Quantum Film could replace CMOS?

[gwhitf]

I can't believe I'm posting this, but I hope David Grover reads it and forwards it to the guys in the back room at Hasselblad, and maybe they someday make an actual full frame 6x6 chip out of this stuff. Or better yet, a 6x8 chip for my old Fuji 680, or a FF chip for the RZ.

I don't understand any of this, but I just saw the phrase "could replace CMOS" and got excited.

http://tinyurl.com/ylyl9yl

Don't sell those old romantic cameras yet, there may still be hope. Roversi might get something for his old Deardorff after all.

[pcunite]

Bring it on I say. Their website:
http://invisageinc.com

[paulmoorestudio]

Bring it on I say. Their website:
http://invisageinc.com

I wonder if they are paying usage for the ansel adams pic on their site

[John.Williams]

If a technology were to replace the photo-electrical design of CCD and CMOS sensors, one could imagine a light sensitive gel or polymer that could 1.) absorb the visible wavelength and 2.) accurately record/transmit the surface position of the photons.

Temperature could be an issue (too hot, turns to liquid - too cold, becomes brittle and shatters) as well as other harmonic frequencies (the microwave oven next door just fried my sensor!)

Regardless, this type of a solution could be an answer to larger sensors and lower production costs due to less complex manufacturing processes.

An interesting find - thanks for sharing.

John

[David Grover / Capture One]

I can't believe I'm posting this, but I hope David Grover reads it and forwards it to the guys in the back room at Hasselblad, and maybe they someday make an actual full frame 6x6 chip out of this stuff. Or better yet, a 6x8 chip for my old Fuji 680, or a FF chip for the RZ.

I don't understand any of this, but I just saw the phrase "could replace CMOS" and got excited.

http://tinyurl.com/ylyl9yl

Don't sell those old romantic cameras yet, there may still be hope. Roversi might get something for his old Deardorff after all.

Looks very interesting.  As usual, can they build in volume with a high enough yield?

Also from reading I don't think you will be able to pop a roll / sheet of it in a conventional camera.  

[EricWHiss]

Just thinking about this.... the benefits seem to be only there for small high density chips with very small sensel sites - perfect for camera phones.  But since the wires don't necessarily scale with the sensel the increased efficiency of light absorption are not going to scale up to bigger sensors like other technology such as back lighting would.  

[John R Smith]

If a technology were to replace the photo-electrical design of CCD and CMOS sensors, one could imagine a light sensitive gel or polymer that could 1.) absorb the visible wavelength and 2.) accurately record/transmit the surface position of the photons.

John

I think, you know, we already had something like this - it might have been called Tri-X or something like that  

John

[Jeremy Payne]

I think, you know, we already had something like this - it might have been called Tri-X or something like that  

John

It had trouble with colors ... came out all grey and stuff ...

[John R Smith]

It had trouble with colors ... came out all grey and stuff ...

 

John

[gwhitf]

Just thinking about this.... the benefits seem to be only there for small high density chips with very small sensel sites - perfect for camera phones.  But since the wires don't necessarily scale with the sensel the increased efficiency of light absorption are not going to scale up to bigger sensors like other technology such as back lighting would.

The little that I understood made it appear that way -- this seems destined for point and shoots and camera phones. This doesn't seem like a new chip; it appears to be a coating for existing CMOS. So if there's gonna be a coating, there's got to be a 6x6 CMOS, and there's not. Yes, it does seem to be like painting Tri-X onto existing CMOS.

Also interesting was their graphic of the evolution of photography -- first Film, then to CCD, then to CMOS, then to QuantumFilm, as if they regarded CMOS more highly than CCD.

So after reading all that, go ahead and make that Fuji 680 lens as a doorstop again. Not much to really get excited about here.

[Ray]

It's interesting how the latest technology in photography often seems to first appear first in the lower grade cameras. The first mass-produced autofocus camera was the Konica C35 AF, a simple point and shoot model released in 1977. Now all cameras have an autofocus feature, even MFDBs, don't they?

The latest feature in P&S cameras is the 'back-illuminated' sensor which allows for an increase in the amount of light each sensel is able to collect because the interconnects, and other electronics in the case of a CMOS sensor, are on the other side of the sensor.

A future back-illuminated Nikon D3s might increase high-ISO performance by yet another stop.

However, the ultimate limit to such increases in performance is photonic shot noise, the random arrival of individual photons which is part of the fabric of quantum reality.

The smaller the sensor, and the smaller the pixel, the more significant shot noise becomes.

[hsmeets]

The little that I understood made it appear that way -- this seems destined for point and shoots and camera phones. This doesn't seem like a new chip; it appears to be a coating for existing CMOS.

From what I read/understand: the coating/film replaces the photodiodes, the silicon beneath the coating is pure interface/readout. So it is a new chip design.

[hsmeets]

It's interesting how the latest technology in photography often seems to first appear first in the lower grade cameras.

