DRIVEN by increasing demand for innovative, self-emissive displays, global shipments of microLED displays are expected to soar to more than 16 million units by 2027, up from negligible levels in 2020.

MicroLED display shipment growth is being led by demand from the smartwatch and television markets, according to the MicroLED Display Technology and Market – 2020 report from Omdia.

Shipments of microLED displays to the smartwatch market are expected to exceed 10 million units in 2027, while shipments to the TV industry are expected to grow to more than 3.3 million units during the same year.

“Because of their superior luminance efficiency compared to rival organic light-emitting diode (OLED) displays, microLED displays are expected to become the next self-emissive display technology,” said Jerry Kang, associate director at Omdia. “Numerous startups, display manufacturers and consumer brands now are developing their own microLED displays, devices and process in various sizes, ranging from medium-to-large sizes to ultra-small dimensions.”

MicroLEDs represent an emerging flat-panel display technology that employs arrays of microscopic LEDs to comprise the individual pixel elements. Compared to conventional LCD displays, microLED displays can deliver superior contrast, faster response times and reduced energy consumption.

As mentioned, microLEDs also offer superior luminance efficiency compared with OLEDs. As a result, microLED technology is expected to emerge as a competitor to LCDs and OLED displays in applications ranging from wearable devices to televisions.

Several major technology players are taking steps to improve microLED technology and manufacturing.

Taiwanese firm AU Optronics recently introduced a prototype 9.4-inch flexible display using blue microLED pixels under red and green colour-conversion filters on low-temperature polysilicon plastic substrate.

Plessey Semiconductors announced it will help Facebook prototype and develop new technologies for potential usage in the augmented reality/virtual reality space.

Samsung Electronics in 2018 introduced a prototype TV with a chip-on-board processed RGB microLED display. The company plans to launch the product in 2020. Samsung Display has also started development of quantum dot nanorod LED technology, which applies nano-tube LEDs onto an oxide TFT glass substrate. Furthermore, many consumer brands are expected to release their own microLED displays or devices in the near future.

“The microLED market is poised for much more rapid growth once the technologies for manufacturing microLED chips – including mass transfer – gain more maturity,” Kang said. “The growing use of microLED display technology will push display makers to evolve away from current LCD and OLED display technologies.”

NAVITAS SEMICONDUCTOR has announced the delivery of its 5 millionth GaN power IC based on its GaNFast technology to OPPO, a fast-charge phone company.

Yingying (Charles) Zha, VP and general manager of Navitas China delivered the 5 millionth IC in the form of an award to Chang Liu, dean of the OPPO Research Institute, indicating OPPO’s affirmation
of Navitas’ GaNFast technology.

OPPO is described as a pioneer in the fast charging market, from the earliest and popular VOOC flash charging protocol, ‘five minutes to charge and two hours to talk’. The next-generation of SuperVOOC has increased the fast charging power of the mobile phone to 125 W. OPPO’s latest generation of lightweight fast-charge products
uses Navitas GaNFast power ICs to shrink up to 12 times compared with silicon-based chargers.

Liu said: “The cooperation with Navitas has perfectly matched the company’s continuous exploration and pursuit of new products, new materials, new processes and new technologies. We are excited to see Navitas’ company vision and excellent technology. We also hope to promote the development of GaN technology through in-depth cooperation and accelerate the commercialisation of the third generation of bandgap semiconductors.”

Zha said: “I am very pleased that OPPO, as a top mobile device manufacturer, has adopted fast charger technology based on GaNFast power ICs. Navitas’ GaN Power ICs with monolithic integration of GaN FET, GaN digital and GaN analog circuits can promote the commercialisation of a new generation of high-frequency, high-efficiency and very high-density power converters in a faster way.”

FOR ITS EXPANSION into the market for GaN-on-silicon high-power electronics and RF epi wafers, Azur Space relies on the AIX G5 + C from Aixtron, a provider of deposition equipment to the semiconductor industry.

Azur Space, a global leader in the development and production of multi-junction solar cells for space and terrestrial concentrated photovoltaic applications, is a long-standing Aixtron customer and has been using the AIX 2800G4-TM and AIX 2600G3 systems for its space solar application.

The now-ordered fully automated AIX G5 + C system featuring, in-situ cleaning, a cassette-to-cassette wafer handler and Auto-Feed Forward individual on-wafer temperature control, guarantees unmatched epitaxial stability and low defect densities.

Furthermore, Aixtron’s Planetary Reactor enables high increases in productivity and performance through highest throughput, lowest cost of ownership and highest yield performance. The state-of-the-art MOCVD platform is used for the production of 150 mm and 200 mm epiwafers.

With the establishment of a second business line leveraging its III-V manufacturing expertise, Azur Space positions itself on the fast growing market for GaN epiwafers for power electronics and RF applications. The demand for these epiwafers with its capacity to operate at higher frequency and in smaller form factor is mainly driven by
the need for energy efficient power systems, rapid charging solutions, renewable energies, server farms and the next generation of wireless networks.

“Market entry will be a challenge. However, our more than 25 years of experience in III-V epitaxy technology with development and mass production is ideally complemented by Aixtron’s system, so we have a very good starting position. Importantly, Aixtron’s Planetary Reactor provides us with the excellent quality level of our epiwafers required to capture the future market for high-performance electronics,” says Azur’s CEO Jürgen Heizmann.

Felix Grawert, president of Aixtron adds: “The market for GaN epiwafers for power electronics and RF applications is very exciting. It is expected to grow significantly driven by numerous applications such as fast charging solutions or the next generation of wireless networks. The high energy efficiency of GaN based power electronics contributes significantly to reducing the climate impact of new technologies”.

5G remains a hot topic in the smartphone market this year, as smartphone brands and mobile processor manufacturers, such as Qualcomm and MediaTek strive to expand their shares in the 5G market. According to the latest investigations by TrendForce, the Chinese government’s 5G commercialisation efforts have been particularly aggressive, leading the country’s 5G base station deployment and network coverage to each score first place in the global 5G industry.

As such, Chinese brands, which were ahead of their competitors in 5G strategies, occupied a 75 percent share in the global 5G smartphone market in the first half of this year.

TrendForce indicates that, aside from the various Android-based smartphone brands, Apple’s new models will also join the ranks of 5G smartphones. Given the total smartphone production forecast of 1.24 billion units in 2020, 5G handset production is expected to reach 235 million units, an 18.9 percent penetration rate.

Chinese brands are expected to occupy four out of the top six spots of 5G smartphone brands ranked by production volume

An analysis of the projected top six smartphone brands ranked by 5G smartphone production volume shows Huawei firmly sitting in first place. Huawei has shifted its focus to the domestic Chinese market under the impact of US sanctions and in preparation for China’s active 5G commercialisation efforts. The company is expected to produce about 74 million 5G smartphones this year.

Apple’s yearly 5G smartphone production is expected to total about 70 million units in 2020, which lands the company in second place. However, 5G functions will increase the production cost of smartphones accordingly. If Apple decides to directly reflect this added cost on the retail prices of the iPhone 12 series, it may lower its consumers’ willingness to purchase, in turn affecting the sales performances of the new iPhones.

Samsung has been experiencing setbacks in the Chinese market in recent years. Although these setbacks have not seriously affected its global market share and revenue, they have considerably slowed Samsung’s growth in the 5G smartphone market.

Samsung’s 5G smartphone production this year is forecasted at 29 million units, placing the company in third place globally. Vivo, OPPO (including OnePlus, OPPO, and Realme) and Xiaomi are tied for fourth place. As Huawei’s aggressive domestic expansion in the past few years has compressed the market shares of the three brands in the Chinese market, they have been actively focusing on increasing overseas market share to maintain their yearly production performances. Vivo, OPPO, and Xiaomi’s yearly 5G smartphone production volumes are projected to reach about 21 million, 20 million, and 19 million units, respectively.

Mid-to-low end 5G chipsets released by AP suppliers are expected to raise the penetration rate of 5G smartphones in 2021

TrendForce’s analysis of future developments in the 5G market shows that an aggressive push by mobile processor manufacturers will lead to the rapidly increasing presence of 5G chipsets in the mid-to-low end market, driving 5G smartphone production to surpass 500 million units in 2021, which will potentially account for about 40 percent of the total smartphone market.

Once 5G chip prices reach a stable level this year, smartphone brands may look to gain additional shares in the 5G smartphone market by sacrificing gross margins. In doing so, they are likely to accelerate the drop of 5G smartphones’ retail prices, and the market may see the arrival of 5G smartphones around the RMB 1000 price level ($1450) by the end of this year. Incidentally, it is worth noting that the penetration rate of 5G smartphones does not equal the usage rate of the 5G network, which depends on the progress of base station construction.

Since the current 5G infrastructure build-out is pushed back as a result of the pandemic, the global 5G network coverage will be unlikely to surpass 50 percent before 2025 at the earliest, with complete coverage taking even longer.

Founded in Taipei, Taiwan in 2000, TrendForce has extended its presence in China since 2004 with offices in Shenzhen and Beijing.

CAMBRIDGE GaN Devices Ltd (CGD), a spin-out from the University of Cambridge Department of Engineering, will lead a €10.3 million project dedicated to the design and development of the most energy-efficient next-generation GaN power modules.

Working alongside a consortium of 13 European partners with expertise across all aspects of power conversion, the GANEXT project, under the PENTA programme, will focus on producing prototypes for low and high power applications such as lighting, motor drives, converter blocks for renewable energies and on-board chargers for electric vehicles.

CGD develops highly efficient power electronics offering major energy savings in applications ranging from power supplies for consumer electronics to LED drives, data centres and wireless chargers. The company was spun out of the Department of Engineering’s High Voltage Microelectronics and Sensors Group in 2016, in order to develop GaN silicon substrate power semiconductors.

Giorgia Longobardi, CGD’s founder and CEO said: “The PENTA project creates a tremendous opportunity for CGD to engage with leading-edge companies in the area of power electronics. Not only will the project advance the knowledge in GaN technology and provide insights into its complex facets, but it will aim at delivering fully-working prototypes in lighting, motor drives, converter blocks for renewable energies and on-board chargers for automotive with record specifications and outstanding performance.”

Florin Udrea, CGD’s CTO and founder, said: “The quality of the PENTA consortium is remarkable and I have no doubt that we will deliver on the promises to make GaN technology a great success in the market. There is also a broader impact in adding our contribution to our ultimate quest for better use of energy resources and a cleaner environment.”

APPLIED OPTOELECTRONICS, a provider of fibre-optic access network products, has announced that production of laser diodes in July 2020 reached an all-time record of over 1.1 million units, nearly 65 percent higher than pre-Covid levels.

