L10 FePd is increasingly recognized as a potential candidate for magnetic tunnel junctions (MTJs), yet there remains room for enhancing device performance. In this work, we fabricated fully-integrated L10 FePd-based perpendicular MTJ devices and achieved a significant increase in tunnel magnetoresistance, reaching ∼65%, compared to the previous record of 25%. Notably, we observed bi-directional switching with a low switching current density of about 1.4 × 105 A/cm2, which outperforms the typical spin-transfer torque (STT) MTJ by about one order of magnitude. We propose two possible mechanisms to elucidate the switching process and associated device performance: (1) The voltage-controlled exchange coupling-driven switching of the bottom CoFeB layer; (2) The STT-driven switching of the exchange-coupled L10 FePd–CoFeB composite. While additional research is necessary, these findings may further advance the integration of L10 FePd into spintronic devices, potentially enabling low-energy memory and logic technologies.

The use of magnetic materials with perpendicular magnetic anisotropy (PMA) in magnetic tunnel junctions (MTJs) is pivotal for advancing spintronics-based memory and logic technologies.1–7 As device sizes shrink to the single-digit nm scale,8,9 the traditional CoFeB material system, featured by its moderate interfacial PMA (Ki,CFB ∼ 1.5 mJ/m2), struggles to offer adequate thermal stability and non-volatility.10,11 Consequently, numerous other PMA materials—ranging from Co (or CoFe)-based multilayers12–15 and rare earth-transition metal compounds1,16,17 to L10-phase alloys18–20—have been explored for use as the free layer in MTJs. Among the various candidates, L10 FePd stands out due to its robust bulk PMA (Kb,FePd ∼ 1.7 MJ/m3)21,22 and low Gilbert damping constant (α = 0.002–0.008).23–26 More importantly, L10 FePd has been identified as a key material for voltage-controlled exchange coupling (VCEC) – an innovative MTJ switching mechanism that boasts reduced energy consumption.27,28 Moreover, the recently discovered interfacial PMA at the L10 FePd-Graphene interface underscores the potential synergies between L10 FePd and 2D material systems.29 Furthermore, by using appropriate buffer and seed layers, high-quality L10 FePd thin films can be produced on industry-ready Si/SiO2 wafers.30 Despite the broad prospect of L10 FePd, integrating it into MTJ is non-trivial. For example, Pd atoms have a propensity to diffuse upon post-annealing, compromising the tunnel magnetoresistance (TMR) ratio of the device.31 In a prior study, we achieved a TMR of ∼25% using L10 FePd-based PMA-MTJs.32 Yet, significant enhancements are needed to further harness the potential of L10 FePd in spintronic applications.

In this study, we fabricated nano-sized L10 FePd PMA-MTJs and achieved a markedly improved TMR ratio of up to 65% at room temperature. To attain such a reasonably high TMR, we employed an ultrathin Ru/Mo bilayer spacer to suppress the diffusion of Pd. Additionally, we recorded an ultralow switching current density, JSW, of ∼1.4 × 105 A/cm2. This value is roughly one order of magnitude lower than what is typically observed with spin-transfer torque (STT) MTJs. Given the stack structure and the observed switching behaviors, we suggest two plausible mechanisms to elucidate the device's switching process. The first hypothesis revolves around VCEC, while the second postulates the magnetic reversal of a composite free layer. Recognizing the importance of both mechanisms, our findings, although requiring further investigation, can be viewed as a significant milestone in the advancement of high-performance L10 FePd PMA-MTJ devices.

