2
ment in the stack has hitherto been lacking. Here
we provide these measurements through an investi-
gation of depth-dependent magnetism in the multi-
layer [Ta(2)/CoFeB(1.5)/MgO(2)]16/Ta(5) (thicknesses
in nm) using low-energy muon-spin rotation (LE-
µ+SR) [24–26]. Previous work [27–30] has shown that
bulk µ+SR [24, 31] is a useful tool in the study of
skyrmion systems and their dynamics, while LE-µ+SR
has probed depth-dependent magnetism at interfaces [32]
and the behavior of superconducting [33] and mag-
netic [34] superlattices. However, these studies do not
address the evolution of the properties with depth within
a repeating structure. Combining our measurements on
[Ta(2)/CoFeB(1.5)/MgO(2)]16 with energy-resolved CD-
REXS, and interpreting them with support from micro-
magnetic simulations, we show here how the domain-wall
angle and magnetic structure varies with depth, and re-
veal a crossover in magnetic properties as a function of
depth.
II. EXPERIMENTAL AND COMPUTATIONAL
DETAILS
The [Ta(2)/CoFeB(1.5)/MgO(2)]16/Ta(5) (thicknesses
in nm) multilayered thin films were grown by magnetron
sputtering onto Si wafers [5, 35]. A Ta layer was always
chosen as the final layer to prevent oxidation. The base
pressure of the deposition system was <3×10−6Pa. The
sputter gas, Ar, was used at a pressure of 0.3 Pa. The
deposition rates for Ta, CoFeB, and MgO were 0.48, 0.30,
and 0.05 ˚
A/s respectively. A sputtering power of 100 W
was used.
The CD-REXS experiments were carried out in the
RASOR diffractometer on beamline I10 at the Diamond
Light Source (UK). The CD-REXS patterns shown are
obtained from the difference between the scattering in-
tensities for left- and right-circularly polarized x-rays for
varying photon energies around the Fe L3absorption
edge. The dichroism extinction condition directly reveals
the twisting angle of the magnetic structure averaged
over the probed depth.
Micromagnetic simulations were performed using the
Ubermag package [36], with oommf [37] as the compu-
tational backend. The same material parameters were
used as in Ref. [5], with the layer thicknesses changed to
be appropriate for our sample. The parameters included
are exchange A= 10 pJm−1, interfacial (Cnv ) DMI
D= 0.4 mJm−2, and out-of-plane uniaxial anisotropy
Ku= 0.9 MJm−3, and the calculations also included de-
magnetisation in the Hamiltonian. In magnetic layers,
the magnetic saturation was Ms= 1.18 MAm−1. Open
boundary conditions were employed on samples of lateral
size 300 ×300 nm, and the system used a discretization
of 4 ×4×0.5 nm. The complete code used can be found
in the Supplemental Material [17].
Low-energy (LE)-µ+SR measurements of were carried
out at the Swiss Muon Source, Paul Scherrer Institut,
Switzerland, using the LEM instrument, which allows the
implantation depth of the muons to be altered by chang-
ing the energy of the incoming muons [25, 26]. Mea-
surements were performed at 290 K, either in the ZF or
TF geometry. Data analysis was carried out using both
the WiMDA program [38], making use of the MINUIT
algorithm [39] via the iminuit [40] Python interface for
global refinement of parameters, and musrfit [41]. Stop-
ping profiles of the muons in the multilayer stack were
modeled using the TRIM.SP software [42]. A model of
the multilayer stack was made in the software by insert-
ing layers matching the density of each of the materials at
the correct thickness to match the experimental system.
TRIM.SP then calculates histograms of the stopping pro-
files for a particular incident muon energy; we varied this
parameter to investigate the changes in stopping profiles.
For these calculations, at all incident energies, the energy
variance was set to 0.45 keV, and muons were incident
perpendicular to the surface with a variance of 15◦.
III. RESULTS & DISCUSSION
A. Circular dichroic resonant elastic x-ray
scattering
We studied the depth-dependent character of the
magnetic domain walls in the multilayer system
[Ta(2)/CoFeB(1.5)/MgO(2)]16/Ta(5) on a Si substrate
using CD-REXS [20–22, 43], which is sensitive to the
twisting angle of spin spirals and skyrmions, and was
previously applied to retrieve χof a magnetic twisted
domain state [21]. Figure 1(a) shows energy-dependent
dichroic scattering patterns, obtained on the multilayer
sample at room temperature and in zero applied mag-
netic field. The ring-like magnetic scattering patterns
were obtained by integrating several azimuthal angles
using the ϕ-axis of the diffractometer. The dichroic pat-
terns feature a dividing vector that separates the blue
and red regions of negative and positive dichroic con-
trast, governed by the dichroism extinction rule [20], see
Fig. S2 in Ref. [17]. The azimuthal direction of the di-
viding vector uniquely reveals the value of χ[21] (Fig. S2
in Ref. [17]). For multilayer systems with a non-uniform
χ(z) profile, the measured χmis the average χfrom all
trilayers with different weightings. Experimentally, by
varying the photon energy across the Fe L3edge, or the
x-rays incidence angle θ, the weight changes due to the
varied sampling depth of the soft x-rays, leading to a dif-
ferent measured average χm[22, 23, 44]. For the investi-
gated [Ta(2)/CoFeB(1.5)/MgO(2)]16/Ta(5) heterostruc-
ture, the attenuation length of the x-rays, Λ, is 79 nm
at the Fe L3edge (707.8 eV) at normal incidence, and
152 nm off-edge (700 eV). The sampling depth varies with
the angle between the x-ray beam and surface normal, α,
as Λ cos(α)/2 [45]. A systematic variation of these pa-
rameters therefore provides a strategy for investigating
the depth-dependent structural property of a 3D hybrid