You might not have heard of them: photonic integrated circuits, or optical chips. But that is about to change. Over the last decade this technology has reached commercial maturity and is now enabling high-bandwidth and energy-efficient transceivers, far over one terabit per second, in our telecom networks and in our datacenters. Why should we care about this? What makes this technology different from so many other technologies that are out there?
The answer reveals itself when we take a look at the historical trend, as shown above. An optical chip is, much like the well-known electronic chip, a piece of semiconductor material on which a set of optical elements are combined together. In electronics we have the transistor, the capacitor and the resistor. In optics we talk about lasers, photodetectors, modulators, and gratings, to name a few. For example, high-capacity transmitters can be realized by combining multiple lasers with modulators on a single chip, including a multiplexer grating to combine all these signals into a single fiber-coupled output. Over the last 25 years, the number of such elements that make up a photonic circuit has doubled every two years, to a few thousand today. Sounds familiar? This is indeed similar to Moore’s Law for electronic chips.
It is hard to comprehend the potential impact of such an explosive growth in complexity and capacity, because this rate of development is hardly seen in any other technology. Notable exceptions include the aforementioned electronic chips and, of course, the internet. Anyone older than 30 will surely recognize the revolution these have caused. Is then the optical chip technology set on the same path of continuing exponential growth over the next decade? Will we have 100,000 elements per optical chip by 2025? That is the same number of elements an Intel 80286 processor had back in the 80s! My hypothesis is that this will happen. There is a clear application pull, as optical chips are key enablers for the internet, which will continue its exponential growth. Photonics will enable the communication between electronic processor chips and memory, or even on-chip communication, the capacity of which will follow the exponential increase of Moore’s Law. Furthermore, the very same fabrication infrastructure that enables us to make complex CMOS-based (1) processors can also be used to make optical chips. So there is really no reason to assume this exponential growth trend will stop in the near future.
It is very important to realize what this means. From a technology point of view we will see an ever growing effort, but at the same time we will see a convergence of the work, a focus on the most promising technologies. This convergence will take place on process level, materials level, component level, and, eventually, circuit level. These lessons we have learned from CMOS. This implies that the cost will decrease and that the technology becomes more widely available. Obviously, this means that photonic integration technology is bound to make an impact in a plethora of fields, besides telecom and datacom. A notable example is the company Genalyte, which uses optical chips to test blood for ebola in 10 minutes only. Start-ups are working on optical-chip-based miniature systems for, e.g., food or gas spectroscopy, and advanced medical imaging, such as optical coherence tomography.
But do we prepare our students well enough for this future? Electro-optics has traditionally been part of the field of electrical engineering. Light is an electro-magnetic wave, laser diodes and photodiodes are, obviously, diodes, and an optical chip is made using exactly the same materials and processes as are used for electronic chips. Photonics is electronics. Or at least, it was. Over the last decades, the field of electrical engineering has slowly moved away from the fundamentals of electronics in many universities, fuelled by the success of CMOS. Why bother about the fundamentals of transistors if the process design kits allow you to simulate your circuit with almost perfect accuracy?
This is not a plea for more photonics students. There are many great universities our there with great programs in photonics. And I am sure these will continue to drive the roadmap. No, this is a plea for making photonic integration technology an integral part of the education of every electrical engineer, and maybe even in other disciplines. We need to find a re-appreciation of the roots and fundamentals of the technology of electronics and, hence, photonics. Not to train students to be a specialist, not to train them to design the next generation of optical chips, but to make them aware of a technology that they will surely have to deal with in their future career.
Such is the nature of exponential growth.
(1) – Complementary metal-oxide-semiconductor (CMOS) is a technology for constructing integrated circuits.
Martijn Heck is an Associate Professor in the Department of Engineering of Aarhus University, where he is setting up a research line on photonic integrated circuits. His focus is on applications in optical interconnects, microwave photonics, sensors, and biomedical imaging and spectroscopy. He received the M.Sc. degree in applied physics and the Ph.D. degree in electrical engineering from the Eindhoven University of Technology, in 2002 and 2008, respectively. As a postdoctoral researcher and research scientist he worked at the COBRA Research Institute in Eindhoven, at the Laser Centre, Vrije Universiteit in Amsterdam, and at the University of California Santa Barbara. In these positions he explored all the major photonic integration technologies and their use in fields like satellite positioning, computer-communication and defense.