From the company's whitepaper:
"NoblePeak Vision has solved the problem of growing thick, pure defect free germanium on silicon with its TriWave® technology. Figure 2 shows schematically our growth technique which traps dislocations by exploiting their characteristic diagonal propagation. A single crystal seed emerges from the narrow aperture and serves as a crystalline template for a large defect-free island. The island is formed after silicon transistor formation. The conventional back-end of the CMOS process then proceeds as usual, with alternating layers of dielectric and metal used to interconnect transistors and to connect the germanium photodiode to the circuitry."
"Figure 3 shows a scanning electron micrograph (SEM) of one such island integrated into a 180 nm CMOS process with three levels of metal and a reflow micro lens used to increase fill factor. TriWave detectors of size from 2 um to 5.6 um and imaging arrays with pixels of size 7 um and 10 um have been fabricated in this technology."
"Figure 5 shows the measured spectral response compared to theoretical predictions for bulk germanium at 220K. The cut off is at longer wavelength than predicted, due to grown-in strain."
VGA Tri-Wave sensor:
- Package Cu W base, Covar lid, 26 pin, AR Sapphire window
- 21mm x 21mm x 13.5mm
- Hermetically sealed, Xenon backfilled
- 4-stage TEC provides cooling to -80 C in ambient temperatures up to +40C
Just to add this in, there are sample images within the white paper, and just to clarify their QE results are normalised to the maximum, hence I had a quick double take on seeing 100% QE...
ReplyDeleteVery nice.
I quickly calculated the dark current for 33 ms at +25 deg.C, and that gives over 20k electrons of dark current. That is not a little bit .... This is almost full well at room temperature.
ReplyDeleteOf course you can correct for the average dark current and you can correct for the dark fixed pattern noise, but unfortunately you can not correct for the dark shot noise that comes together with this large dark current. So cooling will be an absolute must !
Agreed that the well gets quite full at room temperature. To be fair, I set the TEC to 25C for the "room temperature image" (see the TriWave Gallery page) to stabilize the temperature. For any kind of dynamic range, yes cooling is necessary...but 'an absolute must'.....well, have a look at the image.
ReplyDeleteAt best, I think a one stage TEC should be used but would recommend against no cooling at all. Significant packaging cost reduction going from a 4 stage to 1 stage TEC!
We have seen no issues with the microlenses at -80C
Could you please compare Ge based and InGaAs based photodetector in terms of dark current at same temperature ?
ReplyDeleteThanks !
Germanium has a very high Ni, so the diffusion current is high. That's why InGaAs is much better both for lower dark current and higher QE. The on-going chip stacking technology will reduce dramatically the hybridization cost.
ReplyDeleteAnonymous said...I believe the process indicated in Fig.2 is exactly one of the ways of developing standard SOI wafers (the other being oxide-oxide bonding)
ReplyDelete