The L10 FePd PMA-MTJ thin films were prepared on MgO(001) substrates by direct current and radio frequency magnetron sputtering. The stack is MgO subs./Cr (7.5)/Pt (8)/L10 FePd (5)/Ru (0.5)/Mo (0.5–0.9)/Co2Fe6B2 (1)/MgO (1.2)/Co2Fe6B2 (1.5)/Ta (3)/Ru (8), in which the number in parentheses are the layer thickness in nm. In prior to the deposition, we annealed the substrate at 600 °C for 60 min in vacuum and then 30 min in 0.8 mTorr of N2. The sample was then cooled down to 350 °C in the N2 environment. The N2 treatment is performed to heal the Oxygen vacancies generated on MgO surface during vacuum annealing.33 Then, the Cr, Pt, and FePd layers were all grown at 350 °C. After cooling the sample down to room temperature, we deposited a Ru (0.5) cap on top of the FePd layer. Given the constraints of our equipment with limited cathode sites, the vacuum was interrupted after depositing Ru (0.5) to transition the sample between sputtering systems.

Before loading the Ru (0.5)-capped thin films into the second sputter for the following deposition, a small piece of the sample was taken for X-ray diffraction (XRD) and ferromagnetic resonance (FMR) measurements to confirm the microstructure and magnetic properties of the FePd layer. The XRD θ-2θ results were collected using a Rigaku SmartLab (Rigaku Corporation, Japan)34 and plotted in Fig. 1(a). The (002) peaks from the Cr, Pt, and FePd layers as well as the (001) superlattice peak from FePd were observed on this sample. The full-width-at-half maximum was used as an estimate for the structural coherence length of each layer and the integral intensities of the FePd (002) and (001) peaks were used to estimate the degree of L10 order. The estimated structural coherence lengths for the (001) FePd peak and the (002) FePd, Pt, and Cr peaks were (in Å) 44.5(4), 44.3(12), 73.8(11), and 61.2(9), respectively. Numbers in parentheses reflect the one-sigma fitting error for the assembly of symmetric pseudo-Voigt fitting functions used to estimate the full-width-at-half maximum for each peak. The coherence length for the (001) and (002) FePd peaks were nearly identical, and more generally our estimate for the coherence length of each layer came to within 10% of the nominal thickness value, for which the systematically lower values may be related to interfacial roughness. The L10 order parameter for FePd was estimated to be 0.677(22) based on the integrated intensities for the FePd (001) and FePd (002) peaks, revealing a moderate-high degree of superlattice ordering.

FIG. 1.

(a) XRD θ-2θ spectrum of the L10 FePd sample. (b) FMR spectra with f ranging from 18 to 40 GHz. (c) and (d) Curve-fitting results of f vs Hres and ΔH vs f.

FIG. 1.

(a) XRD θ-2θ spectrum of the L10 FePd sample. (b) FMR spectra with f ranging from 18 to 40 GHz. (c) and (d) Curve-fitting results of f vs Hres and ΔH vs f.

Close modal

Broadband FMR spectra, as shown in Fig. 1(b), were recorded using a custom microwave stripline, a broadband generator and a diode detector, using a swept direct current magnetic field out-of-plane and a set of low-frequency (377 Hz) Helmholtz coils for lock-in detection of changes in the microwave absorption on resonance. Based on the frequency-resonance field (f-μ0Hres) spectra shown in Fig. 1(c), the L10 FePd sample showed an effective perpendicular anisotropy field of 0.557(1) T. Measurements of the linewidth (ΔH)-f indicate an α of 0.0151(8), larger than the typical literature values,23,24 which can be understood from the significant spin pumping effect between the underlying Pt layer and the 5 nm thick FePd film. An inhomogeneous linewidth broadening of 0.0361(14) T was estimated as well, consistent with the large PMA and relatively broad mosaic spread of 2.87(12) degrees for the FePd(001) crystallites estimated from the XRD rocking curve (not shown). Mitigating the spin pumping with an insertion layer and/or reducing the FePd(001) mosaic spread are both clear opportunities to explore further materials optimization of the film stacking structure and processing conditions.