“AOI has been investing in capacity by adding additional production equipment, improving our manufacturing processes to increase yield and enhance quality, and adding staff,” said Thompson Lin, Applied Optoelectronics Inc. founder and CEO.

“The vast majority of our current laser production is 25G lasers which are in high demand now in our data center and telecom segments, including 5G wireless. This additional capacity will help us meet this increased demand. Having produced over 1.1 million tested and qualified lasers in the month of July is a significant step to achieve our near-term goal of producing 1.5 million lasers per month, which we expect to reach in Q4 of this year,” added Fred Chang, AOI’s SVP and North America general manager.

“Even more importantly, the significant improvements we’ve made in our manufacturing process have increased our manufacturing yield, which in many cases have also resulted in improved reliability that is critical for our customers.”

ADVANCED MICRO-FABRICATION EQUIPMENT INC. CHINA (AMEC), has launched the Prismo HiT3 system – an MOCVD tool engineered to mass produce deep-UV LEDs.

The system is the newest addition to AMEC’s portfolio of Prismo MOCVD products. With a maximum reactor chamber temperature of 1400 °C, the tool enables the growth of high-quality AlN and high-aluminium component material. Built for high throughput, the system can process up to 18 2-inch epitaxial wafers per run, with extendibility to 4-inch wafers.

Early customers include China-based Jason Semiconductor Co., Ltd. (Jason) a producer of deep-UV LEDs for home applications, medical instruments, and products used for scientific research.

Deep-UV light has been used for decades as a disinfectant for industrial and consumer applications. It works by disrupting the DNA of bacteria and viruses, and killing their reproductive capabilities. The emergence of pathogens like SARS, MERS, and Covid-19 has increased demand for deep-UV LED products with the wavelength power to destroy new microorganisms. Today’s state-of-the-art for this purpose is deep-UV LEDs at wavelengths of 260-280 nm like the products manufactured by Jason.

“Today, we’re seeing the benefits of deep-UV LEDs in real time with the technology now an important tool used to destroy harmful pathogens,” said Jian Kang, chairman at Jason. “As the science of DUV LED technology continues to advance, we are proud to operate at the leading edge with products to disinfect public environments, as well as the home. AMEC’s Prismo HiT3 system can meet the high-volume/low cost-of-ownership manufacturing imperatives for deep-UV LEDs.”

“The Prismo HiT3 system contains all the innovative features of our proven Prismo platform, now with new ultra-high-temperature process capabilities for AlN-based DUV LED applications,” said Zhiyou Du, executive VP and COO at AMEC. “an early user.

SWISS SOLAR START-UP Insolight has announced that a Series A round of just over €4.6 million has successfully been closed. The round was led again by Investiere, with co-investors Zürcher Kantonalbank, Swiss Immo Lab and a number of private business angels.

Following the first Seed Financing round in January 2018, Insolight has established a world record efficiency of more than 29 percent for flat solar panels, as certified by IES-UPM. The company has successfully taken the technology out of the lab and proved its robustness in real weather conditions. The Series A financing round – together with the €10 million HIPERION grant from the EU – will enable Insolight to start producing and selling the modules. The production will be outsourced to external manufacturers. Insolight is based at the EPFL Innovation Park in Lausanne.

Insolight’s technology is based on a patented optical technology and space-grade GaAs-based solar cells.

Insolight intends to sell its first modules to large solar energy corporates for applications in agrivoltaics. There is a unique potential for high efficiency and translucent modules to produce both solar electricity and crops on the land, whether deployed on fields or greenhouses. The nascent agrivoltaic market already represents 5 GWp of installed power with a market size estimated to nearly Ä700 million. It leverages unique advantages of the Insolight technology, differentiating the market entry strongly from mainstream products.

Rainer Isenrich, CEO of Edisun Power Europe AG and Alan Rosling (former executive director at Tata Sons) will remain members of the board of directors. The Insolight team is now on a mission to deliver the modules to the market. Insolight is supported by several programs from the European Commissions (H2020,Solar-ERA.Net, Eurostars, Climate KIC), national programs (Innosuisse, Innovaud, SPEI, FIT, VentureKick, CleanTech Alps), the European Space Agency (ESA BIC) and EPFL (Innogrant).

INFINERA has announced a succession plan to transition its leadership in the coming months. Tom Fallon will be stepping down as Infinera’s CEO and David Heard, the company’s COO, will be succeeding him.

This transition is expected to take place by the end of 2020 on a date to be determined. Fallon, in his 17th year with Infinera and 11th as CEO, will remain on the board of directors.

“I couldn’t be more pleased with the board’s selection of David as Infinera’s next CEO and I have the utmost confidence in his ability to successfully lead the company in its next phase of growth,” said Fallon, Infinera CEO. “When David joined Infinera three years ago, our objective was to bring onboard a COO who would be positioned to take over as CEO at the right time.”

“David’s contributions since joining the company have been substantial, making us more scalable as he led us through a major acquisition, driving synergies through operational improvements, aligning our product and service portfolio and focusing our investments on the highest value areas for our customers and shareholders.”

Having served as Infinera’s COO since October 2018 and led a publicly traded company as CEO in the past, Heard has over 25 years of industry experience in executive leadership roles that span a wide breadth of technology companies, including JDS Uniphase, BigBand Networks (now part of CommScope), Somera (now part of Jabil), Tekelec (now part of Oracle), and Lucent Technologies (now part of Nokia). He holds an MBA from the University of Dayton, an MS in management from the Stanford Graduate School of Business, where he was a Sloan Fellow, and a BA in production and operations management from Ohio State University.

In conjunction with this transition, Infinera is also announcing that Kambiz Hooshmand will be stepping down as chairman of the board on the same date as the CEO transition, while remaining on the board. Hooshmand has served as chairman since October 2010. George Riedel, a current member of the Board, will succeed Hooshmand as chairman.

“Together with the entire Board, I want to thank Kambiz sincerely for his longstanding and dedicated commitment and service to Infinera,” said Tom Fallon.

“We are very fortunate to have someone of George’s industry experience, knowledge and strategic leadership abilities step into the chairman role and partner with David as he assumes the CEO mantle later this year.”

“These succession changes have been thoughtfully planned and we anticipate a smooth transition. I am deeply appreciative of the opportunity to serve as CEO over the past decade and I look forward to helping Infinera in the next phase of our journey.”

US-BASED MATERIALS science researchers, led by Shui-Qing ‘Fisher’ Yu from the University of Arkansas, have demonstrated what they claim is the first electrically injected laser made with GeSn. Used as a semiconducting material for circuits on electronic devices, the diode laser could improve micro-processing speed and efficiency at much lower costs.

In tests, the laser operated in pulsed conditions up to 100 K.

“Our results are a major advance for group-IV-based lasers,” Yu said. “They could serve as the promising route for laser integration on silicon and a major step toward significantly improving circuits for electronics devices.”

The research is sponsored by the Air Force Office of Scientific Research, and the findings have been published in Optica, the journal of The Optical Society. Yiyin Zhou, a University of Arkansas doctoral student in the microelectronics-photonics program authored the article.

Zhou and Yu worked with colleagues at several institutions, including Arizona State University, the University of Massachusetts Boston, Dartmouth College in New Hampshire and Wilkes University in Pennsylvania. The researchers also collaborated with Arktonics, an Arkansas semiconductor equipment manufacturer.

The alloy GeSn is a promising semiconducting material that can be easily integrated into electronic circuits, such as those found in computer chips and sensors. The material could lead to the development of low-cost, lightweight, compact and low power-consuming electronic components that use light for information transmission and sensing.

Yu has worked with GeSn for many years. Researchers in his laboratory have demonstrated the material’s efficacy as a powerful semiconducting alloy. After reporting the fabrication of a first-generation, optically pumped laser, Yu and researchers in his laboratory continue to refine the material.

AT THE virtual International Microwave Symposium (IMS), Macom announced a new GaN-on-SiC power amplifier product line, which it is branding Macom Pure Carbide. The company also announced the introduction of the first two new products in the product line, the MAPC-A1000 and the MAPC-A1100.

“This new product line significantly enhances the capability of our existing RF power product portfolio,” said Stephen Daly, president and CEO. “GaN on SiC is a compelling technology and we are excited to begin offering our customers both standard and custom Macom Pure Carbide power amplifier solutions.”

The MAPC-A1000 is a high power GaN-on-SiC amplifier designed to operate between 30 MHz and 2.7 GHz and is housed in a surface mount plastic package. The easy-to-use general purpose amplifier integrates an input match which simplifies the customer’s design-in effort. The amplifier can deliver more than 25 W (44 dBm) at greater than 50 percent efficiency from 500 MHz to 2.7 GHz when tested in a circuit designed for operation over 2.2 GHz simultaneous bandwidth.

The MAPC-A1100 is a high power GaN-on-SiC amplifier designed to operate up to 3.5 GHz. The device is capable of supporting both CW and pulsed operations with output power levels of at least 65 W (48.1 dBm) in an air cavity ceramic package.

The two new general purpose amplifier products are ideal for use in avionics, high power mobile radios, wireless systems and test instrumentation.

IN JULY THIS YEAR German GaN-on-silicon wafer developer Allos Semiconductors revealed that it had sold its high power electronics and RF business to Azur Space, a provider of III-V epitaxy for solar cells also based in Germany, for an undisclosed sum.

The move comes at a time when the market for GaN-on-silicon power and RF epiwafers is booming, and merger and acquisitions are rife. For example, just over a year ago, France-based Soitec bought GaN wafer supplier EpiGaN for Ä30 million to strengthen its GaN presence in the RF and power markets. Then in spring this year, STMicroelectronics made its mark on the power market when it acquired a majority stake in GaN innovator Exagan.

So why sell now? Clearly the timing works for Azur Space. As one of Allos’ co-founders, Alexander Loesing, says: “Azur Space has a manufacturing capacity of 500,000 150 mm wafers a year and wants to leverage its III-V epitaxy skills for the power market.”

“We see a second wave of power electronics businesses that don’t want to in-source GaN epitaxy,” he adds. “These companies require a reliable and capable source as they don’t want to spend time building up their own epi-supply, so this will be good for Azur Space.”

Allos also has mighty plans for expanding its microLED efforts. Minus its power and RF arm, the company intends to focus its resources on this market, which Loesing describes as ‘incredibly intensifying’. “We have three times as much work here than we did only a year ago,” he says.

Over the years, Allos has been building a compelling case for growing microLEDs on GaN-on-silicon wafers, instead of the industry stalwart, sapphire. But how exactly does the company intend to make its mark on the fast-growing, highly competitive world of microLEDs?

