Smallest possible camera

[zompist]

For sf purposes, what's the smallest possible camera in the next few centuries? I'm not so much interested in the technology as in the physics limitations.

Some quick Googling produced this baby, all of 0.5 mm², which can capture a 200x200 pixel image. But, that's enormous. You can see the thing.

Can you make a camera at the nano scale? Say, 100 nm or less? That's far less than the wavelength of light, so you couldn't see these things with visible light. But for the same reason, I think, you can't create an image using visible light. Maybe you could include a photon detector, but probably there's not enough photons coming in to create an image very quickly. (Leaving your camera there for a week would be useful for certain purposes, of course.) I don't know if there's enough room to actually capture an image, and for that matter storing images might be hard.

If that's too small, what size would allow you to capture an image in visible light?

(Also, I'm aware that taking pictures of things its size will be impossible. Picture it as a spy cam: we want to toss these things on a wall and let them capture an image of the room.)

[Travis B.]

One should remember how pinhole cameras work - the pinhole may be significantly smaller than the photosensor. As a result, it is possible to make a hidden camera that can take a practical image that is essentially invisible except to the very closest inspection.

[Moose-tache]

These designs all depend on taking the image by receiving a two-dimensional field. What about using time instead? If your receptical is just big enough to accept a point source of light, then you just need to move it really fast in a raster pattern and record whether it receives light at a given point in time.

[zompist]

One should remember how pinhole cameras work - the pinhole may be significantly smaller than the photosensor. As a result, it is possible to make a hidden camera that can take a practical image that is essentially invisible except to the very closest inspection.

I don't think that works at the nano scale. Otherwise you could take pictures of cells or molecules just by moving a blocker-with-pinhole around. At a small enough scale, a pinhole turns into a slit experiment, which spreads the beam.

These designs all depend on taking the image by receiving a two-dimensional field. What about using time instead? If your receptical is just big enough to accept a point source of light, then you just need to move it really fast in a raster pattern and record whether it receives light at a given point in time.

You need a lens for that. At any point on the wall, minus a lens, you get light from all points.

[Moose-tache]

These designs all depend on taking the image by receiving a two-dimensional field. What about using time instead? If your receptical is just big enough to accept a point source of light, then you just need to move it really fast in a raster pattern and record whether it receives light at a given point in time.

You need a lens for that. At any point on the wall, minus a lens, you get light from all points.

False. This is true if your recepticle receives light from any angle, which is not the case in our example. If you really need a "lens," create a short tube using your thumb and forefingers such that the hypotenuse of the width and depth of the resulting flesh tube is smaller than the wavelength of light you wish to record.

[Moose-tache]

Now that that's out of the way.

Another issue would be sensitivity. Your eye needs about 29155 photons of typical visible wavelength per second per square millimeter of retina to record information. Our camera will need to be much more sensitive than that if we want to be able to record quickly on a surface area smaller than a square nanometer. I think the best bet is going to be to build an X-ray camera, since the higher energy photons will reduce the time and surface area needed to record an image.

Another possibility is that we forgo detecting EM radiation with our camera and instead detect charged particles. We can bombard our target with a massive electron beam, then use an electromagnet to direct the backscatter toward the collector. A timer will allow us to detect when during its raster the collector registers a spike in electrons, as well as keeping the electron beam and electromagnet from interferring destructively. Then the camera itself can be as small as we want, and it will be able to detect how distant and/or reflective the target is, allowing us to create an image. In fact, if our target is good enough to stand in front of the elctron beam, we don't even need the electromagnet to detect the presence or absence of opaque matter.

[zompist]

False. This is true if your recepticle receives light from any angle, which is not the case in our example. If you really need a "lens," create a short tube using your thumb and forefingers such that the hypotenuse of the width and depth of the resulting flesh tube is smaller than the wavelength of light you wish to record.

Are you imagining that the nanodevice has a short tube in it? That's fine, but it's only 100 nm long. You will get quantum effects at that scale-- you cannot even count on the tube blocking light that doesn't shine straight in the tube. It's all about constructive interference at that scale.

Your eye needs about 29155 photons of typical visible wavelength per second per square millimeter of retina to record information. Our camera will need to be much more sensitive than that if we want to be able to record quickly on a surface area smaller than a square nanometer.

Yeah, that's the sort of limits I'm curious about. My intuition is that nanocameras are going to be pretty frustrating to use.

X-rays are kind of notorious for moving through things we'd like to see. I'm open to making the thing bigger (1 µm?) if that's needed for visible light.

Maybe electrons work, but I consider the whole setup to be the camera... no fair having an entire laboratory involved just to have the imaging element small.

[Moose-tache]

What I learned working at a lab was: if you want a physics answer, you have to ask an engineering question. Otherwise you're asking for pedantry. Technically you never defined where the camera stops, so I assumed it to be the imaging hardware only.

Another possibility, since you balk at my magnetic field apparatus:
The camera has a single photosensitive surface, in the micrometer range you described. A laser sweeps everything around it according to a raster pattern. The camera only detects overall luminosity, but based on how bright its surroundings are, and knowledge of where the laser is pointing at that time, it can create a monochromatic image based on the distance and reflectiveness of the objects around it. If you can spin the laser fast enough, you could photograph the surrounding area every second or two. This would be much slower than the apparatus that redirects an electron beam, but it's smaller and has the added benefit of blasting everyone around you with a laser, and also the collector could use pretty simple light-detecting technology.

[bradrn]

Another possibility, since you balk at my magnetic field apparatus:
The camera has a single photosensitive surface, in the micrometer range you described. A laser sweeps everything around it according to a raster pattern. The camera only detects overall luminosity, but based on how bright its surroundings are, and knowledge of where the laser is pointing at that time, it can create a monochromatic image based on the distance and reflectiveness of the objects around it. If you can spin the laser fast enough, you could photograph the surrounding area every second or two. This would be much slower than the apparatus that redirects an electron beam, but it's smaller and has the added benefit of blasting everyone around you with a laser, and also the collector could use pretty simple light-detecting technology.

This sort of thing is easily possible, and as a bonus you can even get depth information out of it if you add a timer. The problem, however, will be focussing the light — for a Gaussian beam, the smaller the light source, the more divergent the beam. But it looks like there’s been some fairly impressive work done in this field, so I won’t say it’s impossible just yet.

(And before you ask, I do consider the laser focus to be part of the imaging hardware, since without it the camera couldn’t capture anything useful.)

[zompist]

The camera only detects overall luminosity, but based on how bright its surroundings are, and knowledge of where the laser is pointing at that time, it can create a monochromatic image based on the distance and reflectiveness of the objects around it. If you can spin the laser fast enough, you could photograph the surrounding area every second or two.

I like this method... beautifully indirect, and I like the idea of trading size for computation time.

How small can a laser be made?

[bradrn]

How small can a laser be made?

Micrometre or even nanometre scale, according to some searching I did earlier. But as I said, that gives problems with focussing.