All the following layers, starting from Mo (0.5–0.9), were deposited at room temperature in the second sputter. The full stack structure is depicted in Fig. 2(a) and the arrow highlights the ultrathin Ru (0.5)/Mo (0.5–0.9) bilayer. We will further discuss its key role in influencing the device performance and functionality. After the completion of the layer deposition, the sample underwent post-annealing at 300 °C for 10 min to induce the PMA of the CoFeB layers. The magnetization-external magnetic field (M-μ0Hext) loop, taken from a specimen cut from the sample's central region (with a Mo thickness of ∼0.7 nm), was assessed under perpendicular fields using vibrating-sample magnetometry (VSM) and is presented in Fig. 2(b). The stack's PMA is evident, with a distinct two-step magnetization switching pattern emerging. Evaluating the magnetization and switching field for both events, it is plausible to infer that the low field switching (occurring around μ0HSW ∼ 5 mT) is attributed to the magnetization reversal of the top CoFeB layer. Conversely, the high field switching (initiating at ∼20 mT) signifies a synchronous reversal of both the bottom CoFeB and the FePd layers.

FIG. 2.

(a) Schematic illustration of the stack structure of the L10 FePd PMA-MTJ sample. (b) M-Hext loop of the unpatterned thin films (red line and squares) and major R-Hext loop of a patterned device (blue line).

FIG. 2.

(a) Schematic illustration of the stack structure of the L10 FePd PMA-MTJ sample. (b) M-Hext loop of the unpatterned thin films (red line and squares) and major R-Hext loop of a patterned device (blue line).

Close modal

To study the magneto-transport properties of L10 FePd PMA-MTJs, we patterned the sample thin films into nanopillar devices by utilizing the combination of optical and e-beam lithography techniques and Ar+ ion milling. These devices have a pillar diameter of 300 nm. Subsequent contact pads were fashioned through a liftoff process. As illustrated in Fig. 2(b), the resistance (R)-μ0Hext loop reveals a TMR ratio of ∼65%, marking a notable enhancement compared to previous results.32 In this context, the ultrathin Ru/Mo bilayer is leveraged as a diffusion barrier for Pd atoms, contributing to the enhanced TMR. It is worth noting that the switching fields evident in the R-μ0Hext loop are considerably greater than those in the M-μ0Hext loop. This discrepancy can be attributed to alterations in dipolar interaction and shape anisotropy.

To delve deeper into the electrical characteristics of L10 FePd PMA-MTJ devices, we recorded resistance (R) as a function of current (I) under varying offset fields. As depicted by the red line in Fig. 3, pronounced bi-directional switching events emerge when μ0Hext is set to −26.5 mT. Remarkably, we observed a switching current of ∼100 μA, yielding an ultralow switching current density (JSW) of around 1.4 × 105 A/cm2. Such a small JSW is approximately one order of magnitude lower compared with that of typical STT MTJs35 and highly advantageous for the widespread application of spintronic devices. It is worth pointing out that the resistance-area (RA) product of our L10 FePd PMA-MTJ devices exceeds 600 Ω μm2. This is considerably larger than typical STT MTJs whose RA product is in the range of 10 Ω μm2 or so with the MgO barrier of about 1 nm or even thinner. Given this context, it seems improbable that the observed switching is driven purely by STT with such a small JSW.

FIG. 3.

RI loops of a L10 FePd PMA-MTJ device under different Hext. The AP-to-P JSW values are marked by arrows.

FIG. 3.

RI loops of a L10 FePd PMA-MTJ device under different Hext. The AP-to-P JSW values are marked by arrows.

Close modal

It is worth mentioning the resistive switching mechanism, for which the change in device resistance usually stem from the formation and rupture of conducting filaments,36,37 is also reported with MgO as the functional layer.38–40 To further investigate the switching mechanism of the L10 FePd PMA-MTJ devices, we made minor adjustments to the offset field and plotted the resultant resistance-current (RI) loops in Fig. 3 for comparative analysis. As denoted respectively by the green and orange lines, with a lower offset field of −25 and −26 mT, the antiparallel (AP)-to-parallel (P) JSW is significantly reduced. These JSW values are marked by arrows. In contrast, as showcased by the blue line, with a larger offset field of −27 mT, the AP-to-P switching becomes absent. The P-to-AP JSW, however, exhibits a lesser dependency on Hext. In addition, the full AP state is not realized with lower offset fields, which is probably caused by domain wall pinning. Clearly, the switching behavior is highly sensitive to Hext. As a result, the resistive switching mechanism can be ruled out. In other words, the detected R change is attributed to the TMR of the MTJ nanopillar.

