Black Silicon: The Next Generation of Digital Sensors
February 2010
For nearly a decade now, a quiet revolution in silicon technology has been brewing. It is not the revolution that you've been seeing in new digital cameras every eight to sixteen months. Sure, the advances in silicon photo sensor sensitivity and noise control have been remarkable. For example, the Canon 5D MkII can produce clean images at ISO 6400 (photos at left). Improvements to reach this level have been incremental. The revolution taking place, though, is not just another step. It is a giant leap. It will turn your newest DSLR into old school technology in a heartbeat.
The wonderful thing about silicon is that its surface electrons, when struck by photons of light, leave their nuclear orbits and become available to record information. Very cool. This has proved to be very fine and good, except that we have become greedy and we want more information from less light. If we could bump electrons out of orbit with fewer photons, or bump more electrons out of orbit per photon, we would see an increase in ISO senstivity.
Done. SiOnyx is zapping silicon wafers with laser energy. The wafers are placed in a chamber with sulfur, selenium or tellurium (aka "heavy chalcogens" - please don't send me email asking about that) and given a powerful blast with femtosecond pulses of laser power. Don't blink or you'll miss it: a femtosecond is one quadrillionth of a second, or one millionth of a billionth. Such short duration of power is a big deal. More on that later. The result of all this is a change in the surface properties of the silicon such that electrons are 100 to 500 times more easily displaced.
At just 100 times greater sensitivity, what now looks good at ISO 6400 will look good at ISO 640,000. Think of it: ISO six-hundred-forty-thousand! You could photograph your belly button from the inside (and some nut probably will). Even if we are really picky and prefer the image cleanliness that we now get at ISO 1600, and the increase in senstivity of mass-produced black silicon were to be only fifty times that of current chips, we would still be getting images at ISO 80,000 — clean, noise-free images.
This improvement doesn't have to be expensive. It's the same silicon currently in production, but it gets one more step at the end of the line: a blast from the lasers. Thus, the improvement should slide onto the current manufacturing work flow without retooling everything. Also, successful work is being done using longer pulses of laser, in the nanosecond range. The longer pulses are easier to achieve. The hardware drops in size from many cubic yards to a few cubic feet, further suggesting that adding the laser process to the end of the production line would not be physically nor financially daunting.
The process described is not conjecture. It is well-developed. I believe that black silicon sensors could be on the market in six to eight years. Perhaps by then I'll be ready to retire the 5D MkII for the 5D BS. Black silicon: pretty exciting stuff.
Images Copyright © Ed E. Powell
All Rights Reserved