Feynman's double-slit experiment

[pevans]

Here's my puzzle...

Electrons through a single slit one at time - standard distribution
Electrons through a double slit one at a time - interference pattern
Set up a detector to observe which slit the electrons passed through - back to a standard distribution.

This is all well and good. In searching for the origin of this story I found it comes from Feynman's Lectures On Physics. Nice one, Richard.

The problem I have is this. The description given by Feynman appears to be a mere thought experiment. However, the result is quoted in many places as an actual result. I've tried to find the relevant papers which might have shown this result but have so far failed.

My question is then, does anybody know where this result has been shown to hold? And if nobody has actually shown this result, why is the story constantly told (especially in awful popular media - I'm looking at you Wikipedia) that this is the great lesson of the double-slit experiment?

Any ideas?

Pete

 

[Born2bwire]

Right from a link on the Wikipedia article of Young's Double Slit experiment:

http://www.hitachi.com/rd/research/em/doubleslit.html

You can find a video there that explicitly shows the single electron events occurring.

 

[pevans]

I'm terribly sorry, I've been totally unclear!

The result I'm calling into question is the breakdown of the interference pattern once the electrons are being detected as going through one slit or the other.

Feynman's thought experiment suggests a light source hidden behind the metal plate with the slits in it and between the slits, the sort of place you'd expect to be able to bounce photons off each electron as it passes through. My question is, has anybody done this? (and I suspect not!)

 

[cesiumfrog]

The result I'm calling into question is the breakdown of the interference pattern once the electrons are being detected as going through one slit or the other.[...]My question is, has anybody done this? (and I suspect not!)

What do you expect? That the pattern remains regardless (vindicating realism afterall)?

It's been done, in different ways. The ten-dollar way is to replace the electrons with photons, and mark which slit each went through by rotating its polarisation or not; cross-polarised light doesn't interfere, but you can erase the marking and recover the pattern by putting a diagonal polariser in front of the screen. See "quantum eraser" on "google scholar".

 

[Demystifier]

Feynman's thought experiment suggests a light source hidden behind the metal plate with the slits in it and between the slits, the sort of place you'd expect to be able to bounce photons off each electron as it passes through. My question is, has anybody done this? (and I suspect not!)

It has been done many times. It's actually considered too trivial to make explicit reports in the literature.

 

[conway]

I don't know what variations on the experiment have been carried out since Feymann's Lectures were published in 1964, but I'm quite sure that at the time he was describing a thought experiment, as the OP suggests.

 

[DrChinese]

It has been done many times. It's actually considered too trivial to make explicit reports in the literature.

As Demystifier says, you really already know the exact answer.

You knew there was an interference pattern with 2 slits. Now cover up the left one. What do you get? It's just a one slit pattern (as mentioned, no interference). Now cover up the right slit (leaving left open).

Since neither

nor

show any interference, obviously [both open] is not equal to

+

. That's 'bout it.​

 

[Nick89]

But surely there are other ways of detecting which slit the electron went through then simply covering up one slit? When you cover a slit up it's no longer a double slit, so why would you even do that?

I've been wondering this myself... Is the disappearance of an interference pattern a truly quantum effect (because interference cannot occur) or is it merely because we somehow 'block' the electron (by covering up a slit, or by other means)? Or is that really the same thing?

 

[DrChinese]

But surely there are other ways of detecting which slit the electron went through then simply covering up one slit? When you cover a slit up it's no longer a double slit, so why would you even do that?

I've been wondering this myself... Is the disappearance of an interference pattern a truly quantum effect (because interference cannot occur) or is it merely because we somehow 'block' the electron (by covering up a slit, or by other means)? Or is that really the same thing?

Yes, there are other ways too. For light, you use polarizers set orthogonally (crossed). The interference pattern disappears. You can also put a detector behind one slit. But really, if you think the particle went through a specific slit, then closing off one slit entirely should be enough.

[Now, please note that there are interpretations (Bohmian) in which the particle goes through a specific slit while the hypothetical guide wave goes through both slits. These do not make any difference to the predicted outcome, however.]

 

[pevans]

Thanks very much to those that replied. That cross polarisation experiment with photons was the sort of thing I was looking for.

In response to some of the other omments, my question was much more subtle than I originally let on. I have no doubt that detection of which slit the particle went through will destroy the interference pattern. I'm concerned about how one should read Feynman's comments.

If Feynman is quoting a result from an experiment he knew of and then giving an ontological explanation, this is quite different from proposing what he thinks would happen if such an experiment were carried out. If the latter is the case, then any discussion of the interpretation of quantum mechanics would not need to account for such a result in the body of evidence which needs explaining - simply because there would be no evidence indicating that that might be the case.

I suspected the quantum eraser experiments might address the same issue but I was especially interested find out if anyone had used anything but photons to show the effect.