Based on the stack structure, device performance, and switching characteristics, we believe the bi-directional ultralow-JSW switching can be explained by the driven of VCEC, as illustrated in Fig. 4(a). Specifically, the voltage applied instigates an electric field (E-field) capable of modulating the interlayer exchange coupling (IEC) between the L10 FePd layer and the bottom CoFeB layer. Unlike the traditional voltage-controlled magnetic anisotropy effect,41,42 which has been extensively investigated and where the E-field impacts only the local electronic structure within a few atomic layers,43 the VCEC effect manifests a longer range. This is because the Ruderman–Kittel–Kasuya–Yosida (RKKY) interaction is dependent on the electron wave functions in the ferromagnets. Though the E-field is confined to the bottom CoFeB–MgO interface due to Debye screening, it can still modulate the phases of up and down spins across the entire CoFeB layer. This modulation affects the RKKY IEC with the L10 FePd layer.27 With a sufficiently strong VCEC effect, the IEC could even be changed from ferromagnetic to anti-ferromagnetic or vice versa. When the reversed IEC achieves a certain magnitude, the magnetization of the bottom CoFeB can be switched. Previously, the VCEC-driven switching has been reported in the L10 FePd–CoFeB system, showcasing a similarly low JSW of ∼1.1 × 105 A/cm2.25 Considering that the ultrathin Ru layer may mediate a strong anti-ferromagnetic IEC and the insertion of an extra ultrathin spacer can substantially weaken the coupling strength,44,45 the Ru/Mo bilayer spacer in the present study should mediate a relatively weaker IEC, which can have its polarity altered through VCEC.

FIG. 4.

Two possible switching mechanisms of our L10 FePd PMA-MTJ devices. (a) VCEC-driven switching of the bottom CoFeB layer. (b) STT-driven switching of the L10 FePd–CoFeB composite.

FIG. 4.

Two possible switching mechanisms of our L10 FePd PMA-MTJ devices. (a) VCEC-driven switching of the bottom CoFeB layer. (b) STT-driven switching of the L10 FePd–CoFeB composite.

Close modal

Given the large RA product and ultralow JSW, we speculate that STT is insufficient to drive the switching process. In the VCEC switching discussed above, the free layer is the bottom CoFeB layer, and the L10 FePd layer plays a role primarily through its interaction with the CoFeB layer via the ultrathin Ru/Mo bilayer spacer. Nonetheless, quantitative evidence is still needed to support this assumption. By assuming STT as the dominating driven force, the switching process can be understood through the formation of a composite free layer, as shown in Fig. 4(b). When measuring RI loops, the electrons flow from the top contact to the bottom suggests AP-to-P switching, indicating that the free layer is beneath the MgO tunnel barrier. As the bottom CoFeB layer and the FePd layer switch synchronously in the M-Hext loop, it is reasonable to believe that the two layers are tightly coupled through the ultrathin Ru/Mo bilayer spacer and can effectively act as a composite free layer. Compared with the VCEC switching of the CoFeB, the STT switching of the L10 FePd–CoFeB composite requires not merely a large STT at ultralow JSW, but also a strong IEC within the free layer composite. In the future studies, comprehensive research will be essential to conclusively determine the switching mechanism in our L10 FePd PMA-MTJ devices.

