vacancies advertise contact news tip The Vault
facebook rss twitter

IDF Spring 2005: Silicon Photonics as true Interconnects

by David Ross on 1 March 2005, 00:00

Quick Link:

Add to My Vault: x

Follow the light....

What is the fastest traveler known to science? Light. And the fastest defined speed? The speed of light. So what would be the ultimate interconnect over a number of platforms? Light itself..

We have seen the use of light in the form of fibre optics; it is used in data-centers for high speed, low latency interconnections between ISPs and high end servers, as well as for long distance communication – for example across countries or oceans. In the past HEXUS has covered silicon photonics, and we were not surprised with the amount of feedback we got; since this could be huge if the price could be brought down.

Intels ‘vision’ is to be able to use the same technology which they have used to drive down the cost of silicon production to create a silicon based device. Today Intel showed us the first silicon based laser, however, whilst all this sounds simple, Intel had a number of challenges to overcome in order to get this working properly.

Intel have a department called the ‘Communication Technology Lab', this has several key segments of the market to work on, mostly next generation technologies such as digital radio, Wireless lans, WLAN, etc. This department has been responsible for the research and development of this photonics breakthrough.

Of course if Intel can create silicon outside the narrow electronics that they currently work within they could see significant revenue increases.

Why would we want fibre though? Well, Fibre has far more bandwidth available than that of copper, it is more distant independent, and copper has significant problems with any length increase. However one of the major drawbacks of Fibre is that of cost. If someone was able to create a complete silicon solution which was cheaper then surely it would be a ‘winner’.

Photonics is the use of technology with emissions, transmissions and detection of light and manipulations of such.

The flaw, namely that of high cost, is down to the fact that photonic devices are made with exotic materials and are expensive to process and pack.

Intel wishes to deliver photonics using silicon as the base material. This is with the aim of using standard, high volume silicon manufacturing techniques.

With most products the main cost is that of the packaging – this can account for up to 60 to 70% of the cost, however, can silicon micro-machining be used for smart packaging? It is a possibility since fibre can be brought right on to the silicon.

As with chip production Intel have the ability to use a 'pick and place' to produce and package silicon. However, in order to do this a number of photonic building blocks need to be made.

In order to do this you need the following:

- Light Source
- Guide for the light
- Modulation
- Photo-Detector (to transfer back in to electrons)
- Low cost assembly.
- Intelligence.

Intel have managed to create the light source – a silicon laser, this has been very successful, first showing at 1GHz in December, and in the last month this has ramped to 2.5GHz and now 4GHz modulation.

This is a product which can be built in a CMOS lab meaning that we can start to see the convergence of the 2 technologies, this is very close to a convergence of communication and computing on a single chip.

In order to use a light source as a communication method you need to create a Raman wavelength. Previously the solution was that of the Raman amplifier; which was not only costly but needed a large length of fibre to create and straighten the light source; Intel can now do this in a centimeter of silicon rather than kilometers of fibre. This piece of silicon exchanges the energy which vibrates the silicon atom to create the wavelength. Adding mirrors to this bounces the light back and forth to create the laser light.

What Intel have done is to create the major building blocks and this is another step towards higher bandwidth communication within the technology environment; don’t get us wrong – we wont see this done within a matter of months, we are looking years, but this is one of the most exciting technologies which we will see in coming years. The use of light finally gives us very high speed connectivity and hopefully the convergence with silicon will give us a lower price point, bringing the technology, and speed that it allows to the masses.