Discussion about polarisation in ORSO file format

[jfkcooper]

Prompted by the discussion with some of the proposers of a proposal for a modification of the canSAS file format to include polarisation:

https://wiki.cansas.org/index.php?title=NXcanSAS_v1.1

It was decided to try to write down the current (widely accepted) definitions from the reflectometry community about spin/polarisation directions, and then to move forward from this towards an MVP for ORSO

[jfkcooper]

Mainly tagging people to make them aware of this issues existence, though please do comment if you have ideas

[jochenstahn]

suggestion to start the discussion:

polarization:  
    for neutrons:
        Intended orientation of the neutron magnetic moment parallel ('p' for *plus*) 
        or antiparallel ('m' for *minus*) to the magnetic field at the sample.
        (attention: this is opposite to the naming convention used by the SANS community)
        A missing polarization or analysis is denoted by 'o'. 
        Always both, polarization and analysis, are given which leads to the possible combinations
       pp, pm, po, mm, mp, mo, op, om and oo. The last is synonymous with 'unpolarized'.
  for x-rays: no idea

if applicable, the actual polarization degree might be given as a float number in the range
[-1, +1] where -1 corresponds to 100 % 'm' alignment and +1 to 100 % 'p' alignment.

  degree:
     in_p:   0.96
     in_m:   -0.98
     out_p:  0.97
     out_m:  -0.98

[jfkcooper]

I think we need to be a little more explicit about the definition of the +/- states, since it is not possible to determine the neutron moment direction from this definition.
Something like:
We define the moment parallel to the field for the spin state which has the higher reflectivity (or critical edge), for a single, saturated, magnetic Fe layer, grown on a silicon substrate and measured in air from the air side

This may be too specific, and longwinded, but something similar would be helpful to remove ambiguity I think.

[jochenstahn]

I doubt that we are allowd to define what moment parallel to the field means. Maybe we combine the statements 'plus means m parallel to B and minus means m antiparallel to B' and the consequence that the critical edge of total external reflection of a saturated ferromagnetic film is higher in the 'plus' configuration than in the 'minus' configuration of the instrument.

[aglavic]

I think defining the polarization degree is ambigous as there are multiple ways people define it. We should reference a (good) paper and then give the values according to that definition. The data given has to be sufficient for a polarization correction (if correction not already applied.)
It would be very helpful if we provide exact measurement description of how to get the values in the way we define it. We should also think about the situation, where people did/could only measure the direct beam. It is still possible to make some corrections (A.Wildes paper) but then you only have a combined polarizer/analyzer efficiency parameter. (I think that was enough to correct measurements without spin-flip.)

Concernint x-rays, I propose to go with what we already have in orsopy, where single polarization refers to only incident beam (I don't think there is any beamline where they only use analysis with no defined incident polarization, as synchrotron is always polarized):
# x-ray polarizations
pi = "pi" # in scattering plane
sigma = "sigma" # perpendicular to scattering plane
left = "left" # circular left
right = "right" # circular right
pi_pi = "pi_pi"
sigma_sigma = "sigma_sigma"
pi_sigma = "pi_sigma"
sigma_pi = "sigma_pi"

[jfkcooper]

A question (based in ignorance) @aglavic, do the x-ray polarisations cover all cases e.g. for XMCD (I guess left and right?), XMLD (no idea), and other magnetic reflection techniques (not sure what these are either)?

[jfkcooper]

I doubt that we are allowd to define what moment parallel to the field means. Maybe we combine the statements 'plus means m parallel to B and minus means m antiparallel to B' and the consequence that the critical edge of total external reflection of a saturated ferromagnetic film is higher in the 'plus' configuration than in the 'minus' configuration of the instrument.

I like the idea of relating what we define to a physically measurable quantity on a beamline if we can, but unambiguous and concise wording is tricky

[jochenstahn]

1. In the orsopy specs, the field data_sorce.measurement.instrument_settings.polarization allows for only one entry. Up to now this is an abbreviation for the instrument configuration. For neutrons that would be 'pp, pm, ....' and for x-rays 'pi-pi, sigma-pi, ...'. I would keep it this way. Thus, if we want to add quantitative information (e.g. polarization degree or Poincare vector) we need a new field. Or several.
2. I assume that lab x-ray machines are 'unpolarised' as long as no reflecting, refracting or diffracting optics is in the beam. Is this correct?
3. I also like Jos' suggestion to relate whatever we define to some simple measurement. And Fe on Si is an obvious candidate. But I would not use this measurement as a definition.
4. I assume, the words 'plus' and 'minus' for the polarization states were chosen because of the potential $V = V_{nuc} + \mu B$ in the 'plus' state and $V = V_{nuc} - \mu B$ in the 'minus' state. Which means that the relative orientation of $\mu$ and $B$ is a good choice for the definition.