Black silicon enhanced photodetectors: a path to IR CMOS
M. U. Pralle, J. E. Carey, SiOnyx Inc. (United States)
"SiOnyx has developed a next generation silicon based photodetector with spectral sensitivity from 350 to 1300 nm. By doping silicon with sulfur using femtosecond laser processing, we enhance the spectral sensitivity of silicon enabling high performance infrared detection on a CMOS compatible chip well beyond the bandgap cutoff of traditional silicon. These detectors exhibit enhanced QE, photoconductive gain, (with responsivities in excess of 100 A/W) and compelling low noise performance. Absolute noise characterization coupled with spectral responsivity characterization indicate measured detectivity (D*) of 1x10^14 Jones at 940nm, roughly a factor of 10 higher than the best silicon photodetectors. Operating at mere 3V these devices rival avalanche photodiodes at much lower power and bias. When applied to imaging platforms, this detector will enhance visible light imaging and will enable silicon to become the next generation nightvision detector, outperforming incumbent technologies in nearly all nighttime light conditions."
Update: A.T. and F.R. kindly provided me more information on this SPIE paper. SiOnyx reports 68% QE at 940nm wavelength and "greater than 10%" QE at 1107nm. The dark current was reported to be 140pA/sq.cm at unspecified cryogenic temperature. The diode structures of Black Silicon were made in commercial CMOS foundry:
exciting and making a lot of guys trembling !
ReplyDeleteis their detector photovolatic or photoconductor ??
ReplyDeleteThey talk about junction structure in the paper.
ReplyDelete"These detectors exhibit enhanced QE, photoconductive gain, (with responsivities in excess of 100 A/W) ... " have you seen a photoconductive gain in a junction photodiode ??
ReplyDeleteThe article talks about many things possible to do with Black Silicon. Among other things, it mentions photoconductors and junction devices. It also says that the 68% QE @ 940nm was achieved in "extrinsic detector mode".
ReplyDeleteI think it is pretty basic semiconductor physics that when you shrink the energy gap, you increase long wavelength QE and increase the dark current. InGaAs, "black" silicon, SiGe, Ge, nanoparticles, etc., all obey this relationship. It is easy to make a material with high QE and high dark current but this is not so useful. There aren't alot of high volume applications where we can afford the power or cost to cool the device. So, BlackOnyx's of the world, how about reporting dark current at RT? How about reporting the relationship between dark current and temp (e.g. activation energy)? How about dark current in a device of real image sensor pixel dimensions so that edge and corner are properly included? The thickness of the detector layer is also good to know, even if the absorption is a surface effect.
ReplyDeleteSiOnyx paper does answer on some of your questions, albeit in a not so clear manner. The dark current at supposedly room temperature is 120nA/sq.cm combined with responsivity of 32A/W. It's not clear if this was measured on the same detector as the cooled one.
ReplyDeleteThis comment has been removed by the author.
ReplyDeleteEven with a TE cooler, this is interesting for a lot of security, defence and scientific applications. For example, EMCCD made with this technology can extend in the longer wavelength with much more photons at starlight situation. 3-5 detector arrays such InGaAs have to be hrbirdized onto a silicon ROIC with a lot of extra cost and complexity. It's much harder to make small pixel and large array size.
ReplyDeleteUnfortunately the existing paper at the SPIE digital library is only 2 & 1/2 pages long. But some choice quotes:
ReplyDelete* "This high photoresponse was a direct result of photoconductive gain, an amplification mechanism commonly observed in II-VI semiconductors but previously not found in silicon based material systems"
* "What is fundamentally new in Black Silicon is the ability to operate in a traditional junction photodetector architecture."
The way it's written, I'd say that the dark current of 120nA/sq.cm and responsivity of 32A/W are measured on the same detector under the same conditions. But my understanding is that these values are "old", as they are in the intro. They report in this paper a measured dark current value of 140pA/sq.cm.
In neither case it is mentioned the temperatures at which they were measured.
With regards to fabrication, as they have already reported in the past, there is nothing "new" in the semiconductor process itself, so any CMOS foundry is OK. The "magic" is in the laser treatment in SF6 atmosphere. And the question is how this additional step can be translated to a high volume industrial process.
A dark current of 140 pA/cm2 at cryogenic temperatures and 140 nA/cm2 at room temperature ?!?! If these numbers are correct, it looks to me as a kind of solar cell in dark ....
ReplyDeleteMaybe someone with access to the paper can write to ask the authors for some of these details. It's not uncommon for a post about a company or a paper to result in one or more of the principals making a post here.
ReplyDelete