It is worth mentioning that besides the two switching mechanisms discussed above, the STT switching of the bottom CoFeB layer may also be dominating. It is often reported that two layers coupled in the unpatterned sample could decouple from each other after being patterned into nanoscale devices.35 Thus, the bottom CoFeB layer may decouple from the L10 FePd layer after device patterning and undergo STT switching independently. Nonetheless, in this switching process, the magnetization of the FePd layer is expected to impact the switching characteristics, at least in the form of the stray field. As we did not observe any dependence of the offset field on the field-sweeping history, this trivial switching process is less possible.

In conclusion, we have successfully fabricated nano-sized L10 FePd PMA-MTJ devices, achieving an enhanced TMR of up to 65%—a considerable leap from the prior record of 25%. VSM results confirmed the PMA of the stack and showed two switching steps, which respectively correspond to the top CoFeB layer and the composite of the bottom CoFeB layer and the L10 FePd layer. Surprisingly, electrical-driven bi-directional switching was observed with an ultralow JSW of ∼1.4 × 105 A/cm2, which is about one order of magnitude lower than that of typical STT-MTJs. To ensure that the switching is associated with magnetization reversal, we verified the Hext-sensitivity of the switching characteristics and thus possibility of resistive switching should be excluded. Based on the sample stack structure and device performance, two possible switching mechanisms were proposed. One is the VCEC-driven switching of the bottom CoFeB layer, in which its IEC with the L10 FePd layer can be modified via voltage. The other is the STT-driven switching of the L10 FePd–CoFeB composite, in which the two layers switch together and serve as the free layer. These results shed light on the application of L10 FePd as an important material choice in PMA MTJs and offer promising directions for the development of spintronics-based memory and logic devices with ultralow energy consumption.

This work was supported in part by the Defense Advanced Research Projects Agency (DARPA) (Advanced MTJs for computation in and near random access memory) under Grant HR001117S0056-FP-042, in part by the National Institute of Standards and Technology (NIST), in part by the National Science Foundation (NSF) GOALI: Advancement of Heat-Assisted Magnetic Recording Enabled by Time-Resolved Magneto-Optical Kerr Effect Metrology under Grant 2226579, and in part by the Minnesota Nano Center by the NSF through the National Nanotechnology Coordinated Infrastructure (NNCI) under Award ECCS-202512.

The authors have no conflicts to disclose.

Deyuan Lyu: Conceptualization (lead); Investigation (equal); Methodology (equal); Project administration (lead); Writing – original draft (lead); Writing – review & editing (equal). Jenae E. Shoup: Investigation (equal); Methodology (equal). Ali T. Habiboglu: Investigation (equal); Methodology (equal). Qi Jia: Data curation (lead); Investigation (equal); Methodology (supporting). Pravin Khanal: Investigation (supporting); Methodology (supporting). Brandon R. Zink: Investigation (supporting); Methodology (supporting). Yang Lv: Investigation (supporting); Methodology (supporting). Bowei Zhou: Methodology (supporting). Daniel B. Gopman: Funding acquisition (supporting); Investigation (equal); Methodology (supporting); Supervision (equal); Writing – review & editing (supporting). Weigang Wang: Funding acquisition (supporting); Supervision (equal). Jian-Ping Wang: Funding acquisition (lead); Supervision (equal); Writing – review & editing (equal).

The data that support the findings of this study are available from the corresponding author upon reasonable request.