 

[mn4j]

Actually, it is misleading to claim that path knowledge definitely disturbs the interference. At the very least, the jury is still out but here are some experiments which prove otherwise:
* R. Sillitto, and C. Wykes, Phys. Lett. A 39, 333 (1972): two slits with only one open at a time. Interference fringes clearly observed
* E. Fonseca, P. S. Ribeiro, S. Padua, and C. Monken, Phys. Rev. A 60, 1530 (1999): complete path knowledge was obtained for all photons and interference persisted
* L. Basano, and P. Ottonello, Am. J. Phys 68, 245 (2000). Two laser sources, photons from each can only pass through a single slit. Path knowledge complete yet interference obtained.
* C. Santori, et al., Nature 419, 594 (2002): Complete path knowledge, interference obtained

Oh, and the above experiments together with a hand full of others rule out self-interaction which is the current dogma. For example:
* Yu. Dontsov, and A. Baz, Sov. Phys - JETP 25, 1 (1967). Interference disappeared when the photon flux was drastically reduced by placing neutral density filters before the slits but not by placing filters after the slits.
* Y.-H. Kim, et al., Phys. Rev. A 61(R), 051803 (2000). By making sure only one photon was in the system at a given time and controlling the interval between successive photons, the appearance/disappearance of interference was heavily dependent on the time interval.

 

[cesiumfrog]

Actually, it is misleading to claim that path knowledge definitely disturbs the interference. [...] here are some experiments which prove otherwise:
[...PRA 60 p1530 (99):] complete path knowledge was obtained for all photons and interference persisted

mn4j, I picked the reference for which your description sounded the strongest, but it appears your interpretation is false. In that experiment the interference is occurring because there is no way to know which slit of effective-aperture A1A2 the photons go through.

 

[Phrak]

One means that would not destroy phase information until after the passage of the electron is by measuring the recoil imparted to the slit. Good luck with that. As far as the cross polarizers goes, it requires that the observer not be entangled with information as to the orientation of the polarizers until after the electron has passed. I haven't been following along with this stuff. What you might be interested in might go by the name delayed choice.

Earlier than Feynman, Bohr invoked the double slit which-way thought experiment in his reply to EPR.

 

[mn4j]

mn4j, I picked the reference for which your description sounded the strongest, but it appears your interpretation is false. In that experiment the interference is occurring because there is no way to know which slit of effective-aperture A1A2 the photons go through.

Of course you fail to realize that there is no effective aperture. There are two apertures (A1 and A2) and each photon passes only through one but not the other. There are also two separate detectors, one for each aperture and it is obvious from the experiment which aperture the photon went through, yet by considering coincident counts between the apertures interference fringes can be obtained from the results, even though simply adding up the results of both does not give interference. What you call "effective aperture" is not an aspect of the experiment but a method of interpreting the results of the experiments. Unless you have a penchant for mysticism you will not suggest that the photons knew how the results would be interpreted.

This experiment clearly casts doubts to suggestions that knowing the path of the photons somehow disturb the interference. In case you are tempted to suggest that the interference is due to not knowing which side of A2 it the photon passed through, realize that D2 alone did not give an interference pattern without considering the coincidence counts at D1 (see Fig 5).

 

[cesiumfrog]

Of course you fail to realize that there is no effective aperture. There are two apertures (A1 and A2) and each photon passes only through one but not the other. There are also two separate detectors, one for each aperture and it is obvious from the experiment which aperture the photon went through[... mysticism ...] In case you are tempted to suggest that the interference is due to not knowing which side of A2 it the photon passed through, realize that D2 alone did not give an interference pattern without considering the coincidence counts at D1 (see Fig 5).

If there is no effective aperture, why do the experimenters go to such length in reporting its characteristics? (I don't have the article in front of me today, but you'll no doubt see that it is the slit separation of the effective aperture which determines the angles of the interference maxima.)

Are you aware that the pair of photons produced in the crystal are momentum-entangled? This means they both go through complementary positions of their respective apertures. In other words, by selecting only coincidences (where neither photon has been blocked by its respective aperture) it is guaranteed that both photons have gone through the doubly-restricted space of the effective-aperture (and you could confirm this using detection rates).

Of course there is no raw pattern at D2, this is an entanglement experiment after all: If a lens is placed in front of D1 then (Wheeler's CCD at) D1 could tell exactly which part of each aperture was traversed by both photons (including the one to D2, which would contradict quantum interfering of possible paths, a la Cramer's discredited FTL communicator. Another way of viewing this is that the spontaneously emitted photon pair does not have the coherence of a laser: neither beam alone could be used directly to produces raw double-slit interference because it would not reach both sides of the aperture in constant phase). Rather, the geometry has been chosen such that D1 measures the phase difference of the possible paths (e.g., coincidence with D1 selects the photons at D2 which, in equal phase, could have traversed paths through both slits of the effective-aperture, and hence produce a sinc pattern).

 

[Cthugha]

* C. Santori, et al., Nature 419, 594 (2002): Complete path knowledge, interference obtained

I do not know about the other papers, but I know this paper very well and it is about indistinguishability of consecutive photons from a single quantum dot. What they use to show this indistinguishability is Hong-Ou-Mandel interference, which is a consequence of the bosonic nature of photons: two photons incident on a 50-50 beamsplitter with perfect overlap will always leave the beamsplitter together at the same exit ports. This is an intrinsic two-photon interference effect, which should not be mixed up with the single photon interference pattern you get in a double slit experiment. In fact Zeilinger even showed that single photon interference and two photon interference are complementary. You cannot have both at the same time with good contrast. So in fact any demonstration of two-photon interference when the path is known shows directly that there is no single photon interference present.

After a brief look at the other references it seems to me, that they are about two photon interference, too, although I need to take some time to be sure. However, it seems your argument is void. A lot of people do not get the difference between single photon interference and two photon interference. This topic is covered very well in the bible of quantum optics ("Optical coherence and quantum optics" by Mandel and Wolf).

