35

Some members of my lab are performing a polarization-sensitive experiment where they need to use a quarter-waveplate (QWP) with the fast axis in a specific direction. In the process of carefully checking all the polarization optics, they discovered that for QWPs from Special Optics and Thorlabs, different optical axes are labeled as the fast axis. They have done a number of tests (1) and came to the conclusion that the ones from Thorlabs are mislabeled, and what is labeled as the fast-axis is actually the slow axis. For most experiments this wouldn't matter or be easily noticeable, but in our case (and many others) it can cause the sign of the measurement to be incorrect.

After a lot of back-and-forth, Thorlabs now agrees that their QWPs are incorrectly labeled, but would like to confirm this with another research group or industrial lab. Does anyone have a QWP from either of these companies and the ability to determine if the fast axis is correctly labeled? If so, could you please answer here and let us know the model number of the QWP and how you determined the actual fast axis? Thanks!

1) Including looking at the magnetoluminescence from CdTe and the phase shift induced from total internal reflection from a glass prism.

Edit: Just to clarify, this is not any sort of interesting conceptual question about waveplates, but rather a practical matter that may be relevant to a large number of scientists performing polarization-sensitive optics experiments.

Edit2: We would also expect the thicknesses of the two components to be slightly different. We took some high-magnification images today and it does look like the marked edge component is slightly thicker, which also suggests that they are mislabeled.

Thor QWP out of mount from the edge High res image of the edge

ARM
  • 870
  • 2
    When did you get these optics? I was just doing polarization-sensitive realignment with a series of Thorlabs 830nm half and quarter wave plates that we got three years back and found no deviation with how they were labeled and how they responded. –  Jun 30 '14 at 02:57
  • 2
    Could you clarify exactly why you think the fast axis is correctly labeled and what measurements this is based on? I use two WPs in my MOKE setup but even if they were mislabeled, I would never know from symmetry considerations. My coworker has looked at a number of WPs produced by Thor in the last ~2 years, but Thor now thinks that all their WPs produced in the last 11 years are mislabeled. – ARM Jun 30 '14 at 14:20
  • 1
    It would be very interesting if you could describe the experimental setup that, in your mind, confirms the orientation of the fast axis. – Floris Nov 17 '14 at 20:38
  • 3
    My coworkers discussed a bunch of experiments to determine it, but the one he decided was the easiest for us to do and interpret is as follows: "We sent a linearly polarized beam through a right angle glass prism, arranging things so that the beam would be internally reflected off the back surface of the prism. We set the beam polarized at 45deg to the s and p axes of the prism back plane." – ARM Nov 19 '14 at 02:17
  • 2
    "The s and p polarized components of the beam get a different phase shift after TIR. These phase shifts can be calculated from the fresnel equations. This can get a little tricky due to the different possible coordinate systems (and definitions for jones vectors if they are used). Getting these wrong can flip the sign of the phase and/or add Pi." – ARM Nov 19 '14 at 02:17
  • 2
    Roughly speaking (up to factors discussed above), the s and p polarized components will have a phase shift of 45d between them. Adding a quarter waveplate to the beam path will then give a total shift of -45 if the waveplate fast axis is along the slow axis of the prism, and 135deg if it is along the prism fast axis. These two situations can be distinguished easily by analyzing the final polarization state with a rotatable polarizer (the maximum of the signal occurs for orthogonal values of the polarizer rotation). " – ARM Nov 19 '14 at 02:18
  • 2
    "A similar experiment can be done (and we did it) with a Fresnel rhomb which provides a Pi/2 rather than Pi/4 phase shift.

    An experiment demonstrating a very similar idea is shown here: http://video.mit.edu/watch/phase-shifts-in-total-internal-reflection-11864/"

     – ARM Nov 19 '14 at 02:18
  • 3
    I just encountered the same problem. There's a paper Opt. Eng. 41(12) 3316–3318 (December 2002) which describes a method. –  Nov 21 '14 at 16:59
  • 1
    I have found numerous errors in specifications of ThorLabs products in the past. They have always been very receptive and responsive when the errors are pointed out but I think this kind of thing is a problem when you have a very large number of stock lines as ThorLabs does, and it must be a monumental task keeping all the fine details, specifications, drawings up to date in their databases, particularly if they are outsourcing some of their manufacturing. I would politely tell them of your plate's batch number / delivery date, tell them of your suspicions and ask them to repeat your test. ... – Selene Routley Jan 24 '17 at 03:27
  • ...; if there is a problem, my experience is that they will sort it out graciously: they will be keen to know to avoid problems in the future. – Selene Routley Jan 24 '17 at 03:28

2 Answers2

1

We have obtained a Fresnel rhomb (FR) from Thorlabs to test their QWP in the manner suggested above (by ARM) and came to the conclusion that the wave plates are indeed labeled incorrectly. Our procedure is as follows: when the true slow axis of the QWP is collinear with the P-component of the TIR in the rhomb (labeled by Thorlabs as the rhomb's "fast" axis), the polarization of the light passing through the QWP+FR setup should remain unchanged.

user143051
  • 11
1

I also recently found that an entire order of Thorlabs quarter wave plates (WPQ05M-780) had a mislabeled fast axis. I initially suspected this was the case when I saw that they gave an opposite rotation to other wave plates I had, both older ones from Thorlabs and ones from another manufacturer (Meadowlark optics). When I looked closer, the marking on the optic itself (which has a little flat edge indicating the fast axis) was perpendicular to the line indicating the fast axis on the mount.

It's a very strange problem. I can only imagine that some miscommunication led them to think that the axis marked on the optic was actually the slow axis. But it is bizarre that such a trivial problem would last for four years and multiple people (including us) contacting them directly.

user57058
  • 11