1.
N.
Nishimura
,
T.
Hirai
,
A.
Koganei
,
T.
Ikeda
,
K.
Okano
,
Y.
Sekiguchi
, and
Y.
Osada
,
J. Appl. Phys.
91
,
5246
(
2002
).
2.
A. D.
Kent
,
B.
Özyilmaz
, and
E.
del Barco
,
Appl. Phys. Lett.
84
,
3897
(
2004
).
3.
K. J.
Lee
,
O.
Redon
, and
B.
Dieny
,
Appl. Phys. Lett.
86
,
022505
(
2005
).
4.
H.
Meng
and
J.-P.
Wang
,
Appl. Phys. Lett.
88
,
172506
(
2006
).
5.
S.
Mangin
,
D.
Ravelosona
,
J. A.
Katine
,
M. J.
Carey
,
B. D.
Terris
, and
E. E.
Fullerton
,
Nat. Mater.
5
,
210
(
2006
).
6.
J.-P.
Wang
,
S. S.
Sapatnekar
,
C. H.
Kim
,
P.
Crowell
,
S.
Koester
,
S.
Datta
,
K.
Roy
,
A.
Raghunathan
,
X. S.
Hu
,
M.
Niemier
,
A.
Naeemi
,
C.-L.
Chien
,
C.
Ross
, and
R.
Kawakami
, in
Proceedings of the 54th Annual Design Automation Conf. (DAC '17)
(
Austin, TX, USA
,
2017
), p.
16
.
7.
K. C.
Chun
,
H.
Zhao
,
J. D.
Harms
,
T.-H.
Kim
,
J.-P.
Wang
, and
C. H.
Kim
,
IEEE J. Solid-State Circuits
48
,
598
(
2013
).
8.
K.
Watanabe
,
B.
Jinnai
,
S.
Fukami
,
H.
Sato
, and
H.
Ohno
,
Nat. Commun.
9
,
663
(
2018
).
9.
B.
Jinnai
,
J.
Igarashi
,
K.
Watanabe
,
T.
Funatsu
,
H.
Sato
,
S.
Fukami
, and
H.
Ohno
, in
Proceedings of the IEEE Int. Electron. Devices Meeting (IEDM)
(
IEEE
,
San Francisco, CA, USA
,
2020
).
10.
S.
Ikeda
,
K.
Miura
,
H.
Yamamoto
,
K.
Mizunuma
,
H. D.
Gan
,
M.
Endo
,
S.
Kanai
,
J.
Hayakawa
,
F.
Matsukura
, and
H.
Ohno
,
Nat. Mater.
9
,
721
(
2010
).
11.
S.
Peng
,
W.
Kang
,
M.
Wang
,
K.
Cao
,
X.
Zhao
,
L.
Wang
,
Y.
Zhang
,
Y.
Zhang
,
Y.
Zhou
,
K. L.
Wang
, and
W.
Zhao
,
IEEE Magn. Lett.
8
,
3105805
(
2017
).
12.
L. E.
Nistor
,
B.
Rodmacq
,
S.
Auffret
, and
B.
Dieny
,
Appl. Phys. Lett.
94
,
012512
(
2009
).
13.
K.
Yakushiji
,
T.
Saruya
,
H.
Kubota
,
A.
Fukushima
,
T.
Nagahama
,
S.
Yuasa
, and
K.
Ando
,
Appl. Phys. Lett.
97
,
232508
(
2010
).
14.
B.
Carvello
,
C.
Ducruet
,
B.
Rodmacq
,
S.
Auffret
,
E.
Gautier
,
G.
Gaudin
, and
B.
Dieny
,
Appl. Phys. Lett.
92
,
102508
(
2008
).
15.
Z. R.
Tadisina
,
A.
Natarajarathinam
, and
S.
Gupta
,
Journal of Vacuum Science & Technology A: Vacuum, Surfaces, and Films
28
,
973
(
2010
).
16.
M.
Nakayama
,
T.
Kai
,
N.
Shimomura
,
M.
Amano
,
E.
Kitagawa
,
T.
Nagase
,
M.
Yoshikawa
,
T.
Kishi
,
S.
Ikegawa
, and
H.
Yoda
,
J. Appl. Phys.
103
,
07A710
(
2008
).
17.
L.-X.
Ye
,
C.-M.
Lee
,
J.-H.
Lai
,
A.
Canizo-Cabrera
,
W.-J.
Chen
, and
T.-h.
Wu
,
J. Magn. Magn. Mater.
322
,
L9
(
2010
).
18.
P.
de Person
,
P.
Warin
,
M.
Jamet
,
C.
Beigne
, and
Y.
Samson
,
Phys. Rev. B
76
,
184402
(
2007
).
19.
G.
Kim
,
Y.
Sakuraba
,
M.
Oogane