 

[mn4j]

If there is no effective aperture, why do the experimenters go to such length in reporting its characteristics? (I don't have the article in front of me today, but you'll no doubt see that it is the slit separation of the effective aperture which determines the angles of the interference maxima.)

You keep talking about effective aperture as if there is ever a single aperture. There are two, period. When superimposed, they become equivalent to a double slit. So what you call "effective aperture" is not anything the photons care about but a result of post-processing of the data. Every photon that went through A1 went ONLY through A1, and every photon that went through A2 went only through A2. Every photon detected at D1 went ONLY through A1 and every photon detected at D2 went only through A2. Note that, when D2 is kept fixed and D1 is moved, you get interference and vice versa. Consider the former situation, note that A1 is equivalent to a single slit and that D1 only measures photons which have gone through A1. It is therefore obvious that there is complete path knowledge for every photon arriving at D1.

See the following for similar experiments without the "indistinguishability loophole":
* Z. Y. Ou, X. Y. Zou, L. J. Wang, and L. Mandel, Phys. Rev. Lett. 65, 321-324 (1990).
* P. G. Kwiat, et al., Phys. Rev. A 41, 2910-2913 (1990)This clearly casts doubts to the idea that path knowledge has anything to do with appearance or disappearance of interference and also points to the fact that interference is a multi-photon phenomenon. You can argue that entanglement is involved, which doesn't obviate the fact that more than one photon is involved.

 

[mn4j]

This is an intrinsic two-photon interference effect, which should not be mixed up with the single photon interference pattern you get in a double slit experiment.

I challenge you to show me a single experiment where single photon interference was confirmed. In order to prove self-interference, the experiment must meet the following criteria:

1 It must ensure that photons do not overlap in transit
2 The coherence time of the source must exceed the interval between two photon emissions
3 with 1, and 2 valid, interference fringes must be obtained for arbitrary intervals between any two detections. In other words, the interval between durations should not matter.

So my challenge is for you to show me an experiment which met all three criteria above, failing which you can not claim self-interference has been experimentally verified. I've given you two examples which rule out single photon-interference or what is also called (self-interference), see the Dontsov and Baz paper I quoted above for a start. Self-interference is a myth.

A lot of people do not get the difference between single photon interference and two photon interference. This topic is covered very well in the bible of quantum optics ("Optical coherence and quantum optics" by Mandel and Wolf).

I will wait for you to show me convincing evidence of self-interference or what you call single photon interference.

 

[Phrak]

I challenge you to show me a single experiment where single photon interference was confirmed.

If there were no single photon interference, quantum mechanics would be wrong.

 

[Cthugha]

I challenge you to show me a single experiment where single photon interference was confirmed. In order to prove self-interference, the experiment must meet the following criteria:

1 It must ensure that photons do not overlap in transit
2 The coherence time of the source must exceed the interval between two photon emissions
3 with 1, and 2 valid, interference fringes must be obtained for arbitrary intervals between any two detections. In other words, the interval between durations should not matter.

I do not see, why these conditions should be necessary to prove self-interference. In fact 1) and 2) even contradict each other because coherence time is a measure of the time during which several photons are indistinguishable. While I agree that 1) is necessary, 2) must be converted to the opposite: (first order) coherence time must be shorter than the interval between two photon emissions. Otherwise you can never verify that you produced a Fock state. The further process is then very easy. You just need to do a HBT-experiment on your light source to demonstrate that antibunching is present by showing that the equal time intensity correlation function is zero. Then you direct this emission towards a double slit, where the slit separation is still small compared to the coherence length of the emission. Additionally you have to ensure that the recoil of the photon emission does not give any information about the photon path. Although this is rather trivial maybe someone thought of it as interesting enough to publish it. I will check when I have access to the library of my university again.

I've given you two examples which rule out single photon-interference or what is also called (self-interference), see the Dontsov and Baz paper I quoted above for a start. Self-interference is a myth.

This paper is not quoted very often and two of the quotes are articles called "How to judge flawed science" and "'Pseudo-Effects' in Experimental Physics: Some Notes for Case-Studies", which makes this article look very suspicious. However, I will have a look at it when I am back in the office. Now I do not have access to that paper.

edit: If I remember correctly "Experimental Evidence for a Photon Anticorrelation Effect on
a Beam Splitter: A New Light on Single-Photon Interferences." by P. Grangier, G. Roger and A. Aspect should do the trick: Europhys. Lett., 1 (4), pp. 173-179 (1986)

They use a cascaded decay instead of a quantum dot and a Mach-Zehnder interferometer instead of a double slit, but that should not change anything.

 

[mn4j]

I do not see, why these conditions should be necessary to prove self-interference. In fact 1) and 2) even contradict each other because coherence time is a measure of the time during which several photons are indistinguishable. While I agree that 1) is necessary, 2) must be converted to the opposite: (first order) coherence time must be shorter than the interval between two photon emissions. Otherwise you can never verify that you produced a Fock state.

The reason for 2) is to ensure that the source is still coherent from one photon emission to another, so that it is not claimed later that failure to observe interference was due to absence of coherence in the source, rather that absence of self-interference. (Do you disagree with this?) This in no way contradicts the requirement that the two photons not overlap in transit.1) Photons must not overlap in transit. (We agree on this condition)
3) Appearance/disapearance of fringes must not depend on time lag between photons going through the device. If the fringes are due to self-interference, the duration of the time lag between photons should not influence the results. (Do you disagree with this?)

