Ga₂O₃ small size ingot by Vertical Bridgman method, as the material produced by this process is of very high quality. Growth requires a crucible made of a platinum-rhodium alloy and (010) oriented Ga₂O₃ seed crystals. By carefully controlling the movement of the crucible through the temperature gradient created in the furnace, the melted Ga₂O₃ is cooled to grow out the ingots. With this method, NCT has produced the growth of a 2-inch (010) substrate - purported to be the largest size of a (010) oriented Ga₂O₃ substrate reported to date.

      For the growth of the epitaxial layer, NCT used HVPE technology introduced from Tokyo University of Agriculture. NCT collaborated with Saga University to produce Schottky barrier diodes on 100 mm β-Ga₂O₃ epitaxial sheets with chip sizes up to 10 mm by 10 mm. For 10 mm thick films, the uniformity of film thickness is ±5%; When the donor concentration is 1×10¹⁶cm^-3, the change in film thickness is ±7%.

      The diode yield of this wafer with a side length of 10mm×10mm is up to 51%. Based on this figure, the density of fatal defects in epitaxial wafers is about 0.7cm^-2.

Espouse HVPE

      Another advocate for high-voltage power electronics used in Ga₂O₃ is Kyma Technology leader and HVPE expert Heather Splawn. She argued at this year's CS International that this form of epitaxy can produce low-cost and high-performance materials with high growth rates that are ideal for device fabrication. Splawn also notes that HVPE is inherently cleaner than MOCVD due to the high chemical purity of the chemical precursor and the absence of carbon-containing metallic organics. Thanks to this advantage, the HVPE process is able to achieve high purity growth.

      Kyma has introduced to the market a tool called Katharo, which is designed for the growth of Ga₂O₃ devices for high-power switches. The reactor can accommodate wafer up to 200mm in diameter.

Heather Splawn, CEO of Kyma, advocates HVPE for high-voltage power electronics.

Figure 4: NCT uses Vertical Bridgman method to grow high quality Ga₂O₃.

      While there is still some way to go to achieve growth on such a large diameter, the company has already achieved encouraging results on smaller substrates. According to Splawn, excellent doping control can be achieved in a 100mm HVPE reactive growth device when the epitaxial layer thickness can exceed 20μm. Her team also deposited Ga₂O₃ epitaxial layers of uniform doping and thickness on a 50 mm wafer, and X-ray diffraction measurements indicated very high crystal quality - only 25-30 arcseconds in half-height width.

Using HVPE, Kyma has grown Ga₂O₃ layers 16.7 microns thick on a 2-inch self-supporting substrate with a thickness variation of ±3%. The difference in donor and recipient concentrations in this epitaxial layer ranges from 2.7×10¹⁶cm^-3 to 5.8×10¹⁶cm^-3

      Materials produced by Kyma have been used to produce some devices with very encouraging results. These devices have a breakdown field strength of around 5.5mV cm^-1 and a Baliga coefficient of more than 1GW cm^-2, which is very close to the theoretical limit of SiC.

      Such results highlight the huge promise of Ga₂O₃. With SiC and GaN seemingly having a bright future, it will be some time before this oxide really makes a significant impact, but there is no doubt that the omens are in the right direction, both in CS International and in the power electronics industry.