Gallium nitride and efficient transistors
In 2009, those knowledgeable about the forefront of semiconductor research would have forecast dramatic growth in the market for gallium nitride (GaN) devices. Numerous technology news outlets, from PC Magazine to Reuters, reported on the potential advantages this new semiconducting material offered. iSuppli (now IHS) produced a report in 2009 that foretasted growth in the market for GaN to more than $180 million before 2014. Researchers have been experimenting with GaN since the 1980s, and over the past two decades GaN-based light-emitting diodes have achieved widespread usage. In 2005, researchers became more interested in GaN for transistors, another exceedingly common semiconductor application.
We reached out to IHS to obtain industry insight into GaN. According to Marijana Vukicevic, senior principal analyst, power management for IHS and author of the 2009 report, the recent interest in GaN-based transistors for use in power applications is due to the fact that, "Some of the main applications for traditional power devices like MOSFETs (manufactured in Si) have been increasing in power demand." The capabilities of silicon-based transistor components were and are being pushed to new extremes. GaN seemed poised become the 'new' ubiquitous semiconductor because as its material properties are much more favorable in high-power conditions. Unfortunately, the economic downturn has stunted GaN market growth, and the technology remains poised rather than perfected.
Despite being hindered by unfavorable economic conditions, GaN devices inevitably offer significant improvements over older materials, and it is my opinion that the market just needs to want it bad enough to invest in the development of manufacturing technologies.
Better, faster, more efficient
Before gallium nitride was ever reliably produced in a laboratory, researchers anticipated its semiconducting properties. Semiconductors became a significant area of research as interest in transistors and diodes boomed after WWII, and material scientists studied how the atomic-scale composition impacted electronic properties. In particular, these properties are determined by the configuration of electrons within a particular material. In a solid whose electrons are not tightly bound to the atoms that make up its crystal structure, many electrons are freely available to conduct electricity with very little resistance. On the other hand, many materials form crystals where electrons are tightly bound to the their respective atoms, so little or no electricity can flow through these materials, giving electrical insulators. Semiconductors fall somewhere in the middle, and we now understand how different atomic electron configurations produce different crystal structures that provide more or less electrons available for conduction. Due to ease of fabrication, silicon became the first widely used semiconductor in transistor applications, but other types of semiconductors have long been known to exist.
The most important characteristic of semiconducting materials is that the amount of electrons available in the material for conduction can be controlled. Depending on how 'tightly' a particular crystal configuration of atoms is holding on to its electrons, an applied electric field can free more electrons for conduction. All modern electronics are based on this simple property of semiconductors: by putting them in an electric field, we can increase or decrease how much electricity they conduct. With this in mind, there are two properties of semiconducting materials that we need to understand the importance of GaN.
The first is how tightly the electrons are held to the atoms in a crystal, or how much energy it takes to remove electrons from the atomic lattice and make them available for conduction. This energy is a property of a particular material's atomic and crystal structures, and is referred to as the band gap. High band gap materials require stronger electric fields to increase their conductivity, and the electrons are more energetic once removed from the lattice and available for conduction. GaN offers a band gap that is much larger than most silicon crystals. Accordingly, the first GaN application was in light-emitting diodes where its unusually high band gap and highly energetic conducting electrons could produce higher energy electromagnetic radiation. This allowed LED technologists to complete the high-energy end of the visible light spectrum, producing the first blue LEDs in 1989. Scientists were also able to produce higher-frequency, small-wavelength lasers that provide increased spatial resolution in laser-based applications. The most popular application that relies on this property of GaN is data storage, which is how Blu-ray compact disks can store much more information on a small surface area.
The second relevant property is known as electron mobility. This is simply a measure of how easily electrons can move through a particular material. Crystal structure, defects and the presence of foreign atoms (as with 'dopants' in silicon crystals) significantly affect electron mobility. In crystals with low electron mobility, electrons have much greater difficulty flowing through the lattice, and much more frequently lose energy when they 'bump into' atoms in the lattice.