Unless an experiment also fulfills (3), it has not demonstrated self-interference. However, the experiments I quoted above show that the duration of the time lag matters.

edit: If I remember correctly "Experimental Evidence for a Photon Anticorrelation Effect on
a Beam Splitter: A New Light on Single-Photon Interferences." by P. Grangier, G. Roger and A. Aspect should do the trick: Europhys. Lett., 1 (4), pp. 173-179 (1986)

This experiment meets criteria (1) but not (3) therefore it does not prove self-interference.

In case it is still not clear to you how (3) is necessary, realize that it is possible for a photon passing through a device to leave side-effects in the device that will influence the next photon arriving at the device, even if they never overlap in transit. However, such side-effects will decay with time and should be sensitive to the duration of the time lag. Such interference effects are essentially multi-photon interference. That is why (3) is necessary. Note also that the experiments I cited above showed that the appearance of fringes was dependent on the time lag between photons. This certainly rules out self-interference.

 

[cesiumfrog]

You keep talking about effective aperture [...]

I kept giving quantitative predictions, just show them to be false and you win. You, however, seem merely to cite experiments that you do not fully understand as though that were evidence for some vague hand-wavey departure from mainstream QM. Your argument is akin to looking at a http://en.wikipedia.org/wiki/Diffraction#Single-slit_diffraction" pattern and reciting complete knowledge of broadly which aperture it went through: a strawman denial of the precision welcher weg implies.

I challenge you to show me a single experiment where single photon interference was confirmed.

http://www.teachspin.com/instruments/two_slit/index.shtml" photon per hour passes from the collimator to the aperture, the cumulative pattern will be the same as for an unattenuated beam.

 

[mn4j]

I kept giving quantitative predictions, just show them to be false and you win. You, however, seem merely to cite experiments that you do not fully understand as though that were evidence for some vague hand-wavey departure from mainstream QM. Your argument is akin to looking at a http://en.wikipedia.org/wiki/Diffraction#Single-slit_diffraction" pattern and reciting complete knowledge of broadly which aperture it went through: a strawman denial of the precision welcher weg implies.

You are projecting here. I won't waste my time trying to explain to you the difference between a virtual double slit and an actual double slit.

http://www.teachspin.com/instruments/two_slit/index.shtml" photon per hour passes from the collimator to the aperture, the cumulative pattern will be the same as for an unattenuated beam.

Good! Then it should be trivial for you to show me where such an experiment was actually done and the results show that the cumulative pattern was the same. If as you claim a trivial experiment can prove self-interference, your failure to produce one realization of such an experiment will be very telling. I'm waiting.

In the mean time, I have provided multiple rexamples of realization of this so-called trivial experiment, in which interference vanished when the flux was reduced and the time lag was increased.

 

[cesiumfrog]

it should be trivial for you to show me where such an experiment was actually done [single photon at a time twin-slit] and the results show that the cumulative pattern was the same.

It's a common teaching lab experiment, see http://www.physics.brown.edu/physics/demopages/Demo/modern/demo/7a5520.htm" or Am.J.Phys 39 p420 (1971).

I have provided multiple [examples..] of realization of this so-called trivial experiment, in which interference vanished when the flux was reduced and the time lag was increased.

If you don't mind, could you specify exactly which citation(s) you think demonstrate a trivial "single photon at a time" twin-slit experiment in which the pattern alters at very low intensities? And if it isn't too much trouble, could you also explain why it hasn't revolutionised quantum physics?

 

[Cthugha]

The reason for 2) is to ensure that the source is still coherent from one photon emission to another, so that it is not claimed later that failure to observe interference was due to absence of coherence in the source, rather that absence of self-interference. (Do you disagree with this?) This in no way contradicts the requirement that the two photons not overlap in transit.

Yes, I disagree. If the source is coherent from one photon emission to another those two photons are automatically indistinguishable and can by definition interfere. In this case it is not possible to ensure that these photons do not overlap. To have a shorter coherence time automatically rules out the possibility of first order interference coming from two different photons. I see that there might be a loophole left in an experiment, which tries to disprove single photon interference. However for an experiment trying to show single photon interference 2) is anyway by no means necessary.

3) Appearance/disapearance of fringes must not depend on time lag between photons going through the device. If the fringes are due to self-interference, the duration of the time lag between photons should not influence the results. (Do you disagree with this?)

Do you mean the time lag between several detected photons (1) or the time lag between two paths, which lead to the detection of a photon (2)? The experiment I gave you fulfills (1) (the time lag between consecutive detections is stochastic and the time lag is modified drastically by introducing the heralded/gated detection scheme), while it is impossible to fulfill (2) due to the limited coherence time of the emission.

Note also that the experiments I cited above showed that the appearance of fringes was dependent on the time lag between photons. This certainly rules out self-interference.

Which one? The Kwiat paper is about two photon interference. The Kim paper shows that the important point in creating interference is the indistinguishability between different probability amplitudes (or Feynman path amplitudes if you prefer) and shows that an interference pattern in the angular distribution of some SPDC signal vanishes if you have just one out of initially two pump beams left. This vanishing is trivial as very is no interferometer, double slit or anything left, which geven gives the possibility to have two different paths. The "paths" are given solely by the two different pump beams. The Dontsov paper is widely agreed on to be plain wrong. See for example "New Approach to Experiments on Low Intensity Interference" by D. N. PINDER & R. N. GOULD (Nature 221, 460 - 461 (01 February 1969)), which actually gives a detailed discussion.