,
Y.
Ando
, and
T.
Miyazaki
,
Appl. Phys. Lett.
92
,
172502
(
2008
).
20.
T.
Kishi
,
H.
Yoda
,
T.
Kai
,
T.
Nagase
,
E.
Kitagawa
,
M.
Yoshikawa
,
K.
Nishiyama
,
T.
Daibou
,
M.
Nagamine
,
M.
Amano
,
S.
Takahashi
,
M.
Nakayama
,
N.
Shimomura
,
H.
Aikawa
,
S.
Ikegawa
,
S.
Yuasa
,
K.
Yakushiji
,
H.
Kubota
,
A.
Fukushima
,
M.
Oogane
,
T.
Miyazaki
, and
K.
Ando
, in
Proceedings of the IEEE Int. Electron. Devices Meeting (IEDM)
(
IEEE
,
San Francisco, CA, USA
,
2008
).
21.
T.
Klemmer
,
D.
Hoydick
,
H.
Okumura
,
B.
Zhang
, and
W. A.
Soffa
,
Scr. Metall. Mater.
33
,
1793
(
1995
).
22.
H.
Zhao
,
Spin transfer torque induced switching in magnetic tunnel junction for STT-RAM application
,
University of Minnesota Digital Conservancy
,
2013
, https://hdl.handle.net/11299/177055.
23.
S.
Iihama
,
A.
Sakuma
,
H.
Naganuma
,
M.
Oogane
,
S.
Mizukami
, and
Y.
Ando
,
Phys. Rev. B
94
,
174425
(
2016
).
24.
X.
Wang
,
S.
Krylyuk
,
D.
Josell
,
D.
Zhang
,
D.
Lyu
,
J.-P.
Wang
, and
D. B.
Gopman
,
AIP Adv.
11
,
025106
(
2021
).
25.
P.
He
,
X.
Ma
,
J. W.
Zhang
,
H. B.
Zhao
,
G.
Lüpke
,
Z.
Shi
, and
S. M.
Zhou
,
Phys. Rev. Lett.
110
,
077203
(
2013
).
26.
D.
Huang
,
D.
Lyu
,
X.
Wang
,
M. B.
Katz
,
D.
Zhang
,
J.-P.
Wang
,
D. B.
Gopman
, and
X.
Wang
,
Phys. Rev. Mater.
6
,
113402
(
2022
).
27.
D.
Zhang
,
M.
Bapna
,
W.
Jiang
,
D.
Sousa
,
Y.-C.
Liao
,
Z.
Zhao
,
Y.
Lv
,
P.
Sahu
,
D.
Lyu
,
A.
Naeemi
,
T.
Low
,
S. A.
Majetich
, and
J.-P.
Wang
,
Nano Lett.
22
,
622
(
2022
).
28.
B. R.
Zink
,
D.
Zhang
,
H.
Li
,
O. J.
Benally
,
Y.
Lv
,
D.
Lyu
, and
J.-P.
Wang
,
Advanced Electronic Materials
8
,
2200382
(
2022
).
29.
H.
Naganuma
,
M.
Nishijima
,
H.
Adachi
,
M.
Uemoto
,
H.
Shinya
,
S.
Yasui
,
H.
Morioka
,
A.
Hirata
,
F.
Godel
,
M.-B.
Martin
,
B.
Dlubak
,
P.
Seneor
, and
K.
Amemiya
,
ACS Nano
16
,
4139
(
2022
).
30.
D.
Lyu
,
J. E.
Shoup
,
D.
Huang
,
J.
García-Barriocanal
,
Q.
Jia
,
W.
Echtenkamp
,
G. A.
Rojas
,
G.
Yu
,
B. R.
Zink
,
X.
Wang
,
D. B.
Gopman
, and
J.-P.
Wang
,
Adv. Funct. Mater.
33
,
2214201
(
2023
).
31.
D.-L.
Zhang
,
K. B.
Schliep
,
R. J.
Wu
,
P.
Quarterman
,
D.
Reifsnyder Hickey
,
Y.
Lv
,
X.
Chao
,
H.
Li
,
J.-Y.
Chen
,
Z.
Zhao
,
M.
Jamali
,
K. A.
Mkhoyan
, and
J.-P.
Wang
,
Appl. Phys. Lett.
112
,
152401
(
2018
).
32.
D.-L.
Zhang
,
C.
Sun
,
Y.
Lv
,
K. B.
Schliep
,
Z.
Zhao
,
J.-Y.
Chen
,
P. M.
Voyles
, and
J.-P.
Wang
,
Phys. Rev. Appl.
9
,
044028
(
2018
).
33.
I.
Suzuki
,
S.
Kubo
,
H.
Sepehri-Amin
, and
Y. K.
Takahashi
,
ACS Appl. Mater. Interfaces
13
,
16620
(
2021
).
34.

