In my general (layman) reading on physics, certain experiments are often described to illustrate the quantum nature of light… like the ever-popular slit/interference experiment. In this experiment, it is most often mentioned that even if they decrease the intensity of the light down to… say one photon a second… they will still get the typical interference pattern. Okay, I’m cool with that. But how can we be so sure that we are emitting exactly ONE photon a second? Is it just a matter of knowing what the energy of one photon is, and keying our instruments to that level? Or am I missing something extremely obvious? A single photon just seems so infinitesimal.
I have been meaning to ask this question myself for a long time. They tend to do this a lot when you read about physics experiments: “First we isolated one atom of gold and them bombarded it with…”. How do they work with things so small and how do they know what is happening when they do?
Here’s one way: http://www.aip.org/pnu/2002/split/591-3.html
In the more classic experiments you describe, I admit that I’m only going on memory here, but I think they filtered a single photon out of many produced. Not positive about that, though.
If you have a detector which is sensitive enough to distinguish N particles from N+1 particles (for small N), you just keep turning down the relevant knob on your source and watch the detector readings go down in a …3-2-1 (and zero) pattern. That is, you simply count the particles.
An example for photons: When a photon strikes the active face of a photomultiplier tube (PMT), the photon can (via the photoelectric effect) knock a single electron from the low-work-function surface coating. The PMT amplifies that single photoelectron signal in such a way that the final output pulse from a single photoelectron can be distinguished from the output pulse one gets from two near-simultaneous photoelectrons. So…
You flash your light source a bunch of times at a fixed intensity and look at the quantized output from the PMT. Since the PMT will sometimes get one photon, sometimes two, etc., the output should sometimes be Q, sometimes 2Q, etc. (where Q isn’t known before hand. It’s the pattern of intervals that allows the counting.) From this quantized (though resolution limited) spectrum of output sizes, you can determine on a flash-by-flash basis how many photoelectrons were induced in the PMT.
Further, one usually turns down the source until it’s actually highly improbably to get any photons at all. (“Zero” is often very easy to distinguish from “some”.) Then, when something does occur in the detector, you can be pretty sure that it was only one photon. Mathematically, this “Poisson process” looks like this:
If p[sub]0[/sub] = probability of getting zero particles, then
p[sub]1[/sub][given that something occured] = -log(p[sub]0[/sub])p[sub]0[/sub]/(1-p[sub]0[/sub]) .
As p[sub]0[/sub] approaches unity, so does “p[sub]1[/sub][given that something occured]”. For example, 0.99 for the former yields 0.995 for the latter. Thus, if you throw out all your easily identified zero-particle attempts, you’re left with a sample that has 99.5% single particle attempts – even without doing any explicit counting. If the 1% efficiency for getting something to occur at all is problematic (maybe each flash takes a while to set up), then one can sacrifice single-particle purity for efficiency. (For example, 0.75 for p[sub]0[/sub] yields 0.863 for the latter – still mostly single-particle, though now the multi-particle contamination might be a pain in the neck to properly account for (assuming you couldn’t cut these out using the detector output spectrum discussed above.))
It seems that actual two slit experiments using exceedingly weak sources have beendone with electrons, or atoms.
A single photon is actually exceedingly easy to achieve. In a typical well-lit room, at any given time, there is typically only a single photon of visible light, or none at all. There are many photons per second, but in a typical room, it takes much, much less than a second for one of them to hit a wall.
Thanks all for the replys. I’m starting to form a vague idea how it’s done, which is really all I was looking for considering the limited knowledge I personally have on the subject.
But this from Chronos has intrigued me further:
I had always imagined that at every given moment, in a typicaly lit room, there would be bazillions of photons cruising around the room toward their fate. Please enlighten me, cuz you just blew my mind (or whooshed it?).
Are you serious?