Good! Then it should be trivial for you to show me where such an experiment was actually done and the results show that the cumulative pattern was the same. If as you claim a trivial experiment can prove self-interference, your failure to produce one realization of such an experiment will be very telling. I'm waiting.

Reference 8 in the paper I mentioned before already does that. Another good realization can be found in:
Am. J. Phys. Vol. 64, No. 2, Feb 1996, Pages 184-188
A lecture demonstration of single photon interference
Wolfgang Rueckner and Paul Titcomb

This more or less tells how to show this effect in an undergrad lab.

 

[mn4j]

Yes, I disagree. If the source is coherent from one photon emission to another those two photons are automatically indistinguishable and can by definition interfere. In this case it is not possible to ensure that these photons do not overlap. To have a shorter coherence time automatically rules out the possibility of first order interference coming from two different photons. I see that there might be a loophole left in an experiment, which tries to disprove single photon interference. However for an experiment trying to show single photon interference 2) is anyway by no means necessary.



Do you mean the time lag between several detected photons (1) or the time lag between two paths, which lead to the detection of a photon (2)? The experiment I gave you fulfills (1) (the time lag between consecutive detections is stochastic and the time lag is modified drastically by introducing the heralded/gated detection scheme), while it is impossible to fulfill (2) due to the limited coherence time of the emission.



Which one? The Kwiat paper is about two photon interference. The Kim paper shows that the important point in creating interference is the indistinguishability between different probability amplitudes (or Feynman path amplitudes if you prefer) and shows that an interference pattern in the angular distribution of some SPDC signal vanishes if you have just one out of initially two pump beams left. This vanishing is trivial as very is no interferometer, double slit or anything left, which geven gives the possibility to have two different paths. The "paths" are given solely by the two different pump beams. The Dontsov paper is widely agreed on to be plain wrong. See for example "New Approach to Experiments on Low Intensity Interference" by D. N. PINDER & R. N. GOULD (Nature 221, 460 - 461 (01 February 1969)), which actually gives a detailed discussion.



Reference 8 in the paper I mentioned before already does that. Another good realization can be found in:
Am. J. Phys. Vol. 64, No. 2, Feb 1996, Pages 184-188
A lecture demonstration of single photon interference
Wolfgang Rueckner and Paul Titcomb

This more or less tells how to show this effect in an undergrad lab.

Just because the phrase "single-photon interference" is used in an experiment does not mean there is definitive prove of single-photon interference. And just because the authors of a paper set out to measure multi-photon interference does not mean single photon interference should not be visible when the right preconditions are met.

You are also confused about the preconditions, which are not difficult to appreciate at all. In very simple terms:

a) Individual photons must not overlap in transit.
b) Time lag between transit of individual photons must have no impact on fringe visibility

- It is not enough to divide the number of photons passing through per unit time. Measuring two photons per 2 hours does not mean only one photon passed through at a time. They both could pass through at the same time, and if only two photons are measured during the two hours, you will still get an average of 1 photon per hour. The experiment MUST ensure that ONLY one photon is in the device at ANY given time during the experiment. This requirement eliminates all those physic's lab experiments you and others have mentioned above.

- It is also not enough to show that individual photons are passing through one at a time. The experiment must also ensure that side-effects from one photon left behind on the slit/interferometer device are not the cause of the interference obtained. NONE of the experiments mentioned so far in support of self-interference pass this test.

Do you agree that it is possible for two photons which do not overlap in transit to interfere with each due to the fact that the second photon, experiences a device to which momentum has been transferred from the first photon in it's recent history? Please answer this.

If you agree to the above, then explain why it is not necessary to exclude such occurrences when trying to prove self-interference.

 

[Cthugha]

You are also confused about the preconditions, which are not difficult to appreciate at all. In very simple terms:

a) Individual photons must not overlap in transit.
b) Time lag between transit of individual photons must have no impact on fringe visibility

These are ok. I do not havy any problems with that, but a) is still not compatible with your issue of having a coherence time longer than the time between two emissions. If you can leave that point aside, I have no problem with that.

- It is not enough to divide the number of photons passing through per unit time. Measuring two photons per 2 hours does not mean only one photon passed through at a time. They both could pass through at the same time, and if only two photons are measured during the two hours, you will still get an average of 1 photon per hour. The experiment MUST ensure that ONLY one photon is in the device at ANY given time during the experiment.

Sure, I agree that the pseudothermal sources used in the seventies are not enough due to the Bose-Einstein distribution having a standard deviation which is always on the same size as the mean photon number. Thats why I introduces a paper, which used a Fock state.

This requirement eliminates all those physic's lab experiments you and others have mentioned above.

No, the paper by Grangier and Aspect, which I mentioned before is still not eliminated - no matter how hard you try.

- It is also not enough to show that individual photons are passing through one at a time. The experiment must also ensure that side-effects from one photon left behind on the slit/interferometer device are not the cause of the interference obtained. NONE of the experiments mentioned so far in support of self-interference pass this test.

I still see no problem with the Grangier/Aspect paper here.

