It’s official – Silicon Photonics are ACE. Part 1.

Intel and the University of California Santa Barbara recently won the EE Times ACE award for the Most Promising New Technology category. Silicon Photonics technologies use silicon and laser light to transmit and receive data.

And there has been a lot of excitement around the series of recent announcement from Intel, published in publications such as Nature. People generally get excited about technologies that include the word ‘Laser’ and making anything smaller is generally cool, but why all the buzz around Silicon Photonics?

In a nutshell, Silicon Photonics has the potential to enable the computer and communications revolution to continue and indeed hit the gas pedal.

Copper is the traditional medium used to move bits and bytes of data around a computer, and this has been so since the beginning of computing. But copper has its disadvantages – think of a copper trace on a circuit board as a big highway – and the electrons are the cars. The cars have to manoeuvre around the copper atoms, slowing them down, causing friction and heat. The copper trace is rarely perfect – it has tiny deformities – further impeding the electron’s progress, the cars bump into each other, crash about and generally cause mayhem and traffic jams. Essentially this means that there is a limit to how much data you can pump through a bit of copper wire. There are also other issues with copper traces on circuit boards caused by imperfections in the copper traces themselves, and by symmetries in the fibreglass weave used inside the circuit board. These imperfections can be thought of as pot holes and speed bumps on the highway. These cause the signal quality quickly to degrade with distance, and any increase in the amount of data you have travelling up and down the trace only increases the traffic. This is why those high quality audio cables you buy in swanky high-end hi-fi shops are so expensive. Of course copper will stay with us for some time to come and Intel is also working on how to increase the bandwidth of copper connections.

Now light in optical fibre behaves very differently. Think of light as one of those new record breaking French TGV trains (or the Spanish Talgo), but built on magnetic rails. Low friction means that the trains can run very quickly and since photons don’t have to work their way though a lattice of atoms the signal travels very cleanly over much longer distances. In addition, you can send more than one signal down the fibre by using multiple colours (or wavelength) of light. Basically, just add rail tracks and double, triple or quadruple (or more) your total available bandwidth.

So if photonics is such a cool and wonderful solution – why are we not using it yet? Well photonic technologies are in use today – usually to transmit data over very large distances (across campuses, cities and even oceans). But the technology is very expensive. Manipulating light is much harder than manipulating electrons. It has traditionally required the use of very large devices (signal modulators, tranceivers, lasers etc…) using exotic and expensive materials like Gallium Arsenide and Lithium Niobate. Difficult to prounounce, even more difficult to manufacture. They don’t use silicon or the types of manufacturing techniques Intel uses for its chips and thus, these devices are orders of magnitude more expensive to produce than a silicon equivalent. By the way, silicon is basically sand and is the most abundant material in the earth’s crust.

These devices are large, I mean something the size of a matchbox, and by expensive I mean costing up to several thousand dollars. You need hundreds of such devices to make something that can either send or receive a signal.

So why will Silicon Photonics change this? Well very simply – silicon photonics will combine the technologies already in use to make processors out of silicon to make the types of devices small enough and cheap enough to be integrated directly into silicon chips. With all the components necessary built into something the size of your finger nail and costing in the region of tens of dollars.

In my next blog about Silicon Photonics, I’ll talk a bit about what individual components you would need to make a useful device – and what these devices could achieve in the real world.

Comments are closed.