The human eye can see 'invisible' infrared light (2014) | Hacker News Reader

The human eye can see 'invisible' infrared light (2014)

source.wustl.edu

It turns out that photomultiplier tubes can also see into the infrared, even when they're not supposed to. We built a PMT-based detector that was sealed up and so we thought it'd be nice to have a camera on it. We went with an IR camera and IR illumination so we could have the camera on while the detector was running. We chose the wavelengths carefully so that the entire peak of the IR LED emitters would be in the "no response" zone of our PMTs.

Turns out that, no, you could totally tell when those LEDs were on despite that being "impossible". We didn't do any research into this because it was completely feasible to just not run the camera and the detector at the same time, so we did that. (It also wasn't hard at all to choose an IR camera in the first place, so not a lot got wasted here.) PMTs are a little different than many other light detectors (this is how they get single-photon sensitivity!) but the hypothesis of photons doubling up seems completely reasonable. Perhaps we should have placed the IR emitter wavelength at at least twice the maximum sensitive wavelength of the PMTs? But then these detectors probably would have been sensitive enough to notice triple coincidences, so who knows....

The whole thing did in fact get written up here, though it doesn't talk much about the PMTs' IR sensitivity: https://arxiv.org/abs/1204.3500

The money plots for this discussion are Figure 7 ("this will totally work!") and Figure 9 ("this totally doesn't work!").

Everyone can see those cheap security camera IR LEDs glow red? Right?

Yes, but that's more likely to be the LED emitting frequencies outside the range it's supposed to.

Yes and everyone can see 1310nm communication lasers (such as SFP modules) emitting red light when looking directly into one. It's because most IR lasers and LEDs are not really monochromatic. The visible light is very useful for debugging and there's no practical advantage in buying lasers or LEDs emit only one wavelength

There's also humans who can see ultraviolet light—under normal conditions, not just specialized ones set up in laser labs!

https://sci-hub.se/https://doi.org/10.1016/0042-6989(92)9016... ("Scotopic spectral sensitivity of phakic and aphakic observers extending into the near ultraviolet")

Something I also learned not long ago: human eyes can see the polarization of visible light https://en.wikipedia.org/wiki/Haidinger%27s_brush

Okay, so it's about two photons hitting the same molecule in quick succession.

Why don't we see continuous infrared light, then? Surely light without pauses has way more ability to send two photons one after another than light with pauses.

> Surely light without pauses has way more ability to send two photons one after another than light with pauses.

No it's the opposite, SHG[1] requires huge spatial energy density which also means huge temporal density (power). A continuous wave of the required power would obliterate your retina (and your head).

[1] https://en.wikipedia.org/wiki/Second-harmonic_generation

It's not discussed in the linked article, but the paper it's based on argues against second harmonic generation and in favor of a two photon absorbtion process [1]. I'm not enough of an expert to comment on the rest of what you said, but I think I agree (based on skimming the paper) that the process still requires a high photon flux (i.e. high power desity).

[1]: https://www.pnas.org/doi/full/10.1073/pnas.1410162111 ; See second paragraph, and/or search the text for "SHG".

4ad's comment has the right general idea. To explain in more detail:

They compared pulsed light to continuous light with the same average power [1]. This means that the light intensity within the flux was much higher than the intensity of the continuous (since the pulses were short compared to the repetition rate). The proposed mechanism (and their result) depends on the square of the light intensity (because it requires simultaneous absorbtion of two photons), so the effect is scales quickly with increasing peak laser power, but (I'm guessing here) doesn't depend as strongly on average laser power.

Additionally, I expect 4ad was right that sufficiently-intense continuous light would damage our eyes.

My guess is basically that the timescales of two photon absorbtion and human perception are so different (i.e. two photon absorbtion is much faster), that the effects of the pulsed beam (which was repeating at 76 MHz in the experiment) can be treated basically continuous when considering human perception.

[1]: https://www.pnas.org/doi/full/10.1073/pnas.1410162111

Link to paper: https://www.pnas.org/doi/full/10.1073/pnas.1410162111

I presume an infrared laser is just as dangerous as a normal laser, despite the emitted light falls outside the visible spectrum.

IR lasers are in fact more dangerous, for the simple reason that you can't see the beam and so you can't tell if it's on or where the beam path is.

And the most common cutting laser is the ~10um CO2 laser, for what that's worth.

There is also no blink reflex unlike visible light, so you can easily stare at it for longer than you would a visible laser, meaning it would be a lot more dangerous even at the same power.

"Do not look into laser beam with remaining eye"

I can easily see the infrared FaceID illuminator flash when I unlock my phone in a dark room.

My immediate thought about this was for street signs.

Eyeball tracking software, shine a STOP when eyes are going towards a one way or no entry street.

I mean, the thought of hackers uppimg the wattage and blinding people is so real that massive physical safety measures should be put in place.

Unfortunately my next thought was that those cockwombles who make adverts will be salivating at this private show avenue.

I hope I'm wrong.

Reply on news.ycombinator.com