The problem is this: low electron mobility limits the frequency and power at which a transistor can operate. If electrons are not easily able to move back and forth through the semiconductor crystals in transistors, these transistors cannot function with high-frequency alternating currents. When the electrons collide with lattice atoms, they lose energy that is transformed into heat. At high frequencies and powers, many more of these collisions occur, resulting in very low energy efficiency. Furthermore, with all this waste energy going to heat, extensive cooling systems are required to allow transistor operation without damage.
In many applications that continue to push the upper limits of frequency and power in transistors, it is no longer feasible to rely on silicon for the semiconducting material. Scientists have been searching for another solution and, as Ms. Vukicevic said, "GaN came into the picture as a material capable of operating at high frequencies and at lower losses."
Gallium nitride transistors develop
Devices based on gallium nitride have proven much more difficult to fabricate than those based on silicon. In the 1980s, researchers were able to first demonstrate growth of GaN crystals in a process known as 'epitaxy,' which involves growing the crystal layer-by-layer on top of a pre-existing crystalline substrate. Following this development, the first field-effect transistor using a GaN semiconductor was demonstrated in 1993, but this application received minimal attention in the following decade.
By 2006, GaN research had sparked the interest of investors. By 2009, GaN systems began to reach market. Efficient Power Conversion Corporation introduced GaN transistors that year, explained on their website as, " ... replacements in applications such as point-of-load converters, Power over Ethernet (PoE), server and computer DC-DC converters, LED lighting, cell phones, RF transmission, solar micro-inverters, and class-D audio amplifiers with device performance many times greater than the best silicon power MOSFETs." The power management technology company International Rectifier also announced, " ... the successful development of a revolutionary gallium nitride (GaN)-based power device technology platform ... ."
A third company, Transphorm, was later to release its first GaN-based product, but has attracted the attention of numerous investors: Soros Management LLC, Kleiner Perkins Caufield & Byers, Google Ventures, Foundation Capital and Lux Capital. Umesh Mishra, who co-founded Transphorm in 2007, was among the top researchers who worked to develop and demonstrate GaN transistors in high-frequency and high-power applications. In 2011 Transphorm released its "EZ-GaN" transistor, a major release in its platform of efficient power conversion technology, which a press release mentions, " ... customers are already using ... in power suppliers, PV inverters and motor control systems."
Difficulties remain
I reached Mr. Shahin Farchshi, Principal with Lux Capital, a founding investor in Transphorm, by email, who highlighted the current difficulties surrounding the widespread adoption of GaN power conversion technology. He said, "Proving a high voltage GaN transistor is one thing, and building up a manufacturing line that produces products that provide six-sigma quality at scale is another. This is not a trivial task, and Transphorm has done a phenomenal job bringing the technology to market."
Though GaN technology offers staggering improvements on the efficiency of transistors, in the past few years of economic difficulty it has not yet been able to prove itself on the large scale. Ms. Vukicevic echoed Mr. Farchshi's sentiment, emphasizing that the market is still in an early stage in her 2009 report.
"Major obstacle[s] [are] still at the system level and how much OEMs are willing to take risks in taking up the challenges by using GaN devices. Of course, GaN devices that we see today will not be GaN devices that we will see in the market in 5 years as the industry will learn more and make necessary improvements. Still the major challenge lays in working with OEMs to convince them [that] the necessary system changes are to be made in order to see the benefits of high efficiency GaN devices."
GaN technology is just as promising today as it was in 2009. Because of unfortunate economic circumstances, investment and development in manufacturing technology was unexpectedly low, delaying the development and dissemination. Regardless, the potential benefits from power conversion applications remain promising, and plenty of recent interest is sure to usher in new technological developments in this field. The favorable properties of GaN make it conducive to a variety of important uses beyond power conversion as well, from microwave generation to photoelectric solar collection. These applications have not yet reached the same level of interest but are likely to benefit from more developed understanding of GaN manufacturing technology as well.