Photonic ICs and Beyond

Chris Spencer, Chris Bordne,Alex Dunmire, Min Xi

A brief history into photonics might suggest that the sub-discipline of nanophotonics still quiet an untapped market with much potential to be harnessed. A countable number of data points fill the the timeline from 1909 to the present. Inspiring the question that motivated this paper, why are there so few milestone moments in photonics over such a long period of time? Ultimately this question resulted in our review of nanophotonics integration with CMOS integrated circuits in response to Moore’s Law. Therefore, in this review we present our review of the modules in CMOS IC that are causing the slow down in Moore’s law, the nanophotonic analogs that may be viable replacements, and the breakthroughs in nanophotonics that have enabled these technological advancements

In 1909, Arnold Sommerfeld published his proposed analytical proof of surface polarization waves [11] marking in our history of Photonics the cornerstone of the all nanophotonics is motivated. Sixty years following Sommerfeld’s publication, Chinese physicist Charles Kao published a solution for guiding Sommerfeld’s surface excitations using optical fiber [12] which in 2009 he would also receive a Nobel Prize. Today nanophotonic research is being conducted by many countries for many applications, yet their approach is surprising similar. The majority of resources and funding for nanophotonics is the development of better materials. This point will be further evident in following sections, but for now it should be mentioned that of those resources only a marginal portion is allocated in the direction of CMOS integration. Initially, this discovery was quiet shocking for two big reasons. First of all, in recent years Moore’s law’s famous exponential curve of computing performance and affordability over time has become less exponentially improving and we know one major cause of the bottleneck occurring in integrated circuits is interconnects. Illustrated in figure 1 is a comparison of the performance capability of optical fibers vs coaxial cables. In figure 2 is a relation of current nanophotonic waveguide capability compared to optical fiber which has strong implications for what is possible on chips and the potential need for an enhancing technology. Secondly, the CMOS business has been so profitable and so heavily investing in machinery that it seems logical to continue investing as a lot of the infrastructure exists. The answer to my initial shock is illustrated in figures 3. CMOS compatible nanophotonics occupies an extremely narrow space on a wide spectrum of possible use cases and therefore to expect so much of the resources to be allocated so narrowly this early in such a young immature science could greatly delay the achievable possibilities. The following sections, however, will discuss the results of the resources that were allocated for CMOS integrated nanophotonics and the modules that are in development to address Moore’s law.

optical fiber performance vs coaxial [16] and rapid development of nanophotonics vs photonics [10]