About
Us
Indium Phosphide in the Age of Silicon Photonics
In 1996, while with colleagues at the University of Central Florida School of Optics (CREOL), we embarked on an exciting journey into the integration of indium phosphide devices for telecom applications. As the telecom market began to heat up, we founded Advanced Integrated Photonics, later renamed Qusion by our investors. With the support of Princeton Opto-Electronic Materials (POEM), we established our company in Princeton, New Jersey, successfully raising initial venture capital and an additional $12M to build an 8000 square foot indium phosphide fabrication facility South Brunswick—New Jersey's first commercial Indium Phosphide fab. By 2000, we were thrilled produce innovative integrated devices like 40Gb electro-absorption modulators, switches, and lasers, marking a significant milestone in our pioneering efforts.
​
Unfortunately, the dotcom markets were not doing well. Our investors decided that they wanted to exit because they never saw a need for 40gb, even though we were working with Cisco, JDSU and several other tech companies.
​
Today there is an absolute need for indium phosphide devices. When the CHIPS act was enacted, it showed the country the necessity to keep up with current and future technology. We were very excited to finally be able to bring our many years of experience in this field. We formed Combined Semiconductor, Inc. in order to meet today’s needs for integrated photonic devices.
Perhaps Indium Phosphide came too early. Small diameter wafers were good enough to yield lasers – lots of lasers. A 75mm (3”) wafer can yield several thousand lasers. And then came silicon photonics. Silicon photonics promised almost everything that Indium Phosphide could offer with the economics of CMOS – only it wasn’t really true. Today we are at a point in which silicon photonics has reached its limits in terms of modulation rates. Many researchers are trying to integrate III-V lasers onto silicon photonics chips. But that doesn’t solve the limits of silicon photonics modulation rates. There is an opportunity today for a state-of-the-art Indium Phosphide foundry on newly available 6” wafers with advanced photonics packaging features supported with an extensive Process Design Kit (PDK) to enable the next generation of integrated photonics to support 200Gbps per lane and higher.
The chart below compares both hybrid silicon photonics and monolithic silicon photonics to legacy and next gen indium phosphide. This chart makes it clear that Next Gen Indium Phosphide offers a unique combination of advantageous characteristics in almost every category.
Comparison of Commercially Available Photonics Integration Technology
​
​
​
​
​
​
​
​
​
There are several important points regarding silicon photonics and Indium Phosphide for photonic integrated circuits:
• There is only one fab in the world that can truly support Monolithic Silicon Photonics, GlobalFoundries
• Similar to BiCMOS, the CMOS on monolithic processes is many generations behind state of the art and is incapable of supporting DSPs and embedded processors due to integration limitations and excessive power consumption of higher geometry CMOS.
• Legacy InP fabs cannot support integration features and capabilities to enable 2.5D and 3D packaging.
Moving to Larger Wafers
Combined Semiconductor is building a state-of-the-art, advanced Indium Phosphide foundry supporting highly integrated electro-optical platform on 6” wafers. Larger wafers offer a 2.25x advantage over 4” wafers and a 4x advantage over 3” wafers in terms of area. This maps directly to economical advantage because the cost of processing a wafer is roughly the same regardless of the size of the wafer. New, previously unavailable equipment, such as AIXTRON G10 MOVCD brings capabilities that improve throughput as well as yield improvements. All this translates to better economics for everything from lasers to highly integrated transceivers.
Creating an Advanced Photonics Integration Platform
With the availability of 6” InP wafers, and as data rates increase, the suitability of InP as a integration platform becomes much more compelling. InP has always been a better technology for photonic integration optically. However, the convergence of baud rate requirements, higher integration demands, larger wafers and improved yields make it a natural solution to market needs. In order to make an InP fab service into a photonics integration platform, three things are necessary, a comprehensive photonics library including an analog component library and integrated packaging features.
​
