2
In this work, we show that 𝐻𝑧
eff can arise from asymmetric current spreading and thus can widely exist in ST-
FMR bars and Hall bars of magnetic heterostructures. While being microscopically distinct from σz, this current-
spreading-induced 𝐻𝑧
eff shows all the three characteristics that were widely assumed in the literature to signify
the presence of σz, i.e., it can enable external-field-free current switching of perpendicular magnetization and
contributes to HHVR and ST-FMR signals of in-plane magnetization in analogue to the fieldlike SOT of σz.
Neglect of asymmetric current spreading can lead to erroneous analyses of various spin polarizations and SOTs.
Sample Characterizations
For this work, a series of Pt 4/Py 3.3-9.4, Pt 4/FeCoB 2.8-9.6, Pt 4/Ni 2.4-9.2, Pt 4/Co 1.7, and Pt 4/ FeTb 7.5
bilayers are sputter-deposited on oxidized Si substrates (the numbers are layer thicknesses in nanometer, FeCoB
= Fe60Co20B20, Py = Ni81Fe19, FeTb = Fe65Tb35). Each sample is protected by a MgO 1.6 /Ta 1.6 bilayer that is fully
oxidized upon exposure to the atmosphere21. The samples are patterned into microstrips and Hall bars by
photolithography and ion milling, followed by deposition of Ti 5/Pt 150 as the contacts for ST-FMR, HHVR, and
switching measurements.
Effective Perpendicular Magnetic Field within Magnetic Strips
We first perform ST-FMR measurements on the Pt 4/Py 9.4 microstrips using the nominally symmetric 3-
terminal contact configuration in Fig. 1(a). The symmetric (S) and anti-symmetric (A) components of the ST-FMR
responses for the Pt 4/Py 9.4 are plotted in Figs. 1(b) and 1(c) as a function of φ (see Ref. 28 and Supplementary
materials for the method how to determine the S and A). Considering a magnetic strip interacting with a spin
current with arbitrary spin polarization σ = (σx, σy, σz), the S and A values should vary with φ following2,22,23:
S =SDL,y sin2φcosφ + SDL,x sin2φsinφ + SFL,z sin2φ + SSPsinφ, (1)
A = AFL,y sin2φ cosφ + AFL,x sin2φ sinφ + ADL, z sin2φ. (2)
The four terms of Eq. (1) are contributions of the dampinglike SOT of σy, the dampinglike SOT of σx, 𝐻𝑧
eff
(fieldlike SOT of σz and others), and spin pumping, respectively. Equation (2) includes the contributions from the
sum of the fieldlike SOT of σy and the transverse Oersted field, the fieldlike SOT of σx, and the dampinglike SOT
of σz. The fits of the S and A data to Eqs. (1) and (2) untangle different contributions and yield the values of SDL,y,
SDL,x, SFL,z, SSP, AFL,y, AFL,x, and ADL,z. As expected, there is a predominant contribution from σy (SDL,y, AFL,y) but no
indication of σx (SDL,x = 0, AFL,x = 0, see Figs. 1(b) and 1(c)).
However, it is striking that such heavy metal/ferromagnet (HM/FM) samples exhibit a non-negligible SFL,z term
and thus a non-zero 𝐻𝑧
eff, in the case of the 3-terminal geometry (Fig. 1(a)). This 𝐻𝑧
eff is unlikely to arise from
σz because we only measure a negligible dampinglike SOT of σz (i.e., ADL,z ≈ 0). This is reasonable because the
sputter-deposited polycrystalline bilayers do not have a long-range lateral crystal or magnetic symmetry breaking.