From: c.h.thompson

To: ajp@kzoo.edu

CC: Prof Holbrow and others

Subject: Beamsplitters: do they really deal with whole "photons"?

Date: Thursday, April 18, 2002 6:26 PM

 

The Editor

The American Journal of Physics

 

F.A.O. Prof Holbrow

 

Re: C H Holbrow, E Galvez and M E Parks, "Photon quantum mechanics and beam splitters", Am. J. Phys. 70 (3), 260-265 (2002)

 

Dear Sir

 

Might I make a suggestion re the education of the next generation of physicists?  It is entirely possible that within their lifetimes there will be a reversal of the current attitude towards the quantum world.  Could they therefore be set as an exercise the investigation of alternative explanations for each of your demonstrations?  This would be absolutely invaluable to them in the future, giving them insights that will stand them in good stead if and when the bubble of "quantum computing" bursts.  (Nano-scale computing is one thing; computing that depends on entanglement of separated particles quite another.  It is the latter that is not on secure foundations.)

 

The evidence for the existence of photons is by no means as clearcut as is generally believed.  The evidence for entanglement of separated quantum particles is even less so, and, quite apart from any practical implications, in view of the impact that this phenomenon has on our beliefs about the nature of reality it is only right that students should be informed of this fact.  I have made a careful study of the relevant experiments and found (as indeed have others) that ordinary semiclassical, "local realist" explanations have not been ruled out.  I have corresponded with many of the professionals concerned, and discovered that they are aware of the "loopholes".  They have faith that these will eventually be blocked and quantum theory's unbroken record of success continued, but students should surely be prepared for the possibility that they are wrong.

 

The kind of physics you present via your classroom experiments is an idealised version, using mathematical models invented before the computer age.  It is possible, indeed likely, that the limitations of this method will soon become obvious.  Models using unitary matrices etc. are satisfactory so long as it can be assumed that the detectors are "perfect", but in order to model real ones, which do NOT obey exact square laws, I think it will be found that computer simulation assuming something much more general is needed.  It is not enough to assume that a detector can be fully characterised by its "quantum efficiency".

 

Students could investigate the above claim!  They could conduct your basic experiment to show the existence of photons, then conduct further experiments to show that this conclusion is false.  All that is needed is to increase the sensitivity of the detectors[*].

 

As you say, modern avalanche photomultipliers "can register single photons with efficiencies as high as 80%".  This is certainly what is claimed, but ask the students what this really means.  How does anyone measure absolute efficiency?  If such a photomultiplier is used in an experiment in which the results are supposed to reproduce Malus' Law, I would predict failure.  Peaks and troughs would be in the right places but I would not expect to see a sine curve.  I could be wrong, but students could think about the alternative explanation that I have found necessary in order to give local realist explanations for the Bell test experiments.  In this picture, the production of an output is not a matter of the transfer of energy from one complete "photon" to one "electron".  The detector output occurs when the intensity of the input combined with local "noise"  (and, possibly, chance occurrence of favourable "phase relationship") combine to cause a sudden change in the photocathode, conventionally assumed to be the emission of an electron.  The relationship between input and output thus depends on the shape of the noise distribution.  It can only be at best approximately linear, so a sinusoidal variation of input will be distorted on output.

 

I am assuming that light is an electromagnetic wave that, even if it comes in small pulses, has an intensity that varies continuously as those pulses spread.  When light passes through a polariser, however weak it is, part of the energy goes one way, part the other.  It is possible that even with a "50-50" beamsplitter, the division of energy for any given pulse is not even.  It may be affected by local factors -- by the relative phase of the input to that of some pre-existing oscillations in the material of the splitter, for instance.  The appearance of "photons" going one way or the other is no more than an artifact caused by the use of detectors that are unable to detect the weaker part.  (I am assuming also that your statement that there are no coincidences is a slight exaggeration: there are very few, not none at all.)

 

My own approach to these matters is almost entirely intuitive.  For a mathematical version that is for the most part equivalent see Stochastic Electrodynamics (SED) -- a theory that reproduces much of QED but without the photon concept.

 

Further information on local realist explanations of the Bell tests is to be found on my web site, http://users.aber.ac.uk/cat/ , and in the quantum physics archive, e.g.:

"Subtraction of  ‘accidentals’ and the validity of Bell tests", http://arxiv.org/abs/quant-ph/9903066

 

For more on SED see http://homepages.tesco.net/~trevor.marshall and archive papers such as:

"The Myth of the Photon": http://arxiv.org/abs/quant-ph/9711046

 

Yours sincerely

Caroline H Thompson

 

Email: c.h.thompson@pgen.net

Web site: http://www.aber.ac.uk/~cat

 

 

Return to front page

 



[*] (Added 20:04:02)

Holbrow’s experiment consists of using a nonlinear crystal to produce pairs of “photons”, one of which is used to monitor the occurrence of a “down-conversion”, the other passed through a beamsplitter.  The “proof” that we are dealing with photons consists of observing that there are no coincidences for detectors placed at the two outputs.  However (a) are there truly none, or does he mean merely that there are very few, and (b) do we start to get coincidences if we improve the efficiency of the detectors?