Often (but not all) these improvements come from R&D to lower costs of manufacturing. Back Illuminated sensor are in some way cheaper, with improved sensitivity the pixels can become smaller and density/resolotion can be increased again without increasing the actual dimension of the chip (or the chip can be made smaller with same spec's). Cheaper because costs of chip manuf scale with size of the die, basically you pay per square mm.

We'll still have to wait if this new tech really comes to market and is taken up by the market, much will depend on how it will be made available: the company only producing chips themselves or also granting licenses to others and what others will put up against this in competition.

[Bart_van_der_Wolf]

The smaller the sensor, and the smaller the pixel, the more significant shot noise becomes.

Not necessarily. The smaller the sensel, the sooner it's storage capacity overflows, so exposure needs to be reduced to avoid clipping and hence the absolute level of photon shot noise is reduced. Unfortunately, that does nothing for read noise, so all we get is a reduced Signal to Noise ratio, the relative photon shot noise increases. No free lunch.

Read noise is relatively stable regardless of sensel area, because it's more about the supporting electronic components. Photon shot noise scales with the square root of the number of detected photons (which roughly corresponds with Sqrt(surface area) ). In this context, big is beautiful.

Bart

[hsmeets]

Speculative as hell :-)

It could be that the film is applied as a coating on top of the chip surface, no need for having the coating in-registration with the readout circuity below as the Q-dots are so small that multiple Q-dots will be read with one interface pad, the number of these interfaces will set the resolution. So you still have the connection between size and sensitivity.

Challenges in applying the coating: even thickness across the chip surface and uniform distribution of qdots per square mm as variation my change the sensitivity for a small part of the sensor.

Some questions: if the film is a few q-dots thick, how will the excited state of the most outward facing q-dots be propagated through the q-dots beneath to the readout electronics. What is the spectral sensitivty? Will a Q-dot be binary or can it sense mutiple photons hitting it? How stable is the q-dot? can it fall-back in normal state before the readout happened?

Looks very interesting.  As usual, can they build in volume with a high enough yield?

Also from reading I don't think you will be able to pop a roll / sheet of it in a conventional camera.  

[BJL]

Read noise is relatively stable regardless of sensel area, because it's more about the supporting electronic components.

Actual measurements of read noise in electrons say otherwise: it does tend to increase with photosite size. Maybe one cause is that for larger well capacities, the charge handling part of the circuitry needs to have larger capacities and this "upsizing" tends to increase the noise, in electrons RMS. I have seen a trend or noise in e- RMS roughly proportional to photosite linear spacing, corresponding to charge variance proportional to charge handling capacity.

[bradleygibson]

Speculative as hell :-)

+1

Long on hype, short on specifics.

I saw no explanation regarding how they plan to record the frequency of light captured.  Bayer filter array?  So much for the 100% of light claim

And how, exactly will these dots be organized into pixels/sensels?

'Well' capacity?

Performance?  Suitability for video/EVF/live view applications?

I'm as excited about a new breakthrough as the next guy, but this smells more like they're looking for hype to help out with future rounds of funding, to me.

[Ray]

Not necessarily. The smaller the sensel, the sooner it's storage capacity overflows, so exposure needs to be reduced to avoid clipping and hence the absolute level of photon shot noise is reduced. Unfortunately, that does nothing for read noise, so all we get is a reduced Signal to Noise ratio, the relative photon shot noise increases. No free lunch.

Read noise is relatively stable regardless of sensel area, because it's more about the supporting electronic components. Photon shot noise scales with the square root of the number of detected photons (which roughly corresponds with Sqrt(surface area) ). In this context, big is beautiful.

Bart

The absolute level of noise has little meaning until it is compared with the strength of the signal. The fewer the number of photons collected by a sensel, the greater the photonic noise becomes as a proportion of the total signal, but of course it's not greater in an absolute sense.

If 100 photons are supposed to be collected by the sensel, the photonic shot noise is sq rt of 100 or 10% (10 photons misbehave or don't arrive in time). If only 16 photons are supposed to be collected (according to exposure settings), photonic noise rises to 25% although the absolute number of misbehaving photons is less.

As BJL points out, absolute read noise does tend to be less with reduced pixel size. However, photonic noise becomes more significant with reduced pixel size.

[EricWHiss]

+1

Long on hype, short on specifics.

I saw no explanation regarding how they plan to record the frequency of light captured.  Bayer filter array?  So much for the 100% of light claim

And how, exactly will these dots be organized into pixels/sensels?

'Well' capacity?

Performance?  Suitability for video/EVF/live view applications?

I'm as excited about a new breakthrough as the next guy, but this smells more like they're looking for hype to help out with future rounds of funding, to me.

Somewhere on their site you can see a diagram of what they are doing.  The quantum dot film just lays over the circuit and then filters  stack on it like normal.  Mostly what they are doing is moving the sensel wires to a lower layer but apparently the film is twice as efficient at collecting light than current cmos tech.
edit: see here http://invisageinc.com/staging/img/sb_QuantumTechnology.png

[AlexM]

We'll have to wait. Hubble is going to use this technology first

Kind of hard to guess what this technology brings without having a real sensor and seeing images from it.