[lyner]

techmind wrote something which set me thinking.
If photons really are like little squiggles of em wave, then their electric and magnetic field vectors must be of a specific amplitude so that the total energy is hf.
If you take an non polarised source of one frequency of light, then all the photons will be identical in every respect apart from the plane of their E and H fields.

If you then pass the stream of photons through a polarising filter, what does this polariser actually do?
The photons that emerge must all have the same magnitude of E field as before. If it just selected those photons which started off with the E field parallel with its selected plane then you would only get a very few coming through (an infinitesimal fraction of the incoming stream). So that can't be right.

We know that a polariser lets just half of the power through it and that the polarisation is in a particular plane. Has the polariser just picked half of the photons and twisted their polarisation to be all in one plane?

Follow this polariser with a further polariser with its plane at 60 degrees to the first and will produce half brightness. Again, this polariser must have let through a selected half of the photons ( which ones and why?)  and their plane must have been rotated also.

Treating this problem as a wave problem there is no problem at all. You just resolve E fields and discard one of the orthogonal components - resulting in half the power and a particular polarisation. If you wait until then to treat the  light as a shower of photons interacting with the detector, you can predict the result with no nasty questions needing to be asked.

Then, look at radio waves. You can produce any polarisation you like (plane, circular, elliptical) using a combination of waves produced by two totally independent transmitters (just at the same frequency and phase locked) feeding two crossed dipoles. Take the simplest example of the two transmitters being in phase and producing a slant linear polarisation. The 'photons' produced by each dipole will be only oriented on one plane - parallel to each dipole. How, then, can you suddenly get photons produced with an arbitrary polarisation angle out in the space in front of the two dipoles?

Remember, anything which applies to light must apply to radio waves and vice versa.

[lightarrow]

techmind wrote something which set me thinking.
If photons really are like little squiggles of em wave, then their electric and magnetic field vectors must be of a specific amplitude so that the total energy is hf.
If you take an non polarised source of one frequency of light, then all the photons will be identical in every respect apart from the plane of their E and H fields.

If you then pass the stream of photons through a polarising filter, what does this polariser actually do?
The photons that emerge must all have the same magnitude of E field as before. If it just selected those photons which started off with the E field parallel with its selected plane then you would only get a very few coming through (an infinitesimal fraction of the incoming stream). So that can't be right.

We know that a polariser lets just half of the power through it and that the polarisation is in a particular plane. Has the polariser just picked half of the photons and twisted their polarisation to be all in one plane?

Follow this polariser with a further polariser with its plane at 60 degrees to the first and will produce half brightness. Again, this polariser must have let through a selected half of the photons ( which ones and why?)  and their plane must have been rotated also.

Treating this problem as a wave problem there is no problem at all. You just resolve E fields and discard one of the orthogonal components - resulting in half the power and a particular polarisation. If you wait until then to treat the  light as a shower of photons interacting with the detector, you can predict the result with no nasty questions needing to be asked.

Then, look at radio waves. You can produce any polarisation you like (plane, circular, elliptical) using a combination of waves produced by two totally independent transmitters (just at the same frequency and phase locked) feeding two crossed dipoles. Take the simplest example of the two transmitters being in phase and producing a slant linear polarisation. The 'photons' produced by each dipole will be only oriented on one plane - parallel to each dipole. How, then, can you suddenly get photons produced with an arbitrary polarisation angle out in the space in front of the two dipoles?

Remember, anything which applies to light must apply to radio waves and vice versa.

Every attempt to identify photons with em waves has already studied for a lot of decades. If photons had polarization they would violate Bell's theorem and so would behave differently than what they actually behave. What has polarization is the em wave, not photons.

[JP]

I have to read and digest this a bit more, but looking over a book on quantum optics (Optical Coherence and Quantum Optics by Mandel and Wolf), it looks like you can quantize the field in terms of orthogonal linearly polarized photon-number (Fock) states, where you can write a general state as number of photons in one polarization plus number of photons in the other polarization.  The more usual way of dealing with this is to write it in terms of spin, which is like turning these linear polarizations into circular polarizations.  Photons are spin +/- 1, so this makes sense. 

[lyner]

So,  are we saying that the polarisation is not relevant to the photon? Are the photons not to be regarded as existing in the wave? How do the +/-1 spins fit in? It may help me to square things.

[techmind]

I wrote last night on the laser topic (from which this one was spawned)

The stimulated emission concept is quite interesting.

Now we know that inherently the energy "gap" in the excited-state atom is perfectly matched to the rest of the laser-radiation in the cavity.
We also know that the excited state has to be relatively "long lived" (metastable or whatever) to as to "hang around" long enough to be taken to the ground state predominantly by stimulated emission rather than random emission (sorry I can't think of the proper word).
The fact that it "hangs around" could be interpreted/considered as some kind of energy potential-barrier.
So if the excited atom momentarily "borrows" some energy from a passing "photon" then it can return to the ground-state.

