Silicon has served us well over the last 55 years since the first integrated circuit was invented at Texas Instruments. Today, the symphony of chemistry, physics and engineering required to orchestrate the production of 22nm node chips in the latest Intel or TSMC fabs represents the pinnacle of 21st century technology. As great as Silicon may be as the driver of today's digital world, for many applications its properties make it a terrible semiconductor choice. For example, its electron bandgap is not compatible with light emission for LEDs, while its electrical and thermal properties make it an extremely inefficient choice for power electronics. Fortunately, the periodic table has come to the rescue with a vast array of compound semiconductors waiting to fill the gap.
The applications for compound semiconductors are as varied as the different possible alloy combinations. LED lighting, laser diodes, power electronics, sensors and solar come to mind. Each alloy has a unique set of electrical, thermal and optical properties that make it ideal for specific applications. For example, the bandgap of Indium Gallium Nitride (InGaN) matches the wavelength of visible light while Aluminum Nitride (AlN) is better suited for the deep UV spectrum. Silicon Carbide's (SiC) thermal properties make it the ideal choice for high temperature, high voltage power electronics, while Gallium Nitride's (GaN) electrical properties enable higher efficiencies, making GaN the optimum choice below 1000 volts. Cadmium Zinc Telluride (CZT) loves to soak up X and Gama ray radiation while Copper Zinc Tin Sulfide (CZTS) prefers its radiation in solar form.
Despite the broad applications, the compound semiconductor industry can be characterized as a grouping of cottage industries (LED manufacturing and high frequency Gallium Arsenide communication devices are notable exceptions). Collectively, the industry is an order of magnitude smaller than the $300BB silicon device market and, as is often the case in materials, a combination of cost and performance is to blame. But that is about to change. Several weeks ago, the White House recently announced millions of dollars of support to help create a manufacturing hub in North Carolina for "Wide Band Gap" semiconductors – an important subset of compound semiconductors as a whole.
But focusing on device manufacturing innovation and scale may not be the keystone for success. The technical issues are complex but much of the limitation today is rooted in the starting substrates. They generally suffer from some combination of high manufacturing costs, poor scalability and a high concentration of performance-robbing crystal defects. The impact is felt all the way up the value chain.
Let's explore GaN/InGaN as an example. Currently, GaN power devices and InGaN LEDs are generally grown on sapphire or silicon substrates. The crystal lattice mismatch between the substrate and the nitride creates defect densities on the order of a billion per square centimeter. Defects rob efficiency and reduce power, while the brute force method of buffer layer engineering to make the surface "good enough" for device growth adds tremendous cost. Furthermore, the substrates have a different coefficient of thermal expansion compared to the nitride, which causes bowing at high temperature. This "Pringle chip" problem causes a major yield hit, as MOCVD reactors don't conform to curvature in the same way as your tongue. Unfortunately, this gets dramatically worse as substrate diameters rise up to 200mm and 300mm, where highly automated CMOS fabs could be used for low cost wafer processing. The use of a native GaN substrate would solve the two technical problems, but pricing in the thousands of dollars for a 50mm substrate makes this an uneconomic and unscalable choice. Other compound semiconductor materials have their own flavors of the same problems. At Pangaea Ventures, we believe solving these various challenges is the keystone required for the compound semiconductor industry to enjoy the growth that silicon has already seen.
In 2012, we funded proof of concept experiments for a company called Tivra Corporation, which promised to solve the crystal lattice and "Pringle Chip" problem for nitrides such as GaN. Having followed incremental progress in this area for many years, Tivra's entirely different approach to the problem was the first we had seen that promised scalability to large area at a game changing cost. We are excited to say that the experiments worked and the company is now fundraising for its Series A round.
We stepped further out-of-the-box in 2013 investing in a microgravity enabled materials company called Masterson Industries. The first demonstrations of the technology involved healing the defects in SiC wafers and we can say that results exceed even our most optimistic expectations. Customers will benefit from higher yields, higher power and higher voltage devices.
When we invested, these two companies would have been considered too early stage for most VCs. Fortunately, Pangaea's deep understanding of materials technology allowed us to make educated bets on the respective technical risks. The biggest venture returns usually come from left field.
Early stage investing does not always make the most sense. It has been over a decade since we first learned about Redlen Technologies and its revolutionary approach for the tricky process of CZT crystal growth. The need for high quality CZT crystals and detectors has always been obvious, but it wasn't until 2013 that we got comfortable with the maturity of the crystal growing process and adoption cycles of the security and medical imaging markets where they are used. Some parties don't get going until the wee hours, and we are excited to have recently led the company's latest investment round.
With three new portfolio companies working with compound semiconductor technology, if this was poker, you might say we are "all in". But the market applications are huge and with many compound semiconductor technologies at the stage where silicon was 30 years ago, the big commercial breakthroughs and associated venture returns are yet to be made.