Review reflection&refraction and scattering in presence of magnetic field with z component

It seems the z component of the magnetic field is ignored (except for single-layer = SAS samples).

Due to the conservation of the z-component (perpendicular to the multilayer sample) it is set to the same value in the whole stack. Then the field in the fronting medium is subtracted (see polarized implementation document, also valid for non-polarized). Not doing this would mean to have an inhomogeneous field in two or more other layers, or the fronting medium.

Also related to Zeeman effect.

Previously Redmine2417.

added 1 deleted label

changed the description

removed 1 deleted label

added physics label

added polarization r&t labels

assigned to @m.svechnikov

The z-component of any layer other than the upper half-space is ignored.

In SliceStack::setBField(...)

we explictly replace all z-components by fronting + external:

    const double M_z0 = this->at(0).material().magnetization().z();   <---- take fronting z-magnetization
    const double H_z = externalField.z() + M_z0;
    for (Slice& slice : *this)
        slice.initBField(externalField, H_z);      <---- set fronting + external for all slices

External field is, in fact, also ignored.

First, we add it to the magnetization of all layers, including fronting medium in SliceStack::setBField(...) and Slice::initBField(...).

But later, in Resample/Specular/ComputeFluxMagnetic.cpp in computeTR(...) and Compute::SpecularMagnetic::topLayerR(...) we subtract fronting field from all slices:

    const R3& B_0 = slices.front().bField();
    const R3 B = (forward ? +1 : -1) * (slices[i].bField() - B_0);

In "DevRef" (currently 4.3.6.1 Magnetic field in the fronting medium) it is noted that k_z0 in fronting medium remains unchanged with or without field and according to that we just subtract it from everywhere. Reference [57] (Majkrzak, O’Donovan, Berk) illustrates why k_z0 is not affected:

But parallel component k_|| of wave vector should be affected by magnetic field in fronting medium and. this would lead to additional spin precession. This part is just not implemented.

To avoid wrong results obtained by users, I would suggest to block setting external field and z-component of layers magnetization until things change.

mentioned in issue #654

mentioned in merge request !1705 (merged)

Implementation is requested in #654

closed

marked this issue as related to #654