Hmmm… Here’s my back-of-the-envelope calc.:
For T=4000 K, I find that the integral of Planck’s blackbody formula, in frequency space, from 400 nm-equivalent to 700 nm-equivalent yields 20% of the total integral. My room has about 200 W of lightbulbs, so about 40 W of visible light. A representative energy for a visible photon is 4E-19 J, so I’m getting 1E20 visible photons per second. It takes a photon 1E-8 seconds to traverse a 3 meter room, so in one crossing time, I should have about 1E12 photons in the room.
So if we replaced the lightbulb with a 200mW LED we would have 1E9 photons. Given the surface area of a 3m sphere is 113 m^2, on average 1 photon would be hitting a 0.1mm^2 area at any one time. So if our detector was 0.1mm^2, we could distinguish single photons. In practise, we could just poke a 0.1mm^2 hole in the wall and put the detector on the other side.
Would anyone care to boil down what Pasta and Shalmanese are saying? I’d have a much firmer grasp if somebody could even distill the kinds of numbers we’re talking about without all the esoteric lingo. 
Doesn’t a detector stop the interference pattern? And then the experiment becomes just that much more interesting…
Well, a mommy proton and a daddy proton, who love each other very much, get certain urges, and then…
Trying very hard to avoid Gaudere’s law here… the OP asked about photons, not protons, and you’re the only one so far who’s mentioned protons.
Quite a big difference there. Photons are effectively massless, chargeless particles which carry the EM force. Protons are relatively heavy, charged particles that sit inside atomic nuclei.
Also, photon torpedos are from star trek, while proton torpedos are mentioned in star wars.
I think your joke, such as it is, could have been at least as effective with the correct term.
There, that should do it.
The point is, a lightbulb makes photons really fast, but they also move really fast (since they’re going at the speed of light). So in principal, if you have a dim enough bulb and a small enough room, you’d get one photon hitting the wall before the next one was emitted by the lightbulb.
However, I’m not sure if Chronos was assuming that every photon which hits the wall is absorbed. If you can see a wall, clearly some of the photons are coming back to your eye.
:smack:
Like most of my jokes, they seem to die somewhere between the inception and the implimentation.
… implementation …
runs
Which post in particular? My first post can be distilled to this:
-
To get one photon (or other particle), you either count them with some device that is sensitive enough to tell one particle apart from two or three, or you reduce your particle source strength (laser brightness, for example) until it becomes so improbable to get any particles that when something does finally show up in your detector or device, you can be pretty sure there was only one particle involved.
-
Chronos’s post was a somewhat tongue-in-cheek answer to the OP. He pointed out that if you could freeze-frame a well-lit room, you’d find less than one photon (on average) actually in the room at that moment. The idea is that while a light bulb may be putting out zillions of photons every second, it only takes a tiny fraction of a second for a given photon to cross the room and get absorbed, so there aren’t many travelling photons at any given instant. (One photon makes it across and dies before the next one even gets emitted.)
-
I decided to see if this claim made sense using a quick-and-dirty calculation. Given the power (wattage) of the lights in a typical room, and given the frequency spectrum of the light emitted from a typical filament (at, say, 4000 Kelvin), and given a few other sundry bits, I calculated that there would actually be about 10[sup]12[/sup] photons in the room at any given instant (give of take a factor of 10 or so), in disagreement with the “less than one photon” assertion.
-
Shalmanese then extended the calculation to a weaker light source with a detector located 3 meters away. He pointed out that if the detector had a surface area of 0.1 mm[sup]2[/sup] or so (possibly achieved by using an aperture), you’d get about one photon in the detector in any “instant”. He did not say explicitly that “instant” in his calculation must mean “in one room-crossing time”, or about 10 nanoseconds. (After all, the probability of any two photons arriving at the exact same time is zero, regardless of the brightness of the source.) But, two photons arriving 10 nanoseconds apart are easily distinguishable with common devices.
That’s the summary thus far.
Pasta, thank you so much. You made most of the more thorough posts much more clear. Take comfort in the fact that you (and the others) just busted a little ignorance.
So the old adage, opposites attract, isn’t very true after all.
I’m so sorry.