Buuuut... I said before the energy gap is "perfectly matched", which might imply a concept a bit like a high-Q resonant system has some relevance? ... something that would couple to an EM wave of the precisely correct frequency...? And if this was the case you might expect the stimulated "photon" to be phase-matched and polarisation-aligned to the stimulating "photon"?


A bit handwavy I'm afraid - tis late at night, plus a decade or more since I studies undergrad photonics.  []

What do you think? sophiecentaur?

I'm quite attracted to the concept of a photon as a waveicle - some short burst of a classical wave. Presumably quantisation happens in some way, but let's leave that for a moment. Of course this could just be self-delusional wishful thinking.

Does such a model have any merit? More importantly, does it have any irreconcilable flaws?

If we go with the idea for a moment, and try to square it with the E=hc/lambda , what might this imply about the "geometry" of the waveicle of different energies/wavelengths?

If we kept the amplitude (or peak amplitude if its a nice burst that builds up smoothly then decays again) fixed then this might imply that the waveicle has fixed cross-section and fixed length (in the direction of travel) - the energy is proportional to the number of cycles that fit within the "box".
A serious downside of that geometry is that you get a photon with fixed absolute bandwidth (which feels rather unpalettable) - and this leads to serious problems when the wavelength approaches and exceeds the size of the box (radiowaves or whatever).


Another interesting question is what exactly happens when an atom emits a photon by moving an electron between energy states?
Can we have a model which is consistent with generating EM waves at the macro scale (ie radio/micro-waves) by oscillating charges in space?
Any mental models I can come up with would tend to generate a "waveicle" with a sharp initial attack then a slow decay - while something with time-symmetry (and avoiding that broadband frequency burst arising from rapid attack) would be far more satisfactory!!  []
A model which is symmetrical with regard to emission and absorption processes would be nice too.  []


I find the concept of wire-grid polarizers, and the fact that they work from radiowaves down to optical wavelengths very intriguing and reassuring. Can this guide our thoughts in any way on the nature of "photons"?


We can make radio transmitters at 100's GHz with todays technology - we can make EM waves from a small aerial.
Is there any fundamental reason why we couldn't make a light source from a 600THz oscillator feeding a 500nm-scale antenna (eg 250nm long half-wave dipole)? If not, where does it break down, and why?

[techmind]

Parts of my previous post lead up to the traditional 'problem' that you can't generate an EM wave from an 'antenna' which is significantly smaller than a wavelength of the radiation you're hoping to emit. It might be difficult to see how an atom of size ~100pm can emit a photon of wavelength 500nm.

However, we know from radio work that there's a workaround to this problem - you can make antennas smaller by increasing the local dielectric-constant/refractive-index (hence why mobile phone antennas are often made on a ceramic block, or immersed in perspex).
Am I prepared to believe that the dielectric constant in an atom where photons are emitted/absorbed is many times higher than that of free space? What, in the middle of a dense electron cloud? Errr... yes!  []

[techmind]

If you then pass the stream of photons through a polarising filter, what does this polariser actually do?

Can we satisfy all the known results by assuming the polarizer passes photons with a statistical probability proportional to the magnitude of the component of the photon polarisation parallel to the polariser polarisation? ie proportional to the cosine of the angle between the polariser and photon polarisation?

Surely there is no more accurate way to measure polarisation anyway?

[Vern]

It seems obvious to me that "photon" is the name that we give to the smallest possible energy-time we can get from a specific electromagnetic wave. Most of them happen when an electron moves form one energy state to another. It makes one transition. It signals that transition by radiating an electric field and a magnetic field that are in phase with each other. The most simple signal would be one cycle of energy. Observation indicates that the cycle of energy exists as a point surrounded by the fields. The fields diminish away from the point. The fields represent a probability amplitude that the point of action may be detected in a spacial area away from the path of the point.

I don't see anyone here agreeing with that concept, but it solves many problems for me.

[Vern]

I'm quite attracted to the concept of a photon as a waveicle - some short burst of a classical wave. Presumably quantisation happens in some way, but let's leave that for a moment. Of course this could just be self-delusional wishful thinking.

Quantization would naturally happen if the wavicle always goes to electromagnetic saturation. That is one reason it needs to be a determined number of cycles. If the number of wave cycles is undetermined, we have trouble explaining why it always exists as one certain value of energy-time.

[techmind]

It seems obvious to me that "photon" is the name that we give to the smallest possible energy-time we can get from a specific electromagnetic wave. Most of them happen when an electron moves form one energy state to another. It makes one transition. It signals that transition by radiating and electric field and a magnetic field that are in phase with each other. The most simple signal would be one cycle of energy. Observation indicates that the cycle of energy exists as a point surrounded by the fields. The fields diminish away from the point. The fields represent a probability amplitude that the point of action may occur in a spacial area away from the path of the point.

I don't see anyone here agreeing with that concept, but it solves many problems for me.

See the other topic: http://www.thenakedscientists.com/forum/index.php?topic=21378.25

The problem with the concept you describe is that a single, isolated, cycle of a waveform is not analysed as a single frequency. The fractional "bandwidth" is inversely proportional to the number of cycles, if your waveform decays gracefully.
This comes from Fourier analysis.