Making the difference

To reach mass sapphire-based microLED production, industry players are busy addressing manufacturability issues, including yield and cost. For example, GaN-on-sapphire LED chips are processed on 100 mm and 150 mm diameters. Larger wafers are too expensive to be used in LED manufacturing, and also suffer from bow, leading to lower yields.

Meanwhile, inhomogeneities in the emission wavelength of a typical GaN-on-sapphire LED epiwafer mean chips need to be tested and sorted into bins to ensure the microLEDs emit the desired colour. Given a 4K display comprises some 25 million microLEDs, this process of binning clearly isn’t viable at volume production.

“The industry needs to move towards an epi-growth and chip processing environment that delivers large diameters with a high uniformity to avoid the need for binning, and at high yields,” highlights Loesing. And the Allos director is sure that his company’s hetero-epitaxial GaN-on-silicon growth process will make the difference.

According to Loesing, the epi-growth process uses strain engineering and lateral overgrowth to overcome the lattice mismatch and thermal expansion differences between GaN and silicon that would otherwise lead to wafer bow, cracks, poor crystal quality, and associated wavelength non-uniformities and yield loss.

As part of the company’s method to achieve high crystal quality, epitaxial lateral overgrowth is used, which is analogous to the patterned sapphire substrates used in sapphire-based LED manufacture to reduce GaN crystal defects. Careful strain balancing introduced during epi-growth helps to control bow and deliver flat epiwafers after growth. And the company also uses its proprietary and patented strain engineering to precisely control heat across the entire wafer diameter during temperature-sensitive InGaN multi-quantum well growth, promoting wavelength uniformity and reproducibility from wafer to wafer.

“The strain engineering and crystal quality challenges are much more difficult in LEDs compared to power electronics, but this is good for us,” says Loesing. “Very few companies can use GaN-on-silicon to make LEDs with our levels of quality, and I haven’t seen anyone come close to the uniformities that we are showing.”

Indeed, at 2 x 10⁸ cm^–2, the defect density of its GaN-on-silicon substrates is on par with that of market-leading GaN-on-sapphire. Epiwafers are crack-free with minimal bow while wavelength uniformity is stable with nearly 98 percent of wafer areas hitting targets.

As Loesing also points out, silicon substrate removal via grinding and etching, as used in silicon lines, is more straightforward than the laser lift-off necessary to remove sapphire substrates from chips. And epi-growth takes place in standard MOCVD reactors on 200 mm and 300 mm wafers, opening the door to microLED production on low-cost, high yield, silicon processing lines.

Still, without a doubt the microLED world is very much entrenched in sapphire. To tackle tradition, Allos is working to ensure a steady supply of GaN-on-silicon epiwafers. “We need to make sure we can provide enough samples to our customers,” says Loesing.

Given time, the company director is confident that microLED players will make the move from sapphire to GaN-on-silicon. “If we’d spoken four years ago I would have said I don’t believe that the LED industry will ever take-up GaN-on-silicon as companies were focused on throughput,” he says. “But the arrival of microLED displays has changed this, and high yields are crucial to drive costs down – our technology ensures this with wafers, and all subsequent processing steps.”

So what now for Allos Semiconductor? Right now, the company is working on, as Loesing says, ‘results’.

Indeed, since the sale of its power electronics and RF arm, Allos has announced that it is working with researchers from King Abdullah University of Science and Technology (KAUST), Saudi Arabia, on high efficiency nitride-based red LEDs on silicon. Group head Kazuhiro Ohkawa from KAUST has already grown InGaN-based red LEDs on sapphire and working with Allos, intends to do the same on silicon.

“We continue to push forward and look forward to showing more results, and also to making new announcements about customer collaborations very soon,” says Loesing. “This microLED market is exploding and I’m having the most fun at Allos that I’ve ever had.”

WHILE THE WORLD WATCHES the coronavirus crisis unfold, many in the compound semiconductor industry are also alert to some exciting developments underway in South Wales. In June this year, the world’s only compound semiconductor cluster, CSconnected, was awarded a mighty £25.4 million in government funds to push the region’s home-grown technologies into the global marketplace and increase international trade in key sectors such as 5G and autonomous vehicles.

As Drew Nelson, Chief Executive and President of Wales-based epiwafer manufacturer, IQE, tells Compound Semiconductor: “An important aspect of this is to seek partnerships with other regions around the world and make contact with many other companies that may need to access [our technologies] and take part in this cluster.”

The funds will also be used to bolster the region’s existing supply chain so that products can be developed from concept to production, without outsourcing. According to Drew, the project is in its very early stages, but without a doubt, products for future technologies will be delivered.

“For example, we will be working on scaling up compound semiconductor production to include compound semiconductors on silicon for, say, 5G applications, and are looking towards getting some of these technologies onto a 200 mm platform – this is quite a big goal,” he highlights.

And as fellow cluster member, Wyn Meredith, Director of the Compound Semiconductor Centre, a joint IQE-Cardiff University venture to develop and prototype compound semiconductor materials, says: “We’ve been really focused on establishing a coordinated supply chain, in terms of materials, device fabrication and packaging, as well as capital equipment.”

“We will now use this to build high-value added manufacturing,” he adds. “So we’ll be encouraging companies to settle in South Wales, help us to extend our supply chain and scale up compound semiconductor product manufacturing here.”

Building a supply chain

Since its inception in 2015, CSconnected has rapidly expanded to include a host of academic, industry and government partners that now constitute the largest concentration of compound semiconductor activity in the UK. As well as IQE, members include: the Universities of Cardiff and Swansea; etch and deposition process equipment supplier, SPTS; compound semiconductor and silicon foundry, Newport Wafer Fab; and chip maker, Microsemi.

Other key organisations include the Compound Semiconductor Centre, the Compound Semiconductor Applications Catapult, which focuses on research and product development, as well as UK-Welsh government investor, the Cardiff Capital Region City Deal, and Welsh Government.

To date, the cluster has secured more than £600 million in investments, provided around 1480 jobs, and accounted for at least £460 million of sales – with more than 90 percent relating to overseas exports, mostly EU destinations. And while all industry players occupy key places within the UK compound semiconductor supply chain – for example Newport Wafer Fab is the UK’s only silicon foundry, producing 200 mm wafers for silicon and compound semiconductors – industry can expect more players to join soon.

The latest £25.4 million funds come from the UK Research and Innovation’s Strength in Places Fund, and amount to more than half the £43.7 million project cost, with remaining cash provided by CSconnected members. The fund is designed to boost local growth, and current estimates indicate the project will create an extra 1160 jobs in the region come 2025.

At a time when UK industry has been left reeling from coronavirus, not to mention the anticipated effects of Brexit, these figures make welcome reading. As Meredith puts it: “We’re in a really interesting situation here – when most industries are thinking about how to get their staff off furlough and back onto permanent contracts so they can re-start manufacturing we’re thinking about expansion.”

The project is scheduled to start in October this year, so as Drew puts it, ‘precise details are still being worked out’. However, industrial cluster members are poised to lead four key programmes within the project.

Newport Wafer Fab, accompanied by Rockley Photonics, will head up the first programme aimed at developing foundry services for sensing and telecommunications. Meanwhile IQE will lead the second programme that will scale up GaAs-based photonics, including photovoltaics and pixel-based architectures, on large wafer formats. SPTS will head up the third programme that aims to expand capital equipment, including reactors, and in the final programme, Microchip is set to build a power electronics packaging pilot line.

“This is really about upskilling our supply chains and not leaving our infrastructure behind as we grow,” says Meredith.

“If we look at SPTS, for example, the company has always prided itself on using local supply chains in South Wales,” he adds. “This project can now help to grow these supply chains, many of which include SMEs or mid-size businesses, so SPTS can use and support these businesses rather than having to look further afield.”

Indeed, as Meredith also points out, a raft of academic-industry innovation work packages from the cluster are already underway and the businesses that take part will be able to feed into, and grow, future supply chains. Key UK-based industrial partners include measurement and motion control systems developer Renishaw, RF and power semiconductor player, Teledyne E2V, and steel-maker Tata Steel Europe.

“We’ve captured a lot of business here, and many of these guys don’t normally play in the compound semiconductor space,” says Meredith. “But this is deliberate – we are now engaging with companies at the far-ends of the supply chain that are becoming really enlightened to what we are doing.”

So what happens now? Clearly a critical mass of compound semiconductor activity is firmly in place in South Wales, and the latest funds will support more economic growth. As Meredith puts it: “We’re ahead of the game, but if you look at the value that is extracted out of our existing supply chains here, it’s miniscule compared to what we could be doing – our Strength in Places funds will provide a catalyst for this.”

And as the Compound Semiconductor Centre Director points out, more industrial expansion is coming soon.

“Who knows what post-Covid economic conditions are going to bring, and if there is a prolonged economic recession, research, development and innovation budgets are going to come under a lot of pressure,” says Meredith. “But we’ve now got ours ring-fenced for the next four-and-a-half years, so can effectively proceed as we wanted to do pre-Covid.”

BACK IN 1995, when Compound Semiconductor magazine made its debut, Bill Clinton presided over the United States, owners of PCs could start getting their hands on Windows 95, and our industry was far, far smaller than it is today.

Reflecting its diminutive size was its lack of diversity. This is illustrated by the narrowness of the discussion that took place at the 8th ICMOVPE, held in Cardiff in 1996, where delegates pondered the question “MOVPE – Is there any other technology for optoelectronics?”. Back then the manufacture of red and infrared LEDs was well established, but the blue cousin had only just been invented; volumes of GaAs HBT production were tiny, due to very few owners of mobile phones; III-V solar cells were yet to establish themselves as the dominant technology for powering satellites; and the manufacturer of InP lasers and photodetectors to build the information superhighway provided a substantial proportion of total industry revenue.

Although the compound semiconductor industry was in its infancy back then, manufacturers of optoelectronic devices did not have to rely on home-built systems for MOCVD growth. They also had the option of purchasing commercial reactors.

As far back as the 1980s, we had started to consider the limitations of our first commercial reactors, the Aixtron 200 series (see Figure 2, 3 and 4. The latter shows a typical reactor from the Aixtron 200 series, the Aixtron 200/4 system, which features heating with a standard lamp or an RF heating system, a glove box and computer control.). Reactors in the Aixtron 200 series had been prototyped at RWTH, prior to further development and commercialisation by our team. This MOCVD system had a great reputation within the research community, thanks to its combination of simplicity, reliability, versatility and ease of use.

A severe limitation of the horizontal tube reactor is its side walls. When they are hot heterogeneous reactions take place; and when they are cold condensation forms. Another impediment is that the side walls consume source material, impairing the efficiency and the uniformity of the reactor. The side walls also alter the flow pattern perpendicular to the gas flow direction, making the growth of high-quality epilayers more difficult.

