Okay, it’s that time of the year and I have to say it; I love April! Of course, I say this with full realization that I’m writing this blog at the end of April, but I have confidence that April will come again and that thought will tide me over until the next time. So, why do I feel this way and what does this have to do with the compound semiconductor industry?
Well, I see April as the start of nature’s revitalization. Things start to emerge and blossom and at times, I wonder about the health of some of this new growth and I ask “is that going to survive?” Of course in my yard, I often say “what the heck is that?…and what’s it going to look like when it matures?”
These feelings are very similar to ones I had attending the recently concluded Applied Power Electronics Conference (APEC 2017) and Optical Fiber Conference (OFC). These two markets are literally on the opposite ends of the spectrum, with power electronics technologies measured in tens or hundreds of megahertz, while optical technologies reach into the tens or hundreds of gigahertz. Both markets, however, have compound semiconductor applications just starting to emerge, like a fledgling sprout looking to take hold.
I am starting to expand my coverage into the power electronics market and I have just published Compound Semiconductors Increasing Share in Power Electronics Applications as a first look into some of the drivers and trends in this market. This is a large opportunity, currently in the $15- $30 billion range, dominated by silicon technology. The stakes are high because even small improvements in the efficiency of rectifiers, inverters and converters translate to huge energy savings.
Compound semiconductors offer compelling advantages in loss, size, frequency, switching, temperature and breakdown voltage performance. Many of the current applications address high voltage grid and industrial applicatons, so SiC has gotten a head start in this area. In this market, GaN (on silicon) is the new sprout, just trying to push it’s way out of the ground. GaN is early in the product cycle, trying to demonstrate reliability and manufacturability, but proponents are very excited about GaN. They claim that the technology will capture share from silicon in lower voltage applications, as well as enabling new high volume commercial applications like LIDAR, wireless charging and envelop tracking. I will be monitoring how this GaN sprout grows and develops.
At the other end of the frequncy spectrum is the optical market. This market is growing quickly and Compound Semiconductor Trends and Drivers in the Optical Market looks at some of the things I gathered from OFC. This is an established market, slighty smaller than the power electronics market, but with revenue still measured in the billions. As data traffic has increased, equipment data rates have risen from hundreds of Mbps to hundreds of Gbps in response. Even at the lower equipment data rates, the frequency requirements have favored compound semiconductor performance. Revenue in this market has been distributed among GaAs, SiGe and InP devices, but as data rates have increased, InP has become the dominant RF component technology.
The emerging technology sprout in this market is silicon photonics. Silicon photonics has not been quite as resistant as the Corpse Plant (my gardening friends will understand the reference!), which blooms every 30-40 years, but this technology has been “imminent” for quite some time. Based on what I saw at OFC, silicon photonics is finally blooming. This technology has gotten so much nurturing because it offers the integration and installed manufacturing infrastructure advantages of silicon. The main disadvantage of silicon in optics is that it’s an indirect bandgap material, so it can’t easily and effectively be used as a laser. In a semiconductor laser, electrons are excited into a higher energy band and when they decay back to the original energy band, they emit a photon. In silicon, excited electrons drop back to the intermediate band level and emit heat, rather than light in the process. The other major problem is that the performance of other functions (amplification, gain control, photodiodes) may be better in different technologies.
There are still many challenges to solve for silicon photonics to become more widely used, but the technology is generating revenue and growing quickly. At this point, silicon photonics is best suited for short reach applications, but this actually plays nicely into the market conditions. Many of the compound semiconductor companies (and I) tend to focus on network traffic, where Cisco’s Visual Networking Index (VNI) has become the bible for IP network data projections. Cisco also publishes a document that they call the Global Cloud Index (GCI) that forecasts trends and data traffic for data centers and the cloud. This traffic is a multiple of the VNI data; it’s not yet an order of magnitude higher, but it’s getting there. In Cisco’s estimation, close to 80% of this data traffic stays within the data center and this fits well with the short reach capabilities of silicon photonics-enabled transceivers.
As this sprout continues to grow, classification will become very interesting. In its simplest form, a silicon photonics device will be a silicon device with an external laser and perhaps a simple photodetector. To improve performance, the integration gets a bit more exotic. These devices usually employ III-V’s in an epitaxial stack that may be integrated onto InP or silicon, or it may be a heterogeneous integration of different technologies onto an SoI wafer. So, if a silicon photonics device contains III-V functions on a silicon wafer, with some of the functionality in silicon, is it silicon?...is it a compound semiconductor?... if the “III-V” is some quaternary compound like InGaAsP…is that InP?...is that GaAs? Ahhh…the complications!!
So, until it’s time for April to roll around again, I’ll be enjoying this one and diligently monitoring efforts in the power electronics and optical markets to nurture and grow the budding compound semiconductor technologies.