See also: http://www.techmind.org/dsp/

[Vern]

The problem with the concept you describe is that a single, isolated, cycle of a waveform is not analysed as a single frequency.

True; a single cycle can not be analysed as a frequency. However we can know that it may exist. Planck's constant represents some number of cycles of electromagnetic energy. That number needs to be 1 in order for things to make sense.

[LeeE]

Interesting question.

Dunno if this helps or not, but in the n-dimensional model I've been playing with, photons come out as a pair of two-dimensional objects, one following the other.  Having 'width' and 'height', but no 'length', they have orientation about their axis of movement.  Two, one following the other, are required to account for wavelength and the phenomenon of frequency shifting, whilst also accounting for circular polarisation by having different orientations.  This is all pretty irrelevant to the models you're already working with though, and the model is still incomplete.

[Vern]

Interesting question.

Dunno if this helps or not, but in the n-dimensional model I've been playing with, photons come out as a pair of two-dimensional objects, one following the other.  Having 'width' and 'height', but no 'length', they have orientation about their axis of movement.  Two, one following the other, are required to account for wavelength and the phenomenon of frequency shifting, whilst also accounting for circular polarisation by having different orientations.  This is all pretty irrelevant to the models you're already working with though, and the model is still incomplete.

By circular polarisation, are you thinking of the spacial orientation of the electric and magnetic planes? I've heard of experiments in which spin polarized photons convey their angular momentum to the absorbing object.

[JP]

So,  are we saying that the polarisation is not relevant to the photon? Are the photons not to be regarded as existing in the wave? How do the +/-1 spins fit in? It may help me to square things.

From the mathematics, it looks like you can describe photons with a given wave vector (direction), k, completely in terms of photon numbers in two orthogonally polarized states which are perpendicular to the wave vector.  The math to describe these polarized states seems to be the same as classical polarization: you can describe a linearly polarized photon state as a single linear photon, or as a pair of circularly polarized photons (which each have probability 1/2 of appearing if you try to measure them).  To pass photons through a polarizer, I would assume that you just decompose them into the right basis and that gives you a probability that each photon will pass or be absorbed.  In other words,

Can we satisfy all the known results by assuming the polarizer passes photons with a statistical probability proportional to the magnitude of the component of the photon polarisation parallel to the polariser polarisation? ie proportional to the cosine of the angle between the polariser and photon polarisation?

[lyner]

jp
that seems fair enough - the 'resolving' is done, not with the E field but with the statistics of numbers of photons.
Meanwhile, how does this help with ascertaining how big a region the photon actually has an influence? (i.e. its size)

[lyner]

Techmind
There is no reason why you couldn't get an extremely small radiator to transmit. You only have to match it. You are limited only by efficiency of any matching network you use and by the real resistive component of that circuit's impedance.
Many phone antennae now appear to be slot antennae (basically magnetic loops)  rather than  (electric ) monopoles. They don't need a sticky up bit on the top.

[lyner]

The problem with the concept you describe is that a single, isolated, cycle of a waveform is not analysed as a single frequency.

True; a single cycle can not be analysed as a frequency. However we can know that it may exist. Planck's constant represents some number of cycles of electromagnetic energy. That number needs to be 1 in order for things to make sense.

I don't see it making any sense at all, actually. If you are violating something as basic as the relationship between frequency and time functions just to get E = hf to 'make sense' then I think you should look at the problem again.
That value of f must be very well defined and your view of it (a single burst)  produces a completely splurged out sort of a frequency function. For there to be a definite 'f' then there must be a long enough time window for that 'f' to establish itself.

If you listen to a single cycle of a musical note, all you will hear is a click - no note at all.
Why shouldn't the length of the photon include as many cycles as necessary to cause it to interact with a charge system with a fixed (hf) amount of energy?

[Vern]

Yes; absolutely, analysis of one cycle using techniques developed for multiple cycles does not work. However, that does not resolve to a conclusion that one cycle can not exist. I know about Fourier Transforms and have actually experimented with multiple frequencies to produce composite frequencies.

I still am not convinced that a photon must exist as multiple cycles of electromagnetic radiation. I do know that a single cycle can be analysed as a composite of many frequencies. I view that as the nature of the electromagnetic field. But what I think is not important to this discussion, we can move on. []

[lyner]

OK but what is the appeal of your single wiggle?
And what about the width, or is it spherical?

[Vern]

The appeal of the single wiggle is that we can then construct a particle of matter from it by looping it around so that the front catches the back and forms a resonant circle. It is held in place by positive feedback and resonance [] I was trying to write a computer simulation of a neutron when I discovered that one cycle was necessary to develop the charge on a particle. One cycle in a circle places the same polarity on the outside of the circle all the way around.

Neutron model:



Edit: The positive feedback develops from the electric charge which is caused by the asymmetry of the fields in the bend of the curve. The electric plane is shown; the magnetic plane would extend outward toward the viewer and away down into the screen.

The width and height of the fields would extend outward forever. I'm not sure how to make the fields collapse when a photon is absorbed.