Planetary Reactor

Demand for compound semiconductor devices continued to rise throughout the 1990s and beyond. We responded with the introduction of our Planetary Reactor series, which uses two forms of circular motion to ensure the uniformity of many wafers in a reactor. Every wafer rotates around the centre of the reactor, and clusters of them also circulate around their own local centre (see Figure 5).

The Planetary Reactor has many merits. It is a horizontal reactor which is free from side walls, and the temperature of its top wall, also known as the ceiling, can be controlled to make it passive. This type of reactor can ensure excellent, controllable uniformity, making it ideal for large scale production. It is also renowned for its: high precursor utilisation efficiency; its low maintenance requirements; its accommodation of large wafer areas; the opportunity it provides to use automated cassette-to-cassette loading systems; and its high growth rates – values of more than 30 μm/hr. Thanks to this great set of attributes, our Planetary Reactors provide low-cost, high-volume, multi-wafer production using multiple wafers up to 200 mm.

In the 2000s, the increases in yield that drove the cost reductions that are shown in Figure 10 came through accommodating a larger wafer area in the reactor, while keeping uniformity as high as possible. Launched in 2001, our third-generation Planetary Reactor for GaN, the AIX 2400 G3, had a capacity of 11 2-inch wafers. Upgrading this the following year led to the introduction of the AIX 2600 G3, which could accommodate 24 2-inch wafers. By 2006, we could house 42 2-inch wafers, thanks to the launch of the AIX 2800 G4 HT, a fourth-generation GaN Planetary Reactor. The first fifth-generation GaN Planetary Reactor followed in 2009, enabling growth of 8 6-inch epiwafers in a single run. During the last decade increases in productivity have not come from increases in wafer capacity, but from other strategies that trim cycle times, such as the introduction of automated wafer-loading systems.

We have been working on wafer loading for many years. Back in 1996 we developed the first automated MOCVD reactor, featuring a spider-like robot for loading wafers into a cassette. By avoiding the need to cool wafers to room temperature, this handler cut cycle times by up to 30 percent, increasing throughput.

A significant refinement to our reactors came in 2012 when we replaced our conventional, triple injectors with variants featuring five injectors. These ‘penta’ injectors brought knew tuning potential to widen the process window for best uniformities up to 200 mm wafer size.

Another aid to the process engineer has been the introduction of optical in-situ monitoring tools. This form of instrumentation opens the door to advanced process control. It enabled logging and control of far more data, including values for wafer bow, growth rate and local on-wafer temperature. By designing reactors that accommodate an optical path to the growth surface, we established emissivity-corrected pyrometry that provides measurements of on-wafer temperature. Armed with this insight, process engineers can now realise wafer level control and optimise their epitaxial yield.

In addition to our portfolio of Planetary Reactors, we offer those that are based on a Closed Coupled Showerhead design. This alternative reactor architecture, developed in parallel, also offers proven production capability. We offer this design to cater to those customers that are more familiar with the Closed Coupled Showerhead reactor and want to continue to produce epiwafers with this technology.

The Closed Coupled Showerhead

Like our Planetary Reactors, within the suite of Closed Coupled Showerhead designs, throughout the first decade of this century increases in wafer capacity led to breakthroughs in productivity. In 2001 and 2002 we launched reactors for nitride growth that accommodated 6 and then 19 2-inch wafers, and we followed this up with the introduction of the Crius, Crius II and Crius II XL in 2006, 2011 and 2012 – they housed 31 2-inch wafers, 55 2-inch wafers, and 19 4-inch wafers, respectively. In 2015, we introduced a semi-automated reactor with an even larger chamber, capable of producing 31 4-inch epiwafers in a single run. With this class of reactor, growth rates of even up to 280 μm/hr were demonstrated.

We decided to focus on our strength, which is providing the growth technology for producing high-tech devices that demand advanced MOCVD capability. For engineers that are producing lasers, III-V solar cells, advanced transistors, microLEDs for displays, and high-speed InP-based lasers for telecom and datacom networks, our Planetary Reactors are a great choice, combining a high level of performance with excellent yield and productivity. The Close Coupled Showerhead remains the solution of choice for R&D systems or for single systems that then require superior performance once again – which a vertical batch reactor can inherently not deliver well.

For the last quarter of a century, we have focused on ways to enhance productivity. This has been accomplished through the likes of increased reproducibility, superior layer uniformity, wafer level automation, and the introduction of in-situ metrology that enables the acquisition of process data, key to smart system control and predictive maintenance.

Figure 11. The latest (2020) G5+ C dual Planetary Reactor module cluster featuring cassette-to-cassette wafer handling.

Our industry uses silicon substrates as the foundation for manufacturing several different devices. Transphorm employs it as the bedrock for its GaN FETs, now numbering more than 500,000, and Toshiba uses it for producing GaN-on-silicon power devices. Silicon substrates also feature in microLED displays made by Plessey, and in LEDs developed by both Samsung and Bridgelux, now part of Toshiba.

There are many reasons why these chipmakers use silicon as the foundation for device production. Silicon substrates are low in cost, widely available, and allow processing in 200 mm silicon lines, which may offer very competitive production costs, while employing tools with greater levels of performance than those found in many compound semiconductor fabs.

However, producing high-quality devices using silicon is far from easy. The main issue is the significant lattice and thermal mismatch between the nitrides and the substrate. These differences induce stress and strain in the epilayers, causing wafers to bend, bow and even crack. Strain management is essential, as wafers must be flat to within several tens of microns to enable processing in silicon lines.

One way to ensure flat enough epiwafers is to introduce buffer layers that manage stresses and strains. However, this is not easy to do. Successful buffers are complex, requiring years of development. What’s more, their formation often involves depositing many layers, adding to the material costs of the epiwafer, as well as overall production costs, increased by longer growth times. Even with thicker buffer layers, dislocations are present throughout the heterostructure, hindering device performance. Yet another drawback is that the buffer may hamper thermal management, because it can increase the distance between the location of heat generation and any heat spreading technology.

Turning to nano-assembly

Our technology does not involve conventional film growth, with materials deposited from above. Instead, we replace silicon atoms in a special way that retains the original crystal structure of the silicon matrix. Using this ‘nano-assembly’ method, we ‘convert’ a very thin layer of silicon into both cubic and hexagonal SiC.

During this process, we form a special porous buffer layer beneath the SiC. The top nano-SiC layer that is created provides the ideal foundation for the growth of III-Ns, thanks to the compatibility of its properties with the crystal structure of nitride heterostructures.

When epitaxial films are grown on our silicon-based templates, mechanical stress relaxation takes place. This is driven by the preliminary embedding of the substrate lattice through assembly of nano-objects. By ensuring that carbon atoms are positioned at interstitial positions of silicon and silicon vacancies, we create a so-called ‘dilatation dipole’.

Unlike traditional epilayer growth, the orientation of our films is determined by the crystal structure of the original silicon matrix, rather than just the substrate surface. We have found that the best way to produce a cubic crystal is to form a dilatation dipole that is perpendicular to the (111) plane of silicon. When this occurs, almost all the elastic dilation energy of the film relaxes due to dipoles alone, leading to high-quality SiC films.

To demonstrate the promise of our technology, we have used it to produce a wide range of thick, single-crystal, low-defect layers of various III-Ns using HVPE. This includes conventional GaN, AlN, AlGaN, and also a variety of semi-polar forms of GaN. Examples of nano SiC-on-silicon substrates that we have made, reproduced and tested in the lab include: AlN layers with an orientation (0001) and thicknesses ranging from just 40 nm to 700 μm, using silicon orientations (111) and (011); and semi-polar epitaxial AlN layers with the orientations (1120), (1013) and (2023), using silicon orientations (001) and (310).

Additional results include the growth of AlGaN layers with a (0001) orientation that have thicknesses between 100 nm and 400 μm, formed using SiC-on-silicon (001) substrates. We also separated layers with a thickness of 400 μm from the template.

Using our technology, we have produced a portfolio of monocrystal layers that are completely free of cracks. As well as AlGaN layers with a thickness up to 400 μm that we have just mentioned, we have made AlN with a thickness up to 300 μm, GaN with a thickness up to 200 μm and semi-polar (1124) GaN.

We have determined the growth rates for a variety of III-N films grown on our SiC-on-silicon templates, using the growth conditions detailed in Table 1, and layer thicknesses determined by electron microscopy. Films of GaN, AlGaN and AlN had thicknesses of 2.2 μm, 1.82 μm and 0.72 μm, respectively, while the SiC layer is just 70 nm-thick. This implies a growth rate for a GaN layer of 66 μm/hr, that for an AlGaN layer of 36 μm/hr and that for AlN of 42 μm/hr. The growth process still can be further optimized for higher growth rates.

Simple substrate separation

For some devices, such as UV LEDs, it can be useful to remove the silicon substrate, as this can absorb light and hold back efficiency. Thanks to the porous layer between silicon and the incredibly thin layer of SiC, we can employ a very simple chemical etching process, taking a matter of minutes, to separate the thin films or heterostructures from the silicon substrate. The freed material can provide a foundation for further bulk crystal growth, or be attached to another substrate.

Many devices could benefit from the high-quality, III-N layers formed with our silicon template technology. Our breakthrough could aid the manufacture of III-N transistors, as well as visible and UV laser diodes, LEDs and VCSELs. When these optoelectronic devices are grown on the conventional planes of III-Ns, efficient radiative recombination is hampered by strong internal electric fields. Switching to the semi-polar planes we have produced alleviates this, unlocking the door to more powerful light emitters.

To demonstrate the capability of our technology at the device level, we are using our SiC-on-silicon templates as a foundation for HEMTs and deep UV LEDs. For the latter, we start with a layer of AlGaN. We see this as a better choice than AlN, a material that seems incompatible with reasonable levels of p-type doping. Even in layers of AlGaN, realising p-type doping is a challenge, due to the high ionization energy of acceptor impurities. Efforts on this front are also hampered by  the strong interaction of acceptor impurities with various structural AlGaN defects, leading to the formation of deep levels associated with both acceptors and donors. This work is still in its infancy, with successes to date including the fabrication of AlGaN layers with an aluminium content close to 50 percent.

Such efforts are showcasing the capability of our novel templates, and their capability to transform the production of various wide bandgap electronic and optoelectronic devices that benefit from increased material quality. Costs of chip production should also plummet with our templates, due to a drastic reduction in the cost of growth of these structures, a significant simplification of the growth technology, and the scalability on offer by turning to substrates with large diameters.

The history of SiC begins a very long time ago. It is found in the collisions of meteors, which hit our planet to create the only naturally occurring SiC, located in places that include Canyon Diablo in Arizona. Fast-forward to the twentieth century and we had developed the expertise to synthesize this wide bandgap semiconductor material. However, it is only over the course of the last 25 years that SiC has become an integral part of our lives, featuring in cars, cell phones, radar technology and an array of other applications.