Do you agree that it is possible for two photons which do not overlap in transit to interfere with each due to the fact that the second photon, experiences a device to which momentum has been transferred from the first photon in it's recent history? Please answer this.

No, I don't think it is possible unless you use a very strange device or very uncontrolled experimental conditions. While measuring a signal, which is too large might be possible, I see no way how such a disturbance could create destructive interference under the given conditions.

In the Grangier/Aspect paper you have constructive or destructive interference by means of having more or no counts at one out of two photo diodes placed after a beam splitter. This means that the critical element, which already responds to the photons impact must be the beam splitter. There have been no reports of interference of probability amplitudes, which do not overlap, so this should not pose a problem at all. To add on this, the exact timing between two emissions usually follows a statistical distribution.

In addition in the Grangier/Aspect paper the situation is even easier. Avalanche photo diodes are just dead for a while and then work again normally. Afterpulsing as the main error source is pretty easy to identify and does not lead to the interference fringes seen. As the interval between two heralded photon emissions is always larger than this dead time and any other interference effects would already show up if you look at first and second order coherence of a simple CW signal I can assure you that this is not a problem for a photo diode.

If you are still not content with that, write a rebuttal to that paper.

 

[mn4j]

These are ok. I do not havy any problems with that, but a) is still not compatible with your issue of having a coherence time longer than the time between two emissions. If you can leave that point aside, I have no problem with that.

I don't think you understand the correct meaning of coherence. Coherence is a property of the source, not the photon. I don't see how imposing a condition of coherence on the source contradicts the requirement that photons not overlap within the interferometer. But for the sake of argument let us leave that out, as it only comes into play when explaining the absence of interference. So for now let's focus on the two criteria (a) and (b).

No, the paper by Grangier and Aspect, which I mentioned before is still not eliminated - no matter how hard you try.

Where in that paper did they show that fringe visibility was not dependent on time delay between photons. Unless you can show me where, that paper is eliminated. I already mentioned that the paper passed the first criteria but not the second. So show me where the second criteria is met in the paper.

No, I don't think it is possible unless you use a very strange device or very uncontrolled experimental conditions.

I doubt that you understood my question. We are considering 2 photons which go through an interferometer without overlap.
- Photon A goes through the interferometer
- Momentum is transferred from Photon A to the interferometer and vice versa, leaving it in state A'
- Photon B goes through the interferometer which is now in state A'

Are you saying it is impossible for that the new state A' of the interferometer to play a role in the direction of propagation of the photon as it leaves the interferometer?

While measuring a signal, which is too large might be possible, I see no way how such a disturbance could create destructive interference under the given conditions.

Who said anything about destructive interference? You do not need constructive/destructive interference to have interference. You can obtain the same results by coordinated momentum transfer. This was shown since the early 1920s by Duane, Compton, Ehrenfest etc. I can dig up the papers if you are interested. The terms "constructive" and "destructive interference" are misleading simplifications. It doesn't happen like that ontologically.

 

[Phrak]

The coherency time should be less, not more, than the average time between emissions. Otherwise you get overlap between photons--which is somewhat contrary to the experiment...

Could you explain why you are pursuing this. It's somewhat mysterious, considering that it would stand quantum mechanics on it's head and have made headline news, with vanishingly small probability otherwise.

It shouldn't be enough to find a single journal article whos findings are contrary to every other article without reading at least one other article that cites it. Someone else has already, more or less, said as much. Of course, this would bepend on your motives, and I don't know what they are.

 

[Cthugha]

I don't think you understand the correct meaning of coherence. Coherence is a property of the source, not the photon. I don't see how imposing a condition of coherence on the source contradicts the requirement that photons not overlap within the interferometer.

Now you are kidding. I worked on first, second and third order correlation for roughly two years now. I recently got a paper accepted regarding correlation spectroscopy, which I will gladly link here if you are interested as soon as it is published. If you insist on coherence being a sole property of the source, which is not reflected in the em field, I do not think you know much about coherence. Coherence time is also a measure of how well one can define the exact moment of photon emission. This is the inherent reason, why photons from the same source are indistinguishable within coherence time. If the uncertainty of the emission time is large compared to the delay between the emission of two consecutive photons, this pretty much spoils the idea of non-overlapping photons. This is also the reason, why (almost) deterministic photon sources usually have a very short coherence time.

Where in that paper did they show that fringe visibility was not dependent on time delay between photons. Unless you can show me where, that paper is eliminated. I already mentioned that the paper passed the first criteria but not the second. So show me where the second criteria is met in the paper.

The time between successive photons is stochastic. What else do you need? Should they go down to one photon per day?

I doubt that you understood my question. We are considering 2 photons which go through an interferometer without overlap.
- Photon A goes through the interferometer
- Momentum is transferred from Photon A to the interferometer and vice versa, leaving it in state A'
- Photon B goes through the interferometer which is now in state A'

Are you saying it is impossible for that the new state A' of the interferometer to play a role in the direction of propagation of the photon as it leaves the interferometer?

In this case: yes. If you think different, show me a paper, which demonstrates your claim. Even if you work in a darkened room, the number of photons hitting the interferometer due to the em radiation present in the room is several orders larger than the number of signal photons. If you really think about photons manipulating the state of an interferometer, this is probably what you should worry about. However I do not know of any paper, which shows evidence for such behaviour as you mention.

