Analysing polarised neutron reflectometry data (mark2)

refnx recently added the ability to perform polarised neutron reflectometry analysis in v0.1.51, both non-spin flip and spin flip. Here we analyse datasets that uses magnetic films and PNR as an extra contrast. The datasets of interest have the structure:

Si | SiO2 | Permalloy | Au | 2-mercaptoethanol | D2O

# some necessary imports
from importlib import resources

%matplotlib inline
import numpy as np
import matplotlib.pyplot as plt

import refnx
from refnx.analysis import Parameter, Objective, CurveFitter, GlobalObjective
from refnx.reflect import (
    SLD,
    Slab,
    Structure,
    MagneticSlab,
    PolarisedReflectModel,
    SpinChannel,
)
from refnx.dataset import Data1D

# create datasets from the NSF PNR data
with resources.path(refnx.reflect) as pth:
    dd = "c_PLP0007882.dat"
    uu = "c_PLP0007885.dat"

    file_path_uu = pth / f"tests/{uu}"
    file_path_dd = pth / f"tests/{dd}"

data_uu = Data1D(file_path_uu)
data_dd = Data1D(file_path_dd)

# create SLD (Scattering Length Density) objects for each of the materials
si = SLD(2.07, name="Si")
sio2 = SLD(3.47, name="SiO2")
au = SLD(4.66, name="Au")
mercapto = SLD(3.49, name="2-mercaptoethanol")
d2o = SLD(6.35, name="d2o")

py = SLD(9.0, name="permalloy")

# Now make Slabs that describe each layer. These can either be made from SLD objects,
# or by using the `Slab` constructor directly.

# sio2 slab has a thickness of 20 and roughness of 4 with the Si fronting medium
sio2_l = sio2(20, 4)

au_l = au(215, 4)
mercapto_l = mercapto(8, 4)
d2o_l = d2o(0, 4)

Now let’s make the Permalloy layer. To do this we need to utilise a MagneticSlab Component. The value of 1.75 represents a magnetic SLD correction of \(1.75\times 10^{-6}\\A^{-2}\). The value of thetaM=90 (degrees) represents the angle of the magnetic moment in the plane of the sample. For the applied magnetic field to be the plane of the sample Aguide=270 or 90. For the magnetic moment to be parallel or antiparallel to the applied field thetaM=90 or 270 degrees respectively.

# now make the Py layer
py_thickness = Parameter(50, name="Py thickness")
py_roughness = Parameter(5, name="Py roughness")


py_l = MagneticSlab(py_thickness, py, py_roughness, 1.75, 90.0, name="Py slab")

s = si | sio2_l | py_l | au_l | mercapto_l | d2o_l

# Note that we're using the same structure to describe both spin channels.
model_dd = PolarisedReflectModel(s, spin=SpinChannel.DOWN_DOWN, Aguide=270)
model_uu = PolarisedReflectModel(s, spin=SpinChannel.UP_UP, Aguide=270)

objective_dd = Objective(model_dd, data_dd)
objective_uu = Objective(model_uu, data_uu)

global_objective = GlobalObjective([objective_dd, objective_uu])

# select the parameters to be fitted and their bounds
model_uu.scale.setp(vary=True, bounds=(0.9, 1.1))
model_uu.bkg.setp(vary=True, bounds=(1e-7, 5e-6))
model_dd.scale.setp(vary=True, bounds=(0.9, 1.1))
model_dd.bkg.setp(vary=True, bounds=(1e-7, 5e-6))

sio2_l.thick.setp(vary=True, bounds=(10, 25))
sio2_l.rough.setp(vary=True, bounds=(1, 8))

py_thickness.setp(vary=True, bounds=(38, 55))
py_roughness.setp(vary=True, bounds=(1, 8))
py.real.setp(vary=True, bounds=(9, 9.5))
py_l.rhoM.setp(vary=True, bounds=(1.5, 3.0))

au_l.thick.setp(vary=True, bounds=(200, 240))
au_l.rough.setp(vary=True, bounds=(1, 8))
au.real.setp(vary=True, bounds=(4.5, 4.66))

mercapto_l.thick.setp(vary=True, bounds=(5, 15))
mercapto_l.rough.setp(vary=True, bounds=(1, 8))
mercapto.real.setp(vary=True, bounds=(3, 4))

d2o_l.rough.setp(vary=True, bounds=(1, 8))
d2o.real.setp(vary=True, bounds=(6.2, 6.36))

fitter = CurveFitter(global_objective)
fitter.fit("differential_evolution", seed=1);

-1271.4413157228305: : 62it [01:21,  1.31s/it]

fig, ax = plt.subplots()
plt.scatter(data_dd.x, data_dd.y, label="dd", s=4)
plt.plot(data_dd.x, objective_dd.generative())

plt.scatter(data_uu.x, data_uu.y, label="uu", s=4)
plt.plot(data_uu.x, objective_uu.generative())

plt.ylabel("R")
plt.xlabel("Q / $\\AA^{-1}$")
plt.yscale("log")
plt.xscale("log")
plt.legend();