It is fascinating to follow the story of how a material made from meteors’ impacts is now in the palm of our hands. SiC first came into the spotlight in 1907 when it created, by accident, the first LEDs, in the hands of Captain Henry Joseph Round. When this radio pioneer, who moved from the UK to New York to work for the Marconi Company, passed a current through SiC, to his great surprise he witnessed light emission. As well as being deployed in early radios, this material initially found use in abrasives. Throughout the early and middle part of the twentieth century, slow progress occurred, but from then on SiC has undergone a rapid acceleration. 

Driving much of this change has been Cree, the company we co-founded in 1987. Formed by a team of researchers previously employed at North Carolina State University, when we embarked on this venture we still had much to learn about the potential of SiC. Back then we knew how much promise this material held as a semiconductor and LED in commercial applications, but we did not know how to get there, and methods were still to be established for efficiently producing the SiC crystals, a pre-requisite to high-volume device production. 

During those early years we were buoyed by several crucial breakthroughs. Our milestones included the first commercially available blue LED, and the development of the first commercial SiC wafers. Although our substrates were only 1 inch in diameter and quite expensive, we could grow a business from them.

A Wolfspeed 650 V MOSFET.

These successes have provided the bedrock for much more progress over the past 25 years. During this time, SiC has gained commercial momentum that continues to accelerate to this day. We are now in the midst of one of the largest potential transformations in the entire semiconductor industry since the shift in silicon technology from bipolar to CMOS in the 1980s.

Mainstream breakthrough

One of our big, early successes came in 1996, when our blue LEDs were designed into the dashboard of Volkswagen vehicles. Prior to that, we had been selling less-efficient SiC LEDs, but in 1995 we introduced successors based on GaN-on-SiC. Our win with Volkswagen provided the first large-scale use of GaN-on-SiC LEDs in vehicles and signalled an acceptance of this technology – it had matured to a point where it could be adopted by a mainstream automotive player. Key to this acceptance, starting back then and maintained today, is the supply of SiC material. In the early 1990s it was extremely difficult to efficiently and effectively grow SiC crystals, which provide the foundation for LEDs and other devices.

Part of the reason why it is so challenging to produce high-quality SiC substrates is that there are more than 200 different crystal structures associated with various polytypes, and only a few are suitable for the semiconductor market. Adding to the complications, crystal growth requires temperatures nearing 2,500 °C. It’s hard to have a feel for such a high temperature, but to put it in perspective, it is half the temperature of the sun and significantly beyond that of molten lava, which is in the range of 1,700 °C.

Sustaining the needs of a major auto manufacturer drove our first breakthrough in our production of SiC. And this is still a topic of conversation in today’s market, where makers of automobiles are integrating SiC in far higher volumes into the drivetrain and chargers of mass-market, battery electric vehicles.

A lighting revolution

Following that first large-scale automotive use of LEDs, the world experienced a meteoric rise in cell phones. Previously a luxurious, somewhat novelty item for the well-to-do, it quickly became a must-have device for many. Since then, handset makers have been churning out millions of units every year.

When the market ramped in 2001, manufacturers of mobile devices started to incorporate LEDs into their latest models. Initially they backlit keypads, and a few years on, they were the light source for colour screens. They are ideal for that task, due to their low energy consumption, small footprint, robustness, and capability to be driven by a battery. The substantial volume of LEDs produced for this market spawned another huge leap forward for GaN-on-SiC sales.

GaN-based LEDs had now found their first killer application. However, although they had made much progress, they still had a long way to go before they could be viewed as a viable candidate for everyday lighting. Before that could happen, there needed to be an expansion in the colour palette generated by these sources, a hike in their efficiency, and a drop in price so they could be a competitive rival to the incandescent bulb. 

A Cree employee demonstrating the move from 100 mm to 150 mm SiC substrates.

Cree made much progress on all these fronts, introducing, in 2006, the industry’s first lighting-class LEDs. Known as the XLamp portfolio, these devices delivered 100 lumens per watt, exceeding the efficiency of fluorescent lighting. A revolution in lighting had started, after decades witnessing relatively little change.

LED lighting has many virtues. It’s robust, long-lasting, free from mercury, and its great efficiency trims electricity bills, in turn helping to reduce carbon footprints. But initially it did not find favour with the existing lighting infrastructure sector. That’s not that surprising, given the great business model of selling products to the public that could often fail within a year and had to be replaced. To drive a transformation, we started selling our own LED lighting fixtures to show the world what is possible. Our products consisted of a wide variety of indoor and outdoor lighting fixtures, ranging from fluorescent fixture replacements to LED streetlights.

Our final demonstration came in 2013, when we released our first LED light bulb. Rather than looking like something emerging from a science project, it had the style and feel of an incandescent bulb, and it retailed for less than $10. Launched to initiate the mass adoption of LED lighting, we called it ‘the biggest thing since the lightbulb.’ 

Following those major breakthroughs in LED lighting, we began putting additional emphasis in innovations outside the lighting space. This set us on a course for where we are today. Now our focus is to become a global semiconductor powerhouse that leads the transition from silicon to SiC.

Taking on the RF world

In 1998, we had our first major breakthrough in the RF domain, demonstrating the first GaN-on-SiC HEMT. Compared with the GaN-on-sapphire variants of the time, this device delivered increased signal gain and a 400 percent hike in power density in terms of watts per mm, making it an attractive candidate for wireless and broadcast high-power applications.

Further success came in 2008, when we introduced the first GaN RF device that utilised the benefits of GaN-on-SiC to improve RF performance. With its high breakdown field, GaN can operate at very high voltages, and thus high-power densities, without having to compromise reliability. By using SiC as the foundation, the GaN device sits on a very high-thermal-conductivity platform. This base is ideal for dissipating a high-power density and maintaining junction temperatures at a reasonable level. Thanks to this, GaN RF devices deliver higher output powers without having to increase device size. These attributes enable the RF industry to design smaller, more efficient, more powerful RF systems.

During the last decade, we have continued to advance the performance of our RF GaN-on-SiC devices. They are widely adopted by the aerospace and defence industries, which use them to improve the capabilities of their systems as well as in the telecom infrastructure market for higher-efficiency power amplifiers.

SiC powers up

Power electronics started coming of age at the beginning of the twentieth century. One of the earliest examples is the mercury-arc rectifier, developed in 1902 to convert AC power into a DC form. By the early 1930s, selenium had replaced mercury, only to be superseded by silicon in the 1950s. During that era engineers invented the first silicon BJTs and MOSFETs, but the latter would not become suitable for use in power electronics until the late 1970s. The last major breakthrough in silicon power electronics, the development of the IGBT, followed in the 1980s.

Within the power electronics world, the introduction of SiC provided the next big materials breakthrough. The US Army, Air Force, Navy, NASA and DARPA all supported these efforts, funding programs from the early 1990s. For SiC to establish itself as a mainstream technology, SiC wafers had to be cheaper, larger, and contain far fewer defects. Without a substantial fall in defect density, it would not be possible to produce devices with sufficient area to operate at high currents. 

A up-close look at a 150 mm wafer in dicing.

A major impediment was the infamous micropipe defect. At one stage it was thought that micropipes were simply inherent to SiC, so impossible to eliminate. And that was not the only major stumbling block – oxides on SiC were considered to be inherently unreliable, preventing the manufacture of a reliable MOSFET. Even if both these daunting issues could be overcome, maybe that would be to no avail, if SiC wafers were never big enough to allow devices to serve in anything other than the most niche, cost-insensitive applications.

As history attests, all these issues have been addressed. Success on all fronts occurred after the launch of the first small SiC Schottky barrier diodes – introduced in 2001 and 2002 by Infineon and Cree, respectively – with progress paving the way to the introduction of first commercial SiC power MOSFET, which we launched in early 2011.

The introduction of SiC diodes and transistors has had a significant impact on the power electronics industry. It opened the door to faster switching frequencies, higher efficiencies, and superior power densities – all of which are key to the continued development of effective power electronic systems. These attributes allow SiC power electronics to support various emerging industries, including: renewable power conversion, such as solar energy and wind power; energy storage conversion; and electric trains, buses, and other types of public transportation.

SiC power devices can also be found in uninterruptible power supplies, industrial high-frequency power supplies, and motor drives. In all these applications, systems sporting SiC components operate with higher efficiencies take up less space, weigh less, and do not require extensive cooling systems. These valuable merits have taken us to a point where SiC is pushing the boundaries of innovation in a wide variety of industries.

An attitude of innovation

When we look back at these milestones over the past 25 years and think about all the breakthroughs in between, it is clear that Cree’s stubborn pursuit of innovation has been a driving force, both through the company’s history and that of SiC. There have been many obstacles, some considered insurmountable. It has led some to try and then give up, and others to never embark on these challenges.

There were those who argued that you cannot eliminate micropipes in SiC, that you can’t make a wafer larger than two inches, and that it’s impossible to make bipolar devices that don’t degrade. Cree has proved all those fallacies wrong due to a steadfast commitment to overcoming barriers. It is in our DNA that there is no reason to give up, just because someone says it can’t be done.

In some cases, we have not just broken barriers – we have kept on going, setting new benchmarks. After making the first 2 inch SiC substrate, we broke the 4-inch barrier, and now we’re ramping up 200 mm wafers, with mass production slated for the not too distant future. While it has certainly been a challenge and adventure to get this far, there is still so much more to come. As we look ahead, we see a future that has never been brighter. SiC is set to play an ever greater role in a variety of applications.

One of the biggest opportunities for SiC is in electric vehicles. Some of the first cars utilising this material are already on the road, and virtually every global manufacturer is including this technology in future designs that will grace the forecourts in the coming years. By replacing silicon devices with those made from SiC, the range of an electric vehicle can be extended by 5-10 percent, or the cost expended on expensive batteries trimmed by this amount. This most-welcome advantage is quickly positioning SiC as a necessary component in inverters as well as onboard and offboard chargers. As range is a significant factor for those considering the purchase of an electric vehicle, it has a tremendous influence on sales. And the use of SiC doesn’t always command a premium for electric-vehicle makers, as it can lead to an overall decrease in cost compared with the use of silicon technology. Its role looks set to grow further as electric vehicle manufacturers transition from 400 V to 800 V systems, which get an even bigger benefit from the switch to SiC, in terms of efficiency and other measures of performance.