Certain commercial equipment, instruments, or materials are identified in this paper in order to specify the experimental procedure adequately. Such identification is not intended to imply recommendation or endorsement by NIST, nor is it intended to imply that the materials or equipment identified are necessarily the best available for the purpose.

35.
D.
Lyu
,
P.
Khanal
,
Y.
Lv
,
B.
Zhou
,
H.
Yun
,
Q.
Jia
,
B. R.
Zink
,
Y.
Fan
,
K. A.
Mkhoyan
,
W.
Wang
, and
J.-P.
Wang
,
IEEE Electron Device Lett.
43
,
1215
(
2022
).
36.
C.
Hu
,
Q.
Wang
,
S.
Bai
,
M.
Xu
,
D.
He
,
D.
Lyu
, and
J.
Qi
,
Appl. Phys. Lett.
110
,
073501
(
2017
).
37.
Y.
Yang
,
P.
Gao
,
S.
Gaba
,
T.
Chang
,
X.
Pan
, and
W.
Lu
,
Nat. Commun.
3
,
732
(
2012
).
38.
F.-C.
Chiu
,
W.-C.
Shih
, and
J.-J.
Feng
,
J. Appl. Phys.
111
,
094104
(
2012
).
39.
J. M.
Teixeira
,
J.
Ventura
,
R.
Fermento
,
J. P.
Araujo
,
J. B.
Sousa
,
P.
Wisniowski
, and
P. P.
Freitas
,
J. Phys. D: Appl. Phys.
42
,
105407
(
2009
).
40.
A.
Schulman
,
E.
Paz
,
T.
Böhnert
,
A. S.
Jenkins
, and
R.
Ferreira
,
Adv. Funct. Mater.
33
,
2305238
(
2023
).
41.
W.-G.
Wang
,
M.
Li
,
S.
Hageman
, and
C. L.
Chien
,
Nat. Mater.
11
,
64
(
2011
).
42.
Y.
Shiota
,
T.
Nozaki
,
F.
Bonell
,
S.
Murakami
,
T.
Shinjo
, and
Y.
Suzuki
,
Nat. Mater.
11
,
39
(
2012
).
43.
J.
Zhang
,
P. V.
Lukashev
,
S. S.
Jaswal
, and
E. Y.
Tsymbal
,
Phys. Rev. B
96
,
014435
(
2017
).
44.
C.
Yoshida
,
T.
Takenaga
,
Y.
Iba
,
Y.
Yamazaki
,
H.
Noshiro
,
K.
Tsunoda
,
A.
Hatada
,
M.
Nakabayashi
,
A.
Takahashi
,
M.
Aoki
, and
T.
Sugii
,
IEEE Trans. Magn.
49
,
4363
(
2013
).
45.
D.
Lyu
,
D.
Zhang
,
D. B.
Gopman
,
Y.
Lv
,
O. J.
Benally
, and
J.-P.
Wang
,
Appl. Phys. Lett.
120
,
012404
(
2022
).