Who said anything about destructive interference? You do not need constructive/destructive interference to have interference. You can obtain the same results by coordinated momentum transfer. This was shown since the early 1920s by Duane, Compton, Ehrenfest etc. I can dig up the papers if you are interested. The terms "constructive" and "destructive interference" are misleading simplifications. It doesn't happen like that ontologically.

The paper I gave you showed destructive interference (at the right delay) at one of the exit ports of a Mach-Zehnder interferometer. Ideally no photon ever leaves at that exit port for some special delay. I do not see any other good explanation besides single-photon interference for such behaviour.

 

[mn4j]

Now you are kidding. I worked on first, second and third order correlation for roughly two years now. I recently got a paper accepted regarding correlation spectroscopy, which I will gladly link here if you are interested as soon as it is published. If you insist on coherence being a sole property of the source, which is not reflected in the em field.

This still doesn't mean you understand it correctly. We are talking about single photons, where did "em field" suddenly come from. Since you claim to be an expert why don't you explain what it means for a photon to be coherent, or what the coherence of a single photon means, without using the phrase "em field".

Coherence time is also a measure of how well one can define the exact moment of photon emission. This is the inherent reason, why photons from the same source are indistinguishable within coherence time. If the uncertainty of the emission time is large compared to the delay between the emission of two consecutive photons, this pretty much spoils the idea of non-overlapping photons.

This is false. You seem to be sticking to an archaic understanding of what a photon is, which is invalidated by the very Grangier, Roger and Aspect experiment you quoted above. What you describe above may be true for classical waves but not for photons. In any case, this is a rabbit trail for another thread.

The time between successive photons is stochastic. What else do you need? Should they go down to one photon per day?

I need an experiment that satisfies condition (b). This one doesn't. "Stochastic" doesn't cut it. Show me an experiment in which there was no overlap and fringe visibility did not vary with time delay. In other words show me an experiment in which different iterations were carried out with different time delays (fixed for each iteration) and fringe visibility was the same for all iterations.

In this case: yes. If you think different, show me a paper, which demonstrates your claim.

I say it is possible. See the following simulations which demonstrate that it is possible.

Event-based simulation of single-photon beam splitters and Mach-Zehnder interferometers
H. De Raedt, K. De Raedt and K. Michielsen (quant-ph/0501141, Brazilian Journal of Physics, vol. 38, no. 1, March, 2008)

Even if you work in a darkened room, the number of photons hitting the interferometer due to the em radiation present in the room is several orders larger than the number of signal photons. If you really think about photons manipulating the state of an interferometer, this is probably what you should worry about. However I do not know of any paper, which shows evidence for such behaviour as you mention.

If background radiation were an issue, you will never be able to prove self-interference anyhow because there will always be background radiation which might have interfered with your photon. Not that I believe this to be an issue, but I'm just pointing out that you have a different standard of prove for the opposing point of view than for your view. Secondly, there is no paper which disproves this. In fact, I will go the extra distance and claim that this is the explanation of all interference phenomena. Unless you can provide an experiment which meets both criteria (a) and (b) we agreed on above, you can not disprove that it happens this way.

The paper I gave you showed destructive interference (at the right delay) at one of the exit ports of a Mach-Zehnder interferometer. Ideally no photon ever leaves at that exit port for some special delay. I do not see any other good explanation besides single-photon interference for such behaviour.

The requirement is not about the delay within the interferometer due to path length difference between the interferometer arms. The requirement is for time between successive photons reaching the interferometer to be varied without affecting fringe visibility.

 

[Hans de Vries]

* R. Sillitto, and C. Wykes, Phys. Lett. A 39, 333 (1972): two slits with only one open at a time. Interference fringes clearly observed

This would seem impossible but,

http://en.wikipedia.org/wiki/Double-slit_experiment]It[/PLAIN] was shown experimentally in 1972 that in a Young slit system where only one slit was open at any time, interference was nonetheless observed provided the path difference was such that the detected photon could have come from either slit.[15] The experimental conditions were such that the photon density in the system was much less than unity.

Means that both slits were always open simultaneously in
the light cone frames of the detected photons. That, is if
you would "take a picture" of the slits from the place of
the impact, at the time of the impact, then you would see
both slits open due to the difference in propagation time.


Regards, Hans

 

[Cthugha]

This still doesn't mean you understand it correctly. We are talking about single photons, where did "em field" suddenly come from. Since you claim to be an expert why don't you explain what it means for a photon to be coherent, or what the coherence of a single photon means, without using the phrase "em field".

You are still mixing first and second order coherence. First order coherence as given by g1 is always a characteristic of em fields. If you want to define something like coherence of a photon you need to look at second order correlation and can define the timescale on which g2 recovers from 0 to 1 as second order coherence time of a photon if you like to. I can tell you, what the term single photon means in terms of em-fields. A single photon (Fock state) is always second order in terms of fields as it consists of two field operators. If both operators of a detected photon can unambiguously assigned to just one field within the coherence time, you have a single photon. If the operators of several fields can give rise to a photon you have possible interferences present and no single photon state.

This is false. You seem to be sticking to an archaic understanding of what a photon is, which is invalidated by the very Grangier, Roger and Aspect experiment you quoted above. What you describe above may be true for classical waves but not for photons. In any case, this is a rabbit trail for another thread.