Throughout the last 25 years there has been rapid innovation within the communications sector. It is currently undergoing a proliferation of 5G networks and picocells as well as higher volumes of small distributed antennas, radar applications, aerospace and defence tools. For all these applications,

GaN-on-SiC devices have the edge over silicon in efficiency and bandwidth, while offering a far smaller footprint. Drawing on the advances wrought by SiC has increased demand for this material. To bolster global supply, of which we account for 60 percent, we have committed to a $1 billion expansion at our Durham headquarters as well as the construction of an additional manufacturing facility in the Mohawk Valley, New York. When this additional production is available, our company’s capacity will have increased 30-fold from its level in 2017.

In the past 25 years, there has been more change in the SiC and semiconductor industries than we could ever have imagined. Looking ahead, the next 25 years are shaping up to be even more transformational. We expect that they will deliver breakthroughs that seem impossible today. It is an exciting time to participate in these changes, while looking forward to ground-breaking times that lie ahead.

The 1990s will be remembered for the birth of efficient blue and green LEDs based on GaN. In the decade that followed, further improvements came that you literally can’t see. This triumph resulted from extending the emission wavelength of GaN LEDs into the ultraviolet. Replacing the In_xGa_1-xN quantum wells with those made from Al_xGa_1-xN allowed devices to reach the ultraviolet realm, emitting in regions known as the UVB (315 nm to 280 nm) and the UVC (280 nm to 100 nm).

Further progress has followed, and today several companies from around the world are pursuing large-scale manufacturing and commercialisation of these solid-state sources, which are being deployed in many applications. Here, we provide an overview of this progress, including details of the accomplishments of several groups making major contributions. We begin with a historical perspective, before providing an account of the current-status, the emerging applications, and finally some of the latest research directions.

Major milestones

Two trailblazers of the UV LED are our team from the University of South Carolina, and the group at NTT, Japan. At the beginning of 2001 NTT reported a device emitting at 346 nm, and shortly after we unveiled an LED with a peak output at 341 nm. These two devices broke new ground, featuring the first active regions made from the pairing of AlGaN and AlInGaN layers. Before the year was out, we had made further progress, realising 315 nm emission, the edge of the UVB domain. And a year later, we eclipsed that success, reporting an LED with a milliwatt output power, emitting at 278 nm, just inside the UVC band.

Like many first-generation LEDs, these devices had a very low external quantum efficiency. Four factors were to blame: extended defects, resulting from lattice-mismatched growth of AlN/AlGaN layers on sapphire; a low doping efficiency, prevalent in p-AlGaN layers; the absorption of many of the photons exiting the quantum wells by hole-supplying p⁺-GaN; and a low light-extraction efficiency, exacerbated at the very shortest wavelengths.

By tackling these issues, the developers of UV LEDs made much progress over the next two years. For our team, as well as several other US institutions, efforts were supported by DARPA and the US Army. These bodies funded the SUVOS programme, managed by Lt. Col John Carrano. During SUVOS, we improved our materials, their growth, and the design and packaging of our device. These modifications created devices with an external quantum efficiency around 1 percent and an output power of nearly 1 mW at a 25 mA drive current.

Better performance resulted from a combination of: high-temperature, migration-enhanced pulsed epitaxy for the growth of AlN buffer layers, which have a thickness in excess of 2 μm; the introduction of a nested superlattice for the AlGaN layers of the AlN/AlGaN superlattices; inclusion of a magnesium-doped, electron-blocking AlGaN layer; incorporation of polarization-doped, graded-composition p-AlGaN layers (see Figures 2 (a) and 2 (b)); and flip-chip mounting of the devices on metallic carriers (see Figure 2 (c)). 

As just mentioned, one of the features of our second-generation devices is an improved superlattice design that aids defect and strain management and boosts output power (Figure 2 (d)). In this refined architecture, the AlN/Al_xGa_1-xN superlattices are nested short-period superlattices. They are formed using an aluminium-composition of the Al_xGa_1-xN layer of the macro-superlattice that is a thickness-averaged composition of the nested short-period Al_xGa_1-xN/Al_yGa_1-yN superlattices. By adopting this approach, we could grow crack-free, high-quality n⁺-Al_xGa_1-xN (n-contact layers) with thicknesses well over 2 μm. Incorporating this layer into our LEDs increased current spreading, mitigated current crowding and increased the current at which power-saturation kicked in. Thermal management also improved, leading to a longer device lifetime.

Substantial progress followed. By 2005, we were able to report the stability of our devices and their suitability for bio-chemical detection. Our demonstration of high-power, stable UVC LEDs provided a tremendous foundation for their commercialisation in applications that included air and water purification. While these devices are now well off the pace, the level of performance they provided is very effective in killing viruses and bacteria such as Ecoli.

Figure 4: Global Partnerships for deep-UV LED development.  Also shown is the substrate material that each team uses.

In addition to making significant contributions to the development of conventional, large-area UVC LEDs, we have also demonstrated a micropixel device. This configuration, reported in 2004, aids thermal management. Devices run cooler due to a reduction in total device resistance, a measure that combats current crowding and joule heating. Another merit of the micropixel device is that it allows biasing of individual pixels, making it ideal for deep-UV LED displays, UVC optical communication and deep UV lithography systems (see Figure 3). For the growth of UV LEDs, the most common substrate is sapphire. This material is transparent, low in cost, widely availability, and when used in combination with AlN buffer layers, it provides a good foundation for UV emitters. However, two US groups that manufacture bulk AlN substrates – Crystal IS and Hexatech Inc – have explored the potential of this platform for deep-UV LEDs. Note that other groups have not participated in this evaluation, due to the high cost and limited availability of AlN substrates. Although the use of native AlN has promise, one issue with bulk material is that during its growth, impurities are incorporated that have strong absorption in the UVC band.

Commercial conquests

Within the deep-UV LED community, university research groups have driven development. Their knowledge has transferred to start-ups, driving early commercialisation. More recently, several start-ups have formed partnerships, or been acquired by large companies.

Our team started this trend. Our deep-UV LED technology has been behind the launch of SETi and Nitek Inc, two small businesses in Columbia, South Carolina, that have been bought by Seoul Viosys Company and Seoul Semiconductors, respectively. In a similar vein, the technology from Nagoya University and Meijo University spawned UV Craftory Company, a start-up subsequently acquired by Nikkiso/FPG. Meanwhile, Riken’s technology has been directly acquired by DOWA/Panasonic; and more recently, in Germany, the technology from TU Berlin/FBH has been transferred to UV Photonics.

It’s a similar state of affairs for deep-UV LEDs developed on substrates made of AlN. Technology invented at Rensselaer Polytechnic Institute underpinned the launch of Crystal IS, which has been acquired by Asahi Kasei. Likewise, research at NC-State created Hexatech, now part of Stanley. In addition, large corporations, such as Samsung, LG Innotek, QD Jason and Nichia Corporation, have also started to venture into the deep UV.

Many different approaches are adopted for forming relaxed, thick AlN buffers. Such buffers are highly desirable, because their lower threading dislocation density lowers the defects in the active region and increases the LED’s internal quantum efficiency.

We are growing these buffers by high-temperature pulsed epitaxy and hydride vapor phase epitaxy, while researchers at Riken are creating them by ammonia flow modulation epitaxy and a team at TU Berlin is producing them by migration-enhanced lateral epitaxial overgrowth and the use of ex situ high-temperature-annealed sputtered AlN.

Commercial progress with deep UV LEDs has led to the launch of devices emitting between 240 nm and 300 nm. Despite significant progress, the external quantum efficiency of these devices and their wall-plug efficiencies fall well short of the figures for their visible counterparts. While InGaN-based blue LEDs realise figures of around 80 percent for these measures of efficiency, values for deep UV LEDs emitting between 266 nm and 300 nm are typically just 5-6 percent.

As the emission wavelength shortens, the external quantum efficiency of deep-UV LEDs falls, along with its wall-plug efficiency. This is partly due to degradation of material quality, and difficulties in doping nitride-based alloys with increased aluminium composition. In addition, according to a team at Kansas State University, once emission drops below 270-280 nm, transverse magnetic-polarized emission becomes more dominant, due to a reordering of the valence bands. As this form of light travels lateral to the c-plane, it is efficiently trapped in the device, hampering photon extraction.

Today, there is a wide variation in the packages and form factors of commercial deep-UV LEDs. Discrete devices are available in TO-39 packages. This is not suitable for large devices, which require superior thermal management. These chips are housed in SMD packaging with AlN sub-mounts. Recently, there has also been the introduction of devices incorporating UV-transparent epoxy encapsulation (see Figure 5 for images of the different device packaging schemes).

For everyone working in the field of deep UV LEDs, the biggest challenge is to increase device performance to values comparable with visible counterparts. Today’s device design for commercial products has two major drawbacks. First, the p⁺-GaN hole injection layer leads to a loss of light travelling towards the p-contact and the wave-guided light in the active region. Second, light extraction efficiency is very low.

Approaches to mitigate light absorption by the p⁺-GaN layer and extract/collect more light travelling towards the p-electrode include the use of p-AlGaN layers. This approach has been pioneered by a partnership between Sensor Electronic Technology, Rensselear Polytechnic Institute and the US Army Research Laboratory; and also a Japanese collaboration led by RIKEN that has employed reflective p-contacts. Both teams have realised external quantum efficiencies as high as 10 percent.

Even greater success has come from RIKEN, working with Eco Solutions Company, part of Panasonic Corporation. They have reported external quantum efficiencies of 20 percent, realised by combining patterned-sapphire substrates and transparent p-AlGaN contact layers.

Future directions

It not surprising that by and large, the development of AlGaN-based deep UV LEDs is mirroring that of InGaN-based visible LEDs. The latter now incorporates tunnel junctions, a technology starting to be explored in deep-UV LEDs. By adding MBE-grown, UV-transparent AlGaN tunnel junctions contacts made from n⁺⁺/p⁺⁺-Al_0.15Ga_0.85N to MOCVD-grown active layers, researchers at The Ohio State University have doubled the output power and the external quantum efficiency of LEDs emitting in the 280-290 nm range.

By summer 1995, Shuji Nakamura, trailblazer of efficient blue LEDs, had already become an industry celebrity. After inventing this device in 1993, he remained in the limelight by setting a series of ever higher benchmarks for the performance of this light-emitting chip. These successes enriched his standing within the tech pantheon of all-time greats and brought many accolades, including the biggest of all, a Nobel Prize for Physics, coming in 2014. But he got even greater pleasure from seeing his work laying a foundation for a lighting revolution.

While Nakamura is undoubtedly renowned for contributions to the development of the GaN LED, it is by no means the only ubiquitous device he has invented. He is also the originator of the cousin of the GaN LED, the GaN laser diode. The latter is an astonishing triumph. While many manufacturers are making millions of GaN LEDs every month, 25 years on from its introduction, there are still only two high-volume makers of the GaN laser diode: Nichia, where Nakamura worked throughout the 1990s; and Osram Opto Semiconductors of Regensburg, Germany.

