How Does Measurement Affect Photon Behavior in the Double Slit Experiment?

[Jip]

Hi,
My question is the following:
As well known, the interference pattern disappears when one determines which slit the photon has gone through. This also holds for atoms and molecules. Now I understand how the actual experiment works for eg. an atom: They send excited atoms that can decay in some cavity coupled to a photomultiplier; this way one can be sure whether the atom has been through slit 1 or slit 2.

But how does the experiment work with photons? What does it mean here to determine which slit the photon has gone through? (in particular, without destroying the photon!)

Related to this: lasers can be seen by eye if they go through some smoke. What if the double slit experiment is realized in some smokey environment? Does the diffusion on the particles "measure" the photon position? If so, we should see on the screen some interference for photons that have not been measured + two peaks corresponding to photons that have been measured.
Is that correct? If so, does anyone have references to provide?

Many thanks!

 

[mfb]

You can use a polarizer - not all photons will go through, but if they do you know which slit they went through if you measure the polarization.

What if the double slit experiment is realized in some smokey environment? Does the diffusion on the particles "measure" the photon position?

Yes, but only the position of particles that get scattered, they won't make it through the slits anyway (if they get scattered before). Smoke behind the double-slits acts like a three-dimensional detector volume.
Well, a few photons might get scattered very close to a slit, but probably not enough to make a visible difference.

 

[DrChinese]

Just to add to mfb's comment: Typically you would set the left polarizer at perhaps 0 degrees, and the right at 90 degrees. I.e. any angles where they are perpendicular. That eliminates the interference.

If you make them parallel (the same angles), there WILL be interference. In this manner, you can see that the interaction with the polarizer itself does NOT determine the interference. It is the relationship between them that controls.

 

[vanhees71]

Yep, and then you can set the left polarizer to 0 deg and the right one to any angle $\alpha$. If $\alpha \neq 90^{\circ}$, you only get uncertain which-way information but the interference pattern is preserved with lower contrast :-).