You just do not seem to understand anything about what coherence means. See any good book on it starting with Mandel and Wolf. You cannot define the moment of emission of one (out of several) photons better than the coherence time unless you are in the very boring physical situation that you have a single photon state without anything, which could induce interference. In this boring case you could of course backtrack from the detection.

I need an experiment that satisfies condition (b). This one doesn't. "Stochastic" doesn't cut it. Show me an experiment in which there was no overlap and fringe visibility did not vary with time delay. In other words show me an experiment in which different iterations were carried out with different time delays (fixed for each iteration) and fringe visibility was the same for all iterations.

Stochastic is not enough? Well, you claim that there is some esoteric interaction inside the interferometer. So you should show me there is one.

I say it is possible. See the following simulations which demonstrate that it is possible.

Event-based simulation of single-photon beam splitters and Mach-Zehnder interferometers
H. De Raedt, K. De Raedt and K. Michielsen (quant-ph/0501141, Brazilian Journal of Physics, vol. 38, no. 1, March, 2008)

De Raedt is as close to a crackpot as it gets. Even though his papers are mathematically consistent, I see no physical relevance of that work. I could model a beamsplitter as haunted by a demon, which chooses the port a photon will exit, too. This could reproduce experimental data as well, but it would not be of any significance. So this kind of simulation does not show anything as long as it does not give a prediction, which enables us to test, whether it is correct. Do you have an experiment showing us some difference or not?

If background radiation were an issue, you will never be able to prove self-interference anyhow because there will always be background radiation which might have interfered with your photon. Not that I believe this to be an issue, but I'm just pointing out that you have a different standard of prove for the opposing point of view than for your view.

Exactly that was my argument. A lot of background radiation does not change anything, but 1 single signal photon should?. That is absurd.

Secondly, there is no paper which disproves this. In fact, I will go the extra distance and claim that this is the explanation of all interference phenomena. Unless you can provide an experiment which meets both criteria (a) and (b) we agreed on above, you can not disprove that it happens this way.

Again, if that was an issue, random background radiation would spoil any interference. One still sees interference. This already rules out your absurd scenario.

The requirement is not about the delay within the interferometer due to path length difference between the interferometer arms. The requirement is for time between successive photons reaching the interferometer to be varied without affecting fringe visibility.

Well, this one is stochastic and they used their setup once with and once without gate, which reduces the time between consecutive photon detections by roughly 10. This cuts it for me. You will always say that the time delay chosen in any experiment does not match your conditions. If they wait a week you can tell them to wait a year. This is pointless.

I suppose this thread is done for me. Show me some experiment supporting your claim or I said all I need to say.

 

[mn4j]

This would seem impossible but,

Means that both slits were always open simultaneously in
the light cone frames of the detected photons. That, is if
you would "take a picture" of the slits from the place of
the impact, at the time of the impact, then you would see
both slits open due to the difference in propagation time.Regards, Hans

Sillitto and Wykes made sure ONLY one path was open at a given time. Therefore it is impossible for a single photon to have gone through both paths simultaneously. Are you suggesting, that the reason interference persisted is because a single photon accessed both paths at different times (non-simultaneously)?

 

[mn4j]

You are still mixing first and second order coherence. First order coherence as given by g1 is always a characteristic of em fields. If you want to define something like coherence of a photon you need to look at second order correlation and can define the timescale on which g2 recovers from 0 to 1 as second order coherence time of a photon if you like to. I can tell you, what the term single photon means in terms of em-fields. A single photon (Fock state) is always second order in terms of fields as it consists of two field operators. If both operators of a detected photon can unambiguously assigned to just one field within the coherence time, you have a single photon. If the operators of several fields can give rise to a photon you have possible interferences present and no single photon state.

You just do not seem to understand anything about what coherence means. See any good book on it starting with Mandel and Wolf. You cannot define the moment of emission of one (out of several) photons better than the coherence time unless you are in the very boring physical situation that you have a single photon state without anything, which could induce interference. In this boring case you could of course backtrack from the detection.

1) coherence has no meaning in the context of a single photon, just as you can not talk of the correlation of a single variable. A Fock state is NOT a single photon.
2) if self-interference is happening, phase-coherence between different photons should not affect fringe visibility but it does. (see the Sillitto and Wykes paper)
3) Add to that, the results of the Basano and Ottonello paper (L. Basano, and P. Ottonello, Am. J. Phys 68, 245 (2000)) in which two laser sources were used with photons from each only passing through a single slit and interference was obtained. Together they raise a serious question whether self-interference is ever happening at all.
4) Then look (carefully) at the Santori paper. The results rule out the occurence of any self-interference. I know that the authors set out to measure two-photon interference but the stark absence of self-interference is telling.

Again, if that was an issue, random background radiation would spoil any interference. One still sees interference. This already rules out your absurd scenario.

You don't expect random radiation to disturb the interference pattern by interacting with the photon detectors, so it is unreasonable to expect it to disturb the interaction between the photon and the slit. But I'm not surprised because you don't even believe the slits play any role other than introducing uncertainty, which is absurd.

Well, this one is stochastic and they used their setup once with and once without gate, which reduces the time between consecutive photon detections by roughly 10. This cuts it for me. You will always say that the time delay chosen in any experiment does not match your conditions. If they wait a week you can tell them to wait a year. This is pointless.

All I am saying is, don't claim self-interference has been proven. It hasn't. Extraordinary claims require extraordinary evidence.