Nakamura’s laser breakthrough came in late 1995, with news breaking on 12 December. Nichia’s greatest engineer appeared on TV later that month on a national evening news broadcast, demonstrating his wonderful new device.

By early 1996, details of the laser diode started to appear in the scientific press. A paper in the Japanese Journal of Applied Physics described a 30 μm-wide, 1500 μm-long stripe laser grown on a sapphire substrate. Emitting at 417 nm, it had an active region containing 26 quantum wells, each made from a 2.5 nm-thick layer of In_0.2Ga_0.8N. To get this device to lase, the voltage had to be cranked up to 34 V. Driven in pulsed mode at even higher voltages, this chip produced an output of more than 200 mW. Encouragingly, this laser showed no signs of degradation after two hours of operation.

Even before its debut, the entire optoelectronic industry had little doubt regarding the killer application for this short-wavelength laser diode. Back then most music lovers bought CDs, VHS recorders were on the verge of being replaced by DVD players, and even better picture quality was now on the horizon, thanks to a five-fold hike in storage density enabled by blue-violet lasers.

Those keen to splash out on a new generation of disc players that would allow them to watch high-definition movies in their own homes had a really tough time. Early adopters began by getting caught up in a format war between the advocates of HD DVD technology and those giving their backing to Blu-ray. When that had been resolved, allowing customers to pay top whack for one of the first Blu-ray players that hit the shelves in Japan in 2003, further frustration followed – those early adopters had to wait several years for the launch of the first titles.

Fortunately, some good years followed, providing a growing, lucrative market for the blue laser diode. Since then the increasing uptake of streaming services has sent sales of Blu-ray players into terminal decline – but as this door has shut, several others have opened. Blue lasers are now being deployed alongside red and green variants in colour projectors, and they are used also in copper welding systems, where they operate at an absorption sweet spot, enabling excellent welds. There is also the possibility that one day lasers will replace LEDs in general lighting – so maybe, just maybe, there will come a time when society is as thankful to Nakamura for his laser as they are for his LED.

Shuji Nakamura, inventor of efficient GaN LEDs and GaN lasers, sitting next to Nobuo Ogawa, founder of Nichia. Credit: Bob Johnstone. 

In 1999 the merger of UK-based Epitaxial Products International and US-based Quantum Epitaxial Design spawned IQE, which launched on the European equivalent of the NASDAQ.

Efforts to turn the vision of laser-based weapons into reality were supported by a DARPA-funded programme aimed at increasing the efficiency of infrared laser diodes from below 50 percent to 80 percent or more.

Construction of the Institute for Compound Semiconductors is well under way, and should be completed next year.

In spring 2019, Cree, the developer of SiC substrates and various optoelectronic, RF and power devices, re-aligned its business in emphatic fashion. It’s relatively new CEO Greg Lowe had no qualms in carving off the company’s Lighting Products division to Ideal Industries; giving the LED business that brought so many years of success a back seat; and focusing on wide bandgap materials and RF and power devices.

Helping to have made this monumental decision would have been a number of lucrative SiC wafer supply deals: contracts with Infineon, ST Microlectronics and other companies totalled $500 million.

By summer 2019 Cree’s  LED business had softened and it had netted the lion’s share of its $310 million sale to Ideal. These circumstances would have helped Lowe to move forward with his vision, investing $1 billion in massive expansion of SiC capacity. A state-of-the-art, 200 mm facility would be built by 2024, delivering a 30-fold expansion in capacity compared with the first fiscal quarter 2017.

Later that year Cree had an offer it could not turn down – a $500 million grant from the state of New York. Instead of having to spend $450 million retrofitting an existing structure at its headquarters in Durham, NC, it could now invest $170 million in building a new, automotive-qualified 200 mm power and RF wafer fabrication plant in Marcy, New York.

The new building should be up and running in 2022, initially using 150 mm wafers before transitioning to 200 mm wafers two years’ later. Until that comes on line, capacity is being increased at the headquarters.

Since these announcements, the proportion of revenue coming from the less profitable LED products line has continued to decline, while that from the Wolfspeed division – it encompasses SiC wafers, power electronics and RF devices – has seen revenue grow. Sales are currently taking a knock from the Covid-19 epidemic and trading restrictions with China, but the long-term prospects for Cree are very, very good.

Supported by a $500 million grant from the state of New York, Cree/Wolfspeed, is building a new automotive-qualified 200 mm power and RF wafer fabrication plant in Marcy, New York, that should be up and running in 2022.

I EXPECT THAT YOU, like me, are annoyed by the oft-repeated quip ‘gallium arsenide is the material of the future and always will be’. That view surely smacks of ignorance. Those that say such things are clearly unaware of the GaAs-based laser diodes in their CD and DVD players, the high-power variants used to cut and weld metals that form the skeleton of their cars, and the GaAs HBTs that amplify the RF signal within their smartphones.

However, while some of those working within the silicon industry may irk us all – whether we are part of the GaAs industry, or a sector involved in the manufacture of InP, GaN or SiC chips – we have to admit that we can learn a lot from them. The reality is that those of us that work in compound semiconductor fabs dream of replicating the maturity of their technology, their understanding of reliability, and their low levels of defects.

Another lesson we can learn from the silicon industry is that if we plot the rate of success to date and extrapolate this, we see what the future holds. It’s a point well-illustrated by Moore’s Law, a rate of progress that has been maintained for decades. Even when continual success seems unlikely, with further miniaturisation throwing up issues with no obvious solutions, progress has continued at historical rates.

We can also plot progress in our industry. This gives us the opportunity to not only look back at how far we have come, but see where we might be in the future. Using this approach we can take an educated guess at not only where the performance of compound semiconductor chips could be in the future, but also the advances in the technology they support.

In the LED industry, progress follows Haitz’s Law, which states that with the passing of every decade, the cost-per-lumen plummets by 90 percent and the output of a packaged LED climbs by a factor of 20. It is this rapid rate of progress in the bang-per-buck that has enabled white LEDs to move on from their first killer application of providing illumination in the handsets to backlighting screens and eventually providing general lighting. Over the next few years, further falls in the price-per-lumen will help LED lighting become the incumbent technology everywhere. Revenue for this sector is tipped to climb, but the LED light bulb is arguably now a commodity, offering razor-thin margins to chipmakers.

An emerging opportunity for visible LEDs is in displays made from microLEDs. This application, which could net billions and billions of dollars, will be helped by a fall in the cost-per-lumen. However, if this type of display is to become commonplace, a breakthrough is needed in the efficient mass transfer of the tiny chips that create the pixels. It’s a problem that has been grappled with for several years, and there are some promising solutions, such as eLux’s microfluidic technology – but more progress must follow if this is to become the next killer application.

LED on Mars

For deep-UV LEDs, if progress continues to follow Haitz’s Law, these devices will be everywhere by the time Compound Semiconductor celebrates its 50th anniversary. By 2045, packaged deep-UV LEDs could deliver several hundred watts of power, at a cost of just a dollar-per-Watt, thanks to improvements in the chip’s design, its package and the way it is driven.

If that happens, today’s incumbent source of deep-UV emission, the mercury lamp, will be relegated to museums. By then, the UV LED will be the dominant technology for purifying water and disinfecting surfaces and air, and it will also be deployed in numerous processes involving the curing of adhesives. Some devices may even be on Mars – NASA has manned missions slated for the 2030s, and even if there are delays, there is a good chance that within the next 25 years deep-UV LEDs will be purifying an astronaut’s drinking water on the red planet. 

The coming decades should also witness a steady increase in the efficiency of blue lasers. While blue LEDs can operate with wall-plug efficiencies exceeding 70 percent, today’s best single-mode blue lasers are only operating in the high thirties, and their multi-mode variants in the mid-forties. Both figures are on the rise, and even if the rate of improvement is not maintained in the coming decades, efficiencies could still exceed 70 percent by 2045.

What would be the implications of such efficient emitters, which have the added benefit of easing thermal management? There is sure to be a rise in the use of these sources for processing copper and other yellow metals, because these materials have an absorption sweet-spot in the blue. In comparison, when today’s more common infrared lasers are used to process yellow metals, this leads to sputtering and ultimately inferior welds.

In addition to a hike in sales for material processing, driven by the growth in electric-vehicle production, blue lasers are tipped to see increasing use in colour projectors. Insufficient brightness has held back sales, but the opportunity to dispense with a backlit screen is a major asset in both the home and the office. Efficient green lasers are also needed. While they lag those in the blue, they are also improving, and they don’t need to be quite as efficient, because they operate at around the eye’s most sensitive wavelengths.

Given the merits of on-wafer testing, a circular emission profile and a very high modulation rate, the next 25 years should witness growth in shipments and spectral coverage of the VCSEL. Over the next few years, infrared variants will gain greater deployment in smartphones, where they support facial recognition, and further ahead in vehicles, providing a light source for LiDAR. GaN-based blue and green VCSELs are still in development, but there is no doubt that they will move into production, probably initially as sources for car headlights. By 2045, they should have entered other markets, possibly including displays.     

Multi-junction missions

In the next few years many satellites are going to be launched. Contributing to this hike is the SpaceX Starlink programme, aiming to create a network of 12,000 small satellites by 2027. The satellites will form the backbone of global broadband internet.

Such efforts will provide much business for makers of multi-junction solar cells. Although these devices are pricey, the higher efficiency and great robustness to radiation allow them to monopolise this market. The efficiency of these cells is increasing, but gains are not easy to come by. One option currently being explored is to concentrate sunlight, an approach that can deliver a substantial gain.

Efforts at concentrating sunlight previously offered much promise for the creation of a terrestrial industry, before hopes were dashed by a dummy whammy: plummeting silicon prices; and a financial crisis, which starved fledgling firms of much needed investment. Far fewer teams are now battling for the bragging rights that come from having the most efficient cell – a potential solution to cutting costs – but records continue to rise. Leading the way is a team from NREL, pursuing a six-junction design. This device has an efficiency of just over 47 percent, and if its resistance can be reduced to that found in four-junction variants, this architecture could break the 50 percent barrier.

Further gains are possible, with values of just over 60 percent a realistic upper limit. Given the slow progress with silicon cells, that would be more than double the record value for the incumbent technology. But even an advantage that significant is not necessarily going to lead to a re-birth of the terrestrial CPV industry, which is in a catch-22 situation. For costs to come down there needs to be economies-of-scale, alongside refinements coming from insights associated with significant production volumes. Could that come from CPV in space? If it does, and the power generated per unit area comes to the fore as a key metric in sunny climes, CPV stands a chance of being a big industry by 2045.

