With the accidental discovery of "black silicon," Harvard physicists may have very well changed the digital photography, solar power and night vision industries forever. What is black silicon, you say? Well, it's just as it sounds. Black silicon. It's what this revolutionary new material does that's important, starting with light sensitivity. Early indications show black silicon is 100 to 500 times more sensitive to light than a traditional silicon wafer.To create the special silicon, Harvard physicist Eric Mazur shined a super powerful laser onto a silicon wafer. The laser's output briefly matches all the energy produced by the sun falling onto the Earth's entire surface at a given moment in time. To spice the experiment up, he also had researchers apply sulfur hexafluoride, which the semiconductor industry uses to make etchings in silicon for circuitry. Seriously, he did this just for kicks and to secure more funding for an old project. "I got tired of metals and was worrying that my Army funding would dry up," he said. "I wrote the new direction into a research proposal without thinking much about it - I just wrote it in; I don't know why," he said. The new experiment made the silicon black to the naked eye. Under an electron microscope, however, the dark sheen was revealed to be thousands, if not millions, of tiny spikes. As we said above, those spikes had an amazing effect on the light sensitivity of the wafer. Mazur said the material also absorbs about twice as much visible light as traditional silicon, and can detect infrared light that is invisible to today's silicon detectors. And there's no change to the manufacturing process, Mazur said, so existing semiconductor facilities can create black silicon without much additional effort or, more importantly, money. [New York Times]
"100 to 500 times more sensitive to light than a traditional silicon wafer"
Bullshit! Silicon solar panels already convert 20% of the light that falls on them in to electricity. If they were 100 times more efficient they'd convert 2,000% percent of the light that falls of on then (i.e., 20x more light than is actually available)!
Typical small-ish pixel CMOS image sensors have peak quantum efficiencies (QEs) of 70% (for mono sensors) at around 600nm. That means that 70% of the green photos that arrive at the sensor get converted in to a detectable electron. Again, increasing that by 100x would lead to 70x more photons being detected than actually exist!
Of course that's the peak (QE) so it would be possible to enhance the QE at other wavelength, but mathematically it's just not possible to increase the sensitivity of a sensor within the visible range (the part of the spectrum you're generally interested in for conventional photography and video) by 100x, or more.
There is room for a big sensitivity improvement in the near infrared where only something like 1% of light is converted to a signal. But IR is only of interest for security applications.