SiC, GaN and gallium oxide

In the wide bandgap industry, progress is fast, spurred on by Cree’s $1 billion investment in SiC that will increase its capacity by a factor of 30 between 2017 and 2024. Many of its SiC devices will be deployed in electric vehicles, which will become more common as governments all around the world look to trim their carbon footprints. Shipments of GaN-on-silicon power devices are also tipped to take off. Operating at lower voltages – initially 600 V and below – they will enable an increase in the efficiency of numerous forms of power supply. Further ahead, GaN CMOS should start to appear, leading to further gains in circuit efficiency, realised with smaller, lighter-weight units.

However, by 2045, these middleweights – and that is how we should look at GaN and SiC – could well be in decline. By then, they could be overtaken by gallium oxide, a heavyweight with many attractive attributes. Gallium oxide’s potential performance, judged in terms of Baliga’s figure-of-merit, is better by an order of magnitude, and the cost of manufacture will be competitive. Unlike SiC and GaN, native substrates can be grown by melt-based techniques, allowing production of competitively-priced devices with great crystal quality. Like GaN, transistor performance is impeded by difficulties in realising p-type doping, but engineers can turn to a similar trick to ensure success: use the work function of the gate to pinch off an array of fin channels. Another issue with gallium oxide is its low thermal conductivity, but again, the solutions adopted with other compounded semiconductor devices offer routes to success. By following in the footsteps of other parts of the wide bandgap industry, it is possible that gallium oxide devices will take less time to get to market, and undergo a faster ramp to production volumes.

The shape of 7G

Based on historical rates of progress, which indicate that it takes about a decade to migrate from one generation of mobile technology to another, we should have 7G well established by 2045. If this happens, one can envisage incredibly fast data rates, possibly at hundreds of gigabits per second, using transmission in the terahertz domain.

But is it inevitable that we shall have this technology? Some well-informed sceptics suggest not, offering a range of compelling arguments based on market economics. Many of us are upgrading our smartphones far less frequently than we used to, because the latest models offer only incremental benefits – maybe a better camera, or water-resistance. Often we are reluctant to pay high fees for call plans, and so it’s hard for carriers to grow their revenues. The problem is essentially that as time goes on, we expect more while paying less. With that mindset at play, carriers should be reluctant to invest in any new generation of technology. And this concern should be magnified, because the roll-out of successive generations of technology is now getting more expensive. New generations of mobile technology operate at higher frequencies, where attenuation is higher. As increasing the transmission power is not a viable option – regulators are unlikely to approve, and a hike in power would be detrimental to battery life – the solution is to go to smaller cells, and this increases roll-out costs.

One may also wonder why anyone would need these breath-taking transmission speeds. It makes no sense to watch a broadcast in ultra-high-definition on a portable device with a small screen. Maybe ultra-fast data rates would be used for remote robotic surgery, but that is hardly a mainstream application. So, if most of us can’t see the need for super-high speeds, why would we ever be willing to pay extra for them? The threat is that carriers will never recoup their investment.

But maybe we are not looking at this in the right way. Back in the 1990s, when the much-lamented Word 5 provided the jewel in the Microsoft suite, there didn’t seem any need for more computational power, either in speed or memory. But if we were to try and use those machines today, they would feel very clunky and fail to cope with our high-res photos, digital downloads and so on. The extra levels of computational performance have been used in ways that we value today, but had not been envisaged.

Right now, the Covid-19 pandemic has shifted the world of work. While we might now get into the office, if we attend a conference, it’s probably involving logging on and listening to presentations. It’s a poor second to being there, as we can’t interact with the same level of intimacy as before. For example, we can’t pick up on body language. However, suppose we had ultra-fast communication, offering a more immersive experience involving some form of virtual reality. This could be so beneficial that companies would be willing to pick up the tab for the connection, reasoning that it offered a good return for an outlay far less than plane tickets and hotel bills. It would also help to trim carbon footprints, an issue that may be taken more seriously in the 2040s.

If 7G does roll out before this magazine celebrates its fiftieth anniversary, it will surely be good news for our industry. Whether communication occurs high in the gigahertz realm, where there is plenty of space, or in the terahertz domain, compound semiconductor devices are sure to features in handsets and infrastructure. GaAs may still dominate, but it could lose out to InP, the king of speed within the III-Vs.

Successes in this sector, plus those in power electronics, satellites, displays and disinfection, suggests that even those in the silicon industry currently failing to appreciate the benefits of compound semiconductors today should have started to change their mind well before 2045. Compound semiconductor devices are already pervading society, and in 25 years’ time the role they play will be even more significant.

The element with atomic number 49 – indium – is central to modern life. Just consider a typical day: often we will spend hours staring at a flat panel display containing hundreds of thousands of pixels, each turned on and off by thin-film transistors and illuminated by some form of LED. While these transistors and LEDs might contain indium, there is no doubt that this element is encountered by the light traversing the display stack, as it will pass through indium-tin oxide (ITO) electrodes. This oxide combines transparency with conductivity, making it a great material for the display industry. In some of the latest high-colour-gamut TVs, indium crops up again, appearing in the display’s InP quantum dot layer of the colour filter that yields brilliant reds and greens. This makes five potential encounters between photons and indium-based materials before the light hits our retina. And there may be even more. If we watch content stored remotely, such as streaming video, it will reach us through a network of optical fibres, transmitting infrared light generated by an InP-based laser.

For those of us that are reading this on an electronic screen, the chances are that most of the above applies – there is no doubt that indium seems to be everywhere. Yet, despite its ubiquity, the economics of indium are poorly understood. That’s because most of the building blocks of modern technology emerge from supply chains that are rarely discussed, even in the technical press. To put it frankly, most of us don’t have much of a clue where indium comes from.

Let’s go back to the beginning. Like all heavy elements, indium was created during supernovas as planets condensed and formed. Due to its geo-chemical properties, indium appears in sphalerite, the most common ore of zinc, where its concentration can be as high as 100 parts per million. Another common carrier element for indium is lead. Deposits containing relatively high concentrations of indium are found all over the world. According to an exhaustive survey detailed by Ulrich Schwarz-Schampera and Peter Hertzig in Indium: Geology, Mineralogy, and Economics, a large number of indium-bearing deposits have been found around the Pacific Rim and in many other sites in Europe, Africa, and central Asia.

It comes as a shock to many, including quite a few working within the semiconductor industry that there are no indium mines anywhere. Element 49 is almost always co-mined with zinc. That’s not actually too surprising, as there are only a few carrier elements that have their own corresponding production infrastructure, such as copper, iron, and aluminium. When demand swings for most co-mined elements, such as indium, supply at the carrier infrastructure is often relatively inelastic. The real supply for indium is actually set downstream from the mines at the refiners where material is processed to produce an indium content up to 4N (99.99 percent). Indium ore is found all over the globe but refiners are heavily concentrated in China.

The overall global abundance of indium is in the range of 5-7 parts per million. That may not seem like a lot, but it is almost exactly the same value as that for silver. However, unlike that precious metal, which has been in use since ancient times, indium only started its commercial life in the twentieth century and was initially explored for its mechanical properties. One of the first successful applications of indium began in the 1940s when engineers diffused this metal into the surface layer of the bearings in aircraft engine pistons. Indium provided lubrication on the atomic scale, reducing wear and tear on the engine and increasing fuel efficiency – a prime example of our strongly held belief at Indium Corporation that materials science changes the world.

Fast forward to the age of electronics and high-speed communication when InP semiconductor crystals came to the fore, thanks to a bandgap that matches the transmission window of optical fibre. With a light source and a lossless technology for transmitting information, the groundwork was laid for the Internet and the connection of computers. It is a success story that we all know.

The InP-based lasers in optical networks are built on wafers sliced from monocrystalline boules, which are grown using techniques such as vertical freeze gradient and the Czochralski method – both requiring high-purity indium ingots as the starting ingredient. It is not sufficient to use standard commercial indium metal, which has a purity that starts at a 4N. That purity is suitable for making inorganic compounds and ITO sputter targets, but is not nearly enough for the production of compound semiconductors, which require a purity at least 500 times higher, typically 6N5.

What happens to the InP crystals and wafers that are left over from their respective manufacturing process? What about the sustainability of the indium value chain?

From the start, the ITO and flat panel display industry have been reclaiming indium and the compound semiconductor industry is now following suit. This is to be expected, as every mature, high-value-added materials-using industry should have a recycling loop for its valuable input materials. At Indium Corporation we have added this reclaim capability, initially for InP crystals and wafers, and we plan to add capabilities for InSb in the near future.

There is no doubt that it makes sense to do this at the industrial level with end-of-life wafers and unused crystal material. It is harder to accomplish this with packaged electronic InP devices or end-of-life consumer products – this probably explains why we are not aware of any companies doing so. However, if our society is to be a responsible steward of our planet, this is an important objective.

As of today, we are concentrating on output from crystal growers, InP wafer makers, and output from fab and epi-wafer fabs. If you are using InP wafers in a fab and have any recycling needs, we certainly would like to hear from you.

So, what does the InP-consuming industry landscape currently look like and how has it evolved? 25 years ago, this magazine reported that global InP wafer consumption totalled 58,000 wafers, equating to a surface area of 183,000-square inches. Back then, Japan led the world and was responsible for 56 percent of wafer consumption. Today, according to a recent report from SEMI, 64 fabs are capable of processing InP wafers – 40 percent are found in North America, 31 percent in Europe, and the balance in Asia. This report claims that the capacity figure of all these InP fabs, based on 200 mm equivalent wafers, totals 667,000. That’s a figure that equates to just over 11 times that of global consumption in 1995. So, it is clear that there has been an astonishing growth of InP, even with the caveat that the SEMI figures are capacities, rather than actual output. They are total fab capacities; that is, the InP output is in the mix but not the total number.

Another use of indium is, of course, in indium-containing epilayers. They are grown epitaxially, often by MOCVD, using the metal-organic trimethyl-indium. This source is made from high-purity indium trichloride, available as an anhydrous salt. The density of this salt may be tailored to the process of our customers.

When engineers use trimethyl-indium to grow heterostructures by MOCVD, they control the flow of this source of indium. With this lever, they tune the band structure and shift the emission or absorption wavelength of a device. An exciting new development over the last few years is the development of hybrid integration, a technology that allows compound semiconductor layers to be constructed directly on a silicon substrate.

As this technology is still in its infancy, outside the display market the growth of indium consumption is largely driven by data centres. They are the workhorses for the cloud, which has become both the storage and the processor of internet core computing. The cloud holds messages, photos, videos; it hosts entire web sites; and it runs software algorithms like machine learning. And if you are reading this online, the article will be stored there and brought to you via InP lasers to a screen coated in ITO, probably backlit with InGaN-based LEDs.